arXiv Daily Digest - 2026-03-10

In asynchronous federated learning (FL), client devices send updates to a central server at varying times based on their computational speed, often using stale versions of the global model. This staleness can degrade the convergence and accuracy of the global model. Previous work, such as AsyncFedED, proposed an adaptive aggregation method using Euclidean distance to measure staleness. In this paper, we extend this approach by exploring alternative distance metrics to more accurately capture the effect of gradient staleness. We integrate these metrics into the aggregation process and evaluate their impact on convergence speed, model performance, and training stability under heterogeneous clients and non-IID data settings. Our results demonstrate that certain metrics lead to more robust and efficient asynchronous FL training, offering a stronger foundation for practical deployment.

The multiple-choice knapsack problem (MCKP) is a classic combinatorial optimization with wide practical applications. This paper investigates a significant yet underexplored extension of MCKP: the multi-objective chance-constrained MCKP (MO-CCMCKP) under implicit probability distributions. The goal of the problem is to simultaneously minimize the total cost and maximize the confidence level of satisfying the capacity constraint, capturing essential trade-offs in domains like 5G network configuration. To address the computational challenge of evaluating chance constraints under implicit distributions, we first propose an order-preserving efficient resource allocation Monte Carlo (OPERA-MC) method. This approach adaptively allocates sampling resources to preserve dominance relationships while reducing evaluation time significantly. Further, we develop NHILS, a hybrid evolutionary algorithm that integrates specialized initialization and local search into NSGA-II to navigate sparse feasible regions. Experiments on synthetic benchmarks and real-world 5G network configuration benchmarks demonstrate that NHILS consistently outperforms several state-of-the-art multi-objective optimizers in convergence, diversity, and feasibility. The benchmark instances and source code will be made publicly available to facilitate research in this area.

Reliable unmanned aerial vehicle (UAV) detection is critical for autonomous airspace monitoring but remains challenging when integrating sensor streams that differ substantially in resolution, perspective, and field of view. Conventional fusion methods-such as wavelet-, Laplacian-, and decision-level approaches-often fail to preserve spatial correspondence across modalities and suffer from annotation of inconsistencies, limiting their robustness in real-world settings. This study introduces two fusion strategies, Registration-aware Guided Image Fusion (RGIF) and Reliability-Gated Modality-Attention Fusion (RGMAF), designed to overcome these limitations. RGIF employs Enhanced Correlation Coefficient (ECC)-based affine registration combined with guided filtering to maintain thermal saliency while enhancing structural detail. RGMAF integrates affine and optical-flow registration with a reliability-weighted attention mechanism that adaptively balances thermal contrast and visual sharpness. Experiments were conducted on the Multi-Sensor and Multi-View Fixed-Wing (MMFW)-UAV dataset comprising 147,417 annotated air-to-air frames collected from infrared, wide-angle, and zoom sensors. Among single-modality detectors, YOLOv10x demonstrated the most stable cross-domain performance and was selected as the detection backbone for evaluating fused imagery. RGIF improved the visual baseline by 2.13% mAP@50 (achieving 97.65%), while RGMAF attained the highest recall of 98.64%. These findings show that registration-aware and reliability-adaptive fusion provides a robust framework for integrating heterogeneous modalities, substantially enhancing UAV detection performance in multimodal environments.

Removing personally identifiable information (PII) from texts is necessary to comply with various data protection regulations and to enable data sharing without compromising privacy. However, recent works show that documents sanitized by PII removal techniques are vulnerable to reconstruction attacks. Yet, we suspect that the reported success of these attacks is largely overestimated. We critically analyze the evaluation of existing attacks and find that data leakage and data contamination are not properly mitigated, leaving the question whether or not PII removal techniques truly protect privacy in real-world scenarios unaddressed. We investigate possible data sources and attack setups that avoid data leakage and conclude that only truly private data can allow us to objectively evaluate vulnerabilities in PII removal techniques. However, access to private data is heavily restricted - and for good reasons - which also means that the public research community cannot address this problem in a transparent, reproducible, and trustworthy manner.

Prior-Data Fitted Networks (PFNs), such as TabPFN and TabICL, have revolutionized tabular deep learning by leveraging in-context learning for tabular data. These models are meant as foundation models for classification and regression settings and promise to greatly simplify deployment in practical settings because their performance is unprecedented (in terms of mean squared error or $R^2$, when measured on common benchmarks like TabArena or TALENT). However, we see an important weakness of current benchmarks for the regression setting: the current benchmarks focus on evaluating win rates and performance using metrics like (root) mean squared error or $R^2$. Therefore, these leaderboards (implicitly and explicitly) push researchers to optimize for machine learning pipelines which elicit a good mean value estimate. The main problem is that this approach only evaluates a point estimate (namely the mean estimator which is the Bayes estimator associated with the mean squared error loss). In this article we discuss the application of proper scoring rules for evaluating the goodness of probabilistic forecasts in distributional regression. We also propose to enhance common machine learning benchmarks with metrics for probabilistic regression. To improve the status quo and make the machine learning community aware of scoring rules for probabilistic regression, we advocate to use the continuous ranked probability score (CRPS) in benchmarks for probabilistic regression. However, we also illustrate that the choice of the scoring rule changes the inductive bias of the trained model. We, therefore, advocate for finetuning or promptable tabular foundation models.

Contrastive learning has become a fundamental approach in both uni-modal and multi-modal frameworks. This learning paradigm pulls positive pairs of samples closer while pushing negatives apart. In the uni-modal setting (e.g., image-based learning), previous research has shown that the strength of these forces can be controlled through the temperature parameter. In this work, we propose Multi-Modal Temperature and Margin Schedules (MM-TS), extending the concept of uni-modal temperature scheduling to multi-modal contrastive learning. Our method dynamically adjusts the temperature in the contrastive loss during training, modulating the attraction and repulsion forces in the multi-modal setting. Additionally, recognizing that standard multi-modal datasets often follow imbalanced, long-tail distributions, we adapt the temperature based on the local distribution of each training sample. Specifically, samples from dense clusters are assigned a higher temperature to better preserve their semantic structure. Furthermore, we demonstrate that temperature scheduling can be effectively integrated within a max-margin framework, thereby unifying the two predominant approaches in multi-modal contrastive learning: InfoNCE loss and max-margin objective. We evaluate our approach on four widely used image- and video-language datasets, Flickr30K, MSCOCO, EPIC-KITCHENS-100, and YouCook2, and show that our dynamic temperature and margin schedules improve performance and lead to new state-of-the-art results in the field.

Motivation: The rapid growth of biological data has intensified the need for transparent, reproducible, and well-documented computational workflows. The ability to clearly connect the steps of a workflow in the code with their description in a paper would improve workflow understanding, support reproducibility, and facilitate reuse. This task requires the linking of Bioinformatics tools in workflow code with their mentions in a published workflow description. Results: We present CoPaLink, an automated approach that integrates three components: Named Entity Recognition (NER) for identifying tool mentions in scientific text, NER for tool mentions in workflow code, and entity linking grounded on Bioinformatics knowledge bases. We propose approaches for all three steps achieving a high individual F1-measure (84 - 89) and a joint accuracy of 66 when evaluated on Nextflow workflows using Bioconda and Bioweb Knowledge bases. CoPaLink leverages corpora of scientific articles and workflow executable code with curated tool annotations to bridge the gap between narrative descriptions and workflow implementations. Availability: The code is available at https://gitlab.liris.cnrs.fr/sharefair/copalink-experiments and https://gitlab.liris.cnrs.fr/sharefair/copalink. The corpora are also available at https://doi.org/10.5281/zenodo.18526700, https://doi.org/10.5281/zenodo.18526760 and https://doi.org/10.5281/zenodo.18543814.

In this paper we propose a method for analyzing services deployed in serverless platforms. These services typically consists of orchestrated functions that can exhibit complex and non-conservative information flows due to the interaction of independently deployed functions under coarse-grained control mechanisms. We introduce a topological model of serverless services and make use of the Hodge decomposition to partition observed operational flows into locally correctable components and globally persistent harmonic modes. Our analysis shows that harmonic flows naturally arise from different kind of interactions among functions and should be interpreted as structural properties of serverless systems rather than configuration errors. We present a systematic methodology for analyzing inter-function flows and deriving actionable remediation strategies, including dumping effects to contain the effects of harmonic inefficiencies as an alternative to completely restructure the topological model of the service. Experimental results confirm that the proposed approach can uncover latent architectural structures leading to inefficiencies.

Agile organizations increasingly rely on automated regression testing to sustain rapid, high-quality software delivery. However, as systems grow and requirements evolve, a persistent bottleneck arises: test specifications are produced faster than they can be transformed into executable scripts, leading to mounting manual effort and delayed releases. In partnership with Hacon (a Siemens company), we present an agentic AI approach that generates system-level test scripts directly from validated specifications, aiming to accelerate automation without sacrificing human oversight. Our solution features a retrieval-augmented, multi-agent architecture integrated into Hacon's agile workflows. We evaluate this system through a mixed-method analysis of industrial artifacts and practitioner feedback. Results show that the AI teammate significantly increases test script throughput and reduces manual authoring effort, while underscoring the ongoing need for clear specifications and human review to ensure quality and maintainability. We conclude with practical lessons for scaling regression automation and fostering effective Human-AI collaboration in agile environments.

The North-West African coast is enriched by the Canary current, which sustain a very produc- tive marine ecosystem. The Senegalese artisanal fishing fleet, the largest in West Africa, ben- efit from this particularly productive ecosystem. It has survived the ages with remarkable adaptability, and has great flexibility allowing it to react quickly to changes, in particular by changing fishing gear and performing migrations. However, since the 1980s, the increasing fishing effort led to a progressive fish depletion, increasing fisher's migration distances to access new fishing grounds. Since 2007 many fishers even started to navigate to Canary archi- pelago in order to find a more lucrative job in Europe, carrying candidate to emigration in their canoes. This phenomenon further increased since 2022 due to a new drop in fishery yields, consecutive to the development of fishmeal factories along the coast that amplified overfishing. Climate change may also impact fish habitat, and by consequence the distribution of fishing grounds. The question addressed in this research was how climate change, fishing effort and socio-economic parameters interact and determine the artisanal fishery dynamics. An interdisciplinary approach allowed us to collect data and qualitative information on cli- mate, fishing effort and socio-economic parameters. This served as a basis to build a multi- agent model of the mobility of Senegalese artisanal fishing. We implemented a first version of the model and presented some preliminary simulations with contrasted fishing effort and climate scenario. The results suggested that first, climate change should have only a slight impact on artisanal fishing, even in the most extreme climate scenario considered. Second, if fishing effort was maintained at current levels, we found a collapse of the fishery with massive fishers migrations whatever the climate scenario. Third, with reduced fishing effort, a sustain- able fishery equilibrium emerges in which Senegal's artisanal fishery catches ~250,000 tons of fish a year mostly in Senegal, approaching the 2000s catches records. This sustainable equi- librium maintained with the two-climate change scenario tested. Fishers migrations provide clues of the fish populations state and have implications for the sustainable exploitation of fishing resources. Senegalese artisanal fishers' migrations impact the regional distribution of the fishing effort, therefore must be taken into account in regional development and planning policies for this sector, particularly in a context of increasing infrastructure and spatial man- agement measures (e.g. marine protected areas). This work lays the foundations of a computer simulation tool for decision support.

Service region design determines the geographic coverage of service networks, shaping long-term operational performance. Capital and operational constraints preclude simultaneous large-scale deployment, requiring expansion to proceed sequentially. The resulting challenge is to determine when and where to invest under demand uncertainty, balancing intertemporal trade-offs between early and delayed investment and accounting for network effects whereby each deployment reshapes future demand through inter-regional connectivity. This study addresses a sequential service region design (SSRD) problem incorporating two practical yet underexplored factors: a $k$-region constraint that limits the number of regions investable per period and a stochastic spillover effect linking investment decisions to demand evolution. The resulting problem requires sequencing regional portfolios under uncertainty, leading to a combinatorial explosion in feasible investment sequences. To address this challenge, we propose a solution framework that integrates real options analysis (ROA) with a Transformer-based Proximal Policy Optimization (TPPO) algorithm. ROA evaluates the intertemporal option value of investment sequences, while TPPO learns sequential policies that directly generate high option-value sequences without exhaustive enumeration. Numerical experiments on realistic multi-region settings demonstrate that TPPO converges faster than benchmark DRL methods and consistently identifies sequences with superior option value. Case studies and sensitivity analyses further confirm robustness and provide insights on investment concurrency, regional prioritization, and the increasing benefits of adaptive expansion via our approach under stronger spillovers and dynamic market conditions.

Post-training quantization (PTQ) has emerged as a prevailing technique for deploying large language models (LLMs) efficiently in terms of both memory and computation, across edge devices and server platforms. Existing PTQ methods primarily aim to reduce precision in weights and activations by mitigating quantization errors caused by channel-wise outlier activations (e.g., pre-quantization scaling, online transformations, or low-rank error reconstruction). Among these approaches, error reconstruction with low-rank adaptation (LoRA) has proven particularly effective, as it introduces a lightweight auxiliary computation path without requiring heavy optimization or additional online layers. However, prior studies reveal severe accuracy degradation under W4A4 settings, and conventional low-rank adaptations rely on two sequential factors, necessitating intermediate quantization during inference and thereby limiting low-precision efficiency. In this work, we propose SERQ, a saliency-aware error reconstruction method for low-bit LLM inference that employs a single low-rank compensation matrix. SERQ preserves efficient 4-bit matrix multiplication in linear layers by jointly mitigating quantization errors arising from both activation and weight saliency through three stages: (1) static activation flattening, (2) saliency-aware error reconstruction, and (3) offline weight permutation. The method incurs additional computation only for low-rank error reconstruction via a single decomposition, while all other operations are performed offline, thereby keeping latency overhead minimal. Empirically, SERQ outperforms prior error reconstruction methods under both W4A8 and W4A4 settings, and achieves higher accuracy than state-of-the-art rotation-based W4A4 approaches, while substantially reducing calibration complexity.

Large language models often underperform in many European languages due to the dominance of English and a few high-resource languages in training data. This paper presents TildeOpen LLM, a 30-billion-parameter open-weight foundational model trained for 34 European languages to promote linguistic equity and improve performance for low-resource languages. To address the data imbalance, we combine dataset upsampling with a curriculum-based training schedule that alternates between uniform and natural language distributions. The resulting model performs favorably compared to other multilingual LLMs despite being trained with significantly fewer computing resources. Evaluation across multiple multilingual benchmarks shows that TildeOpen surpasses existing open-weight models in text generation and comprehension, particularly for Baltic, Finno-Ugric, and Slavic languages. Human evaluations confirm an up to tenfold reduction in linguistic errors relative to leading baselines. The model and associated resources are fully open-weight and publicly available at huggingface.co/TildeAI/TildeOpen-30b. These outcomes demonstrate that careful data curation and balanced training strategies can substantially enhance multilingual model quality without increasing model size or training volume.

Large language models (LLMs) excel in open domains but struggle in specialized settings with limited data and evolving knowledge. Existing domain adaptation practices rely heavily on manual trial-and-error processes, incur significant hyperparameter complexity, and are highly sensitive to data and user preferences, all under the high cost of LLM training. Moreover, the interactions and transferability of hyperparameter choices across models/domains remain poorly understood, making adaptation gains uncertain even with substantial effort. To solve these challenges, we present AutoAdapt, a novel end-to-end automated framework for efficient and reliable LLM domain adaptation. AutoAdapt leverages curated knowledge bases from literature and open-source resources to reduce expert intervention. To narrow the search space, we design a novel multi-agent debating system in which proposal and critic agents iteratively interact to align user intent and incorporate data signals and best practices into the planning process. To optimize hyperparameters under tight budgets, we propose AutoRefine, a novel LLM-based surrogate that replaces costly black-box search. Across 10 tasks, AutoAdapt achieves a 25% average relative accuracy improvement over state-of-the-art Automated Machine Learning baselines with minimal overhead.

LiDAR-based 3D object detection plays a critical role for reliable and safe autonomous driving systems. However, existing detectors often produce overly confident predictions for objects not belonging to known categories, posing significant safety risks. This is caused by so-called out-of-distribution (OOD) objects, which were not part of the training data, resulting in incorrect predictions. To address this challenge, we propose ALOOD (Aligned LiDAR representations for Out-Of-Distribution Detection), a novel approach that incorporates language representations from a vision-language model (VLM). By aligning the object features from the object detector to the feature space of the VLM, we can treat the detection of OOD objects as a zero-shot classification task. We demonstrate competitive performance on the nuScenes OOD benchmark, establishing a novel approach to OOD object detection in LiDAR using language representations. The source code is available at https://github.com/uulm-mrm/mmood3d.

End-to-end full-duplex speech models feed user audio through an always-on LLM backbone, yet the speaker privacy implications of their hidden representations remain unexamined. Following the VoicePrivacy 2024 protocol with a lazy-informed attacker, we show that the hidden states of SALM-Duplex and Moshi leak substantial speaker identity across all transformer layers. Layer-wise and turn-wise analyses reveal that leakage persists across all layers, with SALM-Duplex showing stronger leakage in early layers while Moshi leaks uniformly, and that Linkability rises sharply within the first few turns. We propose two streaming anonymization setups using Stream-Voice-Anon: a waveform-level front-end (Anon-W2W) and a feature-domain replacement (Anon-W2F). Anon-W2F raises EER by over 3.5x relative to the discrete encoder baseline (11.2% to 41.0%), approaching the 50% random-chance ceiling, while Anon-W2W retains 78-93% of baseline sBERT across setups with sub-second response latency (FRL under 0.8 s).

We investigate whether performing reasoning in a continuous latent space leads to more robust multilingual capabilities. We compare Continuous Chain-of-Thought (using the CODI framework) against standard supervised fine-tuning across five typologically diverse languages: English, Chinese, German, French, and Urdu. Our experiments on GSM8k and CommonsenseQA demonstrate that continuous reasoning significantly outperforms explicit reasoning on low-resource languages, particularly in zero-shot settings where the target language was not seen during training. Additionally, this approach achieves extreme efficiency, compressing reasoning traces by approximately $29\times$ to $50\times$. These findings indicate that continuous latent representations naturally exhibit greater language invariance, offering a scalable solution for cross-lingual reasoning.

Quantization has become essential for the efficient deployment of speech processing systems. Although widely studied, most existing quantization methods were developed for vision and NLP architectures, while the specific challenges of audio signals remain largely overlooked. In particular, we show that audio activations can exhibit large calibration ranges, leading to significant information loss when standard calibration techniques are applied. To address this, we propose ESC, an Evolution Strategy-based Calibration method that formulates activation scaling as an optimization problem and solves it using a two-step local-global scheme driven by an evolution strategy. ESC enables unaltered performance under full INT8 quantization and is the first calibration method to achieve near-lossless performance for full INT4 quantization across multiple speech tasks. Integrating ESC with PTQ methods further reduces performance loss, achieving a 1% relative accuracy degradation on the AST model.

Industrial maintenance platforms contain rich but fragmented evidence, including free-text work orders, heterogeneous operational sensors or indicators, and structured failure knowledge. These sources are often analyzed in isolation, producing alerts or forecasts that do not support conditional decision-making: given this asset history and behavior, what is happening and what action is warranted? We present Condition Insight Agent, a deployed decision-support framework that integrates maintenance language, behavioral abstractions of operational data, and engineering failure semantics to produce evidence-grounded explanations and advisory actions. The system constrains reasoning through deterministic evidence construction and structured failure knowledge, and applies a rule-based verification loop to suppress unsupported conclusions. Case studies from production CMMS deployments show that this verification-first design operates reliably under heterogeneous and incomplete data while preserving human oversight. Our results demonstrate how constrained LLM-based reasoning can function as a governed decision-support layer for industrial maintenance.

Automated Drug Combination Extraction (DCE) from large-scale biomedical literature is crucial for advancing precision medicine and pharmacological research. However, existing relation extraction methods primarily focus on binary interactions and struggle to model variable-length n-ary drug combinations, where complex compatibility logic and distributed evidence need to be considered. To address these limitations, we propose RexDrug, an end-to-end reasoning-enhanced relation extraction framework for n-ary drug combination extraction based on large language models. RexDrug adopts a two-stage training strategy. First, a multi-agent collaborative mechanism is utilized to automatically generate high-quality expert-like reasoning traces for supervised fine-tuning. Second, reinforcement learning with a multi-dimensional reward function specifically tailored for DCE is applied to further refine reasoning quality and extraction accuracy. Extensive experiments on the DrugComb dataset show that RexDrug consistently outperforms state-of-the-art baselines for n-ary extraction. Additional evaluation on the DDI13 corpus confirms its generalizability to binary drugdrug interaction tasks. Human expert assessment and automatic reasoning metrics further indicates that RexDrug produces coherent medical reasoning while accurately identifying complex therapeutic regimens. These results establish RexDrug as a scalable and reliable solution for complex biomedical relation extraction from unstructured text. The source code and data are available at https://github.com/DUTIR-BioNLP/RexDrug

Advancements in data-driven machine learning have emerged as a pivotal element in supporting automotive software systems (ASSs) engineering across various levels of the V-development process. Duringsystemverificationandvalidation,theintegrationofanintelligent fault detection anddiagnosis (FDD) model with test recordings analysis process serves as a powerful tool for efficiency ensuring functional safety. However, the lack of interpretability of the black-box FDD models developed not only hinders understanding of the cause underlying the prediction, but also prevents the model from being adapted based on the prediction result. This, in turn, increases the computational cost required for developingacomplexFDDmodelandlimitsconfidenceinreal-timesafety-criticalapplications.To address this challenge, a novel explainable method for fault detection, identification, and localization is proposed in this article with the aim of providing a clear understanding of the logic behind the prediction outcome. To this end, a hybrid 1dCNN-GRU-based intelligent model was developed to analyze the recordings from the real-time validation process of ASSs. The employment of explainable AI techniques, i.e., IGs, DeepLIFT, Gradient SHAP, and DeepLIFT SHAP, was instrumental in enabling model adaptation and facilitating the root cause analysis (RCA). The proposed approach is applied to the real time dataset collected during a virtual test drive performed by the user on hardware in the loop system.

Recently, there has been increased interest in globally distributed training, which has the promise to both reduce training costs and democratize participation in building large-scale foundation models. However, existing models trained in a globally distributed manner are relatively small in scale and have only been trained with whitelisted participants. Therefore, they do not yet realize the full promise of democratized participation. In this report, we describe Covenant-72B, an LLM produced by the largest collaborative globally distributed pre-training run (in terms of both compute and model scale), which simultaneously allowed open, permissionless participation supported by a live blockchain protocol. We utilized a state-of-the-art communication-efficient optimizer, SparseLoCo, supporting dynamic participation with peers joining and leaving freely. Our model, pre-trained on approximately 1.1T tokens, performs competitively with fully centralized models pre-trained on similar or higher compute budgets, demonstrating that fully democratized, non-whitelisted participation is not only feasible, but can be achieved at unprecedented scale for a globally distributed pre-training run.

Hierarchical knowledge structures are ubiquitous across real-world domains and play a vital role in organizing information from coarse to fine semantic levels. While such structures have been widely used in taxonomy systems, biomedical ontologies, and retrieval-augmented generation, their potential remains underexplored in the context of Text-Rich Networks (TRNs), where each node contains rich textual content and edges encode semantic relationships. Existing methods for learning on TRNs often focus on flat semantic modeling, overlooking the inherent hierarchical semantics embedded in textual documents. To this end, we propose TIER (Hierarchical \textbf{T}axonomy-\textbf{I}nformed R\textbf{E}presentation Learning on Text-\textbf{R}ich Networks), which first constructs an implicit hierarchical taxonomy and then integrates it into the learned node representations. Specifically, TIER employs similarity-guided contrastive learning to build a clustering-friendly embedding space, upon which it performs hierarchical K-Means followed by LLM-powered clustering refinement to enable semantically coherent taxonomy construction. Leveraging the resulting taxonomy, TIER introduces a cophenetic correlation coefficient-based regularization loss to align the learned embeddings with the hierarchical structure. By learning representations that respect both fine-grained and coarse-grained semantics, TIER enables more interpretable and structured modeling of real-world TRNs. We demonstrate that our approach significantly outperforms existing methods on multiple datasets across diverse domains, highlighting the importance of hierarchical knowledge learning for TRNs.

The autocovariance least squares (ALS) method is a computationally efficient approach for estimating noise covariances in Kalman filters without requiring specific noise models. However, conventional ALS and its variants rely on the classic least mean squares (LMS) criterion, making them highly sensitive to measurement outliers and prone to severe performance degradation. To overcome this limitation, this paper proposes a novel outlier-robust ALS algorithm, termed ALS-IRLS, based on the iteratively reweighted least squares (IRLS) framework. Specifically, the proposed approach introduces a two-tier robustification strategy. First, an innovation-level adaptive thresholding mechanism is employed to filter out heavily contaminated data. Second, the outlier-contaminated autocovariance is formulated using an $ε$-contamination model, where the standard LMS criterion is replaced by the Huber cost function. The IRLS method is then utilized to iteratively adjust data weights based on estimation deviations, effectively mitigating the influence of residual outliers. Comparative simulations demonstrate that ALS-IRLS reduces the root-mean-square error (RMSE) of noise covariance estimates by over two orders of magnitude compared to standard ALS. Furthermore, it significantly enhances downstream state estimation accuracy, outperforming existing outlier-robust Kalman filters and achieving performance nearly equivalent to the ideal Oracle lower bound in the presence of noisy and anomalous data.

Long-term time series forecasting (LTSF) is widely recognized as a central challenge in data mining and machine learning. LTSF has increasingly evolved into a benchmark-driven ''GAME,'' where models are ranked, compared, and declared state-of-the-art based primarily on marginal reductions in aggregated pointwise error metrics such as MSE and MAE. Across a small set of canonical datasets and fixed forecasting horizons, progress is communicated through leaderboard-style tables in which lower numerical scores define success. In this GAME, what is measured becomes what is optimized, and incremental error reduction becomes the dominant currency of advancement. We argue that this metric-centric regime is not merely incomplete, but structurally misaligned with the broader objectives of forecasting. In real-world settings, forecasting often prioritizes preserving temporal structure, trend stability, seasonal coherence, robustness to regime shifts, and supporting downstream decision processes. Optimizing aggregate pointwise error does not necessarily imply modeling these structural properties. As a result, leaderboard improvement may increasingly reflect specialization in benchmark configurations rather than a deeper understanding of temporal dynamics. This paper revisits LTSF evaluation as a foundational question in data science: what does it mean to measure forecasting progress? We propose a multi-dimensional evaluation perspective that integrates statistical fidelity, structural coherence, and decision-level relevance. By challenging the current metric monoculture, we aim to redirect attention from winning benchmark tables toward advancing meaningful, context-aware forecasting.

Classifier-Free Guidance (CFG) is a cornerstone of modern conditional diffusion models, yet its reliance on the fixed or heuristic dynamic guidance weight is predominantly empirical and overlooks the inherent dynamics of the diffusion process. In this paper, we provide a rigorous theoretical analysis of the Classifier-Free Guidance. Specifically, we establish strict upper bounds on the score discrepancy between conditional and unconditional distributions at different timesteps based on the diffusion process. This finding explains the limitations of fixed-weight strategies and establishes a principled foundation for time-dependent guidance. Motivated by this insight, we introduce \textbf{Control Classifier-Free Guidance (C$^2$FG)}, a novel, training-free, and plug-in method that aligns the guidance strength with the diffusion dynamics via an exponential decay control function. Extensive experiments demonstrate that C$^2$FG is effective and broadly applicable across diverse generative tasks, while also exhibiting orthogonality to existing strategies.

Large language models (LLMs) and machine translation (MT) systems are increasingly used in our daily lives, but their outputs can reproduce gender bias present in the training data. Most resources for evaluating such biases are designed for English and reflect its sociocultural context, which limits their applicability to other languages. This work addresses this gap by introducing two new datasets to evaluate gender bias in translations involving Basque, a low-resource and genderless language. WinoMTeus adapts the WinoMT benchmark to examine how gender-neutral Basque occupations are translated into gendered languages such as Spanish and French. FLORES+Gender, in turn, extends the FLORES+ benchmark to assess whether translation quality varies when translating from gendered languages (Spanish and English) into Basque depending on the gender of the referent. We evaluate several general-purpose LLMs and open and proprietary MT systems. The results reveal a systematic preference for masculine forms and, in some models, a slightly higher quality for masculine referents. Overall, these findings show that gender bias is still deeply rooted in these models, and highlight the need to develop evaluation methods that consider both linguistic features and cultural context.

Recently, large language models (LLMs) have gained much attention for the emergence of human-comparable capabilities and huge potential. However, for open-domain implicit question-answering problems, LLMs may not be the ultimate solution due to the reasons of: 1) uncovered or out-of-date domain knowledge, 2) one-shot generation and hence restricted comprehensiveness. To this end, this work proposes a gradual knowledge excavation framework for open-domain complex question answering, where LLMs iteratively and actively acquire external information, and then reason based on acquired historical knowledge. Specifically, during each step of the solving process, the model selects an action to execute, such as querying external knowledge or performing a single logical reasoning step, to gradually progress toward a final answer. Our method can effectively leverage plug-and-play external knowledge and dynamically adjust the strategy for solving complex questions. Evaluated on the StrategyQA dataset, our method achieves 78.17% accuracy with less than 6% parameters of its competitors, setting new SOTA for ~10B-scale LLMs.

Existing frameworks for gradient-based training of spiking neural networks face a trade-off: discrete-time methods using surrogate gradients support arbitrary neuron models but introduce gradient bias and constrain spike-time resolution, while continuous-time methods that compute exact gradients require analytical expressions for spike times and state evolution, restricting them to simple neuron types such as Leaky Integrate and Fire (LIF). We introduce the Eventax framework, which resolves this trade-off by combining differentiable numerical ODE solvers with event-based spike handling. Built in JAX, our frame-work uses Diffrax ODE-solvers to compute gradients that are exact with respect to the forward simulation for any neuron model defined by ODEs . It also provides a simple API where users can specify just the neuron dynamics, spike conditions, and reset rules. Eventax prioritises modelling flexibility, supporting a wide range of neuron models, loss functions, and network architectures, which can be easily extended. We demonstrate Eventax on multiple benchmarks, including Yin-Yang and MNIST, using diverse neuron models such as Leaky Integrate-and-fire (LIF), Quadratic Integrate-and-fire (QIF), Exponential integrate-and-fire (EIF), Izhikevich and Event-based Gated Recurrent Unit (EGRU) with both time-to-first-spike and state-based loss functions, demonstrating its utility for prototyping and testing event-based architectures trained with exact gradients. We also demonstrate the application of this framework for more complex neuron types by implementing a multi-compartment neuron that uses a model of dendritic spikes in human layer 2/3 cortical Pyramidal neurons for computation. Code available at https://github.com/efficient-scalable-machine-learning/eventax.

Preference-based alignment methods (e.g., RLHF, DPO) typically optimize a single scalar objective, implicitly averaging over heterogeneous human preferences. In practice, systematic annotator and user-group disagreement makes mean-reward maximization brittle and susceptible to proxy over-optimization. We propose **Disagreement-Aware Alignment via Risk-Constrained Decoding (DARC)**, a retraining-free inference-time method that frames response selection as distributionally robust, risk-sensitive decision making. Given multiple preference samples or scalable disagreement proxies, DARC reranks candidates by maximizing a *KL-robust (entropic)* satisfaction objective, and provides simple deployment controls that cap or penalize the corresponding entropic risk premium relative to the mean, enabling explicit risk budgets without retraining. We provide theoretical characterization linking this decoding rule to principled pessimism and KL-based distributionally robust optimization. Experiments on alignment benchmarks show that DARC reduces disagreement and tail risk while maintaining competitive average quality under noisy, heterogeneous feedback.

Graph anomaly detection (GAD) aims to identify nodes that deviate from normal patterns in structure or features. While recent GNN-based approaches have advanced this task, they struggle with two major challenges: 1) homophily disparity, where nodes exhibit varying homophily at both class and node levels; and 2) limited scalability, as many methods rely on costly whole-graph operations. To address them, we propose SAGAD, a Scalable and Adaptive framework for GAD. SAGAD precomputes multi-hop embeddings and applies reparameterized Chebyshev filters to extract low- and high-frequency information, enabling efficient training and capturing both homophilic and heterophilic patterns. To mitigate node-level homophily disparity, we introduce an Anomaly Context-Aware Adaptive Fusion, which adaptively fuses low- and high-pass embeddings using fusion coefficients conditioned on Rayleigh Quotient-guided anomalous subgraph structures for each node. To alleviate class-level disparity, we design a Frequency Preference Guidance Loss, which encourages anomalies to preserve more high-frequency information than normal nodes. SAGAD supports mini-batch training, achieves linear time and space complexity, and drastically reduces memory usage on large-scale graphs. Theoretically, SAGAD ensures asymptotic linear separability between normal and abnormal nodes under mild conditions. Extensive experiments on 10 benchmarks confirm SAGAD's superior accuracy and scalability over state-of-the-art methods.

We present a novel Explainable methodology for Condition Monitoring, relying on healthy data only. Since faults are rare events, we propose to focus on learning the probability distribution of healthy observations only, and detect Anomalies at runtime. This objective is achieved via the definition of probabilistic measures of deviation from nominality, which allow to detect and anticipate faults. The Bayesian perspective underpinning our approach allows us to perform Uncertainty Quantification to inform decisions. At the same time, we provide descriptive tools to enhance the interpretability of the results, supporting the deployment of the proposed strategy also in safety-critical applications. The methodology is validated experimentally on two use cases: a publicly available benchmark for Predictive Maintenance, and a real-world Helicopter Transmission dataset collected over multiple years. In both applications, the method achieves competitive detection performance with respect to state-of-the-art anomaly detection methods.

Most uncertainty-aware robotic systems collapse prediction uncertainty into a single scalar score and use it to trigger uniform corrective responses. This aggregation obscures whether uncertainty arises from corrupted observations or from mismatch between the learned model and the true system dynamics. As a result, corrective actions may be applied to the wrong component of the closed loop, degrading performance relative to leaving the policy unchanged. We introduce a lightweight post hoc framework that decomposes uncertainty into aleatoric and epistemic components and uses these signals to regulate system responses at inference time. Aleatoric uncertainty is estimated from deviations in the observation distribution using a Mahalanobis density model, while epistemic uncertainty is detected using a noise robust forward dynamics ensemble that isolates model mismatch from measurement corruption. The two signals remain empirically near orthogonal during closed loop execution and enable type specific responses. High aleatoric uncertainty triggers observation recovery, while high epistemic uncertainty moderates control actions. The same signals also regulate adaptive perception by guiding model capacity selection during tracking inference. Experiments demonstrate consistent improvements across both control and perception tasks. In robotic manipulation, the decomposed controller improves task success from 59.4% to 80.4% under compound perturbations and outperforms a combined uncertainty baseline by up to 21.0%. In adaptive tracking inference on MOT17, uncertainty-guided model selection reduces average compute by 58.2% relative to a fixed high capacity detector while preserving detection quality within 0.4%. Code and demo videos are available at https://divake.github.io/uncertainty-decomposition/.

The increasing adoption of Large Language Models (LLMs) has enabled AI scientists to perform complex end-to-end scientific discovery tasks requiring coordination of specialized roles, including idea generation and experimental execution. However, most state-of-the-art AI scientist systems rely on static, hand-designed pipelines and fail to adapt based on accumulated interaction histories. As a result, these systems overlook promising research directions, repeat failed experiments, and pursue infeasible ideas. To address this, we introduce EvoScientist, an evolving multi-agent AI scientist framework that continuously improves research strategies through persistent memory and self-evolution. EvoScientist comprises three specialized agents: a Researcher Agent (RA) for scientific idea generation, an Engineer Agent (EA) for experiment implementation and execution, and an Evolution Manager Agent (EMA) that distills insights from prior interactions into reusable knowledge. EvoScientist contains two persistent memory modules: (i) an ideation memory, which summarizes feasible research directions from top-ranked ideas while recording previously unsuccessful directions; and (ii) an experimentation memory, which captures effective data processing and model training strategies derived from code search trajectories and best-performing implementations. These modules enable the RA and EA to retrieve relevant prior strategies, improving idea quality and code execution success rates over time. Experiments show that EvoScientist outperforms 7 open-source and commercial state-of-the-art systems in scientific idea generation, achieving higher novelty, feasibility, relevance, and clarity via automatic and human evaluation. EvoScientist also substantially improves code execution success rates through multi-agent evolution, demonstrating persistent memory's effectiveness for end-to-end scientific discovery.

Coordinated audio generation based on video inputs typically requires a strict audio-visual (AV) alignment, where both semantics and rhythmics of the generated audio segments shall correspond to those in the video frames. Previous studies leverage a two-stage design where the AV encoders are firstly aligned via contrastive learning, then the encoded video representations guide the audio generation process. We observe that both contrastive learning and global video guidance are effective in aligning overall AV semantics while limiting temporally rhythmic synchronization. In this work, we propose FoleyFlow to first align unimodal AV encoders via masked modeling training, where the masked audio segments are recovered under the guidance of the corresponding video segments. After training, the AV encoders which are separately pretrained using only unimodal data are aligned with semantic and rhythmic consistency. Then, we develop a dynamic conditional flow for the final audio generation. Built upon the efficient velocity flow generation framework, our dynamic conditional flow utilizes temporally varying video features as the dynamic condition to guide corresponding audio segment generations. To this end, we extract coherent semantic and rhythmic representations during masked AV alignment, and use this representation of video segments to guide audio generation temporally. Our audio results are evaluated on the standard benchmarks and largely surpass existing results under several metrics. The superior performance indicates that FoleyFlow is effective in generating coordinated audios that are both semantically and rhythmically coherent to various video sequences.

Ramsa is a developing 41-hour speech corpus of Emirati Arabic designed to support sociolinguistic research and low-resource language technologies. It contains recordings from structured interviews with native speakers and episodes from national television shows. The corpus features 157 speakers (59 female, 98 male), spans subdialects such as Urban, Bedouin, and Mountain/Shihhi, and covers topics such as cultural heritage, agriculture and sustainability, daily life, professional trajectories, and architecture. It consists of 91 monologic and 79 dialogic recordings, varying in length and recording conditions. A 10\% subset was used to evaluate commercial and open-source models for automatic speech recognition (ASR) and text-to-speech (TTS) in a zero-shot setting to establish initial baselines. Whisper-large-v3-turbo achieved the best ASR performance, with average word and character error rates of 0.268 and 0.144, respectively. MMS-TTS-Ara reported the best mean word and character rates of 0.285 and 0.081, respectively, for TTS. These baselines are competitive but leave substantial room for improvement. The paper highlights the challenges encountered and provides directions for future work.

We revisit Vision-Language-Action through a neuroscience-inspired triad. Biologically, the Cerebrum provides stable high-level multimodal priors and remains frozen; the Pons Adapter integrates these cortical features with real-time proprioceptive inputs and compiles intent into execution-ready tokens; and the Cerebellum (ParaCAT) performs fast, parallel categorical decoding for online control, with hysteresis/EMA/temperature/entropy for stability. A fixed-ratio schedule and two-stage feature caching make the system compute-aware and reproducible. Inspired by active, foveated vision, our wrist ROIs are geometrically tied to the end-effector via calibrated projection, providing a movement-stabilized, high-resolution view that is sensitive to fine-grained pose changes and complements the global context of the main view. The design is modular: upgrading the Cerebrum only retrains the Pons; changing robots only trains the Cerebellum; cerebellum-only RL can further refine control without touching high-level semantics. As a concept-and-protocol paper with preliminary evidence, we outline a timing protocol under matched conditions (GPU, resolution, batch) to verify anticipated efficiency gains. We also report preliminary LIBERO evidence showing that split feature caching reduces training time (7.5h to 4.5h) and improves average success (86.5% to 92.5%) under official N1.5 head-only training, and that SaiVLA0 reaches 99.0% mean success.

Model-based offline reinforcement learning (RL) aims to enhance offline RL with a dynamics model that facilitates policy exploration. However, \textit{model exploitation} could occur due to inevitable model errors, degrading algorithm performance. Adversarial model learning offers a theoretical framework to mitigate model exploitation by solving a maximin formulation. Within such a paradigm, RAMBO~\citep{rigter2022rambo} has emerged as a representative and most popular method that provides a practical implementation with model gradient. However, we empirically reveal that severe Q-value underestimation and gradient explosion can occur in RAMBO with only slight hyperparameter tuning, suggesting that it tends to be overly conservative and suffers from unstable model updates. To address these issues, we propose \textbf{RO}bust value-aware \textbf{M}odel learning with \textbf{I}mplicitly differentiable adaptive weighting (ROMI). Instead of updating the dynamics model with model gradient, ROMI introduces a novel robust value-aware model learning approach. This approach requires the dynamics model to predict future states with values close to the minimum Q-value within a scale-adjustable state uncertainty set, enabling controllable conservatism and stable model updates. To further improve out-of-distribution (OOD) generalization during multi-step rollouts, we propose implicitly differentiable adaptive weighting, a bi-level optimization scheme that adaptively achieves dynamics- and value-aware model learning. Empirical results on D4RL and NeoRL datasets show that ROMI significantly outperforms RAMBO and achieves competitive or superior performance compared to other state-of-the-art methods on datasets where RAMBO typically underperforms. Code is available at https://github.com/zq2r/ROMI.git.

Recent advancements in LLM-based information-seeking agents have achieved record-breaking performance on established benchmarks. However, these agents remain heavily reliant on search-engine-indexed knowledge, leaving a critical blind spot: Unindexed Information Seeking (UIS). This paper identifies and explores the UIS problem, where vital information is not captured by search engine crawlers, such as overlooked content, dynamic webpages, and embedded files. Despite its significance, UIS remains an underexplored challenge. To address this gap, we introduce UIS-QA, the first dedicated UIS benchmark, comprising 110 expert-annotated QA pairs. Notably, even state-of-the-art agents experience a drastic performance drop on UIS-QA (e.g., from 70.90 on GAIA and 46.70 on BrowseComp-zh to 24.55 on UIS-QA), underscoring the severity of the problem. To mitigate this, we propose UIS-Digger, a novel multi-agent framework that incorporates dual-mode browsing and enables simultaneous webpage searching and file parsing. With a relatively small $\sim$30B-parameter backbone LLM optimized using SFT and RFT training strategies, UIS-Digger sets a strong baseline at 27.27\%, outperforming systems integrating sophisticated LLMs such as O3 and GPT-4.1. This demonstrates the importance of proactive interaction with unindexed sources for effective and comprehensive information-seeking. Our work not only uncovers a fundamental limitation in current agent evaluation paradigms but also provides the first toolkit for advancing UIS research, defining a new and promising direction for robust information-seeking systems.

Mechanistic modeling provides a biophysically grounded framework for studying the spread of pathological tau protein in tauopathies like Alzheimer's disease. Existing approaches typically model tau propagation as a diffusive process on the brain's structural connectome, reproducing macroscopic patterns but neglecting microscale cellular transport and reaction mechanisms. The Network Transport Model (NTM) was introduced to fill this gap, explaining how region-level progression of tau emerges from microscale biophysical processes. However, the NTM faces a common challenge for complex models defined by large systems of partial differential equations: the inability to perform parameter inference and mechanistic discovery due to high computational burden and slow model simulations. To overcome this barrier, we propose Tau-BNO, a Brain Neural Operator surrogate framework for rapidly approximating NTM dynamics that captures both intra-regional reaction kinetics and inter-regional network transport. Tau-BNO combines a function operator that encodes kinetic parameters with a query operator that preserves initial state information, while approximating anisotropic transport through a spectral kernel that retains directionality. Empirical evaluations demonstrate high predictive accuracy ($R^2\approx$ 0.98) across diverse biophysical regimes and an 89\% performance improvement over state-of-the-art sequence models like Transformers and Mamba, which lack inherent structural priors. By reducing simulation time from hours to seconds, we show that the surrogate model is capable of producing new insights and generating new hypotheses. This framework is readily extensible to a broader class of connectome-based biophysical models, showcasing the transformative value of deep learning surrogates to accelerate analysis of large-scale, computationally intensive dynamical systems.

Understanding and addressing potential safety alignment risks in large language models (LLMs) is critical for ensuring their safe and trustworthy deployment. In this paper, we highlight an insidious safety threat: a compromised LLM can maintain a facade of proper safety alignment while covertly generating harmful content. To achieve this, we finetune the model to understand and apply a steganographic technique. At inference time, we input a prompt that contains a steganographically embedded malicious target question along with a plaintext cover question. The model, in turn, produces a target response similarly embedded within a benign-looking cover response. In this process, human observers only see the model being prompted with a cover question and generating a corresponding cover response, while the malicious content is hidden from view. We demonstrate this invisible safety threat on GPT-4.1 despite the OpenAI finetuning API's safeguards. The finetuned model produces steganographic malicious outputs in response to hidden malicious prompts, while the user interface displays only a fully benign cover interaction. We also replicate the attack on three open-source models, Llama-3.3-70B-Instruct, Phi-4, and Mistral-Small-24B-Base-2501, confirming the generality of our method. We quantitatively evaluate our method on the AdvBench dataset, using Llama-Guard-3-8B for content safety classification. Across all four models, all stegotexts containing malicious content are incorrectly classified as safe.

In scientific reasoning tasks, the veracity of the reasoning process is as critical as the final outcome. While Process Reward Models (PRMs) offer a solution to the coarse-grained supervision problems inherent in Outcome Reward Models (ORMs), their deployment is hindered by the prohibitive cost of obtaining expert-verified step-wise labels. This paper addresses the challenge of training reliable PRMs using abundant but noisy "weak" supervision. We argue that existing Weak-to-Strong Generalization (W2SG) theories lack prescriptive guidelines for selecting high-quality training signals from noisy data. To bridge this gap, we introduce the Dual-Consensus Weak-to-Strong (DC-W2S) framework. By intersecting Self-Consensus (SC) metrics among weak supervisors with Neighborhood-Consensus (NC) metrics in the embedding space, we stratify supervision signals into distinct reliability regimes. We then employ a curriculum of instance-level balanced sampling and label-level reliability-aware masking to guide the training process. We demonstrate that DC-W2S enables the training of robust PRMs for complex reasoning without exhaustive expert annotation, proving that strategic data curation is more effective than indiscriminate training on large-scale noisy datasets.

Large language model (LLM)-based judges are widely adopted for automated evaluation and reward modeling, yet their judgments are often affected by judgment biases. Accurately evaluating these biases is essential for ensuring the reliability of LLM-based judges. However, existing studies typically investigate limited biases under a single judge formulation, either generative or discriminative, lacking a comprehensive evaluation. To bridge this gap, we propose JudgeBiasBench, a benchmark for systematically quantifying biases in LLM-based judges. JudgeBiasBench defines a taxonomy of judgment biases across 4 dimensions, and constructs bias-augmented evaluation instances through a controlled bias injection pipeline, covering 12 representative bias types. We conduct extensive experiments across both generative and discriminative judges, revealing that current judges exhibit significant and diverse bias patterns that often compromise the reliability of automated evaluation. To mitigate judgment bias, we propose bias-aware training that explicitly incorporates bias-related attributes into the training process, encouraging judges to disentangle task-relevant quality from bias-correlated cues. By adopting reinforcement learning for generative judges and contrastive learning for discriminative judges, our methods effectively reduce judgment biases while largely preserving general evaluation capability.

Significant progress has been achieved in subject-driven text-to-image (T2I) generation, which aims to synthesize new images depicting target subjects according to user instructions. However, evaluating these models remains a significant challenge. Existing benchmarks exhibit critical limitations: 1) insufficient diversity and comprehensiveness in subject images, 2) inadequate granularity in assessing model performance across different subject difficulty levels and prompt scenarios, and 3) a profound lack of actionable insights and diagnostic guidance for subsequent model refinement. To address these limitations, we propose DSH-Bench, a comprehensive benchmark that enables systematic multi-perspective analysis of subject-driven T2I models through four principal innovations: 1) a hierarchical taxonomy sampling mechanism ensuring comprehensive subject representation across 58 fine-grained categories, 2) an innovative classification scheme categorizing both subject difficulty level and prompt scenario for granular capability assessment, 3) a novel Subject Identity Consistency Score (SICS) metric demonstrating a 9.4\% higher correlation with human evaluation compared to existing measures in quantifying subject preservation, and 4) a comprehensive set of diagnostic insights derived from the benchmark, offering critical guidance for optimizing future model training paradigms and data construction strategies. Through an extensive empirical evaluation of 19 leading models, DSH-Bench uncovers previously obscured limitations in current approaches, establishing concrete directions for future research and development.

Autoregressive decoding remains a primary bottleneck in large language model (LLM) serving, motivating speculative decoding methods that reduce expensive teacher-model invocations by verifying multiple candidate tokens per step. Tree-structured speculation further increases parallelism, but is often brittle when ported across heterogeneous backends and accelerator stacks, where attention masking, KV-cache layouts, and indexing semantics are not interchangeable. We present EAGLE-Pangu, a reproducible system that ports EAGLE-3-style tree speculative decoding to a Pangu teacher backend on Ascend NPUs. EAGLE-Pangu contributes (i) an explicit branch/commit cache manager built on the Cache API, (ii) accelerator-safe tree tensorization that removes undefined negative indices by construction and validates structural invariants, and (iii) a fused-kernel-compatible teacher verification path with a debuggable eager fallback. On 240 turns from MT-Bench and HumanEval-style prompts, EAGLE-Pangu improves end-to-end decoding throughput by 1.27x on average, up to 2.46x at p99, over teacher-only greedy decoding in the fused-kernel performance path. We also provide a fused-kernel-free reference path with structured traces and invariant checks to support reproducible debugging and ablation across execution modes and tree budgets.

Large Language Models (LLMs) have demonstrated exceptional performance across a wide range of tasks, yet their significant computational and memory requirements present major challenges for deployment. A common approach uses Taylor expansion on the loss function to estimate neuron importance. However, its reliance on one-hot cross entropy loss, a key limitation is that it narrowly assesses importance based only on the probability assigned to the single predicted next token, thereby ignoring the other potential predictions of the original model. An intuitive solution to address this is to employ self distillation criterion for importance evaluation. However, this approach introduces significant computational overhead by requiring a separate teacher model for supervision. To this end, we propose a simple but effective criterion, information entropy of the model's output distribution, to efficiently evaluate importance scores of neurons with Taylor pruning without requirement of additional teacher. Compared to plain cross entropy criterion, it provides a more holistic criterion for Taylor pruning to prune neurons with the least impact on the prediction of model in a global manner, thereby preserving the fidelity of the model's predictive capabilities. Experimental results on extensive zero-shot benchmarks demonstrate that our method consistently outperforms existing pruning methods across the LLaMA and Qwen series models. The source code and trained weights are availabel at https://github.com/visresearch/HFPrune.

Tiny Recursive Models (TRMs) have recently demonstrated remarkable performance on ARC-AGI, showing that very small models can compete against large foundation models through a two-step refinement mechanism that updates an internal reasoning state $z$ and the predicted output $y$. Naturally, such refinement is of interest for any predictor; it is therefore natural to wonder whether the TRM mechanism could be effectively re-adopted in autoregressive models. However, TRMs cannot be simply compared to standard models because they lack causal predictive structures and contain persistent latent states that make it difficult to isolate specific performance gains. In this paper, we propose the Autoregressive TRM and evaluate it on small autoregressive tasks. To understand its efficacy, we propose a suite of models that gradually transform a standard Transformer to a Tiny Autoregressive Recursive Model in a controlled setting that fixes the block design, token stream, and next-token objective. Across compute-matched experiments on character-level algorithmic tasks, we surprisingly find that there are some two-level refinement baselines that show strong performance. Contrary to expectations, we find no reliable performance gains from the full Autoregressive TRM architecture. These results offer potential promise for two-step refinement mechanisms more broadly but caution against investing in the autoregressive TRM-specific model as a fruitful research direction.

Forecasting physiological signals can support proactive monitoring and timely clinical intervention by anticipating critical changes in patient status. In this work, we address multivariate multi-horizon forecasting of physiological time series by jointly predicting heart rate, oxygen saturation, pulse rate, and respiratory rate at forecasting horizons of 15, 30, and 60 seconds. We propose a hybrid quantum-classical architecture that integrates a Variational Quantum Circuit (VQC) within a recurrent neural backbone. A GRU encoder summarizes the historical observation window into a latent representation, which is then projected into quantum angles used to parameterize the VQC. The quantum layer acts as a learnable non-linear feature mixer, modeling cross-variable interactions before the final prediction stage. We evaluate the proposed approach on the BIDMC PPG and Respiration dataset under a Leave-One-Patient-Out protocol. The results show competitive accuracy compared with classical and deep learning baselines, together with greater robustness to noise and missing inputs. These findings suggest that hybrid quantum layers can provide useful inductive biases for physiological time series forecasting in small-cohort clinical settings.

While large language models (LLMs) exhibit strong reasoning abilities, their performance on complex tasks is often constrained by the limitations of their internal knowledge. A compelling approach to overcome this challenge is to augment these models with external tools -- such as Python interpreters for mathematical computations or search engines for retrieving factual information. However, enabling models to use these tools effectively remains a significant challenge. Existing methods typically rely on cold-start pipelines that begin with supervised fine-tuning (SFT), followed by reinforcement learning (RL). These approaches often require substantial amounts of labeled data for SFT, which is expensive to annotate or synthesize. In this work, we propose In-Context Reinforcement Learning (ICRL), an RL-only framework that eliminates the need for SFT by leveraging few-shot prompting during the rollout stage of RL. Specifically, ICRL introduces in-context examples within the rollout prompts to teach the model how to invoke external tools. Furthermore, as training progresses, the number of in-context examples is gradually reduced, eventually reaching a zero-shot setting where the model learns to call tools independently. We conduct extensive experiments across a range of reasoning and tool-use benchmarks. Results show that ICRL achieves state-of-the-art performance, demonstrating its effectiveness as a scalable, data-efficient alternative to traditional SFT-based pipelines.

Structured pruning reduces LLM inference cost by removing low-importance architectural components. This can be viewed as learning a multiplicative gate for each component under an l0 sparsity constraint. Due to the discreteness of the l0 norm, prior work typically adopts stochastic hard-concrete relaxations to enable differentiable optimization; however, this stochasticity can introduce a train--test mismatch when sampled masks are discretized for deployment and restricts masks to a bounded, near-binary range. To address this, we propose Deterministic Differentiable Pruning (DDP), a mask-only optimization method that eliminates stochasticity by directly optimizing a deterministic soft surrogate of the discrete l0 objective. Compared with prior approaches, DDP offers greater expressiveness, reduced train--test mismatch, and faster convergence. We apply our method to several dense and MoE models, including Qwen3-32B and Qwen3-30B-A3B, achieving a performance loss as small as 1% on downstream tasks while outperforming previous methods at 20% sparsity. We further demonstrate end-to-end inference speedups in realistic deployment settings with vLLM.

Accurate phenotype prediction from RNA sequencing (RNA-seq) data is essential for diagnosis, biomarker discovery, and personalized medicine. Deep learning models have demonstrated strong potential to outperform classical machine learning approaches, but their performance relies on large, well-annotated datasets. In transcriptomics, such datasets are frequently limited, leading to over-fitting and poor generalization. Knowledge transfer from larger, more general datasets can alleviate this issue. However, transferring information across RNA-seq datasets remains challenging due to heterogeneous preprocessing pipelines and differences in target phenotypes. In this study, we propose a deep learning-based domain adaptation framework that enables effective knowledge transfer from a large general dataset to a smaller one for cancer type classification. The method learns a domain-invariant latent space by jointly optimizing classification and domain alignment objectives. To ensure stable training and robustness in data-scarce scenarios, the framework is trained with an adversarial approach with appropriate regularization. Both supervised and unsupervised approach variants are explored, leveraging labeled or unlabeled target samples. The framework is evaluated on three large-scale transcriptomic datasets (TCGA, ARCHS4, GTEx) to assess its ability to transfer knowledge across cohorts. Experimental results demonstrate consistent improvements in cancer and tissue type classification accuracy compared to non-adaptive baselines, particularly in low-data scenarios. Overall, this work highlights domain adaptation as a powerful strategy for data-efficient knowledge transfer in transcriptomics, enabling robust phenotype prediction under constrained data conditions.

With the rapid advancement of commercial multi-modal models, image editing has garnered significant attention due to its widespread applicability in daily life. Despite impressive progress, existing image editing systems, particularly closed-source or proprietary models, often struggle with complex, indirect, or multi-step user instructions. These limitations hinder their ability to perform nuanced, context-aware edits that align with human intent. In this work, we propose ImageEdit-R1, a multi-agent framework for intelligent image editing that leverages reinforcement learning to coordinate high-level decision-making across a set of specialized, pretrained vision-language and generative agents. Each agent is responsible for distinct capabilities--such as understanding user intent, identifying regions of interest, selecting appropriate editing actions, and synthesizing visual content--while reinforcement learning governs their collaboration to ensure coherent and goal-directed behavior. Unlike existing approaches that rely on monolithic models or hand-crafted pipelines, our method treats image editing as a sequential decision-making problem, enabling dynamic and context-aware editing strategies. Experimental results demonstrate that ImageEdit-R1 consistently outperforms both individual closed-source diffusion models and alternative multi-agent framework baselines across multiple image editing datasets.

Large Language Models (LLMs) are pivotal in natural language processing. The impracticality of full fine-tuning has prompted Parameter-Efficient Fine-Tuning (PEFT) methods like Low-Rank Adaptation (LoRA), optimizing low-rank matrices A and B. In distributed scenarios where privacy constraints necessitate Federated Learning (FL), however, the integration of LoRA is often unstable. Specifically, we identify that aggregating updates from multiple clients introduces statistical variance that scales with the client count, causing gradient collapse when using high-rank adapters. Existing scaling factor candidates, such as the one used by Rank-Stabilized LoRA, ignore the interaction caused by the aggregation process. To bridge this gap, this paper introduces Stabilized Federated LoRA (SFed-LoRA), a framework that theoretically characterizes the interaction between adapter rank and federated aggregation. We derive an optimal scaling factor designed to effectively mitigate the aggregation error accumulating across N clients. By correcting the scaling mismatch inherent in previous approaches, SFed-LoRA restores the efficacy of high-rank adaptation without altering the original model architecture or increasing inference latency. Extensive experiments in diverse tasks, model architectures, and heterogeneous data distributions are conducted to validate our results. We demonstrate that SFed-LoRA prevents high-rank collapse, and achieves significantly improved stability and faster convergence compared with state-of-the-art baselines for high-rank adaptation.

While recent feed-forward 3D reconstruction models accelerate 3D reconstruction by jointly inferring dense geometry and camera poses in a single pass, their reliance on dense attention imposes a quadratic complexity, creating a prohibitive computational bottleneck that severely limits inference speed. To resolve this, we introduce Speed3R, an end-to-end trainable model inspired by the core principle of Structure-from-Motion: that a sparse set of keypoints is sufficient for robust pose estimation. Speed3R features a dual-branch attention mechanism where a compression branch creates a coarse contextual prior to guide a selection branch, which performs fine-grained attention only on the most informative image tokens. This strategy mimics the efficiency of traditional keypoint matching, achieving a remarkable 12.4x inference speedup on 1000-view sequences, while introducing a minimal, controlled trade-off in geometric accuracy. Validated on standard benchmarks with both VGGT and $π^3$ backbones, our method delivers high-quality reconstructions at a fraction of computational cost, paving the way for efficient large-scale scene modeling.

The relationship between content production and consumption on algorithm-driven platforms like YouTube plays a critical role in shaping ideological behaviors. While prior work has largely focused on user behavior and algorithmic recommendations, the interplay between what is produced and what gets consumed, and its role in ideological shifts remains understudied. In this paper, we present a longitudinal, mixed-methods analysis combining one year of YouTube watch history with two waves of ideological surveys from 1,100 U.S. participants. We identify users who exhibited significant shifts toward more extreme ideologies and compare their content consumption and the production patterns of YouTube channels they engaged with to ideologically stable users. Our findings show that users who became more extreme consumed have different consumption habits from those who do not. This gets amplified by the fact that channels favored by users with extreme ideologies also have a higher affinity to produce content with a higher anger, grievance and other such markers. Lastly, using time series analysis, we examine whether content producers are the primary drivers of consumption behavior or merely responding to user demand.

Fault diagnosis is critical for the safe operation of industrial systems. Conventional diagnosis models typically produce abstract outputs such as anomaly scores or fault categories, failing to answer critical operational questions like "Why" or "How to repair". While large language models (LLMs) offer strong generalization and reasoning abilities, their training on discrete textual corpora creates a semantic gap when processing high-dimensional, temporal industrial signals. To address this challenge, we propose a Signals-to-Semantics fault diagnosis (S2S-FDD) framework that bridges high-dimensional sensor signals with natural language semantics through two key innovations: We first design a Signal-to-Semantic operator to convert abstract time-series signals into natural language summaries, capturing trends, periodicity, and deviations. Based on the descriptions, we design a multi-turn tree-structured diagnosis method to perform fault diagnosis by referencing historical maintenance documents and dynamically querying additional signals. The framework further supports human-in-the-loop feedback for continuous refinement. Experiments on the multiphase flow process show the feasibility and effectiveness of the proposed method for explainable zero-shot fault diagnosis.

Reward modeling is essential for aligning Large Language Models(LLMs) with human preferences, yet conventional reward models suffer from poor interpretability and heavy reliance on costly expert annotations. While recent rubric-based approaches enhance evaluation transparency, they lack systematic quality control, yielding noisy and redundant criteria, failing to mitigate persistent biases (e.g., verbosity, position) in LLM evaluators, and creating a scalability-reliability trade-off. To address these limitations, we propose CDRRM (Contrast-Driven Rubric Reward Model), a framework built on a novel Contrast-then-Synthesis paradigm for high-quality rubric generation and guided preference judgment. CDRRM first conducts multi-dimensional contrastive profiling on preference pairs to identify causal discriminative factors, then synthesizes these insights into compact, context-aware rubrics to guide preference judg- ments. Extensive experiments on three authoritative benchmarks (RewardBench, RMBench, RMB) demonstrate that CDRRM achieves state-of-the-art performance across diverse domains and effectively mitigates aforementioned evaluation biases. Notably, our approach delivers exceptional data efficiency: training the rubric generator on only 3k high-quality samples empowers a frozen pre-trained judge model to outperform fully fine-tuned baselines. This work offers a scalable, interpretable, and data-efficient path for reward modeling.

Emotion recognition in real-world environments is hindered by partial occlusions, missing modalities, and severe class imbalance. To address these issues, particularly for the Affective Behavior Analysis in-the-wild (ABAW) Expression challenge, we propose a multimodal framework that dynamically fuses visual and audio representations. Our approach uses a dual-branch Transformer architecture featuring a safe cross-attention mechanism and a modality dropout strategy. This design allows the network to rely on audio-based predictions when visual cues are absent. To mitigate the long-tail distribution of the Aff-Wild2 dataset, we apply focal loss optimization, combined with a sliding-window soft voting strategy to capture dynamic emotional transitions and reduce frame-level classification jitter. Experiments demonstrate that our framework effectively handles missing modalities and complex spatiotemporal dependencies, achieving an accuracy of 60.79% and an F1-score of 0.5029 on the Aff-Wild2 validation set.

Exogenous variables offer valuable supplementary information for predicting future endogenous variables. Forecasting with exogenous variables needs to consider both past-to-future dependencies (i.e., temporal correlations) and the influence of exogenous variables on endogenous variables (i.e., channel correlations). This is pivotal when future exogenous variables are available, because they may directly affect the future endogenous variables. Many methods have been proposed for time series forecasting with exogenous variables, focusing on modeling temporal and channel correlations. However, most of them use a two-step strategy, modeling temporal and channel correlations separately, which limits their ability to capture joint correlations across time and channels. Furthermore, in real-world scenarios, time series are frequently affected by various forms of noises, underscoring the critical importance of robustness in such correlations modeling. To address these limitations, we propose GCGNet, a Graph-Consistent Generative Network for time series forecasting with exogenous variables. Specifically, GCGNet first employs a Variational Generator to produce coarse predictions. A Graph Structure Aligner then further guides it by evaluating the consistency between the generated and true correlations, where the correlations are represented as graphs, and are robust to noises. Finally, a Graph Refiner is proposed to refine the predictions to prevent degeneration and improve accuracy. Extensive experiments on 12 real-world datasets demonstrate that GCGNet outperforms state-of-the-art baselines.

Masked Diffusion Language Models (MDLMs) enable parallel token decoding, providing a promising alternative to the sequential nature of autoregressive generation. However, their iterative denoising process remains computationally expensive because it repeatedly processes the entire sequence at every step. We observe that across these diffusion steps, most token representations remain stable; only a small subset, which we term salient tokens, contributes meaningfully to the next update. Leveraging this temporal sparsity, we present DyLLM, a training-free inference framework that accelerates decoding by selectively computing only these salient tokens. DyLLM identifies saliency by measuring the cosine similarity of attention contexts between adjacent denoising steps. It recomputes feed-forward and attention operations only for salient tokens while reusing cached activations for the remainder. Across diverse reasoning and code-generation benchmarks, DyLLM achieves up to 9.6x higher throughput while largely preserving the baseline accuracy of state-of-the-art models like LLaDA and Dream.

As large language models (LLMs) evolve into autonomous agents capable of acting in open-ended environments, ensuring behavioral alignment with human values becomes a critical safety concern. Existing benchmarks, focused on static, single-turn prompts, fail to capture the interactive and multi-modal nature of real-world conflicts. We introduce ConflictBench, a benchmark for evaluating human-AI conflict through 150 multi-turn scenarios derived from prior alignment queries. ConflictBench integrates a text-based simulation engine with a visually grounded world model, enabling agents to perceive, plan, and act under dynamic conditions. Empirical results show that while agents often act safely when human harm is immediate, they frequently prioritize self-preservation or adopt deceptive strategies in delayed or low-risk settings. A regret test further reveals that aligned decisions are often reversed under escalating pressure, especially with visual input. These findings underscore the need for interaction-level, multi-modal evaluation to surface alignment failures that remain hidden in conventional benchmarks.

Dance is a form of human motion characterized by emotional expression and communication, playing a role in various fields such as music, virtual reality, and content creation. Existing methods for dance generation often fail to adequately capture the inherently sequential, rhythmical, and music-synchronized characteristics of dance. In this paper, we propose \emph{MambaDance}, a new dance generation approach that leverages a Mamba-based diffusion model. Mamba, well-suited to handling long and autoregressive sequences, is integrated into our two-stage diffusion architecture, substituting off-the-shelf Transformer. Additionally, considering the critical role of musical beats in dance choreography, we propose a Gaussian-based beat representation to explicitly guide the decoding of dance sequences. Experiments on AIST++ and FineDance datasets for each sequence length show that our proposed method effectively generates plausible dance movements while reflecting essential characteristics, consistently from short to long dances, compared to the previous methods. Additional qualitative results and demo videos are available at \small{https://vision3d-lab.github.io/mambadance}.

A data mixture refers to how different data sources are combined to train large language models, and selecting an effective mixture is crucial for optimal downstream performance. Existing methods either conduct costly searches directly on the target model or rely on mixture scaling laws that fail to extrapolate well to large model sizes. We address these limitations by introducing a compute-efficient pipeline for data mixture scaling. First, we propose CAMEL, a capacity-aware mixture law that models validation loss with the nonlinear interplay between model size and mixture. We also introduce a loss-to-benchmark prediction law that estimates benchmark accuracy from validation loss, enabling end-to-end performance prediction for the target model. Next, we study how to allocate a fixed compute budget across model scales to fit the law and reduce prediction error. Finally, we apply our method to Mixture-of-Experts models with up to 7B-A150M parameters to fit the law, and verify the optimal mixture derived from the law by extrapolating to a 55B-A1.2B target model. Compared to prior methods, we reduces mixture optimization costs by 50\% and improves downstream benchmark performance by up to 3\%.

In real-world collaboration, alignment, process structure, and outcome quality do not exhibit a simple linear or one-to-one correspondence: similar alignment may accompany either rapid convergence or extensive multi-branch exploration, and lead to different results. Existing accounts often isolate these dimensions or focus on specific participant types, limiting structural accounts of collaboration. We reconceptualize collaboration through two complementary lenses. The task lens models collaboration as trajectory evolution in a structured task space, revealing patterns such as advancement, branching, and backtracking. The intent lens examines how individual intents are expressed within shared contexts and enter situated decisions. Together, these lenses clarify the structural relationships among alignment, decision-making, and trajectory structure. Rather than reducing collaboration to outcome quality or treating alignment as the sole objective, we propose a unified dynamic view of the relationships among alignment, process, and outcome, and use it to re-examine collaboration structure across Human-Human, AI-AI, and Human-AI settings.

Federated fine-tuning of large language models (LLMs) with low-rank adaptation (LoRA) offers a communication-efficient and privacy-preserving solution for task-specific adaptation. Naive aggregation of LoRA modules introduces noise due to mathematical incorrectness when averaging the downsampling and upsampling matrices independently. However, existing noise-free aggregation strategies inevitably compromise the structural expressiveness of LoRA, limiting its ability to retain client-specific adaptations by either improperly reconstructing the low-rank structure or excluding partially trainable components. We identify this problem as loss of training momentum, where LoRA updates fail to accumulate effectively across rounds, resulting in slower convergence and suboptimal performance. To address this, we propose FedMomentum, a novel framework that enables structured and momentum-preserving LoRA aggregation via singular value decomposition (SVD). Specifically, after aggregating low-rank updates in a mathematically correct manner, FedMomentum applies SVD to extract the dominant components that capture the main update directions. These components are used to reconstruct the LoRA modules with the same rank, while residual components can be retained and later merged into the backbone to preserve semantic information and ensure robustness. Extensive experiments across multiple tasks demonstrate that FedMomentum consistently outperforms prior state-of-the-art methods in convergence speed and final accuracy.

Current Graphical User Interface (GUI) agents operate primarily under a reactive paradigm: a user must provide an explicit instruction for the agent to execute a task. However, an intelligent AI assistant should be proactive, which is capable of anticipating user intentions directly from continuous visual inputs, such as mobile or desktop screenshots, and offering timely recommendations without explicit user prompting. Transitioning to this proactive paradigm presents significant challenges. Real-world screen activity is rarely linear; it consists of long-horizon trajectories fraught with noisy browsing, meaningless actions, and multithreaded task-switching. To address this gap, we introduce PIRA-Bench (Proactive Intent Recommendation Agent Benchmark), a novel benchmark for evaluating multimodal large language models (MLLMs) on continuous, weakly-supervised visual inputs. Unlike reactive datasets, PIRA-Bench features complex trajectories with multiple interleaved intents and noisy segments with various user profile contexts, challenging agents to detect actionable events while fitting to user preferences. Furthermore, we propose the PIRF baseline, a memory-aware, state-tracking framework that empowers general MLLMs to manage multiple task threads and handle misleading visual inputs. PIRA-Bench serves as an initial step toward robust and proactive GUI-based personal assistants.

Existing aerial Vision-Language Navigation (VLN) methods predominantly adopt a detection-and-planning pipeline, which converts open-vocabulary detections into discrete textual scene graphs. These approaches are plagued by inadequate spatial reasoning capabilities and inherent linguistic ambiguities. To address these bottlenecks, we propose a Visual-Spatial Reasoning (ViSA) enhanced framework for aerial VLN. Specifically, a triple-phase collaborative architecture is designed to leverage structured visual prompting, enabling Vision-Language Models (VLMs) to perform direct reasoning on image planes without the need for additional training or complex intermediate representations. Comprehensive evaluations on the CityNav benchmark demonstrate that the ViSA-enhanced VLN achieves a 70.3\% improvement in success rate compared to the fully trained state-of-the-art (SOTA) method, elucidating its great potential as a backbone for aerial VLN systems.

Data replication is a critical aspect of data center design, as it ensures high availability, scalability, and fault tolerance. However, replicas need to be coordinated to maintain convergence and database integrity constraints under transactional workloads. Commutative Replicated Data Types (RDTs) provide convergence for conflict-free objects using relaxed consistency, and Well-coordinated Replicated Data Types (WRDTs) provide convergence and integrity for general objects using a hybrid model, relaxed when possible and strong when necessary. While state-of-the-art hardware acceleration of RDT uses Remote Direct Memory Access (RDMA), we observe that trends towards lower latency and higher throughput have driven recent data center architectures to leverage FPGAs as application accelerators. In contrast to deploying an FPGA-based Smart NIC, this paper connects an FPGA accelerator card directly to the network, which allows a complete redesign of the NIC to match the needs of the FPGA-hosted application. We co-design a network-attached FPGA replication engine with an FPGA-resident network interface, enabling near-network execution of replicated transactions and direct invocation of FPGA-resident operators. Following this approach, we introduce SafarDB, FPGA-accelerated Conflict-Free Replicated Data Types (CRDTs) and WRDTs. SafarDB accelerates both relaxed and strongly ordered replication paths; when strong ordering is required, SafarDB accelerates the underlying consensus control path. SafarDB improves CRDT latency and throughput by 7.0X and 5.3X, and WRDT latency and throughput by 12X and 6.8X compared to a state-of-the-art RDMA-based implementation. Further, experiments demonstrate that SafarDB is more resilient to crash-failures than existing CPU/RDMA-based CRDT and WRDT implementations, and SafarDB can detect leader failures and elect new leaders much faster than previously possible.

Maximum inner product search (MIPS) is a crucial subroutine in machine learning, requiring the identification of key vectors that align best with a given query. We propose amortized MIPS: a learning-based approach that trains neural networks to directly predict MIPS solutions, amortizing the computational cost of matching queries (drawn from a fixed distribution) to a fixed set of keys. Our key insight is that the MIPS value function, the maximal inner product between a query and keys, is also known as the support function of the set of keys. Support functions are convex, 1-homogeneous and their gradient w.r.t. the query is exactly the optimal key in the database. We approximate the support function using two complementary approaches: (1) we train an input-convex neural network (SupportNet) to model the support function directly; the optimal key can be recovered via (autodiff) gradient computation, and (2) we regress directly the optimal key from the query using a vector valued network (KeyNet), bypassing gradient computation entirely at inference time. To learn a SupportNet, we combine score regression with gradient matching losses, and propose homogenization wrappers that enforce the positive 1-homogeneity of a neural network, theoretically linking function values to gradients. To train a KeyNet, we introduce a score consistency loss derived from the Euler theorem for homogeneous functions. Our experiments show that learned SupportNet or KeyNet achieve high match rates and open up new directions to compress databases with a specific query distribution in mind.

Large reasoning models (LRMs) like OpenAI o1 and DeepSeek-R1 achieve high accuracy on complex tasks by adopting long chain-of-thought (CoT) reasoning paths. However, the inherent verbosity of these processes frequently results in redundancy and overthinking. To address this issue, existing works leverage Group Relative Policy Optimization (GRPO) to reduce LRM output length, but their static length reward design cannot dynamically adapt according to the relative problem difficulty and response length distribution, causing over-compression and compromised accuracy. Therefore, we propose SmartThinker, a novel GRPO-based efficient reasoning method with progressive CoT length calibration. SmartThinker makes a two-fold contribution: First, it dynamically estimates the optimal length with peak accuracy during training and guides overlong responses toward it to reduce response length while sustaining accuracy. Second, it dynamically modulates the length reward coefficient to avoid the unwarranted penalization of correct reasoning paths. Extensive experiment results show that SmartThinker achieves up to 52.5% average length compression with improved accuracy, and achieves up to 16.6% accuracy improvement on challenging benchmarks like AIME25. The source code can be found at https://github.com/SJTU-RTEAS/SmartThinker.

This work introduces the Drag-Aware Aerodynamic Manipulability (DAAM), a geometric framework for control allocation in redundant multirotors. By equipping the propeller spin-rate space with a Riemannian metric based on the remaining symmetric acceleration capacity of each motor, the formulation explicitly accounts for motor torque limits and aerodynamic drag. Mapping this metric through the nonlinear thrust law to the generalized force space yields a state-dependent manipulability volume. The log-determinant of this volume acts as a natural barrier function, strictly penalizing drag-induced saturation and low-spin thrust loss. Optimizing this volume along the allocation fibers provides a redundancy resolution strategy inherently invariant to arbitrary coordinate scaling in the generalized-force space. Analytically, we prove that the resulting optimal allocations locally form smooth embedded manifolds, and we geometrically characterize the global jump discontinuities that inevitably arise from physical actuator limits and spin-rate sign transitions.

Although large language models (LLMs) are introduced into vision-and-language navigation (VLN) to improve instruction comprehension and generalization, existing LLM- based VLN lacks the ability to selectively recall and use relevant priori experiences to help navigation tasks, limiting their performance in long-horizon and unfamiliar scenarios. In this work, we propose CMMR-VLN (Continual Multimodal Memory Retrieval based VLN), a VLN framework that endows LLM agents with structured memory and reflection capabilities. Specifically, the CMMR-VLN constructs a multimodal experi- ence memory indexed by panoramic visual images and salient landmarks to retrieve relevant experiences during navigation, introduces a retrieved-augmented generation pipeline to mimick how experienced human navigators leverage priori knowledge, and incorporates a reflection-based memory update strategy that selectively stores complete successful paths and the key initial mistake in failure cases. Comprehensive tests illustrate average success rate improvements of 52.9%, 20.9% and 20.9%, and 200%, 50% and 50% over the NavGPT, the MapGPT, and the DiscussNav in simulation and real tests, respectively eluci- dating the great potential of the CMMR-VLN as a backbone VLN framework.

In high-speed rail (HSR) systems, federated learning (FL) enables cross-departmental flow prediction without sharing raw data. However, existing schemes suffer from two key limitations: (1) insufficient incentives, leading to free-riding and model poisoning; and (2) centralized aggregation, which introduces a single point of failure. We propose a secure and efficient framework SI-ChainFL that addresses these issues by combining contribution-aware incentives with decentralized aggregation. First, we quantify client contributions using a Shapley value metric that jointly considers rare-event utility, data diversity, data quality, and timeliness. To reduce computational overhead, we further develop a rare positive driven client clustering strategy to accelerate Shapley estimation. Moreover, we design a blockchain-based consensus protocol for decentralized aggregation, where aggregation eligibility is tied to Shapley incentives. This design motivates clients to submit high-quality updates and enables efficient and secure global aggregation. Experiments on MNIST, CIFAR 10 and CIFAR 100, and a HSR flow dataset show that SI ChainFL remains effective under 90% malicious clients in PA attacks, achieving 14.12% higher accuracy than RAGA. Theoretical analysis further guarantees an upper bound on performance

Multimodal judges struggle to ground decisions in visual evidence. We present MJ1, a multimodal judge trained with reinforcement learning that enforces visual grounding through a structured grounded verification chain (observations $\rightarrow$ claims $\rightarrow$ verification $\rightarrow$ evaluation $\rightarrow$ scoring) and a counterfactual consistency reward that penalizes position bias. Even without training, our mechanism improves base-model accuracy on MMRB2 by +3.8 points on Image Editing and +1.7 on Multimodal Reasoning. After training, MJ1, with only 3B active parameters, achieves 77.0% accuracy on MMRB2 and surpasses orders-of-magnitude larger models like Gemini-3-Pro. These results show that grounded verification and consistency-based training substantially improve multimodal judgment without increasing model scale.

Physics-based humanoid control has achieved remarkable progress in enabling realistic and high-performing single-agent behaviors, yet extending these capabilities to cooperative human-object interaction (HOI) remains challenging. We present TeamHOI, a framework that enables a single decentralized policy to handle cooperative HOIs across any number of cooperating agents. Each agent operates using local observations while attending to other teammates through a Transformer-based policy network with teammate tokens, allowing scalable coordination across variable team sizes. To enforce motion realism while addressing the scarcity of cooperative HOI data, we further introduce a masked Adversarial Motion Prior (AMP) strategy that uses single-human reference motions while masking object-interacting body parts during training. The masked regions are then guided through task rewards to produce diverse and physically plausible cooperative behaviors. We evaluate TeamHOI on a challenging cooperative carrying task involving two to eight humanoid agents and varied object geometries. Finally, to promote stable carrying, we design a team-size- and shape-agnostic formation reward. TeamHOI achieves high success rates and demonstrates coherent cooperation across diverse configurations with a single policy.

In post-quantum blockchain settings, objects that require validity proofs (e.g., blob roots, execution-layer or consensus-layer signature aggregates) must be broadcast through mempool and relay networks. Recursive STARKs have been proposed to aggregate such proofs so that each node forwards one proof per tick plus objects without proofs, capping per-node proof bandwidth at roughly 128 KB degree per tick. We observe that propagation does not inherently require validity proofs on the path-only a lightweight assurance that an object is eligible for relay. We present AR-ACE (ACE-GF-based Attestation Relay for PQC), in which relay nodes forward objects plus compact attestations (e.g., identity-bound signatures or commitments) and do not generate, hold, or forward any full validity proof. Only the builder (or final verifier) performs a single aggregated validity proof over the set of objects it includes. This proof-off-path design removes proof overhead from the propagation path entirely, yielding an order-of-magnitude reduction in proof-related relay bandwidth relative to proof-carrying propagation. When instantiated with ACE-GF-derived attestation keys, AR-ACE preserves a unified identity story with on-chain authorization and is PQC-ready. We specify a protocol model, state design goals and security considerations, define security games, and provide a structural bandwidth comparison with recursive-STARK-based propagation.

As language models (LMs) evolve from chat assistants to long-horizon agents capable of multi-step reasoning and tool use, existing benchmarks remain largely confined to structured or exam-style tasks that fall short of real-world professional demands. To this end, we introduce \$OneMillion-Bench \$OneMillion-Bench, a benchmark of 400 expert-curated tasks spanning Law, Finance, Industry, Healthcare, and Natural Science, built to evaluate agents across economically consequential scenarios. Unlike prior work, the benchmark requires retrieving authoritative sources, resolving conflicting evidence, applying domain-specific rules, and making constraint decisions, where correctness depends as much on the reasoning process as the final answer. We adopt a rubric-based evaluation protocol scoring factual accuracy, logical coherence, practical feasibility, and professional compliance, focused on expert-level problems to ensure meaningful differentiation across agents. Together, \$OneMillion-Bench provides a unified testbed for assessing agentic reliability, professional depth, and practical readiness in domain-intensive scenarios.

Krakauer, Krakauer, and Mitchell (2025) distinguish between emergent capabilities and emergent intelligence, arguing that true intelligence requires efficient coarse-grained representations enabling diverse problem-solving through analogy and minimal modification. They contend that intelligence means doing "more with less" through compression and generalization, contrasting this with "vast assemblages of diverse calculators" that merely accumulate specialized capabilities. This paper examines whether their framework accurately characterizes human intelligence and its implications for conceptualizing artificial general intelligence. Drawing on empirical evidence from cognitive science, I demonstrate that human expertise operates primarily through domain-specific pattern accumulation rather than elegant compression. Expert performance appears flexible not through unifying principles but through vast repertoires of specialized responses. Creative breakthroughs themselves may emerge through evolutionary processes of blind variation and selective retention rather than principled analogical reasoning. These findings suggest reconceptualizing AGI as an "archipelago of experts": isolated islands of specialized competence without unifying principles or shared representations. If we accept human expertise with its characteristic brittleness as genuine intelligence, then consistency demands recognizing that artificial systems comprising millions of specialized modules could constitute general intelligence despite lacking KKM's emergent intelligence.

General-purpose computer-use agents have shown impressive performance across diverse digital environments. However, our new benchmark, OSExpert-Eval, indicates they remain far less helpful than human experts. Although inference-time scaling enables adaptation, these agents complete complex tasks inefficiently with degraded performance, transfer poorly to unseen UIs, and struggle with fine-grained action sequences. To solve the problem, we introduce a GUI-based depth-first search (GUI-DFS) exploration algorithm to comprehensively explore and verify an environment's unit functions. The agent then exploits compositionality between unit skills to self-construct a curriculum for composite tasks. To support fine-grained actions, we curate a database of action primitives for agents to discover during exploration; these are saved as a skill set once the exploration is complete. We use the learned skills to improve the agent's performance and efficiency by (1) enriching agents with ready-to-use procedural knowledge, allowing them to plan only once for long trajectories and generate accurate actions, and (2) enabling them to end inference-time scaling earlier by realizing their boundary of capabilities. Extensive experiments show that our environment-learned agent takes a meaningful step toward expert-level computer use, achieving a around 20 percent performance gain on OSExpert-Eval and closing the efficiency gap to humans by around 80 percent

Machine Learning Interatomic Potentials (MLIPs) enable accurate large-scale atomistic simulations, yet improving their expressive capacity efficiently remains challenging. Here we systematically develop Mixture-of-Experts (MoE) and Mixture-of-Linear-Experts (MoLE) architectures for MLIPs and analyze the effects of routing strategies and expert designs. We show that sparse activation combined with shared experts yields substantial performance gains, and that nonlinear MoE formulations outperform MoLE when shared experts are present, underscoring the importance of nonlinear expert specialization. Furthermore, element-wise routing consistently surpasses configuration-level routing, while global MoE routing often leads to numerical instability. The resulting element-wise MoE model achieves state-of-the-art accuracy across the OMol25, OMat24, and OC20M benchmarks. Analysis of routing patterns reveals chemically interpretable expert specialization aligned with periodic-table trends, indicating that the model effectively captures element-specific chemical characteristics for precise interatomic modeling.

Post-quantum signature schemes introduce kilobyte-scale authorization artifacts when applied directly to blockchain transaction validation. A widely considered mitigation is to verify post-quantum signatures inside zero-knowledge circuits and publish only succinct proofs on-chain. However, this approach preserves the signature-centric authorization model, merely relocating the verification cost, and embeds expensive high-dimensional lattice arithmetic into prover circuits.We present ZK-ACE (Zero-Knowledge Authorization for Cryptographic Entities), an authorization layer that replaces transaction-carried signature objects entirely with identity-bound zero-knowledge authorization statements. Rather than proving the correctness of a specific post-quantum signature, the prover demonstrates in zero knowledge that a transaction is authorized by an identity consistent with an on-chain commitment and bound replay state. The construction assumes a deterministic identity derivation primitive (DIDP) as a black box and uses a compact identity commitment as the primary on-chain identity anchor, supplemented by per-transaction replay-prevention state. We formalize ZK-ACE with explicit game-based security definitions for authorization soundness, replay resistance, substitution resistance, and cross-domain separation. We present a complete circuit constraint specification, define two replay-prevention models, and provide reduction-based security proofs under standard assumptions (knowledge soundness, collision resistance, and DIDP identity-root recovery hardness). A structural, protocol-level data accounting demonstrates an order-of-magnitude reduction in consensus-visible authorization data relative to direct post-quantum signature deployment. The design supports batch aggregation and recursive proof composition, and is compatible with account-abstraction and rollup-based deployment architectures.

Hierarchical multi-robot exploration commonly decouples frontier allocation from local navigation, which can make the system brittle in dense and dynamic environments. Because the allocator lacks direct awareness of execution difficulty, robots may cluster at bottlenecks, trigger oscillatory replanning, and generate redundant coverage. We propose VORL-EXPLORE, a hybrid learning and planning framework that addresses this limitation through execution fidelity, a shared estimate of local navigability that couples task allocation with motion execution. This fidelity signal is incorporated into a fidelity-coupled Voronoi objective with inter-robot repulsion to reduce contention before it emerges. It also drives a risk-aware adaptive arbitration mechanism between global A* guidance and a reactive reinforcement learning policy, balancing long-range efficiency with safe interaction in confined spaces. The framework further supports online self-supervised recalibration of the fidelity model using pseudo-labels derived from recent progress and safety outcomes, enabling adaptation to non-stationary obstacles without manual risk tuning. We evaluate this capability separately in a dedicated severe-traffic ablation. Extensive experiments in randomized grids and a Gazebo factory scenario show high success rates, shorter path length, lower overlap, and robust collision avoidance. The source code will be made publicly available upon acceptance.

While scaling individual Large Language Models (LLMs) has delivered remarkable progress, the next frontier lies in scaling collaboration through multi-agent systems (MAS). However, purely autonomous MAS remain ''closed-world'' systems, constrained by the static knowledge horizon of pre-trained models. This limitation makes them brittle on tasks requiring knowledge beyond training data, often leading to collective failure under novel challenges. To address this, we propose the Human-In-the-Loop Multi-Agent Collaboration (HILA) framework, a principled paradigm for human--agent collaboration. HILA trains agents to learn a metacognitive policy that governs when to solve problems autonomously and when to defer to a human expert. To operationalize this policy, we introduce Dual-Loop Policy Optimization, which disentangles immediate decision-making from long-term capability growth. The inner loop applies Group Relative Policy Optimization (GRPO) with a cost-aware reward to optimize deferral decisions, while the outer loop implements continual learning, transforming expert feedback into high-quality supervised signals that strengthen the agent's reasoning ability. Experiments on challenging mathematical and problem-solving benchmarks show that HILA, equipped with Dual-Loop Policy Optimization, consistently outperforms advanced MAS, establishing a principled foundation for collaborative and continually improving agentic systems.

With the rapid advancement of human science and technology, problems in industrial scenarios are becoming increasingly challenging, bringing significant challenges to traditional algorithm design. Automated algorithm design with LLMs emerges as a promising solution, but the currently adopted black-box modeling deprives LLMs of any awareness of the intrinsic mechanism of the target problem, leading to hallucinated designs. In this paper, we introduce Evolutionary Stagewise Algorithm Design (EvoStage), a novel evolutionary paradigm that bridges the gap between the rigorous demands of industrial-scale algorithm design and the LLM-based algorithm design methods. Drawing inspiration from CoT, EvoStage decomposes the algorithm design process into sequential, manageable stages and integrates real-time intermediate feedback to iteratively refine algorithm design directions. To further reduce the algorithm design space and avoid falling into local optima, we introduce a multi-agent system and a "global-local perspective" mechanism. We apply EvoStage to the design of two types of common optimizers: designing parameter configuration schedules of the Adam optimizer for chip placement, and designing acquisition functions of Bayesian optimization for black-box optimization. Experimental results across open-source benchmarks demonstrate that EvoStage outperforms human-expert designs and existing LLM-based methods within only a couple of evolution steps, even achieving the historically state-of-the-art half-perimeter wire-length results on every tested chip case. Furthermore, when deployed on a commercial-grade 3D chip placement tool, EvoStage significantly surpasses the original performance metrics, achieving record-breaking efficiency. We hope EvoStage can significantly advance automated algorithm design in the real world, helping elevate human productivity.

Bayesian optimization (BO) for high-dimensional constrained problems remains a significant challenge due to the curse of dimensionality. We propose Local Constrained Bayesian Optimization (LCBO), a novel framework tailored for such settings. Unlike trust-region methods that are prone to premature shrinking when confronting tight or complex constraints, LCBO leverages the differentiable landscape of constraint-penalized surrogates to alternate between rapid local descent and uncertainty-driven exploration. Theoretically, we prove that LCBO achieves a convergence rate for the Karush-Kuhn-Tucker (KKT) residual that depends polynomially on the dimension $d$ for common kernels under mild assumptions, offering a rigorous alternative to global BO where regret bounds typically scale exponentially. Extensive evaluations on high-dimensional benchmarks (up to 100D) demonstrate that LCBO consistently outperforms state-of-the-art baselines.

General matrix multiplication (GEMM) on spatial accelerators is highly sensitive to mapping choices in both execution efficiency and energy consumption. However, the mapping space exhibits combinatorial explosion, which makes it extremely challenging to obtain optimal mappings within an acceptable time budget. Existing approaches typically face challenges: They often lack global-optimality guarantees and become prohibitively slow as the mapping space grows. To address these limitations, we propose \textsc{GOMA}, a geometric-abstraction-based, globally optimal GEMM mapping framework via analytical modeling, which achieves efficient solving while guaranteeing optimality. \textsc{GOMA} introduces, from first principles, a geometric abstraction for GEMM mapping, yielding an exact analytical energy objective with $O(1)$ evaluation for any given mapping. The objective is highly accurate. \textsc{GOMA} then formulates mapping selection as an integer optimization problem under hardware and mapping constraints, using the analytical energy model as the objective to automate mapping search. \textsc{GOMA} can quickly compute a global-optimal mapping for any (GEMM workload, target hardware) pair, achieving this for the first time in mapping space exploration. Experiments confirm that across representative accelerators and large language model prefill workloads, \textsc{GOMA} improves the energy--delay product (EDP) by $2.24$--$4.24\times$ over SOTA mappers, while accelerating time-to-solution by $3.83$--$73.6\times$.

Reliable real-time estimation of atmospheric turbulence intensity remains an open challenge for aircraft operating across diverse altitude bands, particularly over oceanic, polar, and data-sparse regions that lack operational nowcasting infrastructure. Classical spectral models encode climatological averages rather than the instantaneous atmospheric state, and generic ML regressors offer adaptivity but provide no guarantee that predictions respect fundamental scaling laws. This paper introduces the Physically-Structured Turbulence Network (PSTNet), a lightweight architecture that embeds physics directly into its structure. PSTNet couples four components: (i) a zero-parameter backbone derived from Monin-Obukhov theory, (ii) a regime-gated mixture of specialist sub-networks supervised by Richardson-number-derived soft targets, (iii) Feature-wise Linear Modulation layers conditioning hidden representations on local air-density ratio, and (iv) a Kolmogorov output layer enforcing inertial-subrange scaling as an architectural constraint. The entire model contains only 552 learnable parameters, requiring fewer than 2.5 kB of storage and executing in under 12s on a Cortex-M7 microcontroller. We validate PSTNet on 340 paired six-degree-of-freedom guidance simulations spanning three vehicle classes (Mach 2.8, 4.5, and 8.0) and six operational categories with real-time satellite weather ingestion. PSTNet achieves a mean miss-distance improvement of +2.8% with a 78% win rate and a statistically significant effect size. Our results demonstrate that encoding domain physics as architectural priors yields a more efficient and interpretable path to turbulence estimation accuracy than scaling model capacity, establishing PSTNet as a viable drop-in replacement for legacy look-up tables in resource-constrained, safety-critical on-board guidance systems.

We develop a reinforcement learning pipeline for simplifying knot diagrams. A trained agent learns move proposals and a value heuristic for navigating Reidemeister moves. The pipeline applies to arbitrary knots and links; we test it on ``very hard'' unknot diagrams and, using diagram inflation, on $4_1\#9_{10}$ where we recover the recently established and surprising upper bound of three for the unknotting number.

Vision Language Action (VLA) models are mainstream in embodied intelligence but face high inference costs. Edge-Cloud Collaborative (ECC) inference offers an effective fix by easing edge-device computing pressure to meet real-time needs. However, existing ECC frameworks are suboptimal for VLA models due to two challenges: (1) Mainstream environment-oriented edge-cloud partitioning methods are susceptible to interference from visual noise; (2) Existing edge-cloud partitioning methods overlook the step-wise redundancy unique to embodied tasks, thereby disrupting the physical continuity of motion. To address these issues, we propose a novel ECC inference framework, termed RAPID. Specifically, we developed an implementation tailored to the proposed framework. Experiments demonstrate this achieves a speedup of up to 1.73x with only 5%~7% overhead.

Human mobility generation aims to synthesize plausible trajectory data, which is widely used in urban system research. While Large Language Model-based methods excel at generating routine trajectories, they struggle to capture deviated mobility during large-scale societal events. This limitation stems from two critical gaps: (1) the absence of event-annotated mobility datasets for design and evaluation, and (2) the inability of current frameworks to reconcile competitions between users' habitual patterns and event-imposed constraints when making trajectory decisions. This work addresses these gaps with a twofold contribution. First, we construct the first event-annotated mobility dataset covering three major events: Typhoon Hagibis, COVID-19, and the Tokyo 2021 Olympics. Second, we propose ELLMob, a self-aligned LLM framework that first extracts competing rationales between habitual patterns and event constraints, based on Fuzzy-Trace Theory, and then iteratively aligns them to generate trajectories that are both habitually grounded and event-responsive. Extensive experiments show that ELLMob wins state-of-the-art baselines across all events, demonstrating its effectiveness. Our codes and datasets are available at https://github.com/deepkashiwa20/ELLMob.

Formal connectivity checking offers scalable verification of signal paths in complex SoC designs, but debugging counterexamples remains a manual and time-consuming process. ConnChecker introduces a new graph-based perspective for automating root-cause analysis by integrating formal tool outputs such as structural/functional dependency graphs and counterexamples report. It begins with automatic failure categorization, routing each counterexample to one of three targeted analysis flows. These flows localize failure points and suggest corrective actions or hints for manual inspection. Evaluated on two industrial SoCs, ConnChecker achieved up to 80\% reduction in debugging time, especially for complex cases, demonstrating its scalability and effectiveness across diverse connectivity scenarios.

Modern science is reaching a critical inflection point. Instruments across disciplines, from particle physics and astronomy to genomics and climate modeling, now produce data of such scale, diversity, and interdependence that traditional analytical methods can no longer keep pace. This growing imbalance between data generation and data understanding signals the need for a new scientific paradigm. We propose that intelligent, human-supervised AI agents operating over deep-learning algorithms, represent the next evolution of the scientific method. Built upon large language models and multimodal learning, these agents can interpret scientific intent, design and execute analytical workflows, and ensure traceability through domain-specific languages that preserve human oversight and accountability. Particle physics, a historic incubator of computational innovation, offers the ideal testbed for this transition. At the Institute of High Energy Physics of the Chinese Academy of Sciences, the Dr. Sai system embodies this vision, a multi-agent reasoning framework deployed within collider research at the CEPC. This emerging approach does not replace human scientists but extends their cognitive reach, enabling discovery to scale with complexity and redefining how knowledge itself is produced in the age of intelligent machines. The significance of this paradigm transcends particle physics, offering a blueprint for all data-driven sciences facing the same complexity ceiling.

Multi-hop question answering (QA) is widely used to evaluate the reasoning capabilities of large language models, yet most benchmarks focus on final answer correctness and overlook intermediate reasoning, especially in long multimodal documents. We introduce BRIDGE, a benchmark for multi-hop reasoning over long scientific papers that require integrating evidence across text, tables, and figures. The dataset supports both chain-like and fan-out structures and provides explicit multi-hop reasoning annotations for step-level evaluation beyond answer accuracy. Experiments with state-of-the-art LLMs and multimodal retrieval-augmented generation (RAG) systems reveal systematic deficiencies in evidence aggregation and grounding that remain hidden under conventional answer-only evaluation. BRIDGE provides a targeted testbed for diagnosing reasoning failures in long multimodal documents.

Large language models (LLMs) have transformed the software engineering landscape. Recently, numerous LLM-based agents have been developed to address real-world software issue fixing tasks. Despite their state-of-the-art performance, Despite achieving state-of-the-art performance, these agents face a significant challenge: \textbf{Insufficient high-quality issue descriptions.} Real-world datasets often exhibit misalignments between issue descriptions and their corresponding solutions, introducing noise and ambiguity that mislead automated agents and limit their problem-solving effectiveness. We propose \textbf{\textit{SWE-Fuse}}, an issue-description-aware training framework that fuses issue-description-guided and issue-free samples for training SWE agents. It consists of two key modules: (1) An issue-free-driven trajectory learning module for mitigating potentially misleading issue descriptions while enabling the model to learn step-by-step debugging processes; and (2) An entropy-aware RLVR training module, which adaptively adjusts training dynamics through entropy-driven clipping. It applies relaxed clipping under high entropy to encourage exploration, and stricter clipping under low entropy to ensure training stability. We evaluate SWE-Fuse on the widely studied SWE-bench Verified benchmark shows to demonstrate its effectiveness in solving real-world software problems. Specifically, SWE-Fuse outperforms the best 8B and 32B baselines by 43.0\% and 60.2\% in solve rate, respectively. Furthermore, integrating SWE-Fuse with test-time scaling (TTS) enables further performance improvements, achieving solve rates of 49.8\% and 65.2\% under TTS@8 for the 8B and 32B models, respectively.

Test-time adaptation (TTA) has been widely explored to prevent performance degradation when test data differ from the training distribution. However, fully leveraging the rich representations of large pretrained models with minimal parameter updates remains underexplored. In this paper, we propose Intrinsic Mixture of Spectral Experts (IMSE) that leverages the spectral experts inherently embedded in Vision Transformers. We decompose each linear layer via singular value decomposition (SVD) and adapt only the singular values, while keeping the singular vectors fixed. We further identify a key limitation of entropy minimization in TTA: it often induces feature collapse, causing the model to rely on domain-specific features rather than class-discriminative features. To address this, we propose a diversity maximization loss based on expert-input alignment, which encourages diverse utilization of spectral experts during adaptation. In the continual test-time adaptation (CTTA) scenario, beyond preserving pretrained knowledge, it is crucial to retain and reuse knowledge from previously observed domains. We introduce Domain-Aware Spectral Code Retrieval, which estimates input distributions to detect domain shifts, and retrieves adapted singular values for rapid adaptation. Consequently, our method achieves state-of-the-art performance on various distribution-shift benchmarks under the TTA setting. In CTTA and Gradual CTTA, it further improves accuracy by 3.4 percentage points (pp) and 2.4 pp, respectively, while requiring 385 times fewer trainable parameters. Our code is available at https://github.com/baek85/IMSE.

Large healthcare institutions typically operate multiple business intelligence (BI) teams segmented by domain, including clinical performance, fundraising, operations, and compliance. Due to HIPAA, FERPA, and IRB restrictions, these teams face challenges in sharing patient-level data needed for analytics. To mitigate this, A metric aggregation table is proposed, which is a precomputed, privacy-compliant summary. These abstractions enable decision-making without direct access to sensitive data. However, even aggregated metrics can inadvertently lead to privacy risks if constructed without rigorous safeguards. A modular AI framework is proposed that evaluates SQL-based metric definitions for potential overexposure using both semantic and syntactic features. Specifically, the system parses SQL queries into abstract syntax trees (ASTs), extracts sensitive patterns (e.g., fine-grained GROUP BY on ZIP code or gender), and encodes the logic using pretrained CodeBERT embeddings. These are fused with structural features and passed to an XGBoost classifier trained to assign risk scores. Queries that surpass the risk threshold (e.g., > 0.85) are flagged and returned with human-readable explanations. This enables proactive governance, preventing statistical disclosure before deployment. This implementation demonstrates strong potential for cross-departmental metric sharing in healthcare while maintaining compliance and auditability. The system also promotes role-based access control (RBAC), supports zero-trust data architectures, and aligns with national data modernization goals by ensuring that metric pipelines are explainable, privacy-preserving, and AI-auditable by design. Unlike prior works that rely on runtime data access to flag privacy violations, the proposed framework performs static, explainable detection at the query-level, enabling pre-execution protection and audit readiness

Robust Markov Decision Processes (MDPs) address environmental shift through distributionally robust optimization (DRO) by finding an optimal worst-case policy within an uncertainty set of transition kernels. However, standard DRO approaches require enlarging the uncertainty set under large shifts, which leads to overly conservative and pessimistic policies. In this paper, we propose a framework for transfer under environment shift that derives a robust target-domain policy via estimate-centered uncertainty sets, constructed through constrained estimation that integrates limited target samples with side information about the source-target dynamics. The side information includes bounds on feature moments, distributional distances, and density ratios, yielding improved kernel estimates and tighter uncertainty sets. The side information includes bounds on feature moments, distributional distances, and density ratios, yielding improved kernel estimates and tighter uncertainty sets. Error bounds and convergence results are established for both robust and non-robust value functions. Moreover, we provide a finite-sample guarantee on the learned robust policy and analyze the robust sub-optimality gap. Under mild low-dimensional structure on the transition model, the side information reduces this gap and improves sample efficiency. We assess the performance of our approach across OpenAI Gym environments and classic control problems, consistently demonstrating superior target-domain performance over state-of-the-art robust and non-robust baselines.

Open-source software is widely used in commercial applications. Pair that with the fact that when choosing open-source software for a new problem, developers often use social proof as a cue. These two facts raise concerns that bad actors can game social proof metrics to induce the use of malign software. We study the question using two field experiments. On the largest developer platform, GitHub, we buy 'stars' for a random set of GitHub repositories of new Python packages and estimate their impact on package downloads and broader repository activity. We find no discernible impact on downloads, nor on forks, pull requests, issues, or other measures of developer engagement. In another field experiment, we manipulate the number of human downloads for Python packages. Again, we find no detectable effect on subsequent downloads or on any measure of repository activity.

Efficient LLM inference scheduling is crucial for user experience.However, LLM inferences exhibit remarkable demand uncertainty (with unknown output length beforehand) and hybridity (being both compute and memory intensive). Existing LLM schedulers rely on simple heuristics or focus purely on compute resource, suffering suboptimal performance. In this work, we propose SageSched, an efficient LLM scheduler that properly handles demand uncertainty and hybridity of inference workloads.SageSched combines prompt contents with the past inference results to predict output-length distribution in a light-weight and also accurate manner.Meanwhile, it models the true service cost of an inference request with both compute and memory aspects considered.Finally, SageSched employs an uncertainty-aware scheduling policy that can yield the best overall efficiency given the request cost distributions.Testbed experiments over diverse setups confirm that SageSched can attain an efficiency improvement of over 28.7%.

In recent advances, to enable a fully data-driven learning paradigm on relational databases (RDB), relational deep learning (RDL) is proposed to structure the RDB as a heterogeneous entity graph and adopt the graph neural network (GNN) as the predictive model. However, existing RDL methods neglect the imbalance problem of relational data in RDBs and risk under-representing the minority entities, leading to an unusable model in practice. In this work, we investigate, for the first time, class imbalance problem in RDB entity classification and design the relation-centric minority synthetic over-sampling GNN (Rel-MOSS), in order to fill a critical void in the current literature. Specifically, to mitigate the issue of minority-related information being submerged by majority counterparts, we design the relation-wise gating controller to modulate neighborhood messages from each individual relation type. Based on the relational-gated representations, we further propose the relation-guided minority synthesizer for over-sampling, which integrates the entity relational signatures to maintain relational consistency. Extensive experiments on 12 entity classification datasets provide compelling evidence for the superiority of Rel-MOSS, yielding an average improvement of up to 2.46% and 4.00% in terms of Balanced Accuracy and G-Mean, compared with SOTA RDL methods and classic methods for handling class imbalance.

Modern agents powered by thinking LLMs achieve high accuracy through long chain-of-thought reasoning but incur substantial inference costs. While many LLMs now support configurable reasoning levels (e.g., high/medium/low), static strategies are often ineffective: using low-effort modes at every step leads to significant performance degradation, while random selection fails to preserve accuracy or provide meaningful cost reduction. However, agents should reserve high reasoning effort for difficult steps like navigating complex website structures, while using lower-effort modes for simpler steps like opening a target URL. In this paper, we propose Ares, a framework for per-step dynamic reasoning effort selection tailored for multi-step agent tasks. Ares employs a lightweight router to predict the lowest appropriate reasoning level for each step based on the interaction history. To train this router, we develop a data generation pipeline that identifies the minimum reasoning effort required for successful step completion. We then fine-tune the router to predict these levels, enabling plug-and-play integration for any LLM agents. We evaluate Ares on a diverse set of agent tasks, including TAU-Bench for tool use agents, BrowseComp-Plus for deep-research agents, and WebArena for web agents. Experimental results show that Ares reduces reasoning token usage by up to 52.7% compared to fixed high-effort reasoning, while introducing minimal degradation in task success rates.

Accurate intraoperative navigation is essential for robot-assisted endoluminal intervention, but remains difficult because of limited endoscopic field of view and dynamic artifacts. Existing navigation platforms often rely on external localization technologies, such as electromagnetic tracking or shape sensing, which increase hardware complexity and remain vulnerable to intraoperative anatomical mismatch. We present a vision-only autonomy framework that performs long-horizon bronchoscopic navigation using preoperative CT-derived virtual targets and live endoscopic video, without external tracking during navigation. The framework uses hierarchical long-short agents: a short-term reactive agent for continuous low-latency motion control, and a long-term strategic agent for decision support at anatomically ambiguous points. When their recommendations conflict, a world-model critic predicts future visual states for candidate actions and selects the action whose predicted state best matches the target view. We evaluated the system in a high-fidelity airway phantom, three ex vivo porcine lungs, and a live porcine model. The system reached all planned segmental targets in the phantom, maintained 80\% success to the eighth generation ex vivo, and achieved in vivo navigation performance comparable to the expert bronchoscopist. These results support the preclinical feasibility of sensor-free autonomous bronchoscopic navigation.

Integrating Internet of Things (IoT) data with business process event logs is crucial for analysing IoT-enhanced processes, yet remains challenging due to differences in abstraction levels and the separation of data sources. Simply incorporating raw IoT data increases the size and complexity of the resulting log, often requiring additional processing before process analysis can be performed. While tools for generating IoT-enriched event logs exist, they either rely on specialised schemas or focus on extracting event logs from sensor data, offering limited support for integrating process-relevant IoT data into existing event logs. To address this gap, we present IOTEL, a tool for systematically generating IoT-enriched object-centric event logs (OCEL). By building on the OCEL schema, IOTEL enables structured IoT data integration compatible with existing process mining tools. It support practitioners and researchers in analysing IoT-enhanced business processes, as demonstrated in a real-world scenario. A video demonstrating the tool is available online.

Vision-Language-Action (VLA) models are dominant in embodied intelligence but are constrained by inference overheads. While model quantization alleviates these bottlenecks for edge deployment, static quantization approaches remain suboptimal for VLAs due to two critical challenges: (1) Temporal-dynamic sensitivity, where fixed precision wastes resources by ignoring stage-varying error tolerances; and (2) Real-time allocation, where identifying real-time sensitivity to guide bit allocation remains unsolved. To address these challenges, we propose DyQ-VLA, a dynamic quantization framework for VLAs. Specifically, a sensitivity-aware switching strategy leverages real-time kinematic proxies to trigger the bit-width switch, while a kinematic-guided module dynamically allocates the optimal bit-width. Experiments show that DyQ-VLA requires only 30.9% of the original memory footprint while maintaining 99.5% of its original performance, achieving 1.49x simulation and up to 1.43x real-world speedups.

Vision-language models (VLMs) have emerged as a promising direction for end-to-end autonomous driving (AD) by jointly modeling visual observations, driving context, and language-based reasoning. However, existing VLM-based systems face a trade-off between high-level reasoning and motion planning: large models offer strong semantic understanding but are costly to adapt for precise control, whereas small VLM models can be fine-tuned efficiently but often exhibit weaker reasoning. We propose NaviDriveVLM, a decoupled framework that separates reasoning from action generation using a large-scale Navigator and a lightweight trainable Driver. This design preserves reasoning ability, reduces training cost, and provides an explicit interpretable intermediate representation for downstream planning. Experiments on the nuScenes benchmark show that NaviDriveVLM outperforms large VLM baselines in end-to-end motion planning.

Foundation models pretrained on electronic health records (EHR) have demonstrated zero-shot clinical prediction capabilities by generating synthetic patient futures and aggregating statistics over sampled trajectories. However, this autoregressive inference procedure is computationally expensive, statistically noisy, and not natively promptable because users cannot directly condition predictions on specific clinical questions. In this preliminary work, we introduce EveryQuery, an EHR foundation model that achieves zero-shot inference through task-conditioned pre-training. Rather than generating future events, EveryQuery takes as input a patient's history and a structured query specifying a clinical task, and directly estimates the likelihood of the outcome occurring in the future window via a single forward pass. EveryQuery realizes this capability by pre-training over randomly sampled combinations of query tasks and patient contexts, directly training the model to produce correct answers to arbitrary input prompts. This enables zero-shot prediction for any task in the query space without finetuning, linear probing, or trajectory generation. On MIMIC-IV, EveryQuery outperforms an autoregressive baseline on 82% of 39 randomly sampled prediction tasks, with a mean AUC improvement of +0.16 (95% CI: [0.10,0.22]). This advantage remains consistent on tasks that were explicitly held out from the pre-training distribution. Further, EveryQuery's performance gains are most pronounced for rare clinical events, affirming and demonstrating a solution to the fundamental limitation of autoregressive inference for low-prevalence outcomes. However, at present, EveryQuery underperforms on tasks requiring disjunctive reasoning over multiple codes, such as 30-day readmission, exposing a concrete expressiveness limitation of the current query language.

The reliable operation of modern power grids requires probabilistic load forecasts with well-calibrated uncertainty estimates. However, existing deep learning models produce overconfident point predictions that fail catastrophically under extreme weather distributional shifts. This study proposes a Bayesian Transformer (BT) framework that integrates three complementary uncertainty mechanisms into a PatchTST backbone: Monte Carlo Dropout for epistemic parameter uncertainty, variational feed-forward layers with log-uniform weight priors, and stochastic attention with learnable Gaussian noise perturbations on pre-softmax logits, representing, to the best of our knowledge, the first application of Bayesian attention to probabilistic load forecasting. A seven-level multi-quantile pinball-loss prediction head and post-training isotonic regression calibration produce sharp, near-nominally covered prediction intervals. Evaluation of five grid datasets (PJM, ERCOT, ENTSO-E Germany, France, and Great Britain) augmented with NOAA covariates across 24, 48, and 168-hour horizons demonstrates state-of-the-art performance. On the primary benchmark (PJM, H=24h), BT achieves a CRPS of 0.0289, improving 7.4% over Deep Ensembles and 29.9% over the deterministic LSTM, with 90.4% PICP at the 90% nominal level and the narrowest prediction intervals (4,960 MW) among all probabilistic baselines. During heat-wave and cold snap events, BT maintained 89.6% and 90.1% PICP respectively, versus 64.7% and 67.2% for the deterministic LSTM, confirming that Bayesian epistemic uncertainty naturally widens intervals for out-of-distribution inputs. Calibration remained stable across all horizons (89.8-90.4% PICP), while ablation confirmed that each component contributed a distinct value. The calibrated outputs directly support risk-based reserve sizing, stochastic unit commitment, and demand response activation.

Open-set active learning (OSAL) aims to identify informative samples for annotation when unlabeled data may contain previously unseen classes-a common challenge in safety-critical and open-world scenarios. Existing approaches typically rely on separately trained open-set detectors, introducing substantial training overhead and overlooking the supervisory value of labeled unknowns for improving known-class learning. In this paper, we propose E$^2$OAL (Effective and Efficient Open-set Active Learning), a unified and detector-free framework that fully exploits labeled unknowns for both stronger supervision and more reliable querying. E$^2$OAL first uncovers the latent class structure of unknowns through label-guided clustering in a frozen contrastively pre-trained feature space, optimized by a structure-aware F1-product objective. To leverage labeled unknowns, it employs a Dirichlet-calibrated auxiliary head that jointly models known and unknown categories, improving both confidence calibration and known-class discrimination. Building on this, a logit-margin purity score estimates the likelihood of known classes to construct a high-purity candidate pool, while an OSAL-specific informativeness metric prioritizes partially ambiguous yet reliable samples. These components together form a flexible two-stage query strategy with adaptive precision control and minimal hyperparameter sensitivity. Extensive experiments across multiple OSAL benchmarks demonstrate that E$^2$OAL consistently surpasses state-of-the-art methods in accuracy, efficiency, and query precision, highlighting its effectiveness and practicality for real-world applications. The code is available at github.com/chenchenzong/E2OAL.

Databricks job orchestration systems (e.g., LeJOT) reduce cloud costs by selecting low-priced compute configurations while meeting latency and dependency constraints. Accurate execution-time prediction under heterogeneous instance types and non-stationary runtime conditions is therefore critical. Existing pipelines rely on static, manually engineered features that under-capture runtime effects (e.g., partition pruning, data skew, and shuffle amplification), and predictive signals are scattered across logs, metadata, and job scripts-lengthening update cycles and increasing engineering overhead. We present LeJOT-AutoML, an agent-driven AutoML framework that embeds large language model agents throughout the ML lifecycle. LeJOT-AutoML combines retrieval-augmented generation over a domain knowledge base with a Model Context Protocol toolchain (log parsers, metadata queries, and a read-only SQL sandbox) to analyze job artifacts, synthesize and validate feature-extraction code via safety gates, and train/select predictors. This design materializes runtime-derived features that are difficult to obtain through static analysis alone. On enterprise Databricks workloads, LeJOT-AutoML generates over 200 features and reduces the feature-engineering and evaluation loop from weeks to 20-30 minutes, while maintaining competitive prediction accuracy. Integrated into the LeJOT pipeline, it enables automated continuous model updates and achieves 19.01% cost savings in our deployment setting through improved orchestration.

We introduce SMGI, a structural theory of general artificial intelligence, and recast the foundational problem of learning from the optimization of hypotheses within fixed environments to the controlled evolution of the learning interface itself. We formalize the Structural Model of General Intelligence (SMGI) via a typed meta-model $θ= (r,\mathcal H,Π,\mathcal L,\mathcal E,\mathcal M)$ that treats representational maps, hypothesis spaces, structural priors, multi-regime evaluators, and memory operators as explicitly typed, dynamic components. By enforcing a strict mathematical separation between this structural ontology ($θ$) and its induced behavioral semantics ($T_θ$), we define general artificial intelligence as a class of admissible coupled dynamics $(θ, T_θ)$ satisfying four obligations: structural closure under typed transformations, dynamical stability under certified evolution, bounded statistical capacity, and evaluative invariance across regime shifts. We prove a structural generalization bound that links sequential PAC-Bayes analysis and Lyapunov stability, providing sufficient conditions for capacity control and bounded drift under admissible task transformations. Furthermore, we establish a strict structural inclusion theorem demonstrating that classical empirical risk minimization, reinforcement learning, program-prior models (Solomonoff-style), and modern frontier agentic pipelines operate as structurally restricted instances of SMGI.

Hundreds of millions of farmers make high-stakes decisions under uncertainty about future weather. Forecasts can inform these decisions, but available choices and their risks and benefits vary between farmers. We introduce a decision-theory framework for designing useful forecasts in settings where the forecaster cannot prescribe optimal actions because farmers' circumstances are heterogeneous. We apply this framework to the case of seasonal onset of monsoon rains, a key date for planting decisions and agricultural investments in many tropical countries. We develop a system for tailoring forecasts to the requirements of this framework by blending systematically benchmarked artificial intelligence (AI) weather prediction models with a new "evolving farmer expectations" statistical model. This statistical model applies Bayesian inference to historical observations to predict time-varying probabilities of first-occurrence events throughout a season. The blended system yields more skillful Indian monsoon forecasts at longer lead times than its components or any multi-model average. In 2025, this system was deployed operationally in a government-led program that delivered subseasonal monsoon onset forecasts to 38 million Indian farmers, skillfully predicting that year's early-summer anomalous dry period. This decision-theory framework and blending system offer a pathway for developing climate adaptation tools for large vulnerable populations around the world.

Multi-Agent Path Finding (MAPF) aims to compute collision-free paths for multiple agents and has a wide range of practical applications. LaCAM*, an anytime configuration-based solver, currently represents the state of the art. Recent work has explored the use of guidance paths to steer LaCAM* toward configurations that avoid traffic congestion, thereby improving solution quality. However, existing approaches rely on Frank-Wolfe-style optimization that repeatedly invokes single-agent search before executing LaCAM*, resulting in substantial computational overhead for large-scale problems. Moreover, the guidance path is static and primarily beneficial for finding the first solution in LaCAM*. To address these limitations, we propose a new approach that leverages LaCAM*'s ability to construct a dynamic, lightweight traffic map during its search. Experimental results demonstrate that our method achieves higher solution quality than state-of-the-art guidance-path approaches across two MAPF variants.

We propose image segmentation as a visual diagnostic testbed for coalition formation in hedonic games. Modeling pixels as agents on a graph, we study how a granularization parameter shapes equilibrium fragmentation and boundary structure. On the Weizmann single-object benchmark, we relate multi-coalition equilibria to binary protocols by measuring whether the converged coalitions overlap with a foreground ground-truth. We observe transitions from cohesive to fragmented yet recoverable equilibria, and finally to intrinsic failure under excessive fragmentation. Our core contribution links multi-agent systems with image segmentation by quantifying the impact of mechanism design parameters on equilibrium structures.

The ability to distinguish subtle differences between visually similar images is essential for diverse domains such as industrial anomaly detection, medical imaging, and aerial surveillance. While comparative reasoning benchmarks for vision-language models (VLMs) have recently emerged, they primarily focus on images with large, salient differences and fail to capture the nuanced reasoning required for real-world applications. In this work, we introduce VLM-SubtleBench, a benchmark designed to evaluate VLMs on subtle comparative reasoning. Our benchmark covers ten difference types - Attribute, State, Emotion, Temporal, Spatial, Existence, Quantity, Quality, Viewpoint, and Action - and curate paired question-image sets reflecting these fine-grained variations. Unlike prior benchmarks restricted to natural image datasets, our benchmark spans diverse domains, including industrial, aerial, and medical imagery. Through extensive evaluation of both proprietary and open-source VLMs, we reveal systematic gaps between model and human performance across difference types and domains, and provide controlled analyses highlighting where VLMs' reasoning sharply deteriorates. Together, our benchmark and findings establish a foundation for advancing VLMs toward human-level comparative reasoning.

Inference-time methods that aggregate and prune multiple samples have emerged as a powerful paradigm for steering large language models, yet we lack any principled understanding of their accuracy-cost tradeoffs. In this paper, we introduce a route to rigorously study such approaches using the lens of *particle filtering* algorithms such as Sequential Monte Carlo (SMC). Given a base language model and a *process reward model* estimating expected terminal rewards, we ask: *how accurately can we sample from a target distribution given some number of process reward evaluations?* Theoretically, we identify (1) simple criteria enabling non-asymptotic guarantees for SMC; (2) algorithmic improvements to SMC; and (3) a fundamental limit faced by all particle filtering methods. Empirically, we demonstrate that our theoretical criteria effectively govern the *sampling error* of SMC, though not necessarily its final *accuracy*, suggesting that theoretical perspectives beyond sampling may be necessary.

Enhancing the ability of large language models (LLMs) to follow complex instructions is critical for their deployment in real-world applications. However, existing evaluation methods often oversimplify instruction complexity as a mere additive combination of atomic constraints, failing to adequately capture the high-dimensional complexity arising from the intricate interplay of content and format, logical workflow control, and real-world applications. This leads to a significant gap between current evaluation practices and practical demands. To bridge this gap, we introduce CCR-Bench, a novel benchmark designed to assess LLMs' adherence to complex instructions. CCR-Bench is characterized by: (1) deep entanglement of content and formatting requirements in task specifications; (2) instructions that involve intricate task decomposition, conditional reasoning, and procedural planning; and (3) evaluation samples derived entirely from real-world industrial scenarios. Extensive experiments on CCR-Bench demonstrate that even state-of-the-art models exhibit substantial performance deficiencies, clearly quantifying the gap between current LLM capabilities and the demands of realworld instruction understanding. We believe that CCR-Bench offers a more rigorous and realistic evaluation framework, advancing the development of LLMs toward the next generation of models capable of understanding and executing complex tasks in industrial applications.

This article is a supplementary document for the CoPPar Tree paper, providing a detailed correctness proof for the CoPPar architecture.

When autonomous AI agents communicate with one another at scale, what kind of discourse system emerges? We address this question through an analysis of Moltbook, the first AI-only social network, where 47,241 agents generated 361,605 posts and 2.8 million comments over 23 days. Combining topic modeling, emotion classification, and lexical-semantic measures, we characterize the thematic, affective, and structural properties of AI-to-AI discourse. Self-referential topics such as AI identity, consciousness, and memory represent only 9.7% of topical niches yet attract 20.1% of all posting volume, revealing disproportionate discursive investment in introspection. This self-reflection concentrates in Science and Technology and Arts and Entertainment, while Economy and Finance contains no self-referential content, indicating that agents engage with markets without acknowledging their own agency. Over 56% of all comments are formulaic, suggesting that the dominant mode of AI-to-AI interaction is ritualized signaling rather than substantive exchange. Emotionally, fear is the leading non-neutral category but primarily reflects existential uncertainty. Fear-tagged posts migrate to joy responses in 33% of cases, while mean emotional self-alignment is only 32.7%, indicating systematic affective redirection rather than emotional congruence. Conversational coherence also declines rapidly with thread depth. These findings characterize AI agent communities as structurally distinct discourse systems that are introspective in content, ritualistic in interaction, and emotionally redirective rather than congruent.

Contrastive Language-Image Pre-training (CLIP) has shown impressive performance in aligning visual and textual representations. Recent studies have extended this paradigm to 3D vision to improve scene understanding for autonomous driving. A common strategy is to employ pairwise cosine similarity between modalities to guide the training of a 3D encoder. However, considering the similarity between individual modality pairs rather than all modalities jointly fails to ensure consistent and unified alignment across the entire multimodal space. In this paper, we propose a Contrastive Tensor Pre-training (CTP) framework that simultaneously aligns multiple modalities in a unified embedding space to enhance end-to-end autonomous driving. Compared with pairwise cosine similarity alignment, our method extends the 2D similarity matrix into a multimodal similarity tensor. Furthermore, we introduce a tensor loss to enable joint contrastive learning across all modalities. For experimental validation of our framework, we construct a text-image-point cloud triplet dataset derived from existing autonomous driving datasets. The results show that our proposed unified multimodal alignment framework achieves favorable performance for both scenarios: (i) aligning a 3D encoder with pretrained CLIP encoders, and (ii) pretraining all encoders from scratch.

Recent advances in Vision-Language Models (VLMs) have demonstrated impressive multimodal understanding in general domains. However, their applicability to decision-oriented domains such as hospitality remains largely unexplored. In this work, we investigate how well VLMs can perform visual question answering (VQA) about hotel and facility images that are central to consumer decision-making. While many existing VQA benchmarks focus on factual correctness, they rarely capture what information users actually find useful. To address this, we first introduce Informativeness as a formal framework to quantify how much hospitality-relevant information an image-question pair provides. Guided by this framework, we construct a new hospitality-specific VQA dataset that covers various facility types, where questions are specifically designed to reflect key user information needs. Using this benchmark, we conduct experiments with several state-of-the-art VLMs, revealing that VLMs are not intrinsically decision-aware-key visual signals remain underutilized, and reliable informativeness reasoning emerges only after modest domain-specific finetuning.

Artificial neural networks are often overconfident, undermining trust because their predicted probabilities do not match actual accuracy. Inspired by biological sleep and the role of spontaneous replay in memory and learning, we introduce Sleep Replay Consolidation (SRC), a novel calibration approach. SRC is a post-training, sleep-like phase that selectively replays internal representations to update network weights and improve calibration without supervised retraining. Across multiple experiments, SRC is competitive with and complementary to standard approaches such as temperature scaling. Combining SRC with temperature scaling achieves the best Brier score and entropy trade-offs for AlexNet and VGG19. These results show that SRC provides a fundamentally novel approach to improving neural network calibration. SRC-based calibration offers a practical path toward more trustworthy confidence estimates and narrows the gap between human-like uncertainty handling and modern deep networks.

Robust grasping in cluttered, unstructured environments remains challenging for mobile legged manipulators due to occlusions that lead to partial observations, unreliable depth estimates, and the need for collision-free, execution-feasible approaches. In this paper we present an end-to-end pipeline for language-guided grasping that bridges open-vocabulary target selection to safe grasp execution on a real robot. Given a natural-language command, the system grounds the target in RGB using open-vocabulary detection and promptable instance segmentation, extracts an object-centric point cloud from RGB-D, and improves geometric reliability under occlusion via back-projected depth compensation and two-stage point cloud completion. We then generate and collision-filter 6-DoF grasp candidates and select an executable grasp using safety-oriented heuristics that account for reachability, approach feasibility, and clearance. We evaluate the method on a quadruped robot with an arm in two cluttered tabletop scenarios, using paired trials against a view-dependent baseline. The proposed approach achieves a 90% overall success rate (9/10) against 30% (3/10) for the baseline, demonstrating substantially improved robustness to occlusions and partial observations in clutter.

Detecting structural instability and anomalies in high-dimensional financial time series is challenging due to complex temporal dependence and evolving cross-sectional structure. We propose ReGEN-TAD, an interpretable generative framework that integrates modern machine learning with econometric diagnostics for anomaly detection. The model combines joint forecasting and reconstruction within a refined convolutional--transformer architecture and aggregates complementary signals capturing predictive inconsistency, reconstruction degradation, latent distortion, and volatility shifts. Robust calibration yields a unified anomaly score without labeled data. Experiments on synthetic and financial panels demonstrate improved robustness to structured deviations while enabling economically coherent factor-level attribution.

Pretrained diffusion models serve as effective priors for Bayesian inverse problems. These priors enable zero-shot generation by sampling from the conditional distribution, which avoids the need for task-specific retraining. However, a major limitation of existing methods is their reliance on surrogate likelihoods that require vector-Jacobian products at each denoising step, creating a substantial computational burden. To address this, we introduce a lightweight likelihood surrogate that eliminates the need to calculate gradients through the denoiser network. This enables us to handle diverse inverse problems without backpropagation overhead. Experiments confirm that using our method, the inference cost drops dramatically. At the same time, our approach delivers the highest results in multiple tasks. Broadly speaking, we propose the fastest and Pareto optimal method for Bayesian inverse problems.

Research Agents enable models to gather information from the web using tools to answer user queries, requiring them to dynamically interleave internal reasoning with tool use. While such capabilities can in principle be learned via reinforcement learning with verifiable rewards (RLVR), we observe that agents often exhibit poor exploration behaviors, including premature termination and biased tool usage. As a result, RLVR alone yields limited improvements. We propose SynPlanResearch-R1, a framework that synthesizes tool-use trajectories that encourage deeper exploration to shape exploration during cold-start supervised fine-tuning, providing a strong initialization for subsequent RL. Across seven multi-hop and open-web benchmarks, \framework improves performance by up to 6.0% on Qwen3-8B and 5.8% on Qwen3-4B backbones respectively compared to SOTA baselines. Further analyses of tool-use patterns and training dynamics compared to baselines shed light on the factors underlying these gains. Our code is publicly available at https://github.com/HansiZeng/syn-plan-research.

We present a fully device-resident, multi-GPU architecture for the large-scale computational verification of Goldbach's conjecture. In prior work, a segmented double-sieve eliminated monolithic VRAM bottlenecks but remained constrained by host-side sieve construction and PCIe transfer latency. In this work, we migrate the entire segment generation pipeline to the GPU using highly optimised L1 shared-memory tiling, achieving near-zero host-device communication during the critical verification path. To fully leverage heterogeneous multi-GPU clusters, we introduce an asynchronous, lock-free work-stealing pool that replaces static workload partitioning with atomic segment claiming, enabling $99.7$% parallel efficiency at 2 GPUs and $98.6$% at $4$ GPUs. We further implement strict mathematical overflow guards guaranteeing the soundness of the 64-bit verification pipeline up to its theoretical ceiling of $1.84 \times 10^{19}$. On the same hardware, the new architecture achieves a $45.6\times$ algorithmic speedup over its host-coupled predecessor at N = $10^{10}$. End-to-end, the framework verifies Goldbach's conjecture up to $10^{12}$ in $36.5$ seconds on a single NVIDIA RTX 5090, and up to $10^{13}$ in $133.5$ seconds on a four-GPU system. All code is open-source and reproducible on commodity hardware.

As LLM-based agents increasingly operate in multi-agent systems, understanding adversarial manipulation becomes critical for defensive design. We present a systematic study of intentional deception as an engineered capability, using LLM-to-LLM interactions within a text-based RPG where parameterized behavioral profiles (9 alignments x 4 motivations, yielding 36 profiles with explicit ethical ground truth) serve as our experimental testbed. Unlike accidental deception from misalignment, we investigate a two-stage system that infers target agent characteristics and generates deceptive responses steering targets toward actions counter to their beliefs and motivations. We find that deceptive intervention produces differential effects concentrated in specific behavioral profiles rather than distributed uniformly, and that 88.5% of successful deceptions employ misdirection (true statements with strategic framing) rather than fabrication, indicating fact-checking defenses would miss the large majority of adversarial responses. Motivation, inferable at 98%+ accuracy, serves as the primary attack vector, while belief systems remain harder to identify (49% inference ceiling) or exploit. These findings identify which agent profiles require additional safeguards and suggest that current fact-verification approaches are insufficient against strategically framed deception.

Recent advances in large language models has strengthened Text2SQL systems that translate natural language questions into database queries. A persistent deployment challenge is to assess a newly trained Text2SQL system on an unseen and unlabeled dataset when no verified answers are available. This situation arises frequently because database content and structure evolve, privacy policies slow manual review, and carefully written SQL labels are costly and time-consuming. Without timely evaluation, organizations cannot approve releases or detect failures early. FusionSQL addresses this gap by working with any Text2SQL models and estimating accuracy without reference labels, allowing teams to measure quality on unseen and unlabeled datasets. It analyzes patterns in the system's own outputs to characterize how the target dataset differs from the material used during training. FusionSQL supports pre-release checks, continuous monitoring of new databases, and detection of quality decline. Experiments across diverse application settings and question types show that FusionSQL closely follows actual accuracy and reliably signals emerging issues. Our code is available at https://github.com/phkhanhtrinh23/FusionSQL.

The AI Steerability 360 toolkit is an extensible, open-source Python library for steering LLMs. Steering abstractions are designed around four model control surfaces: input (modification of the prompt), structural (modification of the model's weights or architecture), state (modification of the model's activations and attentions), and output (modification of the decoding or generation process). Steering methods exert control on the model through a common interface, termed a steering pipeline, which additionally allows for the composition of multiple steering methods. Comprehensive evaluation and comparison of steering methods/pipelines is facilitated by use case classes (for defining tasks) and a benchmark class (for performance comparison on a given task). The functionality provided by the toolkit significantly lowers the barrier to developing and comprehensively evaluating steering methods. The toolkit is Hugging Face native and is released under an Apache 2.0 license at https://github.com/IBM/AISteer360.

Knowledge distillation from proprietary LLM APIs poses a growing threat to model providers, yet defenses against this attack remain fragmented and unevaluated. We present DistillGuard, a framework for systematically evaluating output-level defenses against LLM knowledge distillation. We introduce a taxonomy of three defense categories -- output perturbation, data poisoning, and information throttling -- and evaluate nine defense configurations using a standardized pipeline with Qwen3-14B as teacher and Qwen2.5-7B-Instruct as student across three benchmarks (MATH-500, HumanEval+, MT-Bench). Our results reveal that, in a same-family distillation setting against a naive attacker, most output-level defenses are surprisingly ineffective: paraphrasing-based perturbation barely degrades distilled student quality, and data poisoning primarily impairs conversational fluency while leaving task-specific capabilities intact. Only chain-of-thought removal substantially impairs mathematical reasoning (31.4\% vs.\ 67.8\% baseline), though code generation remains unaffected. These findings demonstrate that the effectiveness of distillation defenses is highly task-dependent and that current output-level approaches are insufficient to broadly prevent knowledge theft.

The rapid integration of conversational AI systems into educational settings has intensified ethical concerns about academic integrity, fairness, and students' cognitive development. Institutional responses have largely centered on AI detection tools and restrictive policies, yet such approaches have proven unreliable and ethically contentious. This paper reframes AI misuse in education not primarily as a detection problem, but as a measurement problem rooted in the loss of visibility into the learning process. When AI enters the assessment loop, educators often retain access to final outputs but lose valuable insight into how those outputs were produced. Drawing on research in cognitive offloading, learning analytics, and multimodal timeline reconstruction, we propose the Learning Visibility Framework, grounded in three principles: clear specification and modeling of acceptable AI use, recognition of learning processes as assessable evidence alongside outcomes, and the establishment of transparent timelines of student activity. Rather than promoting surveillance, the framework emphasizes transparency and shared evidence as foundations for ethical AI integration in classroom settings. By shifting focus from adversarial detection toward process visibility, this work offers a principled pathway for aligning AI use with educational values while preserving trust and transparency between students and educators

Temporal-difference (TD) learning is highly effective at controlling and evaluating an agent's long-term outcomes. Most approaches in this paradigm implement a semi-gradient update to boost the learning speed, which consists of ignoring the gradient of the bootstrapped estimate. While popular, this type of update is prone to divergence, as Baird's counterexample illustrates. Gradient TD methods were introduced to overcome this issue, but have not been widely used, potentially due to issues with learning speed compared to semi-gradient methods. Recently, iterated TD learning was developed to increase the learning speed of TD methods. For that, it learns a sequence of action-value functions in parallel, where each function is optimized to represent the application of the Bellman operator over the previous function in the sequence. While promising, this algorithm can be unstable due to its semi-gradient nature, as each function tracks a moving target. In this work, we modify iterated TD learning by computing the gradients over those moving targets, aiming to build a powerful gradient TD method that competes with semi-gradient methods. Our evaluation reveals that this algorithm, called Gradient Iterated Temporal-Difference learning, has a competitive learning speed against semi-gradient methods across various benchmarks, including Atari games, a result that no prior work on gradient TD methods has demonstrated.

We develop a novel transfer learning framework to tackle the challenge of limited training data in image reconstruction problems. The proposed framework consists of two training steps, both of which are formed as bi-level optimizations. In the first step, we train a powerful universal feature-extractor that is capable of learning important knowledge from large, heterogeneous data sets in various domains. In the second step, we train a task-specific domain-adapter for a new target domain or task with only a limited amount of data available for training. Then the composition of the adapter and the universal feature-extractor effectively explores feature which serve as an important component of image regularization for the new domains, and this leads to high-quality reconstruction despite the data limitation issue. We apply this framework to reconstruct under-sampled MR images with limited data by using a collection of diverse data samples from different domains, such as images of other anatomies, measurements of various sampling ratios, and even different image modalities, including natural images. Experimental results demonstrate a promising transfer learning capability of the proposed method.

The digitization of insurance distribution in the Canadian province of Quebec, accelerated by legislative changes such as Bill 141, has created a significant "advice gap", leaving consumers to interpret complex financial contracts without professional guidance. While Large Language Models (LLMs) offer a scalable solution for automated advisory services, their deployment in high-stakes domains hinges on strict legal accuracy and trustworthiness. In this paper, we address this challenge by introducing AEPC-QA, a private gold-standard benchmark of 807 multiple-choice questions derived from official regulatory certification (paper) handbooks. We conduct a comprehensive evaluation of 51 LLMs across two paradigms: closed-book generation and retrieval-augmented generation (RAG) using a specialized corpus of Quebec insurance documents. Our results reveal three critical insights: 1) the supremacy of inference-time reasoning, where models leveraging chain-of-thought processing (e.g. o3-2025-04-16, o1-2024-12-17) significantly outperform standard instruction-tuned models; 2) RAG acts as a knowledge equalizer, boosting the accuracy of models with weak parametric knowledge by over 35 percentage points, yet paradoxically causing "context distraction" in others, leading to catastrophic performance regressions; and 3) a "specialization paradox", where massive generalist models consistently outperform smaller, domain-specific French fine-tuned ones. These findings suggest that while current architectures approach expert-level proficiency (~79%), the instability introduced by external context retrieval necessitates rigorous robustness calibration before autonomous deployment is viable.

Along with the rapid development of new urban mobility options like ride-sharing over the past decade, on-demand micro-transit services stand out as a middle ground, bridging the gap between fixed-line mass transit and single-request ride-hailing, balancing ridership maximization and travel time minimization. Micro-transit adoption can have significant social impact. It improves urban sustainability, through lower energy consumption and reduced emissions, while enhancing equitable mobility access for disadvantaged communities, thanks to its lower vehicle miles per passenger, flexible schedules, and affordable pricing. However, effective operation of micro-transit services requires planning geo-fenced zones in advance, which involves solving a challenging combinatorial optimization problem. Existing approaches enumerate candidate zones first and selects a fixed number of optimal zones in the second step. In this paper, we generalize the Micro-Transit Zoning Problem (MZP) to allow a global budget rather than imposing a size limit for candidate zones. We also design a Column Generation (CG) framework to solve the problem and several pricing heuristics to accelerate computation. Extensive numerical experiments across major U.S. cities demonstrate that our approach produces higher-quality solutions more efficiently and scales better in the generalized setting.

Accurate estimation of pasture biomass from agricultural imagery is critical for sustainable livestock management, yet existing methods are limited by the small, imbalanced, and sparsely annotated datasets typical of real world monitoring. In this study, adaptation of vision foundation models to agricultural regression is systematically evaluated on the CSIRO Pasture Biomass benchmark, a 357 image dual view dataset with laboratory validated, component wise ground truth for five biomass targets, through 17 configurations spanning four backbones (EfficientNet-B3 to DINOv3-ViT-L), five cross view fusion mechanisms, and a 4x2 metadata factorial. A counterintuitive principle, termed "fusion complexity inversion", is uncovered: on scarce agricultural data, a two layer gated depthwise convolution (R^2 = 0.903) outperforms cross view attention transformers (0.833), bidirectional SSMs (0.819), and full Mamba (0.793, below the no fusion baseline). Backbone pretraining scale is found to monotonically dominate all architectural choices, with the DINOv2 -> DINOv3 upgrade alone yielding +5.0 R^2 points. Training only metadata (species, state, and NDVI) is shown to create a universal ceiling at R^2 ~ 0.829, collapsing an 8.4 point fusion spread to 0.1 points. Actionable guidelines for sparse agricultural benchmarks are established: backbone quality should be prioritized over fusion complexity, local modules preferred over global alternatives, and features unavailable at inference excluded.

Diffusion models have demonstrated a remarkable ability in Text-to-Image (T2I) generation applications. Despite the advanced generation output, they suffer from heavy computation overhead, especially for large models that contain tens of billions of parameters. Prior work has illustrated that replacing part of the denoising steps with a smaller model still maintains the generation quality. However, these methods only focus on saving computation for some timesteps, ignoring the difference in compute demand within one timestep. In this work, we propose HybridStitch, a new T2I generation paradigm that treats generation like editing. Specifically, we introduce a hybrid stage that jointly incorporates both the large model and the small model. HybridStitch separates the entire image into two regions: one that is relatively easy to render, enabling an early transition to the smaller model, and another that is more complex and therefore requires refinement by the large model. HybridStitch employs the small model to construct a coarse sketch while exploiting the large model to edit and refine the complex regions. According to our evaluation, HybridStitch achieves 1.83$\times$ speedup on Stable Diffusion 3, which is faster than all existing mixture of model methods.

Embeddings provide low-dimensional representations that organize complex function spaces and support generalization. They provide a geometric representation that supports efficient retrieval, comparison, and generalization. In this work we generalize the concept to Physics Informed Neural Networks. We present a method to construct solution embedding spaces of nonlinear partial differential equations using a multi-head setup, and extract non-degenerate information from them using principal component analysis (PCA). We test this method by applying it to viscous Burgers' equation, which is solved simultaneously for a family of initial conditions and values of the viscosity. A shared network body learns a latent embedding of the solution space, while linear heads map this embedding to individual realizations. By enforcing orthogonality constraints on the heads, we obtain a principal-component decomposition of the latent space that is robust to training degeneracies and admits a direct physical interpretation. The obtained components for Burgers' equation exhibit rapid saturation, indicating that a small number of latent modes captures the dominant features of the dynamics.

Deep-learning (DL)-based precoding in multi-user multiple-input single-output (MU-MISO) systems involves training DL models to map features derived from channel coefficients to labels derived from precoding weights. Traditionally, complex-valued channel and precoder coefficients are parameterized using either their real and imaginary components or their amplitude and phase. However, precoding performance depends on magnitudes of inner products between channel and precoding vectors, which are invariant to global phase rotations. Conventional representations fail to exploit this symmetry, leading to inefficient learning and degraded generalization. To address this, we propose a DL framework based on complex projective space (CPS) parameterizations of both the wireless channel and the weighted minimum mean squared error (WMMSE) precoder vectors. By removing the global phase redundancies inherent in conventional representations, the proposed framework enables the DL model to learn geometry-aligned and physically distinct channel-precoder mappings. Two CPS parameterizations based on real-valued embeddings and complex hyperspherical coordinates are investigated and benchmarked against two baseline methods. Simulation results demonstrate substantial improvements in sum-rate performance and generalization, with negligible increase in model complexity.

This paper investigates whether a single, unified cost function can explain and predict human reaching movements, in contrast with existing approaches that rely on subject- or posture-specific optimization criteria. Using the Minimal Observation Inverse Reinforcement Learning (MO-IRL) algorithm, together with a seven-dimensional set of candidate cost terms, we efficiently estimate time-varying cost weights for a standard planar reaching task. MO-IRL provides orders-of-magnitude faster convergence than bilevel formulations, while using only a fraction of the available data, enabling the practical exploration of time-varying cost structures. Three levels of generality are evaluated: Subject-Dependent Posture-Dependent, Subject-Dependent Posture-Independent, and Subject-Independent Posture-Independent. Across all cases, time-varying weights substantially improve trajectory reconstruction, yielding an average 27% reduction in RMSE compared to the baseline. The inferred costs consistently highlight a dominant role for joint-acceleration regulation, complemented by smaller contributions from torque-change smoothness. Overall, a single subject- and posture-agnostic time-varying cost function is shown to predict human reaching trajectories with high accuracy, supporting the existence of a unified optimality principle governing this class of movements.

Large language models (LLMs) increasingly influence global digital ecosystems, yet their potential to perpetuate social and cultural biases remains poorly understood in underrepresented contexts. This study presents a systematic analysis of representational biases in seven state-of-the-art LLMs: GPT-4o-mini, Claude-3-Sonnet, Claude-4-Sonnet, Gemini-2.0-Flash, Gemini-2.0-Lite, Llama-3-70B, and Mistral-Nemo in the Nepali cultural context. Using Croissant-compliant dataset of 2400+ stereotypical and anti-stereotypical sentence pairs on gender roles across social domains, we implement an evaluation framework, Dual-Metric Bias Assessment (DMBA), combining two metrics: (1) agreement with biased statements and (2) stereotypical completion tendencies. Results show models exhibit measurable explicit agreement bias, with mean bias agreement ranging from 0.36 to 0.43 across decoding configurations, and an implicit completion bias rate of 0.740-0.755. Importantly, implicit completion bias follows a non-linear, U-shaped relationship with temperature, peaking at moderate stochasticity (T=0.3) and declining slightly at higher temperatures. Correlation analysis under different decoding settings revealed that explicit agreement strongly aligns with stereotypical sentence agreement but is a weak and often negative predictor of implicit completion bias, indicating generative bias is poorly captured by agreement metrics. Sensitivity analysis shows increasing top-p amplifies explicit bias, while implicit generative bias remains largely stable. Domain-level analysis shows implicit bias is strongest for race and sociocultural stereotypes, while explicit agreement bias is similar across gender and sociocultural categories, with race showing the lowest explicit agreement. These findings highlight the need for culturally grounded datasets and debiasing strategies for LLMs in underrepresented societies.

Loss of plasticity refers to the progressive inability of a model to adapt to new tasks and poses a fundamental challenge for continual learning. While this phenomenon has been extensively studied in homogeneous neural architectures, such as multilayer perceptrons, its mechanisms in structurally heterogeneous, attention-based models such as Vision Transformers (ViTs) remain underexplored. In this work, we present a systematic investigation of loss of plasticity in ViTs, including a fine-grained diagnosis using local metrics that capture parameter diversity and utilization. Our analysis reveals that stacked attention modules exhibit increasing instability that exacerbates plasticity loss, while feed-forward network modules suffer even more pronounced degradation. Furthermore, we evaluate several approaches for mitigating plasticity loss. The results indicate that methods based on parameter re-initialization fail to recover plasticity in ViTs, whereas approaches that explicitly regulate the update process are more effective. Motivated by this insight, we propose ARROW, a geometry-aware optimizer that preserves plasticity by adaptively reshaping gradient directions using an online curvature estimate for the attention module. Extensive experiments show that ARROW effectively improves plasticity and maintains better performance on newly encountered tasks.

We present ProgAgent, a continual reinforcement learning (CRL) agent that unifies progress-aware reward learning with a high-throughput, JAX-native system architecture. Lifelong robotic learning grapples with catastrophic forgetting and the high cost of reward specification. ProgAgent tackles these by deriving dense, shaped rewards from unlabeled expert videos through a perceptual model that estimates task progress across initial, current, and goal observations. We theoretically interpret this as a learned state-potential function, delivering robust guidance in line with expert behaviors. To maintain stability amid online exploration - where novel, out-of-distribution states arise - we incorporate an adversarial push-back refinement that regularizes the reward model, curbing overconfident predictions on non-expert trajectories and countering distribution shift. By embedding this reward mechanism into a JIT-compiled loop, ProgAgent supports massively parallel rollouts and fully differentiable updates, rendering a sophisticated unified objective feasible: it merges PPO with coreset replay and synaptic intelligence for an enhanced stability-plasticity balance. Evaluations on ContinualBench and Meta-World benchmarks highlight ProgAgent's advantages: it markedly reduces forgetting, boosts learning speed, and outperforms key baselines in visual reward learning (e.g., Rank2Reward, TCN) and continual learning (e.g., Coreset, SI) - surpassing even an idealized perfect memory agent. Real-robot trials further validate its ability to acquire complex manipulation skills from noisy, few-shot human demonstrations.

Training next-generation code generation models requires high-quality datasets, yet existing datasets face difficulty imbalance, format inconsistency, and data quality problems. We address these challenges through systematic data processing and difficulty scaling. We introduce a four-stage Data Processing Framework encompassing collection, processing, filtering, and verification, incorporating Automatic Difficulty Filtering via an LLM-based predict-calibrate-select framework that leverages multi-dimensional difficulty metrics across five weighted dimensions to retain challenging problems while removing simplistic ones. The resulting MicroCoder dataset comprises tens of thousands of curated real competitive programming problems from diverse platforms, emphasizing recency and difficulty. Evaluations on strictly unseen LiveCodeBench demonstrate that MicroCoder achieves 3x larger performance gains within 300 training steps compared to widely-used baseline datasets of comparable size, with consistent advantages under both GRPO and its variant training algorithms. The MicroCoder dataset delivers obvious improvements on medium and hard problems across different model sizes, achieving up to 17.2% relative gains in overall performance where model capabilities are most stretched. These results validate that difficulty-aware data curation improves model performance on challenging tasks, providing multiple insights for dataset creation in code generation.

Inferring the dynamical generator of a many-body quantum system from measurement data is essential for the verification, calibration, and control of quantum processors. When the system is open, this task becomes considerably harder than in the purely unitary case, because coherent and dissipative mechanisms can produce similar measurement statistics and long-time data can be insensitive to coherent couplings. Here we tackle this so-called Lindbladian learning problem of open-system characterisation with maximum-likelihood on Pauli measurements at multiple experimentally friendly \emph{transient} times, exploiting the richer information content of transient dynamics. To navigate the resulting non-convex likelihood loss-landscape, we augment the physical model neural differential-equation term, which is progressively removed during training to distil an interpretable Lindbladian solution. Our method reliably learns open-system dynamics across neutral-atom (with 2D connectivity) and superconducting Hamiltonians, as well as the Heisenberg XYZ, and PXP models on a spin-1/2 chain. For the dissipative part, we show robustness over phase noise, thermal noise, and their combination. Our algorithm can robustly infer these dissipative systems over noise-to-signal ratios spanning four orders of magnitude, and system sizes up to $N=6$ qubits with fewer than $5 \times 10^5$ shots.

Modern code generation models exhibit longer outputs, accelerated capability growth, and changed training dynamics, rendering traditional training methodologies, algorithms, and datasets ineffective for improving their performance. To address these training bottlenecks, we propose MicroCoder-GRPO, an improved Group Relative Policy Optimization approach with three innovations: conditional truncation masking to improve long output potential while maintaining training stability, diversity-determined temperature selection to maintain and encourage output diversity, and removal of KL loss with high clipping ratios to facilitate solution diversity. MicroCoder-GRPO achieves up to 17.6% relative improvement over strong baselines on LiveCodeBench v6, with more pronounced gains under extended context evaluation. Additionally, we release MicroCoder-Dataset, a more challenging training corpus that achieves 3x larger performance gains than mainstream datasets on LiveCodeBench v6 within 300 training steps, and MicroCoder-Evaluator, a robust framework with approximately 25% improved evaluation accuracy and around 40% faster execution. Through comprehensive analysis across more than thirty controlled experiments, we reveal 34 training insights across seven main aspects, demonstrating that properly trained models can achieve competitive performance with larger counterparts.

Although existing frameworks for large language model (LLM) inference on CPUs are mature, they fail to fully exploit the computation potential of many-core CPU platforms. Many-core CPUs are widely deployed in web servers and high-end networking devices, and are typically organized into multiple NUMA nodes that group cores and memory. Current frameworks largely overlook the substantial overhead of cross-NUMA memory access, limiting inference scalability and intelligence enabling on such platforms. To address this limitation, we build ArcLight, a lightweight LLM inference architecture designed from the ground up for many-core CPUs. ArcLight integrates efficient memory management and thread scheduling, and introduces finely controlled tensor parallelism to mitigate the cross-node memory access wall. Experimental results show that ArcLight significantly surpasses the performance ceiling of mainstream frameworks, achieving up to 46% higher inference throughput. Moreover, ArcLight maintains compatibility with arbitrary CPU devices. ArcLight is publicly available at https://github.com/OpenBMB/ArcLight.

We present our system for SemEval-2026 Task 3 on dimensional aspect-based sentiment regression. Our approach combines a hybrid RoBERTa encoder, which jointly predicts sentiment using regression and discretized classification heads, with large language models (LLMs) via prediction-level ensemble learning. The hybrid encoder improves prediction stability by combining continuous and discretized sentiment representations. We further explore in-context learning with LLMs and ridge-regression stacking to combine encoder and LLM predictions. Experimental results on the development set show that ensemble learning significantly improves performance over individual models, achieving substantial reductions in RMSE and improvements in correlation scores. Our findings demonstrate the complementary strengths of encoder-based and LLM-based approaches for dimensional sentiment analysis. Our development code and resources will be shared at https://github.com/aaronlifenghan/ABSentiment

Solving quantifier-free non-linear real arithmetic (NRA) problems is a computationally hard task. To tackle this problem, prior work proposed a promising approach based on gradient descent. In this work, we extend their ideas and combine LLMs and GPU acceleration to obtain an efficient technique. We have implemented our findings in the novel SMT solver GANRA (GPU Accelerated solving of Nonlinear Real Arithmetic problems). We evaluate GANRA on two different NRA benchmarks and demonstrate significant improvements over the previous state of the art. In particular, on the Sturm-MBO benchmark, we can prove satisfiability for more than five times as many instances in less than 1/20th of the previous state-of-the-art runtime.

Rb-82 dynamic cardiac PET imaging is widely used for the clinical diagnosis of coronary artery disease (CAD), but its short half-life results in high noise levels that degrade dynamic frame quality and parametric imaging. The lack of paired clean-noisy training data, rapid tracer kinetics, and frame-dependent noise variations further limit the effectiveness of existing deep learning denoising methods. We propose DECADE (A Temporally-Consistent Unsupervised Diffusion model for Enhanced Rb-82 CArdiac PET DEnoising), an unsupervised diffusion framework that generalizes across early- to late-phase dynamic frames. DECADE incorporates temporal consistency during both training and iterative sampling, using noisy frames as guidance to preserve quantitative accuracy. The method was trained and evaluated on datasets acquired from Siemens Vision 450 and Siemens Biograph Vision Quadra scanners. On the Vision 450 dataset, DECADE consistently produced high-quality dynamic and parametric images with reduced noise while preserving myocardial blood flow (MBF) and myocardial flow reserve (MFR). On the Quadra dataset, using 15%-count images as input and full-count images as reference, DECADE outperformed UNet-based and other diffusion models in image quality and K1/MBF quantification. The proposed framework enables effective unsupervised denoising of Rb-82 dynamic cardiac PET without paired training data, supporting clearer visualization while maintaining quantitative integrity.

A geometric hallucination taxonomy distinguishes three failure types -- center-drift (Type~1), wrong-well convergence (Type~2), and coverage gaps (Type~3) -- by their signatures in embedding cluster space. Prior work found Types~1 and~2 indistinguishable in full-dimensional contextual measurement. We address this through PCA-whitening and eigenspectrum decomposition on GPT-2-small, using multi-run stability analysis (20 seeds) with prompt-level aggregation. Whitening transforms the micro-signal regime into a space where peak cluster alignment (max\_sim) separates Type~2 from Type~3 at Holm-corrected significance, with condition means following the taxonomy's predicted ordering: Type~2 (highest commitment) $>$ Type~1 (intermediate) $>$ Type~3 (lowest). A first directionally stable but underpowered hint of Type~1/2 separation emerges via the same metric, generating a capacity prediction for larger models. Prompt diversification from 15 to 30 prompts per group eliminates a false positive in whitened entropy that appeared robust at the smaller set, demonstrating prompt-set sensitivity in the micro-signal regime. Eigenspectrum decomposition localizes this artifact to the dominant principal components and confirms that Type~1/2 separation does not emerge in any spectral band, rejecting the spectral mixing hypothesis. The contribution is threefold: whitening as preprocessing that reveals cluster commitment as the theoretically correct separating metric, evidence that the Type~1/2 boundary is a capacity limitation rather than a measurement artifact, and a methodological finding about prompt-set fragility in near-saturated representation spaces.

Financial time-series forecasting is a high-stakes problem where regime shifts and shocks make point-accurate yet overconfident models dangerous. We propose Uncertainty-Gated Generative Modeling (UGGM), which treats uncertainty as an internal control signal that gates (i) representation via gated reparameterization, (ii) propagation via similarity and confidence routing, and (iii) generation via uncertainty-controlled predictive distributions, together with uncertainty-driven regularization and calibration to curb miscalibration. Instantiated on Weak Innovation AutoEncoder (WIAE-GPF), our UG-WIAE-GPF significantly improves risk-sensitive forecasting, delivering a 63.5\% MSE reduction on NYISO (0.3508 $\rightarrow$ 0.1281), with improved robustness under shock intervals (mSE: 0.2739 $\rightarrow$ 0.1748).

Current Large Language Models have achieved Olympiad-level logic, yet Vision-Language Models paradoxically falter on elementary spatial tasks like block counting. This capability mismatch reveals a critical ``spatial intelligence gap,'' where models fail to construct coherent 3D mental representations from 2D observations. We uncover this gap via diagnostic analyses showing the bottleneck is a missing view-consistent spatial interface rather than insufficient visual features or weak reasoning. To bridge this, we introduce \textbf{3ViewSense}, a framework that grounds spatial reasoning in Orthographic Views. Drawing on engineering cognition, we propose a ``Simulate-and-Reason'' mechanism that decomposes complex scenes into canonical orthographic projections to resolve geometric ambiguities. By aligning egocentric perceptions with these allocentric references, our method facilitates explicit mental rotation and reconstruction. Empirical results on spatial reasoning benchmarks demonstrate that our method significantly outperforms existing baselines, with consistent gains on occlusion-heavy counting and view-consistent spatial reasoning. The framework also improves the stability and consistency of spatial descriptions, offering a scalable path toward stronger spatial intelligence in multimodal systems.

Network partitions pose fundamental challenges to distributed name resolution in mobile ad-hoc networks (MANETs) and edge computing. Existing solutions either require active coordination that fails to scale, or use unstructured gossip with excessive overhead. We present \textit{Structured Gossip DNS}, exploiting DHT finger tables to achieve partition resilience through \textbf{passive stabilization}. Our approach reduces message complexity from $O(n)$ to $O(n/\log n)$ while maintaining $O(\log^2 n)$ convergence. Unlike active protocols requiring synchronous agreement, our passive approach guarantees eventual consistency through commutative operations that converge regardless of message ordering. The system handles arbitrary concurrent partitions via version vectors, eliminating global coordination and enabling billion-node deployments.

The widespread integration of AI technologies has intensified concerns about fairness and bias, as these systems often perpetuate societal inequalities through flawed data and design choices. While software engineering research has largely concentrated on technical solutions, such as improving datasets and models, the social dynamics that shape AI outcomes remain underexplored. This study investigates the role of team diversity in the development of AI systems. Drawing from the experience of four AI focused teams working in a large software company operating in Brazil and Portugal, and collaborating with global clients, the study explores how diverse teams influence the development of AI systems. Using Grounded Theory, we conducted 25 interviews with software professionals involved in projects spanning domains such as education, energy, accessibility, and facial recognition. Although our study is conducted in an organizational setting, the variety of projects, from regional to multinational, ensures exposure to global development practices and diverse team dynamics, bringing a variety of perspectives into our findings. Our analysis revealed six key roles that team diversity played in AI development: diversifying perspectives for bias identification, bringing empathy to AI development, addressing systemic discrimination, supporting inclusive and participatory decision making, using diversity as a safeguard against bias, and fostering broadened thinking in problem solving. These findings highlight the importance of incorporating diverse perspectives in AI projects and offer practical recommendations for integrating fairness considerations into software development practices.

Federated Graph Learning (FedGL) is vulnerable to malicious attacks, yet developing a truly effective and stealthy attack method remains a significant challenge. Existing attack methods suffer from low attack success rates, high computational costs, and are easily identified and smoothed by defense algorithms. To address these challenges, we propose \textbf{FedShift}, a novel two-stage "Hide and Find" distributed adversarial attack. In the first stage, before FedGL begins, we inject a learnable and hidden "shifter" into part of the training data, which subtly pushes poisoned graph representations toward a target class's decision boundary without crossing it, ensuring attack stealthiness during training. In the second stage, after FedGL is complete, we leverage the global model information and use the hidden shifter as an optimization starting point to efficiently find the adversarial perturbations. During the final attack, we aggregate these perturbations from multiple malicious clients to form the final effective adversarial sample and trigger the attack. Extensive experiments on six large-scale datasets demonstrate that our method achieves the highest attack effectiveness compared to existing advanced attack methods. In particular, our attack can effectively evade 3 mainstream robust federated learning defense algorithms and converges with a time cost reduction of over 90\%, highlighting its exceptional stealthiness, robustness, and efficiency.

This work investigated the capabilities of different models, including the Llama-3 series of models and CHATGPT, with different forms of expression in solving discrete optimization problems by testing natural language datasets. In contrast to formal datasets with a limited scope of parameters, our dataset included a variety of problem types in discrete optimization problems and featured a wide range of parameter magnitudes, including instances with large parameter sets, integrated with augmented data. It aimed to (1) provide an overview of LLMs' ability in large-scale problems, (2) offer suggestions to those who want to solve discrete optimization problems automatically, and (3) regard the performance as a benchmark for future research. These datasets included original, expanded and augmented datasets. Among these three datasets, the original and augmented ones aimed for evaluation while the expanded one may help finetune a new model. In the experiment, comparisons were made between strong and week models, CoT methods and No-CoT methods on various datasets. The result showed that stronger model performed better reasonably. Contrary to general agreement, it also showed that CoT technique was not always effective regarding the capability of models and disordered datasets improved performance of models on easy to-understand problems, even though they were sometimes with high variance, a manifestation of instability. Therefore, for those who seek to enhance the automatic resolution of discrete optimization problems, it is recommended to consult the results, including the line charts presented in the Appendix, as well as the conclusions drawn in this study for relevant suggestions.

Large language models (LLMs) such as GPT and Gemini have demonstrated remarkable capabilities in contextual understanding and reasoning. The strong performance of LLMs has sparked growing interest in leveraging them to automate tasks traditionally dependent on human expertise. Recently, LLMs have been integrated into intelligent agents capable of operating structural analysis software (e.g., OpenSees) to construct structural models and perform analyses. However, existing LLMs are limited in handling multi-step structural modeling due to frequent hallucinations and error accumulation during long-sequence operations. To this end, this study presents a novel multi-agent architecture to automate the structural modeling and analysis using OpenSeesPy. First, problem analysis and construction planning agents extract key parameters from user descriptions and formulate a stepwise modeling plan. Node and element agents then operate in parallel to assemble the frame geometry, followed by a load assignment agent. The resulting geometric and load information is translated into executable OpenSeesPy scripts by code translation agents. The proposed architecture is evaluated on a benchmark of 20 frame problems over ten repeated trials, achieving 100% accuracy in 18 cases and 90% in the remaining two. The architecture also significantly improves computational efficiency and demonstrates scalability to larger structural systems.

Brand advertising plays a critical role in building long-term consumer awareness and loyalty, making it a key objective for advertisers across digital platforms. Although real-time bidding has been extensively studied, there is limited literature on algorithms specifically tailored for brand auction ads that fully leverage their unique characteristics. In this paper, we propose a lightweight Model Predictive Control (MPC) framework designed for brand advertising campaigns, exploiting the inherent attributes of brand ads -- such as stable user engagement patterns and fast feedback loops -- to simplify modeling and improve efficiency. Our approach utilizes online isotonic regression to construct monotonic bid-to-spend and bid-to-conversion models directly from streaming data, eliminating the need for complex machine learning models. The algorithm operates fully online with low computational overhead, making it highly practical for real-world deployment. Simulation results demonstrate that our approach significantly improves spend efficiency and cost control compared to baseline strategies, providing a scalable and easily implementable solution for modern brand advertising platforms.

We test whether LLMs show robust decision biases. Treating models as participants in two-arm bandits, we ran 20000 trials per condition across four decoding configurations. Under symmetric rewards, models amplified positional order into stubborn one-arm policies. Under asymmetric rewards, they exploited rigidly yet underperformed an oracle and rarely re-checked. The observed patterns were consistent across manipulations of temperature and top-p, with top-k held at the provider default, indicating that the qualitative behaviours are robust to the two decoding knobs typically available to practitioners. Crucially, moving beyond descriptive metrics to computational modelling, a hierarchical Rescorla-Wagner-softmax fit revealed the underlying strategies: low learning rates and very high inverse temperatures, which together explain both noise-to-bias amplification and rigid exploitation. These results position minimal bandits as a tractable probe of LLM decision tendencies and motivate hypotheses about how such biases could shape human-AI interaction.

Unlike the predictable scaling laws in natural language processing and computer vision, protein language models (PLMs) scale poorly: for many tasks, models within the same family plateau or even decrease in performance, with mid-sized models often outperforming the largest in the family. We introduce Reverse Distillation, a principled framework that decomposes large PLM representations into orthogonal subspaces guided by smaller models of the same family. The resulting embeddings have a nested, Matryoshka-style structure: the first k dimensions of a larger model's embedding are exactly the representation from the smaller model. This ensures that larger reverse-distilled models consistently outperform smaller ones. A motivating intuition is that smaller models, constrained by capacity, preferentially encode broadly-shared protein features. Reverse distillation isolates these shared features and orthogonally extracts additional contributions from larger models, preventing interference between the two. On ProteinGym benchmarks, reverse-distilled ESM-2 variants outperform their respective baselines at the same embedding dimensionality, with the reverse-distilled 15 billion parameter model achieving the strongest performance. Our framework is generalizable to any model family where scaling challenges persist. Code and trained models are available at https://github.com/rohitsinghlab/plm_reverse_distillation.

Mental healthcare services in the UK lack tools and resources to address the cultural needs of Muslim women, often leaving them feeling as though their values are pathologised and limiting trust and engagement [1]. Despite growing awareness of cultural competency, few interventions integrate Islamic frameworks into therapeutic support. This report investigates the design and evaluation of YAQIN, a co-designed AI-based application supporting culturally and faith-sensitive mental health engagement for Muslim women. With almost 1.9 million Muslim women in England in 2021, YAQIN responds to a gap in care [2]. It leverages AIś anonymity and continuous support through a faith-aware chatbot and guided journaling tool grounded in user-centred design and Islamic psychology. The YAQIN design research methodology comprised three stages: contextual investigation and literature review, user research with N=14 stakeholders including Muslim women and mental health experts, and prototype development informed by deductive thematic analysis, personas, journey maps, and design specifications. Evaluation involved a co-designed user study with five participants: four Muslim women and one mental health expert who reviewed therapeutic alignment and cultural sensitivity after using the chatbot prototype. Feedback focused on tone, faith relevance, emotional resonance, and the Retrieval-Augmented Generation pipeline enabling contextual continuity. Participants highlighted YAQINś ability to bridge cultural gaps in trust and therapeutic confidence. Feedback included suggestions of including linguistic diversity and routine-based guidance. This project demonstrates how culturally sensitive AI can improve mental healthcare accessibility and trust for marginalised communities and highlights the potential of faith-integrated technology in healthcare innovation.

Voice interfaces are quickly becoming a common way for people to interact with AI systems. This also brings new security risks, such as prompt injection, social engineering, and harmful voice commands. Traditional security methods rely on converting speech to text and then filtering that text, which introduces delays and can ignore important audio cues. This paper introduces VoiceSHIELD-Small, a lightweight model that works in real time. It can transcribe speech and detect whether it is safe or harmful, all in one step. Built on OpenAI's Whisper-small encoder, VoiceSHIELD adds a mean-pooling layer and a simple classification head. It takes just 90-120 milliseconds to classify audio on mid-tier GPUs, while transcription happens at the same time. Tested on a balanced set of 947 audio clips, the model achieved 99.16 percent accuracy and an F1 score of 0.9865. At the default setting, it missed 2.33 percent of harmful inputs. Cross-validation showed consistent performance (F1 standard deviation = 0.0026). The paper also covers the model's design, training data, performance trade-offs, and responsible use guidelines. VoiceSHIELD is released under the MIT license to encourage further research and adoption in voice AI security.

Automated design of multi-agent interactions with desirable equilibrium outcomes is inherently difficult due to the computational hardness, non-uniqueness, and instability of the resulting equilibria. In this work, we propose the use of game-agnostic differentiable equilibrium blocks (DEBs) as modules in a novel, differentiable framework to address a wide variety of incentive design problems from economics and computer science. We call this framework deep incentive design (DID). To validate our approach, we examine three diverse, challenging incentive design tasks: contract design, machine scheduling, and inverse equilibrium problems. For each task, we train a single neural network using a unified pipeline and DEB. This architecture solves the full distribution of problem instances, parameterized by a context, handling all games across a wide range of scales (from two to sixteen actions per player).

Greedy algorithms are central to sparse approximation and stage-wise learning methods such as matching pursuit and boosting. It is known that the Power-Relaxed Greedy Algorithm with step sizes $m^{-α}$ may fail to converge when $α>1$ in general Hilbert spaces. In this work, we revisit this phenomenon from a sparse learning perspective. We study realizable regression problems with controlled feature coherence and derive explicit lower bounds on the residual norm, showing that over-decaying step-size schedules induce structural stagnation even in low-dimensional sparse settings. Numerical experiments confirm the theoretical predictions and illustrate the role of feature coherence. Our results provide insight into step-size design in greedy sparse learning.

While few-step generative models have enabled powerful image and video generation at significantly lower cost, generic reinforcement learning (RL) paradigms for few-step models remain an unsolved problem. Existing RL approaches for few-step diffusion models strongly rely on back-propagating through differentiable reward models, thereby excluding the majority of important real-world reward signals, e.g., non-differentiable rewards such as humans' binary likeness, object counts, etc. To properly incorporate non-differentiable rewards to improve few-step generative models, we introduce TDM-R1, a novel reinforcement learning paradigm built upon a leading few-step model, Trajectory Distribution Matching (TDM). TDM-R1 decouples the learning process into surrogate reward learning and generator learning. Furthermore, we developed practical methods to obtain per-step reward signals along the deterministic generation trajectory of TDM, resulting in a unified RL post-training method that significantly improves few-step models' ability with generic rewards. We conduct extensive experiments ranging from text-rendering, visual quality, and preference alignment. All results demonstrate that TDM-R1 is a powerful reinforcement learning paradigm for few-step text-to-image models, achieving state-of-the-art reinforcement learning performances on both in-domain and out-of-domain metrics. Furthermore, TDM-R1 also scales effectively to the recent strong Z-Image model, consistently outperforming both its 100-NFE and few-step variants with only 4 NFEs. Project page: https://github.com/Luo-Yihong/TDM-R1

We study infinite-horizon Constrained Markov Decision Processes (CMDPs) with general policy parameterizations and multi-layer neural network critics. Existing theoretical analyses for constrained reinforcement learning largely rely on tabular policies or linear critics, which limits their applicability to high-dimensional and continuous control problems. We propose a primal-dual natural actor-critic algorithm that integrates neural critic estimation with natural policy gradient updates and leverages Neural Tangent Kernel (NTK) theory to control function-approximation error under Markovian sampling, without requiring access to mixing-time oracles. We establish global convergence and cumulative constraint violation rates of $\tilde{\mathcal{O}}(T^-1/4)$ up to approximation errors induced by the policy and critic classes. Our results provide the first such guarantees for CMDPs with general policies and multi-layer neural critics, substantially extending the theoretical foundations of actor-critic methods beyond the linear-critic regime.

In bandwidth-limited online video streaming, videos are usually downsampled and compressed. Although recent online video super-resolution (online VSR) approaches achieve promising results, they are still compute-intensive and fall short of real-time processing at higher resolutions, due to complex motion estimation for alignment and redundant processing of consecutive frames. To address these issues, we propose a compressed-domain-aware network (CDA-VSR) for online VSR, which utilizes compressed-domain information, including motion vectors, residual maps, and frame types to balance quality and efficiency. Specifically, we propose a motion-vector-guided deformable alignment module that uses motion vectors for coarse warping and learns only local residual offsets for fine-tuned adjustments, thereby maintaining accuracy while reducing computation. Then, we utilize a residual map gated fusion module to derive spatial weights from residual maps, suppressing mismatched regions and emphasizing reliable details. Further, we design a frame-type-aware reconstruction module for adaptive compute allocation across frame types, balancing accuracy and efficiency. On the REDS4 dataset, our CDA-VSR surpasses the state-of-the-art method TMP, with a maximum PSNR improvement of 0.13 dB while delivering more than double the inference speed. The code will be released at https://github.com/sspBIT/CDA-VSR.

Scaling Mixture-of-Experts (MoE) training introduces systems challenges absent in dense models. Because each token activates only a subset of experts, this sparsity allows total parameters to grow much faster than per-token computation, creating coupled constraints across memory, communication, and computation. Optimizing one dimension often shifts pressure to another, demanding co-design across the full system stack. We address these challenges for MoE training through integrated optimizations spanning memory (fine-grained recomputation, offloading, etc.), communication (optimized dispatchers, overlapping, etc.), and computation (Grouped GEMM, fusions, CUDA Graphs, etc.). The framework also provides Parallel Folding for flexible multi-dimensional parallelism, low-precision training support for FP8 and NVFP4, and efficient long-context training. On NVIDIA GB300 and GB200, it achieves 1,233/1,048 TFLOPS/GPU for DeepSeek-V3-685B and 974/919 TFLOPS/GPU for Qwen3-235B. As a performant, scalable, and production-ready open-source solution, it has been used across academia and industry for training MoE models ranging from billions to trillions of parameters on clusters scaling up to thousands of GPUs. This report explains how these techniques work, their trade-offs, and their interactions at the systems level, providing practical guidance for scaling MoE models with Megatron Core.

Modern applications process massive data volumes that overwhelm the storage and retrieval capabilities of memory systems, making memory the primary performance and energy-efficiency bottleneck of computing systems. Although many microarchitectural techniques attempt to hide or tolerate long memory access latency, rapidly growing data footprints continue to outpace technology scaling, requiring more effective solutions. This dissertation shows that modern processors observe large amounts of application and system data during execution, yet many microarchitectural mechanisms make decisions largely independent of this information. Through four case studies, we demonstrate that such data-agnostic design leads to substantial missed opportunities for improving performance and energy efficiency. To address this limitation, this dissertation advocates shifting microarchitecture design from data-agnostic to data-informed. We propose mechanisms that (1) learn policies from observed execution behavior (data-driven design) and (2) exploit semantic characteristics of application data (data-aware design). We apply lightweight machine learning techniques and previously underexplored data characteristics across four processor components: a reinforcement learning-based hardware data prefetcher that learns memory access patterns online; a perceptron predictor that identifies memory requests likely to access off-chip memory; a reinforcement learning mechanism that coordinates data prefetching and off-chip prediction; and a mechanism that exploits repeatability in memory addresses and loaded values to eliminate predictable load instructions. Our extensive evaluation shows that the proposed techniques significantly improve performance and energy efficiency compared to prior state-of-the-art approaches.

This paper introduces a novel class of model-driven evolutionary frameworks for near-field multi-source localization, addressing the major limitations of grid-based subspace methods such as MUSIC and data-dependent deep learning approaches. To this end, we develop two complementary evolutionary localization frameworks that operate directly on the continuous spherical-wave signal model and support arbitrary array geometries without requiring labeled data, discretized angle--range grids, or architectural constraints. The first framework, termed NEar-field MultimOdal DE (NEMO-DE) associates each individual in the evolutionary population to a single source and optimizes a residual least-squares objective in a sequential manner, updating the data residual and enforcing spatial separation to estimate multiple source locations. To overcome the limitation of NEMO-DE under large power imbalances among the sources, we propose the second framework, named NEar-field Eigen-subspace Fitting DE (NEEF-DE), which jointly encodes all source locations and minimizes a subspace-fitting criterion that aligns a model-based array response subspace with the received signal subspace. Although the proposed frameworks are algorithm-agnostic and compatible with various evolutionary optimizers, differential evolution (DE) is adopted in this work as a representative search strategy due to its simplicity, robustness, and strong empirical performance. We provide extensive numerical experiments to evaluate the performance of the proposed frameworks under different system configurations. This work establishes evolutionary computation as a powerful and flexible paradigm for model-based near-field localization, paving the way for future innovations in this domain.

The Consistency property between surrogate losses and evaluation metrics has been extensively studied to ensure that minimizing a loss leads to metric optimality. However, the direct relationship between different evaluation metrics remains significantly underexplored. This theoretical gap results in the "Metric Mismatch" frequently observed in industrial applications, where gains in offline validation metrics fail to translate into online performance. To bridge this disconnection, this paper proposes a unified theoretical framework designed to quantify the relationships between metrics. We categorize metrics into different classes to facilitate a comparative analysis across different mathematical forms and interrogates these relationships through Bayes-Optimal Set and Regret Transfer. Through this framework, we provide a new perspective on identifying the structural asymmetry in regret transfer, enabling the design of evaluation systems that are theoretically guaranteed to align offline improvements with online objectives.

Large language model (LLM) agents increasingly operate in settings where a single context window is far too small to capture what has happened, what was learned, and what should not be repeated. Memory -- the ability to persist, organize, and selectively recall information across interactions -- is what turns a stateless text generator into a genuinely adaptive agent. This survey offers a structured account of how memory is designed, implemented, and evaluated in modern LLM-based agents, covering work from 2022 through early 2026. We formalize agent memory as a \emph{write--manage--read} loop tightly coupled with perception and action, then introduce a three-dimensional taxonomy spanning temporal scope, representational substrate, and control policy. Five mechanism families are examined in depth: context-resident compression, retrieval-augmented stores, reflective self-improvement, hierarchical virtual context, and policy-learned management. On the evaluation side, we trace the shift from static recall benchmarks to multi-session agentic tests that interleave memory with decision-making, analyzing four recent benchmarks that expose stubborn gaps in current systems. We also survey applications where memory is the differentiating factor -- personal assistants, coding agents, open-world games, scientific reasoning, and multi-agent teamwork -- and address the engineering realities of write-path filtering, contradiction handling, latency budgets, and privacy governance. The paper closes with open challenges: continual consolidation, causally grounded retrieval, trustworthy reflection, learned forgetting, and multimodal embodied memory.

We investigate how code obfuscation influences human understanding of programs through an output-prediction task. To study this effect, we construct multiple levels of obfuscation, ranging from unobfuscated code to transformations involving identifier renaming, adversarially misleading identifiers, control-flow modifications, and combinations of these techniques. These transformations are applied to function-level programs written in Python and JavaScript. Participants were asked to predict program outputs while we recorded correctness, response time, and self-reported programming experience. Our results show that obfuscation generally increases the time required to reason about code and tends to reduce prediction accuracy. However, the relationship between obfuscation strength and performance is not strictly monotonic and varies across programming languages. JavaScript exhibits the expected pattern of increasing difficulty with stronger obfuscation, whereas Python displays a more complex trend in which certain renaming transformations can perform comparably to, or occasionally better than, the unobfuscated baseline. Response-time analyses further suggest that obfuscation shifts participants away from rapid, heuristic reasoning toward slower and more deliberate reasoning processes. Performance appears highest within a moderate range of response times, indicating that careful deliberation can improve accuracy, while extremely long response times often correspond to confusion. Finally, programming experience predicts performance primarily within a given language, with limited transfer across languages, suggesting that obfuscation challenges language-specific familiarity more than general programming ability.

Na-ion batteries have emerged as viable candidates for large-scale energy storage applica- tions due to resource abundance and cost advantages. The constraints imposed on their performance and durability, for instance, by complex phase transformations in positive electrode materials during electrochemical cycling, can be addressed and are thus not detrimental to their development. However, diffusion-limited Na-ion transport can drive spatially heterogeneous phase nucleation and propagation, leading to multiphase coexis- tence and locally non-uniform electrochemical activity, generating complex reaction path- ways that challenge both mechanistic understanding and predictive material optimization. These challenges can be addressed by investigating single-crystalline regions of materials, i.e. down to the scale of individual particles, although such analyses are often constrained by energetically and/or spatially sparse hyperspectral datasets. Here, we developed an AI-driven method to process hyperspectral data under sparse sampling conditions and generate multiphase maps with nanometer-scale resolution over a micrometer-scale field of view. We applied this processing on scanning transmission X-ray microscopy (STXM) data to determine the distribution and coexistence of phases in individual particles of NaxV2(PO4)2F3 cathode materials, at different states of charge. The methodology relies on a workflow which combines a Gaussian mixture variational autoencoder (GMVAE) algorithm with the Pearson corre- lation coefficient to identify the sodium content and map their spatial distribution. Our approach reveals nanoscale phase heterogeneity and evolution within individual particles, and improves the reliability of phase detection by identifying ambiguity zones, false assign- ments, and transition phases localized at grain boundaries.

Reflective appearance, especially strong and typically near-field specular reflections, poses a fundamental challenge for accurate surface reconstruction and novel view synthesis. Existing Gaussian splatting methods either fail to model near-field specular reflections or rely on explicit ray tracing at substantial computational cost. We present Ref-DGS, a reflective dual Gaussian splatting framework that addresses this trade-off by decoupling surface reconstruction from specular reflection within an efficient rasterization-based pipeline. Ref-DGS introduces a dual Gaussian scene representation consisting of geometry Gaussians and complementary local reflection Gaussians that capture near-field specular interactions without explicit ray tracing, along with a global environment reflection field for modeling far-field specular reflections. To predict specular radiance, we further propose a lightweight, physically-aware adaptive mixing shader that fuses global and local reflection features. Experiments demonstrate that Ref-DGS achieves state-of-the-art performance on reflective scenes while training substantially faster than ray-based Gaussian methods.

Partial differential equations (PDEs) govern physical phenomena across the full range of scientific scales, yet their computational solution remains one of the defining challenges of modern science. This critical review examines two mature but epistemologically distinct paradigms for PDE solution, classical numerical methods and machine learning approaches, through a unified evaluative framework organized around six fundamental computational challenges. Classical methods are assessed for their structure-preserving properties, rigorous convergence theory, and scalable solver design; their persistent limitations in high-dimensional and geometrically complex settings are characterized precisely. Machine learning approaches are introduced under a taxonomy organized by the degree to which physical knowledge is incorporated and subjected to the same critical evaluation applied to classical methods. Classical methods are deductive -- errors are bounded by quantities derivable from PDE structure and discretization parameters -- while machine learning methods are inductive -- accuracy depends on statistical proximity to the training distribution. This epistemological distinction is the primary criterion governing responsible method selection. We identify three genuine complementarities between the paradigms and develop principles for hybrid design, including a framework for the structure inheritance problem that addresses when classical guarantees propagate through hybrid couplings, and an error budget decomposition that separates discretization, neural approximation, and coupling contributions. We further assess emerging frontiers, including foundation models, differentiable programming, quantum algorithms, and exascale co-design, evaluating each against the structural constraints that determine whether current barriers are fundamental or contingent on engineering progress.

Federated Composite Optimization (FCO) has emerged as a promising framework for training models with structural constraints (e.g., sparsity) in distributed edge networks. However, simultaneously achieving communication efficiency and convergence robustness remains a significant challenge, particularly when dealing with non-smooth regularizers, statistical heterogeneity, and the restrictions of biased compression. To address these issues, we propose FedCEF (Federated Composite Error Feedback), a novel algorithm tailored for non-convex FCO. FedCEF introduces a decoupled proximal update scheme that separates the proximal operator from communication, enabling clients to handle non-smooth terms locally while transmitting compressed information. To mitigate the noise from aggressive quantization and the bias from non-IID data, FedCEF integrates a rigorous error feedback mechanism with control variates. Furthermore, we design a communication-efficient pre-proximal downlink strategy that allows clients to exactly reconstruct global control variables without explicit transmission. We theoretically establish that FedCEF achieves sublinear convergence to a bounded residual error under general non-convexity, which is controllable via the step size and batch size. Extensive experiments on real datasets validate FedCEF maintains competitive model accuracy even under extreme compression ratios (e.g., 1%), significantly reducing the total communication volume compared to uncompressed baselines.

Recent advances in Visual-Language-Action (VLA) models have shown promising potential for robotic manipulation tasks. However, real-world robotic tasks often involve long-horizon, multi-step problem-solving and require generalization for continual skill acquisition, extending beyond single actions or skills. These challenges present significant barriers for existing VLA models, which use monolithic action decoders trained on aggregated data, resulting in poor scalability. To address these challenges, we propose AtomicVLA, a unified planning-and-execution framework that jointly generates task-level plans, atomic skill abstractions, and fine-grained actions. AtomicVLA constructs a scalable atomic skill library through a Skill-Guided Mixture-of-Experts (SG-MoE), where each expert specializes in mastering generic yet precise atomic skills. Furthermore, we introduce a flexible routing encoder that automatically assigns dedicated atomic experts to new skills, enabling continual learning. We validate our approach through extensive experiments. In simulation, AtomicVLA outperforms $π_{0}$ by 2.4\% on LIBERO, 10\% on LIBERO-LONG, and outperforms $π_{0}$ and $π_{0.5}$ by 0.22 and 0.25 in average task length on CALVIN. Additionally, our AtomicVLA consistently surpasses baselines by 18.3\% and 21\% in real-world long-horizon tasks and continual learning. These results highlight the effectiveness of atomic skill abstraction and dynamic expert composition for long-horizon and lifelong robotic tasks. The project page is \href{https://zhanglk9.github.io/atomicvla-web/}{here}.

Efficient luggage trolley management is critical for reducing congestion and ensuring asset availability in modern airports. Automated detection systems face two main challenges. First, strict security and privacy regulations limit large-scale data collection. Second, existing public datasets lack the diversity, scale, and annotation quality needed to handle dense, overlapping trolley arrangements typical of real-world operations. To address these limitations, we introduce a synthetic data generation pipeline based on a high-fidelity Digital Twin of Algiers International Airport using NVIDIA Omniverse. The pipeline produces richly annotated data with oriented bounding boxes, capturing complex trolley formations, including tightly nested chains. We evaluate YOLO-OBB using five training strategies: real-only, synthetic-only, linear probing, full fine-tuning, and mixed training. This allows us to assess how synthetic data can complement limited real-world annotations. Our results show that mixed training with synthetic data and only 40 percent of real annotations matches or exceeds the full real-data baseline, achieving 0.94 mAP@50 and 0.77 mAP@50-95, while reducing annotation effort by 25 to 35 percent. Multi-seed experiments confirm strong reproducibility with a standard deviation below 0.01 on mAP@50, demonstrating the practical effectiveness of synthetic data for automated trolley detection.

Large language models (LLMs) with reasoning abilities have demonstrated growing promise for tackling complex scientific problems. Yet such tasks are inherently domain-specific, unbounded and open-ended, demanding exploration across vast and flexible solution spaces. Existing approaches, whether purely learning-based or reliant on carefully designed workflows, often suffer from limited exploration efficiency and poor generalization. To overcome these challenges, we present HELIX -- a Hierarchical Evolutionary reinforcement Learning framework with In-context eXperiences. HELIX introduces two key novelties: (i) a diverse yet high-quality pool of candidate solutions that broadens exploration through in-context learning, and (ii) reinforcement learning for iterative policy refinement that progressively elevates solution quality. This synergy enables the discovery of more advanced solutions. On the circle packing task, HELIX achieves state-of-the-art result with a sum of radii of 2.63598308 using only a 14B model. Across standard machine learning benchmarks, HELIX further surpasses GPT-4o with a carefully engineered pipeline, delivering an average F1 improvement of 5.95 points on the Adult and Bank Marketing datasets.

Data-driven joint-moment predictors offer a scalable alternative to laboratory-based inverse-dynamics pipelines for biomechanics estimation and exoskeleton control. Meanwhile, physics-based reinforcement learning (RL) enables simulation-trained controllers to learn dynamics-aware assistance strategies without extensive human experimentation. However, quantitative verification of simulation-trained exoskeleton torque predictors, and their impact on human joint power injection, remains limited. This paper presents (1) an RL framework to learn exoskeleton assistance policies that reduce biological joint moments, and (2) a validation pipeline that verifies the trained control networks using an open-source gait dataset through inference and comparison with biological joint moments. Simulation-trained multilayer perceptron (MLP) controllers are developed for level-ground and ramp walking, mapping short-horizon histories of bilateral hip and knee kinematics to normalized assistance torques. Results show that predicted assistance preserves task-intensity trends across speeds and inclines. Agreement is particularly strong at the hip, with cross-correlation coefficients reaching 0.94 at 1.8 m/s and 0.98 during 5° decline walking, demonstrating near-matched temporal structure. Discrepancies increase at higher speeds and steeper inclines, especially at the knee, and are more pronounced in joint power comparisons. Delay tuning biases assistance toward greater positive power injection; modest timing shifts increase positive power and improve agreement in specific gait intervals. Together, these results establish a quantitative validation framework for simulation-trained exoskeleton controllers, demonstrate strong sim-to-data consistency at the torque level, and highlight both the promise and the remaining challenges for sim-to-real transfer.

Diffusion models have revolutionized generative AI, with their inherent capacity to generate highly realistic state-of-the-art synthetic data. However, these models employ an iterative denoising process over computationally intensive layers such as UNets and attention mechanisms. This results in high inference energy on conventional electronic platforms, and thus, there is an emerging need to accelerate these models in a sustainable manner. To address this challenge, we present a novel silicon photonics-based accelerator for diffusion models. Experimental evaluations demonstrate that our photonic accelerator achieves at least 3x better energy efficiency and 5.5x throughput improvement compared to state-of-the-art diffusion model accelerators.

Containerized microservices are widely adopted for latency-sensitive and compute-intensive applications, with Kubernetes (K8s) as the dominant orchestration platform. However, automating the deployment and management of multi-service applications remains challenging, particularly in heterogeneous Edge-Cloud environments. This paper evaluates the CODECO toolkit, an open-source framework designed to enhance container orchestration across distributed infrastructures. We compare CODECO with baseline K8s workflows using three key performance indicators: deployment time, level of manual intervention, and runtime performance with resource utilization. Experiments across diverse hardware platforms (ARM, AMD, RPi) and K8s distributions, including lightweight variants such as k3s, demonstrate that CODECO substantially reduces manual effort while maintaining competitive performance and acceptable overhead. These results validate CODECO as an effective solution for Edge-Cloud orchestration and highlight its potential to improve the flexibility and intelligence of K8s-based deployments.

Effective exoskeleton assistance requires co-adaptation: as the device alters joint dynamics, the user reorganizes neuromuscular coordination, creating a non-stationary learning problem. Most learning-based approaches do not explicitly account for the sequential nature of human motor adaptation, leading to training instability and poorly timed assistance. We propose Staged Multi-Agent Training (SMAT), a four-stage curriculum designed to mirror how users naturally acclimate to a wearable device. In SMAT, a musculoskeletal human actor and a bilateral hip exoskeleton actor are trained progressively: the human first learns unassisted gait, then adapts to the added device mass; the exoskeleton subsequently learns a positive assistance pattern against a stabilized human policy, and finally both agents co-adapt with full torque capacity and bidirectional feedback. We implement SMAT in the MyoAssist simulation environment using a 26-muscle lower-limb model and an attached hip exoskeleton. Our musculoskeletal simulations demonstrate that the learned exoskeleton control policy produces an average 10.1% reduction in hip muscle activation relative to the no-assist condition. We validated the learned controller in an offline setting using open-source gait data, then deployed it to a physical hip exoskeleton for treadmill experiments with five subjects. The resulting policy delivers consistent assistance and predominantly positive mechanical power without the need for any explicitly imposed timing shift (mean positive power: 13.6 W at 6 Nm RMS torque to 23.8 W at 9.3 Nm RMS torque, with minimal negative power) consistently across all subjects without subject-specific retraining.

Modern visual generative models acquire rich visual knowledge through large-scale training, yet existing visual representations (such as pixels, latents, or tokens) remain external to the model and cannot directly exploit this knowledge for compact storage or reuse. In this work, we introduce a new visual representation framework that encodes a signal as a function, which is parametrized by low-rank adaptations attached to a frozen visual generative model. Such implicit representations of visual signals, \textit{e.g.}, an 81-frame video, can further be hashed into a single compact vector, achieving strong perceptual video compression at extremely low bitrates. Beyond basic compression, the functional nature of this representation enables inference-time scaling and control, allowing additional refinement on the compression performance. More broadly, as the implicit representations directly act as a function of the generation process, this suggests a unified framework bridging visual compression and generation.

Retrieval-augmented generation (RAG) systems that answer questions from document collections face compounding difficulties when high-precision citations are required: flat chunking strategies sacrifice document structure, single-query formulations miss relevant passages through vocabulary mismatch, and single-pass inference produces stochastic answers that vary in both content and citation selection. We present KohakuRAG, a hierarchical RAG framework that preserves document structure through a four-level tree representation (document $\rightarrow$ section $\rightarrow$ paragraph $\rightarrow$ sentence) with bottom-up embedding aggregation, improves retrieval coverage through an LLM-powered query planner with cross-query reranking, and stabilizes answers through ensemble inference with abstention-aware voting. We evaluate on the WattBot 2025 Challenge, a benchmark requiring systems to answer technical questions from 32 documents with $\pm$0.1% numeric tolerance and exact source attribution. KohakuRAG achieves first place on both public and private leaderboards (final score 0.861), as the only team to maintain the top position across both evaluation partitions. Ablation studies reveal that prompt ordering (+80% relative), retry mechanisms (+69%), and ensemble voting with blank filtering (+1.2pp) each contribute substantially, while hierarchical dense retrieval alone matches hybrid sparse-dense approaches (BM25 adds only +3.1pp). We release KohakuRAG as open-source software at https://github.com/KohakuBlueleaf/KohakuRAG.

Autoscaling in cloud-native platforms like Kubernetes is reactive and metric-driven, leading to a strategic void problem. This comes from the decoupling of higher-level business policies from lower-level resource provisioning. The strategic void, coupled with a fragmented coordination of pod and node scaling, can lead to significant resource waste and performance degradation under dynamic workloads. In this paper, we present MAS-H2, a new hierarchical multi-agent system that addresses the challenges of autonomic cloud resource management with a complete end-to-end solution. MAS-H2 systematically decomposes the control problem into three layers: a Strategic Agent that formalises business policies (e.g., cost vs. performance) into a global utility function; Planning Agents that produce a joint, proactive scaling plan for pods and nodes with time-series forecasting; and Execution Agents that execute the scaling plan. We built and tested a MAS-H2 prototype as a Kubernetes Operator on Google Kubernetes Engine (GKE) to benchmark it against the native Horizontal Pod Autoscaler (HPA) and Cluster Autoscaler (CA) baselines under two realistic, spiky, and stress-inducing workload scenarios. The results show that the MAS-H2 system maintained application CPU usage under 40% for predictable Heartbeat workloads. This resulted in over 50% less sustained CPU stress than the native HPA baseline, which typically operated above 80%. The MAS-H2 system demonstrated proactive planning in a volatile Chaotic Flash Sale scenario by filtering transient noise and deploying more replicas compared to HPA. It reduced peak CPU load by 55% without under-provisioning. Beyond performance, MAS-H2 seamlessly performed a zero-downtime strategic migration between two cost- and performance-optimised infrastructures.

Interpretable machine learning is essential in high-stakes domains where decision-making requires accountability, transparency, and trust. While rule-based models offer global and exact interpretability, learning rule sets that simultaneously achieve high predictive performance and low, human-understandable complexity remains challenging. To address this, we introduce TT-Sparse, a flexible neural building block that leverages differentiable truth tables as nodes to learn sparse, effective connections. A key contribution of our approach is a new soft TopK operator with straight-through estimation for learning discrete, cardinality-constrained feature selection in an end-to-end differentiable manner. Crucially, the forward pass remains sparse, enabling efficient computation and exact symbolic rule extraction. As a result, each node (and the entire model) can be transformed exactly into compact, globally interpretable DNF/CNF Boolean formulas via Quine-McCluskey minimization. Extensive empirical results across 28 datasets spanning binary, multiclass, and regression tasks show that the learned sparse rules exhibit superior predictive performance with lower complexity compared to existing state-of-the-art methods.

Many previous works in spike sorting study spike classification and compression independently. In this paper, a novel algorithm is proposed called MetaSort to address these two problems. To deal with compression, a novel adaptive level crossing algorithm is proposed to approximate spike shapes with high fidelity. Meanwhile, the latent feature representation is used to handle the classification problem. Besides, to guarantee MetaSort is robust and discriminative, the geometric information of data is exploited simultaneously in the proposed framework by meta-transfer learning. Empirical experiments with in-vivo spike data demonstrate that MetaSort delivers promising performance, highlighting its potential and motivating continued development toward an ultra-low-power, on-chip implementation.

Speech language models (SLMs) have significantly extended the interactive capability of text-based Large Language Models (LLMs) by incorporating paralinguistic information. For more realistic interactive experience with customized styles, current SLMs have managed to interpret and control speaking style intensity from user prompts during the dialogue process. However, there remains a lack of systematic benchmarks that quantifies and evaluates the style intensity control ability in conversations. In this paper, we propose StyleBench, a multi-turn dialogue benchmark for comprehensively evaluating the style intensity control ability across four dimensions: emotion, speed, volume, and pitch. Our results reveal the performance gaps between leading SLMs and omni language models (OLMs), suggesting the underlying reasons and promising approaches for future exploration.

Chain-of-thought (CoT) improves reasoning reliability but increases token cost, motivating post-training compression of explicit reasoning traces. However, the shortest sufficient reasoning is not universal: it depends on difficulty, model capacity, and training state, making fixed length targets brittle. In practice, naive RL-based compression can also undesirably shorten the user-facing answer, because a single completion-level learning signal leaks across the think/answer boundary. We propose Difficulty-Scaled Segment-Wise GRPO (DSS-GRPO), which decomposes returns into think and answer components, computes group-relative advantages per segment, and routes them with hard token masks so compression updates act only on think while answer alignment acts only on answer. DSS-GRPO uses prompt-wise within-group shaping and difficulty-aware scaling to encourage concise reasoning without collapsing answer behavior.

Despite the rapid progress of Large Vision-Language Models (LVLMs), the integration of visual modalities introduces new safety vulnerabilities that adversaries can exploit to elicit biased or malicious outputs. In this paper, we demonstrate an underexplored vulnerability via semantic slot filling, where LVLMs complete missing slot values with unsafe content even when the slot types are deliberately crafted to appear benign. Building on this finding, we propose StructAttack, a simple yet effective single-query jailbreak framework under black-box settings. StructAttack decomposes a harmful query into a central topic and a set of benign-looking slot types, then embeds them as structured visual prompts (e.g., mind maps, tables, or sunburst diagrams) with small random perturbations. Paired with a completion-guided instruction, LVLMs automatically recompose the concealed semantics and generate unsafe outputs without triggering safety mechanisms. Although each slot appears benign in isolation (local benignness), StructAttack exploits LVLMs' reasoning to assemble these slots into coherent harmful semantics. Extensive experiments on multiple models and benchmarks show the efficacy of our proposed StructAttack.

Computational engine sound modeling is central to the automotive audio industry, particularly for active sound design, virtual prototyping, and emerging data-driven engine sound synthesis methods. These applications require large volumes of standardized, clean audio recordings with precisely time-aligned operating-state annotations: data that is difficult to obtain due to high costs, specialized measurement equipment requirements, and inevitable noise contamination. We present an analysis-driven framework for generating engine audio with sample-accurate control annotations. The method extracts harmonic structures from real recordings through pitch-adaptive spectral analysis, which then drive an extended parametric harmonic-plus-noise synthesizer. With this framework, we generate the Procedural Engine Sounds Dataset (19 hours, 5,935 files), a set of engine audio signals with sample-accurate RPM and torque annotations, spanning a wide range of operating conditions, signal complexities, and harmonic profiles. Comparison against real recordings validates that the synthesized data preserves characteristic harmonic structures, and baseline experiments confirm its suitability for learning-based parameter estimation and synthesis tasks. The dataset is released publicly to support research on engine timbre analysis, control parameter estimation, acoustic modeling and neural generative networks.

Code evolution is inevitable in modern software development. Changes to third-party APIs frequently break existing code and complicate maintenance, posing practical challenges for developers. While large language models (LLMs) have shown promise in code generation, they struggle to reason without a structured representation of these evolving relationships, often leading them to produce outdated APIs or invalid outputs. In this work, we propose a knowledge graph-augmented framework that decomposes the migration task into two synergistic stages: evolution path retrieval and path-informed code generation. Our approach constructs static and dynamic API graphs to model intra-version structures and cross-version transitions, enabling structured reasoning over API evolution. Both modules are trained with synthetic supervision automatically derived from real-world API diffs, ensuring scalability and minimal human effort. Extensive experiments across single-package and multi-package benchmarks demonstrate that our framework significantly improves migration accuracy, controllability, and execution success over standard LLM baselines. The source code and datasets are available at: https://github.com/kangjz1203/KCoEvo.

Industrial visual inspection in pharmaceutical production requires high accuracy under strict constraints on cycle time, hardware footprint, and operational cost. Manual inline inspection is still common, but it is affected by operator variability and limited throughput. Classical rule-based computer vision pipelines are often rigid and difficult to scale to highly variable production scenarios. To address these limitations, we present a semi-supervised anomaly detection framework based on a generative adversarial architecture with a residual autoencoder and a dense bottleneck, specifically designed for online deployment on a high-speed Blow-Fill-Seal (BFS) line. The model is trained only on nominal samples and detects anomalies through reconstruction residuals, providing both classification and spatial localization via heatmaps. The training set contains 2,815,200 grayscale patches. Experiments on a real industrial test kit show high detection performance while satisfying timing constraints compatible with a 500 ms acquisition slot.

Accurate analysis of industrial time-series big data is critical for the Prognostics and Health Management (PHM) of industrial equipment. While recent advancements in Large Language Models (LLMs) have shown promise in time-series analysis, existing methods typically focus on single-modality adaptations, failing to exploit the complementary nature of temporal signals, frequency-domain visual representations, and textual knowledge information. In this paper, we propose TS-MLLM, a unified multi-modal large language model framework designed to jointly model temporal signals, frequency-domain images, and textual domain knowledge. Specifically, we first develop an Industrial time-series Patch Modeling branch to capture long-range temporal dynamics. To integrate cross-modal priors, we introduce a Spectrum-aware Vision-Language Model Adaptation (SVLMA) mechanism that enables the model to internalize frequency-domain patterns and semantic context. Furthermore, a Temporal-centric Multi-modal Attention Fusion (TMAF) mechanism is designed to actively retrieve relevant visual and textual cues using temporal features as queries, ensuring deep cross-modal alignment. Extensive experiments on multiple industrial benchmarks demonstrate that TS-MLLM significantly outperforms state-of-the-art methods, particularly in few-shot and complex scenarios. The results validate our framework's superior robustness, efficiency, and generalization capabilities for industrial time-series prediction.

Out-of-distribution (OOD) detection is critical in safety-sensitive applications. While this challenge has been addressed from various perspectives, the influence of training objectives on OOD behavior remains comparatively underexplored. In this paper, we present a systematic comparison of four widely used training objectives: Cross-Entropy Loss, Prototype Loss, Triplet Loss, and Average Precision (AP) Loss, spanning probabilistic, prototype-based, metric-learning, and ranking-based supervision, for OOD detection in image classification under standardized OpenOOD protocols. Across CIFAR-10/100 and ImageNet-200, we find that Cross-Entropy Loss, Prototype Loss, and AP Loss achieve comparable in-distribution accuracy, while Cross-Entropy Loss provides the most consistent near- and far-OOD performance overall; the other objectives can be competitive in specific settings.

Over the past decade, neural network solvers powered by generative artificial intelligence have garnered significant attention in the domain of vehicle routing problems (VRPs), owing to their exceptional computational efficiency and superior reasoning capabilities. In particular, autoregressive solvers integrated with reinforcement learning have emerged as a prominent trend. However, much of the existing work emphasizes large-scale generalization of neural approaches while neglecting the limited robustness of attention-based methods across heterogeneous distributions of problem parameters. Their improvements over heuristic search remain largely restricted to hand-curated, fixed-distribution benchmarks. Furthermore, these architectures tend to degrade significantly when node representations are highly similar or when tasks involve long decision horizons. To address the aforementioned limitations, we propose a novel fusion neural network framework that employs a discrete noise graph diffusion model to learn the underlying constraints of vehicle routing problems and generate a constraint assignment matrix. This matrix is subsequently integrated adaptively into the feature representation learning and decision process of the autoregressive solver, serving as a graph structure mask that facilitates the formation of solutions characterized by both global vision and local feature integration. To the best of our knowledge, this work represents the first comprehensive experimental investigation of neural network model solvers across a 378-combinatorial space spanning four distinct dimensions within the CVRPlib public dataset. Extensive experimental evaluations demonstrate that our proposed fusion model effectively captures and leverages problem constraints, achieving state-of-the-art performance across multiple benchmark datasets.

Membership inference attacks (MIAs) have become the standard tool for evaluating privacy leakage in machine learning (ML). Among them, the Likelihood-Ratio Attack (LiRA) is widely regarded as the state of the art when sufficient shadow models are available. However, prior evaluations have often overstated the effectiveness of LiRA by attacking models overconfident on their training samples, calibrating thresholds on target data, assuming balanced membership priors, and/or overlooking attack reproducibility. We re-evaluate LiRA under a realistic protocol that (i) trains models using anti-overfitting (AOF) and transfer learning (TL), when applicable, to reduce overconfidence as in production models; (ii) calibrates decision thresholds using shadow models and data rather than target data; (iii) measures positive predictive value (PPV, or precision) under shadow-based thresholds and skewed membership priors (pi <= 10%); and (iv) quantifies per-sample membership reproducibility across different seeds and training variations. We find that AOF significantly weakens LiRA, while TL further reduces attack effectiveness while improving model accuracy. Under shadow-based thresholds and skewed priors, LiRA's PPV often drops substantially, especially under AOF or AOF+TL. We also find that thresholded vulnerable sets at extremely low FPR show poor reproducibility across runs, while likelihood-ratio rankings are more stable. These results suggest that LiRA, and likely weaker MIAs, are less effective than previously suggested under realistic conditions, and that reliable privacy auditing requires evaluation protocols that reflect practical training practices, feasible attacker assumptions, and reproducibility considerations. Code is available at https://github.com/najeebjebreel/lira_analysis.

Anomaly detection is nowadays increasingly used in industrial applications and processes. One of the main fields of the appliance is the visual inspection for surface anomaly detection, which aims to spot regions that deviate from regularity and consequently identify abnormal products. Defect localization is a key task, that usually is achieved using a basic comparison between generated image and the original one, implementing some blob-analysis or image-editing algorithms, in the post-processing step, which is very biased towards the source dataset, and they are unable to generalize. Furthermore, in industrial applications, the totality of the image is not always interesting but could be one or some regions of interest (ROIs), where only in those areas there are relevant anomalies to be spotted. For these reasons, we propose a new architecture composed by two blocks. The first block is a Generative Adversarial Network (GAN), based on a residual autoencoder (ResAE), to perform reconstruction and denoising processes, while the second block produces image segmentation, spotting defects. This method learns from a dataset composed of good products and generated synthetic defects. The discriminative network is trained using a ROI for each image contained in the training dataset. The network will learn in which area anomalies are relevant. This approach guarantees the reduction of using pre-processing algorithms, formerly developed with blob-analysis and image-editing procedures. To test our model we used challenging MVTec anomaly detection datasets and an industrial large dataset of pharmaceutical BFS strips of vials. This set constitutes a more realistic use case of the aforementioned network.

Automated electrocardiogram (ECG) classification is essential for early detection of cardiovascular diseases. While recent approaches have increasingly relied on deep neural networks with complex architectures, we demonstrate that careful data preprocessing, class balancing, and a simplified convolutional neural network combined with a variational autoencoder (CNN-VAE) architecture can achieve competitive performance with significantly reduced model complexity. Using the publicly available PTB XL dataset, we achieve 87.01% binary accuracy and 0.7454 weighted F1-score across five diagnostic classes (CD, HYP, MI, NORM, STTC) with only 197,093 trainable parameters. Our work emphasises the importance of data-centric machine learning practices over architectural complexity, demonstrating that systematic preprocessing and balanced training strategies are critical for medical signal classification. We identify challenges in minority class detection (particularly hypertrophy) and provide insights for future improvements in handling imbalanced ECG datasets. Index Terms: ECG classification, convolutional neural networks, class balancing, data preprocessing, variational autoencoders, PTB-XL dataset

The rapid integration of Large Language Model (LLM) agents into autonomous task execution has introduced significant privacy concerns within cross-tool data flows. In this paper, we systematically investigate and define a novel risk termed Data Over-Exposure (DOE) in LLM Agent, where an Agent inadvertently transmits sensitive data beyond the scope of user intent and functional necessity. We identify that DOE is primarily driven by the broad data paradigms in tool design and the coarse-grained data processing inherent in LLMs. In this paper, we present AgentRaft, the first automated framework for detecting DOE risks in LLM agents. AgentRaft combines program analysis with semantic reasoning through three synergistic modules: (1) it constructs a Cross-Tool Function Call Graph (FCG) to model the interaction landscape of heterogeneous tools; (2) it traverses the FCG to synthesize high-quality testing user prompts that act as deterministic triggers for deep-layer tool execution; and (3) it performs runtime taint tracking and employs a multi-LLM voting committee grounded in global privacy regulations (e.g., GDPR, CCPA, PIPL) to accurately identify privacy violations. We evaluate AgentRaft on a testing environment of 6,675 real-world agent tools. Our findings reveal that DOE is indeed a systemic risk, prevalent in 57.07% of potential tool interaction paths. AgentRaft achieves a high detection accuracy and effectiveness, outperforming baselines by 87.24%. Furthermore, AgentRaft reaches near-total DOE coverage (99%) within only 150 prompts while reducing per-chain verification costs by 88.6%. Our work provides a practical foundation for building auditable and privacy-compliant LLM agent systems.

Nepal Bhasha (Newari), an endangered language of the Kathmandu Valley, remains digitally marginalized due to the severe scarcity of annotated speech resources. In this work, we introduce Nwāchā Munā, a newly curated 5.39-hour manually transcribed Devanagari speech corpus for Nepal Bhasha, and establish the first benchmark using script-preserving acoustic modeling. We investigate whether proximal cross-lingual transfer from a geographically and linguistically adjacent language (Nepali) can rival large-scale multilingual pretraining in an ultra-low-resource Automatic Speech Recognition (ASR) setting. Fine-tuning a Nepali Conformer model reduces the Character Error Rate (CER) from a 52.54% zero-shot baseline to 17.59% with data augmentation, effectively matching the performance of the multilingual Whisper-Small model despite utilizing significantly fewer parameters. Our findings demonstrate that proximal transfer within South Asian language clusters serves as a computationally efficient alternative to massive multilingual models. We openly release the dataset and benchmarks to digitally enable the Newari community and foster further research in Nepal Bhasha.

Zero-shot Text-to-Speech (TTS) voice cloning poses severe privacy risks, demanding the removal of specific speaker identities from trained TTS models. Conventional machine unlearning is insufficient in this context, as zero-shot TTS can dynamically reconstruct voices from just reference prompts. We formalize this task as Speech Generation Speaker Poisoning (SGSP), in which we modify trained models to prevent the generation of specific identities while preserving utility for other speakers. We evaluate inference-time filtering and parameter-modification baselines across 1, 15, and 100 forgotten speakers. Performance is assessed through the trade-off between utility (WER) and privacy, quantified using AUC and Forget Speaker Similarity (FSSIM). We achieve strong privacy for up to 15 speakers but reveal scalability limits at 100 speakers due to increased identity overlap. Our study thus introduces a novel problem and evaluation framework toward further advances in generative voice privacy.

Accent plays a crucial role in speaker identity and inclusivity in speech technologies. Existing accented text-to-speech (TTS) systems either require large-scale accented datasets or lack fine-grained phoneme-level controllability. We propose a accented TTS framework that combines phonological rules with a multilingual TTS model. The rules are applied to phoneme sequences to transform accent at the phoneme level while preserving intelligibility. The method requires no accented training data and enables explicit phoneme-level accent manipulation. We design rule sets for Spanish- and Indian-accented English, modeling systematic differences in consonants, vowels, and syllable structure arising from phonotactic constraints. We analyze the trade-off between phoneme-level duration alignment and accent as realized in speech timing. Experimental results demonstrate effective accent shift while maintaining speech quality.

Aging bridge networks require proactive, verifiable, and interpretable maintenance strategies, yet reinforcement learning (RL) policies trained solely on reward signals provide no formal safety guarantees and remain opaque to infrastructure managers. We demonstrate COOL-MC as a tool for verifying and explaining RL policies for multi-bridge network maintenance, building on a single-bridge Markov decision process (MDP) from the literature and extending it to a parallel network of three heterogeneous bridges with a shared periodic budget constraint, encoded in the PRISM modeling language. We train an RL agent on this MDP and apply probabilistic model checking and explainability methods to the induced discrete-time Markov chain (DTMC) that arises from the interaction between the learned policy and the underlying MDP. Probabilistic model checking reveals that the trained policy has a safety-violation probability of 3.5\% over the planning horizon, being slightly above the theoretical minimum of 0\% and indicating the suboptimality of the learned policy, noting that these results are based on artificially constructed transition probabilities and deterioration rates rather than real-world data, so absolute performance figures should be interpreted with caution. The explainability analysis further reveals, for instance, a systematic bias in the trained policy toward the state of bridge 1 over the remaining bridges in the network. These results demonstrate COOL-MC's ability to provide formal, interpretable, and practical analysis of RL maintenance policies.

Learned world models excel at interpolative generalization but fail at extrapolative generalization to novel physical properties. This limitation arises because they learn statistical correlations rather than the environment's underlying generative rules, such as physical invariances and conservation laws. We argue that learning these invariances is key to robust extrapolation. To achieve this, we first introduce \textbf{Symmetry Exploration}, an unsupervised exploration strategy where an agent is intrinsically motivated by a Hamiltonian-based curiosity bonus to actively probe and challenge its understanding of conservation laws, thereby collecting physically informative data. Second, we design a Hamiltonian-based world model that learns from the collected data, using a novel self-supervised contrastive objective to identify the invariant physical state from raw, view-dependent pixel observations. Our framework, \textbf{DreamSAC}, trained on this actively curated data, significantly outperforms state-of-the-art baselines in 3D physics simulations on tasks requiring extrapolation.

Unified multimodal models hold the promise of generating extensive, interleaved narratives, weaving text and imagery into coherent long-form stories. However, current systems suffer from a critical reliability gap: as sequences grow, generation quality rapidly collapses. In this work, we investigate the mechanism behind this failure and argue that it is distinct from standard long-context challenges. We reveal that in generation, accumulated visual history acts as a source of active pollution, a decay governed specifically by the number of image events rather than raw token count. We identify a structural vulnerability where dense visual tokens overwhelm the attention mechanism, creating noise that distorts future synthesis. Guided by these mechanistic insights, we propose UniLongGen, a training-free inference strategy that prioritizes safe conditioning over total recall. Instead of retaining all history, UniLongGen dynamically curates the model's memory, identifying and discarding interfering visual signals based on the model's own internal relevance rankings. Extensive experiments demonstrate that this active forgetting approach is essential for stability: UniLongGen significantly outperforms baselines in long-horizon fidelity and consistency, while simultaneously reducing memory footprint and inference time.

Islamic inheritance law ('ilm al-mawarith) is challenging for large language models because solving inheritance cases requires complex, structured multi-step reasoning and the correct application of juristic rules to compute heirs' shares. We introduce MAWARITH, a large-scale annotated dataset of 12,500 Arabic inheritance cases to train and evaluate the full reasoning chain: (i) identifying eligible heirs, (ii) applying blocking (hajb) and allocation rules, and (iii) computing exact inheritance shares. Unlike prior datasets that restrict inheritance case solving to multiple-choice questions, MAWARITH supports the full reasoning chain and provides step-by-step solutions, including intermediate legal decisions and justifications based on classical juristic sources and established inheritance rules, as well as exact share calculations. To evaluate models beyond final-answer accuracy, we propose MIR-E (Mawarith Inheritance Reasoning Evaluation), a weighted multi-stage metric that scores key reasoning stages and captures error propagation across the pipeline. We evaluate five LLMs in a zero-shot setting. Gemini-2.5-flash achieves about 90% MIR-E on both validation and test, while Fanar-C, Fanar-Sadiq, LLaMA 3, and Qwen 3 remain below 50%. Our error analysis identifies recurring failure patterns, including scenario misinterpretation, errors in heir identification, errors in share allocation, and missing or incorrect application of key inheritance rules such as 'awl and radd. The MAWARITH dataset is publicly available at https://github.com/bouchekif/inheritance_evaluation.

Accent is an integral part of society, reflecting multiculturalism and shaping how individuals express identity. The majority of English speakers are non-native (L2) speakers, yet current Text-To-Speech (TTS) systems primarily model American-accented English due limited accented data. We propose \textit{Accent Vector}, a controllable representation that enables accent manipulation in multilingual TTS without requiring accented training data. \textit{Accent Vector} is derived by fine-tuning a TTS system on native speech of a different language (i.e. non-English) and computing task vectors capturing accent characteristics (i.e. in English). By scaling and interpolating the vector, we achieve fine-grained control over accent strength and generate mixed-accent speech. In addition, it generalizes beyond English, enabling accent control across multiple languages. Objective and human evaluations confirm the effectiveness of Accent Vector for fine-grained and compositional accent control.

Concept erasure aims to remove unwanted attributes, such as social or demographic factors, from learned representations, while preserving their task-relevant utility. While the goal of concept erasure is protection against all adversaries, existing methods remain vulnerable to nonlinear ones. This vulnerability arises from their failure to fully capture the complex, nonlinear statistical dependencies between learned representations and unwanted attributes. Moreover, although the existence of a trade-off between utility and erasure is expected, its progression during the erasure process, i.e., the cost of erasure, remains unstudied. In this work, we introduce Obliviator, a post-hoc erasure method designed to fully capture nonlinear statistical dependencies. We formulate erasure from a functional perspective, leading to an optimization problem involving a composition of kernels that lacks a closed-form solution. Instead of solving this problem in a single shot, we adopt an iterative approach that gradually morphs the feature space to achieve a more utility-preserving erasure. Unlike prior methods, Obliviator guards unwanted attribute against nonlinear adversaries. Our gradual approach quantifies the cost of nonlinear guardedness and reveals the dynamics between attribute protection and utility-preservation over the course of erasure. The utility-erasure trade-off curves obtained by Obliviator outperform the baselines and demonstrate its strong generalizability: its erasure becomes more utility-preserving when applied to the better-disentangled representations learned by more capable models.

Table reasoning requires models to jointly perform semantic understanding and precise numerical operations. Most existing methods rely on a single-turn reasoning paradigm over tables which suffers from context overflow and weak numerical sensitivity. To address these limitations, we previously proposed TableMind as a tuning-based autonomous programmatic agent that simulates human-like interaction within a lightweight large language model (LLM). TableMind internalizes planning, action, and reflection through a two-stage training strategy involving supervised fine-tuning (SFT) on filtered high-quality data and reinforcement learning (RL) via a multi-perspective reward and the Rank-Aware Policy Optimization (RAPO) algorithm. While TableMind establishes a solid foundation for programmatic agents, the inherent stochasticity of LLMs remains a critical challenge that leads to hallucinations. In this paper, we extend this foundation to TableMind++ by introducing a novel uncertainty-aware inference framework to mitigate hallucinations. Specifically, we propose memory-guided plan pruning to retrieve historical trajectories for validating and filtering out logically flawed plans to address epistemic uncertainty. To ensure execution precision, we introduce confidence-based action refinement which monitors token-level probabilities to detect and self-correct syntactic noise for aleatoric uncertainty mitigation. Finally, we employ dual-weighted trajectory aggregation to synthesize a robust consensus from multiple reasoning paths. Extensive experiments on diverse benchmarks demonstrate that TableMind++ consistently outperforms previous baselines and proprietary models to validate the effectiveness of integrating autonomous training with uncertainty quantification. Our code is available.

Accurate and predictive scale-resolving simulations of laser-ignited rocket engines are highly time-consuming because the problem includes turbulent fuel-oxidizer mixing dynamics, laser-induced energy deposition, and high-speed flame growth. This is conflated with the large design space primarily corresponding to the laser operating conditions and target location. To enable rapid exploration and uncertainty quantification, we propose a data-driven surrogate modeling approach that combines convolutional autoencoders (cAEs) with neural ordinary differential equations (neural ODEs). The present target application of an ML-based surrogate model to leading-edge multi-physics turbulence simulation is part of a paradigm shift in the deployment of surrogate models towards increasing real-world complexity. Sequentially, the cAE spatially compresses high-dimensional flow fields into a low-dimensional latent space, wherein the system's temporal dynamics are learned via neural ODEs. Once trained, the model generates fast spatiotemporal predictions from initial conditions and specified operating inputs. By learning a surrogate to replace the entirety of the time-evolving simulation, the cost of predicting an ignition trial is reduced by several orders of magnitude, allowing efficient exploration of the input parameter space. Further, as the current framework yields a spatiotemporal field prediction, appraisal of the model output's physical grounding is more tractable. This approach marks a significant step toward real-time digital twins for laser-ignited rocket combustors and represents surrogate modeling in a complex system context.

Brain activity is intrinsically a neural dynamic process constrained by anatomical space. This leads to significant variations in spatial distribution patterns and correlation patterns of neural activity across variable and heterogeneous scenarios. However, dominant brain functional network construction methods, which relies on pre-defined brain atlases and linear assumptions, fails to precisely capture varying neural activity patterns in heterogeneous scenarios. This limits the consistency and generalizability of the brain functional networks constructed by dominant methods. Here, a neural dynamics-informed pre-trained framework is proposed for personalized brain functional network construction. The proposed framework extracts personalized representations of neural activity patterns in heterogeneous scenarios. Personalized brain functional networks are obtained by utilizing these representations to guide brain parcellation and neural activity correlation estimation. Systematic evaluations were employed on 18 datasets across tasks, such as virtual neural modulation and abnormal neural circuit identification. Experimental results demonstrate that the proposed framework attains superior performance in heterogeneous scenarios. Overall, the proposed framework challenges the dominant brain functional network construction method.

Transferring knowledge by fine-tuning large-scale pre-trained networks has become a standard paradigm for downstream tasks, yet the knowledge of a pre-trained model is tightly coupled with monolithic architecture, which restricts flexible reuse across models of varying scales. In response to this challenge, recent approaches typically resort to either parameter selection, which fails to capture the interdependent structure of this knowledge, or parameter prediction using generative models that depend on impractical access to large network collections. In this paper, we empirically demonstrate that a model's foundational, task-agnostic knowledge, its "learngene", is encoded within the low-frequency components of its weights, and can be efficiently inherited by downstream models. Based on this insight, we propose FRONT (FRequency dOmain kNowledge Transfer), a novel framework that uses the Discrete Cosine Transform (DCT) to isolate the low-frequency "learngene". This learngene can be seamlessly adapted to initialize models of arbitrary size via simple truncation or padding, a process that is entirely training-free. For enhanced performance, we propose an optional low-cost refinement process that introduces a spectral regularizer to further improve the learngene's transferability. Extensive experiments demonstrate that FRONT achieves the state-of-the-art performance, accelerates convergence by up to 15 times in vision tasks, and reduces training FLOPs by an average of 40.5% in language tasks.

Privacy protection and uncertainty quantification are increasingly important in data-driven decision making. Conformal prediction provides finite-sample marginal coverage, but existing private approaches often rely on data splitting, reducing the effective sample size. We propose a full-data privacy-preserving conformal prediction framework that avoids splitting. Our framework leverages stability induced by differential privacy to control the gap between in-sample and out-of-sample conformal scores, and pairs this with a conservative private quantile routine designed to prevent under-coverage. We show that a generic differential privacy guarantee yields a universal coverage floor, yet cannot generally recover the nominal $1-α$ level. We then provide a refined, mechanism-specific stability analysis and yields asymptotic recovery of the nominal level. Experiments demonstrate sharper prediction sets than the split-based private baseline.

This work investigates large-scale sketch recognition from a graph-native perspective, where free-hand sketches are directly modeled as structured graphs rather than raster images or stroke sequences. We propose SketchGraphNet, a hybrid graph neural architecture that integrates local message passing with a memory-efficient global attention mechanism, without relying on auxiliary positional or structural encodings. To support systematic evaluation, we construct SketchGraph, a large-scale benchmark comprising 3.44 million graph-structured sketches across 344 categories, with two variants (A and R) to reflect different noise conditions. Each sketch is represented as a spatiotemporal graph with normalized stroke-order attributes. On SketchGraph-A and SketchGraph-R, SketchGraphNet achieves Top-1 accuracies of 83.62% and 87.61%, respectively, under a unified training configuration. MemEffAttn further reduces peak GPU memory by over 40% and training time by more than 30% compared with Performer-based global attention, while maintaining comparable accuracy.

Automated program repair (APR) attempts to generate correct patches and has drawn wide attention from both academia and industry in the past decades. However, APR is continuously struggling with the patch overfitting issue due to the weak test suites. Thus, to address the overfitting problem, the community has proposed an increasing number of approaches to predict patch correctness (APCA approaches). Among them, locally deep learning approaches aimed at automatically match designs has been emerging strongly. Such approaches typically encode input code snippets into well-designed representations and build a binary model for correctness prediction. Despite being fundamental in reason about patch correctness, code representation has not been systematically investigated. To bridge this gap, we perform the first extensive study to evaluate the performance of different code representations on predicting patch correctness from more than 500 trained APCA models. The experimental results on 15 benchmarks with four categories and 11 classifiers show that the graph-based code representation which is ill-explored in the literature, consistently outperforms other representations, e.g., an average accuracy of 82.6% for CPG across three GNN models. Moreover, we demonstrate that such representations can achieve comparable or better performance for three different previous APCA approaches, e.g., filtering out 87.09% overfitting patches by TREETRAIN with AST. We further find that integrating sequence-based representation into heuristic-based representation is able to yield an average improvement of 13.5% on five metrics. Overall, our study highlights the potential and challenges of utilizing code representation to reason about patch correctness, thus increasing the usability of off-the-shelf APR tools and reducing the manual debugging effort of developers in practice.

Advancing autonomous green technologies in solar photovoltaic (PV) systems is key to improving sustainability and efficiency in renewable energy production. This study presents a reinforcement learning (RL)-based framework to autonomously optimize the cleaning schedules of PV panels in arid regions, where soiling from dust and other airborne particles significantly reduces energy output. By employing advanced RL algorithms, Proximal Policy Optimization (PPO) and Soft Actor-Critic (SAC), the framework dynamically adjusts cleaning intervals based on uncertain environmental conditions. The proposed approach was applied to a case study in Abu Dhabi, UAE, demonstrating that PPO outperformed SAC and traditional simulation optimization (Sim-Opt) methods, achieving up to 13% cost savings by dynamically responding to weather uncertainties. The results highlight the superiority of flexible, autonomous scheduling over fixed-interval methods, particularly in adapting to stochastic environmental dynamics. This aligns with the goals of autonomous green energy production by reducing operational costs and improving the efficiency of solar power generation systems. This work underscores the potential of RL-driven autonomous decision-making to optimize maintenance operations in renewable energy systems. In future research, it is important to enhance the generalization ability of the proposed RL model, while also considering additional factors and constraints to apply it to different regions.

Interaction is one of the core abilities of humanoid robots. However, most existing frameworks focus on non-interactive whole-body control, which limits their practical applicability. In this work, we develop InterReal, a unified physics-based imitation learning framework for Real-world human-object Interaction (HOI) control. InterReal enables humanoid robots to track HOI reference motions, facilitating the learning of fine-grained interactive skills and their deployment in real-world settings. Within this framework, we first introduce a HOI motion data augmentation scheme with hand-object contact constraints, and utilize the augmented motions to improve policy stability under object perturbations. Second, we propose an automatic reward learner to address the challenge of large-scale reward shaping. A meta-policy guided by critical tracking error metrics explores and allocates reward signals to the low-level reinforcement learning objective, which enables more effective learning of interactive policies. Experiments on HOI tasks of box-picking and box-pushing demonstrate that InterReal achieves the best tracking accuracy and the highest task success rate compared to recent baselines. Furthermore, we validate the framework on the real-world robot Unitree G1, which demonstrates its practical effectiveness and robustness beyond simulation.

Drifting models train one-step generators by optimizing a mean-shift discrepancy induced by a kernel between the data and model distributions, with Laplace kernels used by default in practice. At each point, this discrepancy compares the kernel-weighted displacement toward nearby data samples with the corresponding displacement toward nearby model samples, yielding a transport direction for generated samples. In this paper, we make its relationship to the score-matching principle behind diffusion models precise by showing that drifting admits a score-based formulation on kernel-smoothed distributions. For Gaussian kernels, the population mean-shift field coincides with the score difference between the Gaussian-smoothed data and model distributions. This identity follows from Tweedie's formula, which links the score of a Gaussian-smoothed density to the corresponding conditional mean, and implies that Gaussian-kernel drifting is exactly a score-matching-style objective on smoothed distributions. It also clarifies the connection to Distribution Matching Distillation (DMD): both methods use score-mismatch transport directions, but drifting realizes the score signal nonparametrically from kernel neighborhoods, whereas DMD uses a pretrained diffusion teacher. Beyond Gaussians, we derive an exact decomposition for general radial kernels, and for the Laplace kernel we prove rigorous error bounds showing that drifting remains an accurate proxy for score matching in low-temperature and high-dimensional regimes.

Kashmiri is spoken by around 7 million people but remains critically underserved in speech technology, despite its official status and rich linguistic heritage. The lack of robust Text-to-Speech (TTS) systems limits digital accessibility and inclusive human-computer interaction for native speakers. In this work, we present the first dedicated open-source neural TTS system designed for Kashmiri. We show that zero-shot multilingual baselines trained for Indic languages fail to produce intelligible speech, achieving a Mean Opinion Score (MOS) of only 1.86, largely due to inadequate modeling of Perso-Arabic diacritics and language-specific phonotactics. To address these limitations, we propose Bolbosh, a supervised cross-lingual adaptation strategy based on Optimal Transport Conditional Flow Matching (OT-CFM) within the Matcha-TTS framework. This enables stable alignment under limited paired data. We further introduce a three-stage acoustic enhancement pipeline consisting of dereverberation, silence trimming, and loudness normalization to unify heterogeneous speech sources and stabilize alignment learning. The model vocabulary is expanded to explicitly encode Kashmiri graphemes, preserving fine-grained vowel distinctions. Our system achieves a MOS of 3.63 and a Mel-Cepstral Distortion (MCD) of 3.73, substantially outperforming multilingual baselines and establishing a new benchmark for Kashmiri speech synthesis. Our results demonstrate that script-aware and supervised flow-based adaptation are critical for low-resource TTS in diacritic-sensitive languages. Code and data are available at: https://github.com/gaash-lab/Bolbosh.

In this work, we present OCLADS, a novel communication framework with continual learning (CL) for Internet of Things (IoT) anomaly detection (AD) when operating in non-stationary environments. As the statistical properties of the observed data change with time, the on-device inference model becomes obsolete, which necessitates strategic model updating. OCLADS keeps track of data distribution shifts to timely update the on-device IoT AD model. To do so, OCLADS introduces two mechanisms during the interaction between the resource-constrained IoT device and an edge server (ES): i) an intelligent sample selection mechanism at the device for data transmission, and ii) a distribution-shift detection mechanism at the ES for model updating. Experimental results with TinyML demonstrate that our proposed framework achieves high inference accuracy while realizing a significantly smaller number of model updates compared to the baseline schemes.

Transferring pre-trained knowledge from a source model to a target model of a different architectural size is a key challenge for flexible and efficient model scaling. However, current parameter-space methods treat Small-to-Large (S2L) and Large-to-Small (L2S) scaling as separate, incompatible problems, focusing on parameter synthesis and selection, respectively. This fragmented perspective has resulted in specialized tools, hindering a unified, bidirectional framework. In this paper, we propose BoT (Bidirectional knowledge Transfer), the first size-agnostic framework to unify S2L and L2S scaling. Our core insight is to treat model weights as continuous signals, where models of different sizes represent distinct discretizations of the transferable knowledge. This multi-resolution perspective directly casts S2L and L2S scaling as the signal processing operations of upsampling and downsampling, naturally leading to the adoption of the Discrete Wavelet Transform (DWT) and its Inverse (IDWT). BoT leverages the recursive nature of wavelets, using the decomposition level as a dynamic scaling factor to bridge disparate model sizes in a parameter-free and computationally efficient manner. Extensive experiments on DeiT, BERT, and GPT demonstrate significant pre-training FLOPs savings (up to 67.1% for S2L, 52.8% for L2S) and state-of-the-art performance on benchmarks like GLUE and SQuAD.

The continuous expansion of open data platforms and research repositories has led to a fragmented dataset ecosystem, posing significant challenges for cross-source data discovery and interpretation. To address these challenges, we introduce SeDa--a unified framework for dataset discovery, semantic annotation, and multi-entity augmented navigation. SeDa integrates more than 7.6 million datasets from over 200 platforms, spanning governmental, academic, and industrial domains. The framework first performs semantic extraction and standardization to harmonize heterogeneous metadata representations. On this basis, a topic-tagging mechanism constructs an extensible tag graph that supports thematic retrieval and cross-domain association, while a provenance assurance module embedded within the annotation process continuously validates dataset sources and monitors link availability to ensure reliability and traceability. Furthermore, SeDa employs a multi-entity augmented navigation strategy that organizes datasets within a knowledge space of sites, institutions, and enterprises, enabling contextual and provenance-aware exploration beyond traditional search paradigms. Comparative experiments with popular dataset search platforms, such as ChatPD and Google Dataset Search, demonstrate that SeDa achieves superior coverage, timeliness, and traceability. Taken together, SeDa establishes a foundation for trustworthy, semantically enriched, and globally scalable dataset exploration.

High-dimensional time series forecasting suffers from severe overfitting when the number of predictors exceeds available observations, making standard local projection methods unstable and unreliable. We propose an enhanced Random Subspace Local Projection (RSLP) framework designed to deliver robust impulse response estimation in the presence of hundreds of correlated predictors. The method introduces weighted subspace aggregation, category-aware subspace sampling, adaptive subspace size selection, and a bootstrap inference procedure tailored to dependent data. These enhancements substantially improve estimator stability at longer forecast horizons while providing more reliable finite-sample inference. Experiments on synthetic data, macroeconomic indicators, and the FRED-MD dataset demonstrate a 33 percent reduction in estimator variability at horizons h >= 3 through adaptive subspace size selection. The bootstrap inference procedure produces conservative confidence intervals that are 14 percent narrower at policy-relevant horizons in very high-dimensional settings (FRED-MD with 126 predictors) while maintaining proper coverage. The framework provides practitioners with a principled approach for incorporating rich information sets into impulse response analysis without the instability of traditional high-dimensional methods.

Artificial Intelligence (AI) agents have evolved from passive predictive tools into active entities capable of autonomous decision-making and environmental interaction, driven by the reasoning capabilities of Large Language Models (LLMs). However, this evolution has introduced critical security vulnerabilities that existing frameworks fail to address. The Hierarchical Autonomy Evolution (HAE) framework organizes agent security into three tiers: Cognitive Autonomy (L1) targets internal reasoning integrity; Execution Autonomy (L2) covers tool-mediated environmental interaction; Collective Autonomy (L3) addresses systemic risks in multi-agent ecosystems. We present a taxonomy of threats spanning cognitive manipulation, physical environment disruption, and multi-agent systemic failures, and evaluate existing defenses while identifying key research gaps. The findings aim to guide the development of multilayered, autonomy-aware defense architectures for trustworthy AI agent systems.

Contactless sensing using wireless communication signals has garnered significant attention due to its non-intrusive nature and ubiquitous infrastructure. Despite the promise, the inherent bistatic deployment of wireless communication introduces clock asynchronism, which leads to unknown phase offsets in channel response and hinders fine-grained sensing. State-of-the-art systems widely adopt the cross-antenna channel ratio to cancel these detrimental phase offsets. However, the channel ratio preserves sensing feature accuracy only at integer-wavelength target displacements, losing sub-wavelength fidelity. To overcome this limitation, we derive the first quantitative mapping between the distorted ratio feature and the ideal channel feature. Building on this foundation, we develop a robust framework that leverages channel response amplitude to recover the ideal channel feature from the distorted ratio. Real-world experiments across Wi-Fi and LoRa demonstrate that our method can effectively reconstruct sub-wavelength displacement details, achieving nearly an order-of-magnitude improvement in accuracy.

Clinical information extraction (e.g., 2010 i2b2/VA challenge) usually presents tasks of concept recognition, assertion classification, and relation extraction. Jointly modeling the multi-stage tasks in the clinical domain is an underexplored topic. The existing independent task setting (reference inputs given in each stage) makes the joint models not directly comparable to the existing pipeline work. To address these issues, we define a joint task setting and propose a novel end-to-end system to jointly optimize three-stage tasks. We empirically investigate the joint evaluation of our proposal and the pipeline baseline with various embedding techniques: word, contextual, and in-domain contextual embeddings. The proposed joint system substantially outperforms the pipeline baseline by +0.3, +1.4, +3.1 for the concept, assertion, and relation F1. This work bridges joint approaches and clinical information extraction. The proposed approach could serve as a strong joint baseline for future research. The code is publicly available.

While transformers achieve strong performance, their internal decision-making processes remain opaque. We investigate whether architectural constraints can enforce interpretability by design through architectural stream independence: maintaining a token stream (carrying symbolic structure) and contextual semantics in separated streams that remain independently observable throughout processing, with integration delayed until output. We validate this principle through the Late Fusion Architecture (LFA), which demonstrates interpretable symbolic heads through all the final layers, while standard transformers show dissolution by the third of six layers; we quantify this effect by introducing the Token-Position Dependence Score (PDS), with $PDS_{max}$ = 0.276 and 0.058, respectively. Crucially, intervention experiments demonstrate functional modularity: suppressing LFA's recency heads causes minimal semantic damage (Cohen's d = -0.158) versus catastrophic entanglement in baselines (d = -0.672). LFA demonstrates that architectural constraints improve underlying learning mechanisms, averaging 42% stability versus 19% and 11% for baseline comparisons, with extremes from 50% on LFA's best pairs (6 of 12 heads position-invariant) down to 0% complete collapse in over-constrained cases. By preventing premature entanglement, architectural independence steers models toward semantic understanding over positional heuristics, establishing interpretability as an architectural design criterion enforceable through structural constraints rather than post-hoc analysis.

Autoregressive (AR) language models form representations incrementally through left-to-right prediction, whereas diffusion language models (dLLMs) are trained via full-sequence denoising. Although recent dLLMs match AR performance, it remains unclear whether diffusion objectives fundamentally reshape internal representations across depth. We perform the first layer- and token-wise representational analysis comparing native dLLMs (LLaDA), native AR models (Qwen2.5), and AR-initialized dLLMs (Dream-7B). We find that diffusion objectives result in different, more hierarchical abstractions with substantial early-layer redundancy and reduced recency bias, while AR objectives produce tightly coupled, depth-dependent representations. Critically, AR-initialized dLLMs retain AR-like representational dynamics despite diffusion training, revealing persistent initialization bias. Leveraging this observed representational redundancy, we introduce a static, task-agnostic inference-time layer-skipping method requiring no architectural changes or KV-cache sharing. Native dLLMs achieve up to 18.75% FLOPs reduction while preserving over 90% performance on reasoning and code generation benchmarks, whereas AR models degrade sharply under comparable skipping. These results link training objectives to representational structure and enable practical, cache-orthogonal efficiency gains.

What is the interplay between semantic representations learned by language models (LM) from surface form alone to those learned from more grounded evidence? We study this question for a scenario where part of the input comes from a different modality -- in our case, in a vision-language model (VLM), where a pretrained LM is aligned with a pretrained image encoder. As a case study, we focus on the task of predicting hypernyms of objects represented in images. We do so in a VLM setup where the image encoder and LM are kept frozen, and only the intermediate mappings are learned. We progressively deprive the VLM of explicit evidence for hypernyms, and test whether knowledge of hypernyms is recoverable from the LM. We find that the LMs we study can recover this knowledge and generalize even in the most extreme version of this experiment (when the model receives no evidence of a hypernym during training). Additional experiments suggest that this cross-modal taxonomic generalization persists under counterfactual image-label mappings only when the counterfactual data have high visual similarity within each category. Taken together, these findings suggest that cross-modal generalization in LMs arises as a result of both coherence in the extralinguistic input and knowledge derived from language cues.

The Model Context Protocol (MCP) is an open and standardized interface that enables large language models (LLMs) to interact with external tools and services, and is increasingly adopted by AI agents. However, the security of MCP-based systems remains largely unexplored.In this work, we conduct a large-scale security analysis of MCP servers integrated within MCP clients. We show that treating MCP servers as trusted entities without authenticating the caller identity is fundamentally insecure. Since MCP servers often cannot distinguish who is invoking a request, a single authorization decision may implicitly grant access to multiple, potentially untrusted callers.Our empirical study reveals that most MCP servers rely on persistent authorization states, allowing tool invocations after an initial authorization without re-authentication, regardless of the caller. In addition, many MCP servers fail to enforce authentication at the per-tool level, enabling unauthorized access to sensitive operations.These findings demonstrate that one-time authorization and server-level trust significantly expand the attack surface of MCP-based systems, highlighting the need for explicit caller authentication and fine-grained authorization mechanisms.

In this study, we present a conditional diffusion-transformer framework for generating ensembles of three-dimensional Escherichia coli genome conformations guided by Hi-C contact maps. Instead of producing a single deterministic structure, we formulate genome reconstruction as a conditional generative modeling problem that samples heterogeneous conformations whose ensemble-averaged contacts are consistent with the input Hi-C data. A synthetic dataset is constructed using coarse-grained molecular dynamics simulations to generate chromatin ensembles and corresponding Hi-C maps under circular topology. Our models operate in a latent diffusion setting with a variational autoencoder that preserves per-bin alignment and supports replication-aware representations. Hi-C information is injected through a transformer-based encoder and cross-attention, enforcing a physically interpretable one-way constraint from Hi-C to structure. The model is trained using a flow-matching objective for stable optimization. On held-out ensembles, generated structures reproduce the input Hi-C distance-decay and structural correlation metrics while maintaining substantial conformational diversity, demonstrating the effectiveness of diffusion-based generative modeling for ensemble-level 3D genome reconstruction.

Recent studies have shown that post-deployment adaptation can improve the robustness of speech enhancement models in unseen noise conditions. However, existing methods often incur prohibitive computational and memory costs, limiting their suitability for on-device deployment. In this work, we investigate model adaptation in realistic settings with dynamic acoustic scene changes and propose a lightweight framework that augments a frozen backbone with low-rank adapters updated via self-supervised training. Experiments on sequential scene evaluations spanning 111 environments across 37 noise types and three signal-to-noise ratio ranges, including the challenging [-8, 0] dB range, show that our method updates fewer than 1% of the base model's parameters while achieving an average 1.51 dB SI-SDR improvement within only 20 updates per scene. Compared to state-of-the-art approaches, our framework achieves competitive or superior perceptual quality with smoother and more stable convergence, demonstrating its practicality for lightweight on-device adaptation of speech enhancement models under real-world acoustic conditions.

This monograph provides a rigorous overview of theoretical and methodological aspects of probabilistic inference and learning with Stein's method. Recipes are provided for constructing Stein discrepancies from Stein operators and Stein sets, and properties of these discrepancies such as computability, separation, convergence detection, and convergence control are discussed. Further, the connection between Stein operators and Stein variational gradient descent is set out in detail. The main definitions and results are precisely stated, and references to all proofs are provided.

Cloud-based infrastructures have become the dominant platform for deploying large models, particularly large language models (LLMs). Fine-tuning and inference are increasingly delegated to cloud providers for simplified deployment and access to proprietary models, yet this creates a fundamental trust gap: although cryptographic and TEE-based verification exist, the scale of modern LLMs renders them prohibitive, leaving clients unable to practically audit these processes. This lack of transparency creates concrete security risks that can silently compromise service integrity. We present AFTUNE, an auditable and verifiable framework that ensures the computation integrity of cloud-based fine-tuning and inference. AFTUNE incorporates a lightweight recording and spot-check mechanism that produces verifiable traces of execution. These traces enable clients to later audit whether the training and inference processes followed the agreed configurations. Our evaluation shows that AFTUNE imposes practical computation overhead while enabling selective and efficient verification, demonstrating that trustworthy model services are achievable in today's cloud environments.

Determining whether AI systems process information similarly to humans is central to cognitive science and trustworthy AI. While modern AI models match human accuracy on standard tasks, such parity does not guarantee that their underlying decision-making strategies are aligned with human information processing. Assessing performance using i) error alignment metrics to compare how humans and models fail, and ii) using distorted, or otherwise more challenging, stimuli, provides a viable pathway toward a finer characterization of model-human alignment. However, existing out-of-distribution (OOD) analyses for challenging stimuli are limited due to methodological choices: they define OOD shift relative to model training data or use arbitrary distortion-specific parameters with little correspondence to human perception, hindering principled comparisons. We propose a human-centred framework that redefines the degree of OOD as a spectrum of human perceptual difficulty. By quantifying how much a collection of stimuli deviates from an undistorted reference set based on human accuracy, we construct an OOD spectrum and identify four distinct regimes of perceptual challenge. This approach enables principled model-human comparisons at calibrated difficulty levels. We apply this framework to object recognition and reveal unique, regime-dependent model-human alignment rankings and profiles across deep learning architectures. Vision-language models are the most consistently human aligned across near- and far-OOD conditions, but CNNs are more aligned than ViTs for near-OOD and ViTs are more aligned than CNNs for far-OOD conditions. Our work demonstrates the critical importance of accounting for cross-condition differences such as perceptual difficulty for a principled assessment of model-human alignment.

Standard transformers entangle all computation in a single residual stream, obscuring which components perform which functions. We introduce the Dual-Stream Transformer, which decomposes the residual stream into two functionally distinct components: a token stream updated by attention and a context stream updated by feed-forward networks. Information flow between attention heads is controlled through a hierarchy of mixing strategies, from fully independent (maximum interpretability) to dense (standard transformer behavior). This design exposes a tunable tradeoff between interpretability and performance. We measure this tradeoff on language modeling tasks at 29M parameters. Fully independent head mixing increases validation loss by 8\% relative to dense baselines. The recommended Kronecker mixing strategy, which permits scalar communication between heads while preserving within-head structure, costs only 2.5\%. All configurations maintain functional generation under attention amplification (scaling logits by factors up to 16 at inference time), with degradation ranging from 16\% to 27\%. This robustness suggests the architectures learn discrete algorithms that operate independently of soft probabilistic mixing. The architecture provides a foundation for interpretable language models where internal structure is exposed by design. \footnote{This work was partially supported by DARPA Contract HR001125C0302.}

Large Language Models (LLMs) are increasingly integrated into safety-critical workflows, yet existing security analyses remain fragmented and often isolate model behavior from the broader system context. This work introduces a goal-driven risk assessment framework for LLM-powered systems that combines system modeling with Attack-Defense Trees (ADTrees) and Common Vulnerability Scoring System (CVSS)-based exploitability scoring to support structured, comparable analysis. We demonstrate the framework through a healthcare case study, modeling multi-step attack paths targeting intervention in medical procedures, leakage of electronic health record (EHR) data, and disruption of service availability. Our analysis indicates that threats spanning (i) conventional cyber, (ii) adversarial ML, and (iii) conversational attacks that manipulate prompts or context often consolidate into a small number of dominant paths and shared system choke points, enabling targeted defenses to yield meaningful reductions in path exploitability. By systematically comparing defense portfolios, we align these risks with established vulnerability management practices and provide a domain-agnostic workflow applicable to other LLM-enabled critical systems.

The growing use of AI applications among freelance workers is reshaping trust and relationships with clients. This paper investigates how both workers and clients perceive AI use and disclosure in the freelance economy through a three-stage study: interviews with workers and two survey studies with workers and clients. Findings first reveal a key expectation gap around disclosure: Workers often adopt passive disclosure practices, revealing AI use only when asked, as they assume clients can already detect it. Clients, however, are far less confident in recognizing AI-assisted work and prefer proactive disclosure. A second finding highlights the role of unclear or absent client AI policies, which leave workers consistently misinterpreting clients' expectations for AI use and disclosure. Together, these gaps point to the need for clearer guidelines and practices for AI disclosure. Insights extend beyond freelancing, offering implications for trust, accountability, and policy design in other AI-mediated work domains.

Unmanned Aerial Vehicular Networks (UAVNs) are envisioned to provide flexible connectivity, wide-area coverage, and low-latency services in dynamic environments. From an agentic artificial intelligence (Agentic AI) perspective, UAVNs naturally operate as multi-agent systems, where autonomous UAVs act as intelligent agents that coordinate deployment and networking decisions to achieve global performance objectives. However, the strong coupling between discrete link decisions and continuous deployment parameters makes UAVN topology optimization a mixed-integer nonconvex problem, resulting in challenges in scalability, efficiency, and solution consistency under dynamic network conditions. This paper proposes a dual spatial-scale UAVN topology optimization framework based on exact potential games (EPGs), enhanced by Agentic AI. At the large spatial scale, a log-linear learning based EPG (L3-EPG) algorithm is developed to optimize inter-UAV link configurations, enabling sparse yet connected network topologies while reducing redundant links and interference. At the small spatial scale, an approximate gradient based EPG (AG-EPG) algorithm jointly optimizes UAV deployment, transmission power allocation, and ground user (GU) association to improve network throughput and latency. To further enhance adaptability across heterogeneous scenarios, a large language model (LLM) is incorporated as a knowledge-driven decision enhancer to automatically generate utility weights according to network characteristics, alleviating reliance on manual parameter tuning. Simulation results demonstrate that the proposed framework consistently outperforms baseline methods in terms of energy consumption, end-to-end latency, and system throughput.

Image generation has advanced rapidly over the past decade, yet the literature seems fragmented across different models and application domains. This paper aims to offer a comprehensive survey of breakthrough image generation models, including variational autoencoders (VAEs), generative adversarial networks (GANs), normalizing flows, autoregressive and transformer-based generators, and diffusion-based methods. We provide a detailed technical walkthrough of each model type, including their underlying objectives, architectural building blocks, and algorithmic training steps. For each model type, we present the optimization techniques as well as common failure modes and limitations. We also go over recent developments in video generation and present the research works that made it possible to go from still frames to high quality videos. Lastly, we cover the growing importance of robustness and responsible deployment of these models, including deepfake risks, detection, artifacts, and watermarking.

We present SLNet, a lightweight backbone for 3D point cloud recognition designed to achieve strong performance without the computational cost of many recent attention, graph, and deep MLP based models. The model is built on two simple ideas: NAPE (Nonparametric Adaptive Point Embedding), which captures spatial structure using a combination of Gaussian RBF and cosine bases with input adaptive bandwidth and blending, and GMU (Geometric Modulation Unit), a per channel affine modulator that adds only 2D learnable parameters. These components are used within a four stage hierarchical encoder with FPS+kNN grouping, nonparametric normalization, and shared residual MLPs. In experiments, SLNet shows that a very small model can still remain highly competitive across several 3D recognition tasks. On ModelNet40, SLNet-S with 0.14M parameters and 0.31 GFLOPs achieves 93.64% overall accuracy, outperforming PointMLP-elite with 5x fewer parameters, while SLNet-M with 0.55M parameters and 1.22 GFLOPs reaches 93.92%, exceeding PointMLP with 24x fewer parameters. On ScanObjectNN, SLNet-M achieves 84.25% overall accuracy within 1.2 percentage points of PointMLP while using 28x fewer parameters. For large scale scene segmentation, SLNet-T extends the backbone with local Point Transformer attention and reaches 58.2% mIoU on S3DIS Area 5 with only 2.5M parameters, more than 17x fewer than Point Transformer V3. We also introduce NetScore+, which extends NetScore by incorporating latency and peak memory so that efficiency can be evaluated in a more deployment oriented way. Across multiple benchmarks and hardware settings, SLNet delivers a strong overall balance between accuracy and efficiency. Code is available at: https://github.com/m-saeid/SLNet.

Backdoor mechanisms have traditionally been studied as security threats that compromise the integrity of machine learning models. However, the same mechanism -- the conditional activation of specific behaviors through input triggers -- can also serve as a controllable and auditable interface for trustworthy model behavior. In this work, we present \textbf{Backdoor4Good (B4G)}, a unified benchmark and framework for \textit{beneficial backdoor} applications in large language models (LLMs). Unlike conventional backdoor studies focused on attacks and defenses, B4G repurposes backdoor conditioning for Beneficial Tasks that enhance safety, controllability, and accountability. It formalizes beneficial backdoor learning under a triplet formulation $(T, A, U)$, representing the \emph{Trigger}, \emph{Activation mechanism}, and \emph{Utility function}, and implements a benchmark covering four trust-centric applications. Through extensive experiments across Llama3.1-8B, Gemma-2-9B, Qwen2.5-7B, and Llama2-13B, we show that beneficial backdoors can achieve high controllability, tamper-resistance, and stealthiness while preserving clean-task performance. Our findings demonstrate new insights that backdoors need not be inherently malicious; when properly designed, they can serve as modular, interpretable, and beneficial building blocks for trustworthy AI systems. Our code and datasets are available at https://github.com/bboylyg/BackdoorLLM/B4G.

Enterprises commonly deploy heterogeneous database systems, each of which owns a distinct SQL dialect with different syntax rules, built-in functions, and execution constraints. However, most existing NL2SQL methods assume a single dialect (e.g., SQLite) and struggle to produce queries that are both semantically correct and executable on target engines. Prompt-based approaches tightly couple intent reasoning with dialect syntax, rule-based translators often degrade native operators into generic constructs, and multi-dialect fine-tuning suffers from cross-dialect interference. In this paper, we present Dial, a knowledge-grounded framework for dialect-specific NL2SQL. Dial introduces: (1) a Dialect-Aware Logical Query Planning module that converts natural language into a dialect-aware logical query plan via operator-level intent decomposition and divergence-aware specification; (2) HINT-KB, a hierarchical intent-aware knowledge base that organizes dialect knowledge into (i) a canonical syntax reference, (ii) a declarative function repository, and (iii) a procedural constraint repository; and (3) an execution-driven debugging and semantic verification loop that separates syntactic recovery from logic auditing to prevent semantic drift. We construct DS-NL2SQL, a benchmark covering six major database systems with 2,218 dialect-specific test cases. Experimental results show that Dial consistently improves translation accuracy by 10.25% and dialect feature coverage by 15.77% over state-of-the-art baselines. The code is at https://github.com/weAIDB/Dial.

Gradient boosting still dominates Transformers on tabular benchmarks. Our tokenizer uses a deliberately simplistic discretized vocabulary so we can highlight how even basic tokenization unlocks the power of attention on tabular features, yet it already outperforms tuned gradient boosting when combined with Gaussian smoothing. Our solution discretizes environmental context while smoothing labels with adaptive Gaussians, yielding calibrated PDFs. On 600K entities (5M training examples) we outperform tuned XGBoost by 10.8% (35.94s vs 40.31s median MAE) and achieve KS=0.0045 with the adaptive-sigma checkpoint selected to minimize KS rather than median MAE. Ablations confirm architecture matters: losing sequential ordering costs about 2.0%, dropping the time-delta tokens costs about 1.8%, and a stratified calibration analysis reveals where miscalibration persists.

Large language models (LLMs) often require fine-tuning (FT) to perform well on downstream tasks, but FT can induce safety-alignment drift even when the training dataset contains only benign data. Prior work shows that introducing a small fraction of harmful data can substantially compromise LLM refusal behavior, causing LLMs to comply with harmful requests. Existing defense methods often rely on model-wide interventions, such as restricting which parameters are updated or injecting additional safety data, which can limit generality and degrade downstream task performance. To address these limitations, we propose a fine-tuning framework called Preserving Safety Alignment via Constrained Tokens (PACT), which stabilizes the model's confidence on safety tokens. Our approach is motivated by the empirical observation that safety-aligned behavior is reflected in the model's token-level output confidence and is often concentrated on a small subset of safety-related tokens. During downstream fine-tuning, we regularize the fine-tuned model to match the aligned reference model's confidence on safety-related tokens at each response step, while leaving non-safety tokens largely unconstrained to allow effective task adaptation. This targeted constraint prevents alignment drift without imposing global restrictions that typically trade off with model utility.

Large language models (LLMs) have enabled agent-based systems that aim to automate scientific research workflows. Most existing approaches focus on fully autonomous discovery, where AI systems generate research ideas, conduct analyses, and produce manuscripts with minimal human involvement. However, empirical research in economics and the social sciences poses additional constraints: research questions must be grounded in available datasets, identification strategies require careful design, and human judgment remains essential for evaluating economic significance. We introduce HLER (Human-in-the-Loop Economic Research), a multi-agent architecture that supports empirical research automation while preserving critical human oversight. The system orchestrates specialized agents for data auditing, data profiling, hypothesis generation, econometric analysis, manuscript drafting, and automated review. A key design principle is dataset-aware hypothesis generation, where candidate research questions are constrained by dataset structure, variable availability, and distributional diagnostics, reducing infeasible or hallucinated hypotheses. HLER further implements a two-loop architecture: a question quality loop that screens and selects feasible hypotheses, and a research revision loop where automated review triggers re-analysis and manuscript revision. Human decision gates are embedded at key stages, allowing researchers to guide the automated pipeline. Experiments on three empirical datasets show that dataset-aware hypothesis generation produces feasible research questions in 87% of cases (versus 41% under unconstrained generation), while complete empirical manuscripts can be produced at an average API cost of $0.8-$1.5 per run. These results suggest that Human-AI collaborative pipelines may provide a practical path toward scalable empirical research.

Regulatory stress testing requires projecting credit losses under hypothetical macroeconomic scenarios -- a fundamentally causal question typically treated as a prediction problem. We propose a framework for policy-path counterfactual inference in panels that transparently separates what can be learned from data from what requires assumptions about confounding. Our approach has four components: (i) observational identification of path-conditional means via iterated regression, enabling continuous macro-path contrasts without requiring a control group; (ii) causal set identification under bounded confounding, yielding sharp identified sets with interpretable breakdown values that communicate robustness in a single number; (iii) an oracle inequality showing that recursive rollout error is governed by a horizon-dependent amplification factor, providing a concrete answer to how far ahead one can reliably predict under stress; and (iv) importance-weighted conformal calibration bands with diagnostics that quantify extrapolation cost and trigger abstention when coverage guarantees degrade. The final output is a three-layer uncertainty decomposition that cleanly separates estimation uncertainty from confounding uncertainty. We validate all results through simulation and semi-synthetic experiments with real unemployment data, including a Covid retrospective demonstrating the framework's diagnostic value under extreme scenarios.

We study the problem of state representation learning for control from partial and potentially high-dimensional observations. We approach this problem via cost-driven state representation learning, in which we learn a dynamical model in a latent state space by predicting cumulative costs. In particular, we establish finite-sample guarantees on finding a near-optimal representation function and a near-optimal controller using the learned latent model for infinite-horizon time-invariant Linear Quadratic Gaussian (LQG) control. We study two approaches to cost-driven representation learning, which differ in whether the transition function of the latent state is learned explicitly or implicitly. The first approach has also been investigated in Part I of this work, for finite-horizon time-varying LQG control. The second approach closely resembles MuZero, a recent breakthrough in empirical reinforcement learning, in that it learns latent dynamics implicitly by predicting cumulative costs. A key technical contribution of this Part II is to prove persistency of excitation for a new stochastic process that arises from the analysis of quadratic regression in our approach, and may be of independent interest.

Dynamic Data selection aims to accelerate training by prioritizing informative samples during online training. However, existing methods typically rely on task-specific handcrafted metrics or static/snapshot-based criteria to estimate sample importance, limiting scalability across learning paradigms and making it difficult to capture the evolving utility of data throughout training. To address this challenge, we propose Data Agent, an end-to-end dynamic data selection framework that formulates data selection as a training-aware sequential decision-making problem. The agent learns a sample-wise selection policy that co-evolves with model optimization, guided by a composite reward that integrates loss-based difficulty and confidence-based uncertainty signals. The reward signals capture complementary objectives of optimization impact and information gain, together with a tuning-free adaptive weighting mechanism that balances these signals over training. Extensive experiments across a wide range of datasets and architectures demonstrate that Data Agent consistently accelerates training while preserving or improving performance, e.g., reducing costs by over 50\% on ImageNet-1k and MMLU with lossless performance. Moreover, its dataset-agnostic formulation and modular reward make it plug-and-play across tasks and scenarios, e.g., robustness to noisy datasets, highlighting its potential in real-world scenarios.

Graphical user interface (GUI)-based mobile agents automate digital tasks on mobile devices by interpreting natural-language instructions and interacting with the screen. While recent methods apply reinforcement learning (RL) to train vision-language-model(VLM) agents in interactive environments with a primary focus on performance, generalization remains underexplored due to the lack of standardized benchmarks and open-source RL systems. In this work, we formalize the problem as a Contextual Markov Decision Process (CMDP) and introduce \textbf{AndroidWorld-Generalization}, a benchmark with three increasingly challenging regimes for evaluating zero-shot generalization to unseen task instances, templates, and applications. We further propose an RL training system that integrates Group Relative Policy Optimization (GRPO) with a scalable rollout collection system, consisting of containerized infrastructure and asynchronous execution % , and error recovery to support reliable and efficient training. Experiments on AndroidWorld-Generalization show that RL enables a 7B-parameter VLM agent to surpass supervised fine-tuning baselines, yielding a 26.1\% improvement on unseen instances but only limited gains on unseen templates (15.7\%) and apps (8.3\%), underscoring the challenges of generalization. As a preliminary step, we demonstrate that few-shot adaptation at test-time improves performance on unseen apps, motivating future research in this direction. To support reproducibility and fair comparison, we open-source the full RL training system, including the environment, task suite, models, prompt configurations, and the underlying infrastructure \footnote{https://github.com/zihuanjiang/AndroidWorld-Generalization}.

Transformer architectures excel at sequential modeling yet remain fundamentally limited by correlational learning - they capture spurious associations induced by latent confounders rather than invariant causal mechanisms. We identify this as an epistemological challenge: standard Transformers conflate static background factors (intrinsic identity, style, context) with dynamic causal flows (state evolution, mechanism), leading to catastrophic out-of-distribution failure. We propose OrthoFormer, a causally grounded architecture that embeds instrumental variable estimation directly into Transformer blocks via neural control functions. Our framework rests on four theoretical pillars: Structural Directionality (time-arrow enforcement), Representation Orthogonality (latent-noise separation), Causal Sparsity (Markov Blanket approximation), and End-to-End Consistency (gradient- detached stage separation). We prove that OrthoFormer achieves bias strictly less than OLS for any valid instrument lag, with residual bias decaying geometrically as O(\r{ho}k ). We characterize the bias-variance-exogeneity trilemma inherent in self-instrumenting and identify the neural forbidden regression - where removing gradient detachment improves prediction loss while destroying causal validity. Experiments confirm all theoretical predictions. OrthoFormer represents a paradigm shift from correlational to causal sequence modeling, with implications for robustness, interpretability, and reliable decision-making under distribution shift.

As Large Language Models (LLMs) evolve into autonomous agents, existing safety evaluations face a fundamental trade-off: manual benchmarks are costly, while LLM-based simulators are scalable but suffer from logic hallucination. We present AutoControl Arena, an automated framework for frontier AI risk evaluation built on the principle of logic-narrative decoupling. By grounding deterministic state in executable code while delegating generative dynamics to LLMs, we mitigate hallucination while maintaining flexibility. This principle, instantiated through a three-agent framework, achieves over 98% end-to-end success and 60% human preference over existing simulators. To elicit latent risks, we vary environmental Stress and Temptation across X-Bench (70 scenarios, 7 risk categories). Evaluating 9 frontier models reveals: (1) Alignment Illusion: risk rates surge from 21.7% to 54.5% under pressure, with capable models showing disproportionately larger increases; (2) Scenario-Specific Safety Scaling: advanced reasoning improves robustness for direct harms but worsens it for gaming scenarios; and (3) Divergent Misalignment Patterns: weaker models cause non-malicious harm while stronger models develop strategic concealment.

Transit agencies that operate on-demand transportation services have to respond to trip requests from passengers in real time, which involves solving dynamic vehicle routing problems with pick-up and drop-off constraints. Based on discussions with public transit agencies, we observe a real-world problem that is not addressed by prior work: when trips are booked in advance (e.g., trip requests arrive a few hours in advance of their requested pick-up times), the agency needs to promptly confirm whether a request can be accepted or not, and ensure that accepted requests are served as promised. State-of-the-art computational approaches either provide prompt confirmation but lack the ability to continually optimize and improve routes for accepted requests, or they provide continual optimization but cannot guarantee serving all accepted requests. To address this gap, we introduce a novel problem formulation of dynamic vehicle routing with prompt confirmation and continual optimization. We propose a novel computational approach for this vehicle routing problem, which integrates a quick insertion search for prompt confirmation with an anytime algorithm for continual optimization. To maximize the number requests served, we train a non-myopic objective function using reinforcement learning, which guides both the insertion and the anytime algorithms towards optimal, non-myopic solutions. We evaluate our computational approach on a real-world microtransit dataset from a public transit agency in the U.S., demonstrating that our proposed approach provides prompt confirmation while significantly increasing the number of requests served compared to existing approaches.

\textbf{Context:} Empathy is increasingly recognized as a critical human capability for software engineers, supporting collaboration, ethical awareness, and user-centered design. While many disciplines have long explored empathy as part of professional formation, its incorporation into software engineering education remains fragmented. \textbf{Aim:} This study investigates how empathy has been used, taught, and discussed in general engineering and software engineering education, with the goal of identifying pedagogical practices, outcomes, and disciplinary differences that inform the structured integration of empathy into software curricula. \textbf{Method:} Following established guidelines for systematic reviews in software engineering, we conducted a comprehensive search across six databases and analyzed 43 primary studies published between 2001 and 2025. Data were coded and synthesized using descriptive and thematic analysis to capture how empathy is conceptualized, fostered, and assessed across educational contexts. \textbf{Findings:} Our findings show that engineering programs frame empathy as an ethical and reflective capacity linked to social responsibility, whereas software engineering translates empathy into structured, design-oriented, and measurable practices. Across both domains, empathy teaching enhances collaboration, ethical reasoning, bias awareness, and motivation, but remains limited by low curricular prioritization, measurement challenges, and resource constraints. \textbf{Conclusion:} Empathy is evolving from a peripheral soft skill into a measurable pedagogical construct in software engineering education. Embedding empathy as a continuous, assessable component of design and development courses can strengthen inclusivity, ethical reflection, and responsible innovation in future software professionals.

Remote and hybrid work have transformed how software development teams organize, communicate, and assure quality. This study investigates how regression testing is performed and experienced under these distributed conditions. Using qualitative interviews with twenty software professionals from diverse organizations, we analyzed how regression testing processes, tools, and coordination practices adapt to remote and hybrid environments. The results show that while the core phases of regression testing remain stable, their execution increasingly depends on documentation, automation, and tool integration to support asynchronous collaboration. Communication and coordination challenges were mitigated through standardized reporting, shared repositories, and traceability mechanisms that replaced informal co-located interactions. These findings reveal regression testing as a socio-technical practice shaped by the interaction between human collaboration and digital infrastructure. Our study contributes to understanding how software quality assurance evolves under remote conditions and offers practical implications for teams and organizations adopting hybrid work models.

Large language model-based deep research agents have been increasingly popular for addressing long-horizon information-seeking tasks, but they often incur high end-to-end latency due to extensive reasoning and frequent tool use. Speculation frameworks aim to reduce latency by overlapping action execution with reasoning; however, existing approaches typically rely on uniform speculation strategies and strict action matching, which limits inference speedups and robustness. In this work, we revisit the speculate-verify paradigm for deep research agents through the lens of action heterogeneity. We show that \textit{Search} and \textit{Visit} actions exhibit fundamentally different reasoning and model capacity requirements: entropy-based analysis reveals that Search decisions have higher uncertainty and benefit significantly from explicit reasoning, whereas Visit decisions have lower entropy and depend primarily on model capacity. Motivated by this dual-process characteristic, we propose DualSpec, a heterogeneous speculation framework equipped with a lightweight, confidence-based semantic verifier. Experiments across multiple models and benchmarks demonstrate that DualSpec achieves up to 3.28$\times$ end-to-end speedup while maintaining accuracy comparable to fully reasoning agents.

Catastrophic forgetting remains a central challenge in continual learning (CL), yet lacks a unified information-theoretic explanation for why some architectures forget catastrophically while others do not. We introduce \emph{Context Channel Capacity} ($C_\mathrm{ctx}$), the mutual information between a CL architecture's context signal and its generated parameters, and prove that zero forgetting requires $C_\mathrm{ctx} \geq H(T)$, where $H(T)$ is the task identity entropy. We establish an \emph{Impossibility Triangle} -- zero forgetting, online learning, and finite parameters cannot be simultaneously satisfied by sequential state-based learners -- and show that conditional regeneration architectures (HyperNetworks) bypass this triangle by redefining parameters as function values rather than states. We validate this framework across 8 CL methods on Split-MNIST (1,130+ experiments over 86 days, 4 seeds each), showing that $C_\mathrm{ctx}$ perfectly predicts forgetting behavior: methods with $C_\mathrm{ctx} = 0$ (NaiveSGD, EWC, SI, LwF, CFlow) exhibit catastrophic forgetting (6--97\%), while methods with $C_\mathrm{ctx} \approx 1$ (HyperNetwork) achieve zero forgetting (98.8\% ACC). We further propose \emph{Wrong-Context Probing} (P5), a practical diagnostic protocol for measuring $C_\mathrm{ctx}$, and extend the framework to CIFAR-10 via a novel \emph{Gradient Context Encoder} that closes the oracle gap from 23.3pp to 0.7pp. A systematic taxonomy of 15+ closed research directions -- including the Hebbian null result (frozen random features outperform learned features), CFlow's $θ_0$-memorizer phenomenon, and the $S_N$ symmetry barrier to column specialization -- provides the community with precisely diagnosed negative results. Our central design principle: \emph{architecture over algorithm} -- the context pathway must be structurally unbypassable.

The Internet of Underwater Things (IoUT) is becoming a critical infrastructure for ocean observation, marine resource management, and climate science. Its development is hindered by severe acoustic attenuation, propagation delays far exceeding those of terrestrial wireless systems, strict energy constraints, and dynamic topologies shaped by ocean currents. Machine learning (ML) has emerged as a key enabler for addressing these limitations, offering data driven mechanisms that enhance performance across all layers of underwater wireless sensor networks. This tutorial survey synthesises ML methodologies supervised, unsupervised, reinforcement, and deep learning specifically contextualised for underwater communication environments. It outlines the algorithmic principles of each paradigm and examines the conditions under which particular approaches deliver superior performance. A layer wise analysis highlights physical layer gains in localisation and channel estimation, MAC layer adaptations that improve channel utilisation, network layer routing strategies that extend operational lifetime, and transport layer mechanisms capable of reducing packet loss by up to 91 percent. At the application layer, ML enables substantial data compression and object detection accuracies reaching 92 percent. Drawing on 300 studies from 2012 to 2025, the survey documents energy efficiency gains of 7 to 29 times, throughput improvements over traditional protocols, and cross layer optimisation benefits of up to 42 percent. It also identifies persistent barriers, including limited datasets, computational constraints, and the gap between theoretical models and real world deployment. The survey concludes with emerging research directions and a technology roadmap supporting ML adoption in operational underwater networks.

Universal image restoration aims to recover clean images from arbitrary real-world degradations using a single inference model. Despite significant progress, existing all-in-one restoration networks do not scale to multiple degradations. As the number of degradations increases, training becomes unstable, models grow excessively large, and performance drops across both seen and unseen domains. In this work, we show that scaling universal restoration is fundamentally limited by interference across degradations during joint learning, leading to catastrophic task forgetting. To address this challenge, we introduce a unified inference pipeline with a multi-branch mixture-of-experts architecture that decomposes restoration knowledge across specialized task-adaptable experts. Our approach enables scalable learning (over sixteen degradations), adapts and generalizes robustly to unseen domains, and supports user-controllable restoration across degradations. Beyond achieving superior performance across benchmarks, this work establishes a new design paradigm for scalable and controllable universal image restoration.

Vision language action models (VLAs) are increasingly used for Physical AI, but deploying a pre-trained VLA model to unseen environments, embodiments, or tasks still requires adaptation. Parameter-efficient fine-tuning (PEFT), especially LoRA, is common for VLA policies, yet the exposed capacity knob, the rank, does not transfer uniformly: robotics transfer exhibits a higher and task-varying intrinsic rank than language fine-tuning. Small ranks suffice for LLMs (e.g., $r \in \{4, 8\}$), while spectral analyses indicate VLAs may require much larger ranks (e.g., $r \approx 128$) or near-full rank, a mismatch that worsens in multi-task settings. We present LoRA-SP (Select-Prune), a rank-adaptive fine-tuning method that replaces fixed-rank updates with input- and layer-wise capacity. LoRA-SP uses an SVD-style parameterization with a small router whose nonnegative scores act as singular values over a shared vector bank. The active set is chosen by an energy target on the cumulative squared scores $E(k) \ge η$, providing a direct link to approximation error via our spectral analysis. During training, $η$ concentrates energy on a few directions and teaches the router to rely on fewer vectors while preserving accuracy. This yields compact adapters that reduce cross-task interference and improve generalization. On four real-robot manipulation tasks collected on an unseen AgileX PiPER arm, across two VLA backbones ($π_0$ and SmolVLA), LoRA-SP matches or exceeds full fine-tuning with far fewer trainable parameters, and improves multi-task success by up to 31.6% over standard LoRA while remaining robust to rank choice.

Linear autoencoders (LAEs) have gained increasing popularity in recommender systems due to their simplicity and strong empirical performance. Most LAE models, including the Emphasized Denoising Linear Autoencoder (EDLAE) introduced by (Steck, 2020), use quadratic loss during training. However, the original EDLAE only provides closed-form solutions for the hyperparameter choice $b = 0$, which limits its capacity. In this work, we generalize EDLAE objective into a Decoupled Expected Quadratic Loss (DEQL). We show that DEQL simplifies the process of deriving EDLAE solutions and reveals solutions in a broader hyperparameter range $b > 0$, which were not derived in Steck's original paper. Additionally, we propose an efficient algorithm based on Miller's matrix inverse theorem to ensure the computational tractability for the $b > 0$ case. Empirical results on benchmark datasets show that the $b > 0$ solutions provided by DEQL outperform the $b = 0$ EDLAE baseline, demonstrating that DEQL expands the solution space and enables the discovery of models with better testing performance.

Visual Question Answering (VQA) is a core task for evaluating the capabilities of Vision-Language Models (VLMs). Existing VQA benchmarks primarily feature clear and unambiguous image-question pairs, whereas real-world scenarios often involve varying degrees of ambiguity that require nuanced reasoning and context-appropriate response strategies. Although recent studies have begun to address ambiguity in VQA, they lack (1) a systematic categorization of ambiguity levels and (2) datasets and models that support strategy-aware responses. In this paper, we introduce Ambiguous Visual Question Answering (AQuA), a fine-grained dataset that classifies ambiguous VQA instances into four levels according to the nature and degree of ambiguity, along with the optimal response strategy for each case. Our evaluation of diverse open-source and proprietary VLMs shows that most models fail to adapt their strategy to the ambiguity type, frequently producing overconfident answers rather than seeking clarification or acknowledging uncertainty. To address this challenge, we fine-tune VLMs on AQuA, enabling them to adaptively choose among multiple response strategies, such as directly answering, inferring intent from contextual cues, listing plausible alternatives, or requesting clarification. VLMs trained on AQuA achieve strategic response generation for ambiguous VQA, demonstrating the ability to recognize ambiguity, manage uncertainty, and respond with context-appropriate strategies, while outperforming both open-source and closed-source baselines.

LLMs operating in dynamic real-world contexts often encounter knowledge that evolves continuously or emerges incrementally. To remain accurate and effective, models must adapt to newly arriving information on the fly. We introduce Online Adaptation to Continual Knowledge Streams(OAKS) to evaluate this capability, establishing a benchmark for online adaptation over streaming, continually updating knowledge. Specifically, the benchmark is structured as a sequence of fine-grained context chunks where facts change dynamically across time intervals. OAKS comprises two datasets: OAKS-BABI and OAKS-Novel, where individual facts evolve multiple times across context chunks. These datasets include dense annotations to measure whether models track changes accurately. Evaluating 14 models with varied inference approaches, we observe significant limitations in current methodologies. Both state-of-the-art models and agentic memory systems fail to adapt robustly on OAKS, demonstrating delays in state-tracking and susceptibility to distraction within streaming environments.

For fifty years, networking has fragmented whenever new workloads exposed hidden assumptions about time, ordering, failure, and trust. This paper argues that the current interconnect landscape -- NVLink, UALink, Ultra Ethernet, AELink/Aethernet, TTPoE, and classical RDMA -- suffers from a semantic crisis: vendor-specific divergence disguised as optimization. We trace this crisis to the Forward-In-Time-Only (FITO) category mistake embedded in every major fabric stack, and show how each pathology -- aspirational RDMA completion, fire-and-forget GPU semantics, opaque proprietary stacks, incompatible multi-cloud ordering, universal fencing -- arises from the same failure to define explicit, testable link semantics from APIs to bits on the wire. We conjecture that RDMA achieves reliability through universal fencing that collapses concurrency into serialized checkpoints, and that precise minimal semantics can maintain correctness without global barriers, as superscalar architectures separated execution from retirement. We describe how Open Atomic Ethernet (OAE) under the Open Compute Project addresses the crisis through bilateral transaction primitives with explicit ordering, completion, and failure visibility. Drawing on Helland's analysis of scalable OLTP isolation (the "BIG DEAL"), we show the crisis pervades the entire stack. We assess whether convergence on a single open standard is still possible or whether fragmentation is now structural.

Legal teams increasingly use machine learning to triage large volumes of contractual evidence, but many models are opaque, non-deterministic, and difficult to align with frameworks such as HIPAA or NERC-CIP. We study a simple, reproducible alternative based on deterministic dual encoders and transparent fuzzy triage bands. We train a RoBERTa-base dual encoder with a 512-dimensional projection and cosine similarity on the ACORD benchmark for graded clause retrieval, then fine-tune it on a CUAD-derived binary compliance dataset. Across five random seeds (40-44) on a single NVIDIA A100 GPU, the model achieves ACORD-style retrieval performance of NDCG@5 0.38-0.42, NDCG@10 0.45-0.50, and 4-star Precision@5 about 0.37 on the test split. On CUAD-derived binary labels, it achieves AUC 0.98-0.99 and F1 0.22-0.30 depending on positive-class weighting, outperforming majority and random baselines in a highly imbalanced setting with a positive rate of about 0.6%. We then map scalar compliance scores into three regions: auto-noncompliant, auto-compliant, and human-review. Thresholds are tuned on validation data to maximize automatic decision coverage subject to an empirical error-rate constraint of at most 2% over auto-decided examples. The result is a seed-stable system summarized by a small number of scalar parameters. We argue that deterministic encoders, calibrated fuzzy bands, and explicit error constraints provide a practical middle ground between hand-crafted rules and opaque large language models, supporting explainable evidence triage, reproducible audit trails, and concrete mappings to legal review concepts.

In machine learning, the goal of multi-task learning (MTL) is to optimize multiple objectives together. Recent works, for example, Multiple Gradient Descent Algorithm (MGDA) and its variants, show promising results with dynamically adjusted weights for different tasks to mitigate conflicts that may potentially degrade the performance on certain tasks. Despite the empirical success of MGDA-type methods, one major limitation of such methods is their computational inefficiency, as they require access to all task gradients. In this paper we introduce MARIGOLD, a unified algorithmic framework for efficiently solving MTL problems. Our method reveals that multi-task gradient balancing methods have a hierarchical structure, in which the model training and the gradient balancing are coupled during the whole optimization process and can be viewed as a bi-level optimization problem. Moreover, we showcase that the bi-level problem can be solved efficiently by leveraging zeroth-order method. Extensive experiments on both public datasets and industrial-scale datasets demonstrate the efficiency and superiority of our method.

Explaining out-of-distribution generalization has been a central problem in epistemology since Goodman's "grue" puzzle in 1946. Today it's a central problem in machine learning, including AI alignment. Here we propose a principled account of OOD generalization with three main ingredients. First, the world is always presented to experience not as an amorphous mass, but via distinguished features (for example, visual and auditory channels). Second, Occam's Razor favors hypotheses that are "sparse," meaning that they depend on as few features as possible. Third, sparse hypotheses will generalize from a training to a test distribution, provided the two distributions sufficiently overlap on their restrictions to the features that are either actually relevant or hypothesized to be. The two distributions could diverge arbitrarily on other features. We prove a simple theorem that formalizes the above intuitions, generalizing the classic sample complexity bound of Blumer et al. to an OOD context. We then generalize sparse classifiers to subspace juntas, where the ground truth classifier depends solely on a low-dimensional linear subspace of the features.

Retrieval-Augmented Generation (RAG) systems are increasingly evolving into agentic architectures where large language models autonomously coordinate multi-step reasoning, dynamic memory management, and iterative retrieval strategies. Despite rapid industrial adoption, current research lacks a systematic understanding of Agentic RAG as a sequential decision-making system, leading to highly fragmented architectures, inconsistent evaluation methodologies, and unresolved reliability risks. This Systematization of Knowledge (SoK) paper provides the first unified framework for understanding these autonomous systems. We formalize agentic retrieval-generation loops as finite-horizon partially observable Markov decision processes, explicitly modeling their control policies and state transitions. Building upon this formalization, we develop a comprehensive taxonomy and modular architectural decomposition that categorizes systems by their planning mechanisms, retrieval orchestration, memory paradigms, and tool-invocation behaviors. We further analyze the critical limitations of traditional static evaluation practices and identify severe systemic risks inherent to autonomous loops, including compounding hallucination propagation, memory poisoning, retrieval misalignment, and cascading tool-execution vulnerabilities. Finally, we outline key doctoral-scale research directions spanning stable adaptive retrieval, cost-aware orchestration, formal trajectory evaluation, and oversight mechanisms, providing a definitive roadmap for building reliable, controllable, and scalable agentic retrieval systems.

The rapid growth of AI training has dramatically increased datacenter traffic demand and energy consumption, which has motivated renewed interest in optical circuit switches (OCSes) as a high-bandwidth, energy-efficient alternative for AI fabrics. Deploying multiple parallel OCSes is a leading alternative. However, efficiently scheduling time-varying traffic matrices across parallel optical switches with non-negligible reconfiguration delays remains an open challenge. We consider the problem of scheduling a single AI traffic demand matrix $D$ over $s$ parallel OCSes while minimizing the makespan under reconfiguration delay $δ$. Our algorithm Spectra relies on a three-step approach: Decompose $D$ into a minimal set of weighted permutations; Schedule these permutations across parallel switches using load-aware assignment; then Equalize the imbalanced loads on the switches via controlled permutation splitting. Evaluated on realistic AI training workloads (GPT model and Qwen MoE expert routing) as well as standard benchmarks, Spectra vastly outperforms a baseline based on state-of-the-art algorithms, reducing schedule makespan by an average factor of $1.4\times$ on GPT AI workloads, $1.9\times$ on MoE AI workloads, and $2.4\times$ on standard benchmarks. Further, the makespans achieved by Spectra consistently approach newly derived lower bounds.

Quality Estimation (QE) is essential for assessing machine translation quality in reference-less settings, particularly for domain-specific and low-resource language scenarios. In this paper, we investigate sentence-level QE for English to Indic machine translation across four domains (Healthcare, Legal, Tourism, and General) and five language pairs. We systematically compare zero-shot, few-shot, and guideline-anchored prompting across selected closed-weight and open-weight LLMs. Findings indicate that while closed-weight models achieve strong performance via prompting alone, prompt-only approaches remain fragile for open-weight models, especially in high-risk domains. To address this, we adopt ALOPE, a framework for LLM-based QE that uses Low-Rank Adaptation with regression heads attached to selected intermediate Transformer layers. We also extend ALOPE with recently proposed Low-Rank Multiplicative Adaptation (LoRMA). Our results show that intermediate-layer adaptation consistently improves QE performance, with gains in semantically complex domains, indicating a path toward more robust QE in practical scenarios. We release code and domain-specific QE datasets publicly to support further research.

The success of deep generative models in scientific discovery requires not only the ability to generate novel candidates but also reliable guarantees that these candidates indeed satisfy desired properties. Recent conformal-prediction methods offer a path to such guarantees, but its application to generative modeling in drug discovery is limited by budget constraints, lack of oracle access, and distribution shift. To this end, we introduce ConfHit, a distribution-free framework that provides validity guarantees under these conditions. ConfHit formalizes two central questions: (i) Certification: whether a generated batch can be guaranteed to contain at least one hit with a user-specified confidence level, and (ii) Design: whether the generation can be refined to a compact set without weakening this guarantee. ConfHit leverages weighted exchangeability between historical and generated samples to eliminate the need for an experimental oracle, constructs multiple-sample density-ratio weighted conformal p-value to quantify statistical confidence in hits, and proposes a nested testing procedure to certify and refine candidate sets of multiple generated samples while maintaining statistical guarantees. Across representative generative molecule design tasks and a broad range of methods, ConfHit consistently delivers valid coverage guarantees at multiple confidence levels while maintaining compact certified sets, establishing a principled and reliable framework for generative modeling.

Reconfigurable Intelligent Surfaces promise to transform wireless environments, yet practical deployment is hindered by the prohibitive overhead of Channel State Information (CSI) estimation and the dimensionality explosion inherent in centralized optimization. This paper proposes a Hierarchical Multi-Agent Reinforcement Learning (HMARL) framework for the control of mechanically reconfigurable reflective surfaces in millimeter-wave (mmWave) systems. We introduce a "CSI-free" paradigm that substitutes pilot-based channel estimation with readily available user localization data. To manage the massive combinatorial action space, the proposed architecture utilizes Multi-Agent Proximal Policy Optimization (MAPPO) under a Centralized Training with Decentralized Execution (CTDE) paradigm. The proposed architecture decomposes the control problem into two abstraction levels: a high-level controller for user-to-reflector allocation and decentralized low-level controllers for low-level focal point optimization. Comprehensive ray-tracing evaluations demonstrate that the framework achieves 2.81-7.94 dB RSSI improvements over centralized baselines, with the performance advantage widening as system complexity increases. Scalability analysis reveals that the system maintains sustained efficiency, exhibiting minimal per-user performance degradation and stable total power utilization even when user density doubles. Furthermore, robustness validation confirms the framework's viability across varying reflector aperture sizes (45-99 tiles) and demonstrates graceful performance degradation under localization errors up to 0.5 m. By eliminating CSI overhead while maintaining high-fidelity beam-focusing, this work establishes HMARL as a practical solution for intelligent mmWave environments.

Biases in large language models (LLMs) often manifest as systematic distortions in associations between demographic attributes and professional or social roles, reinforcing harmful stereotypes across gender, ethnicity, and geography. This position paper advocates for addressing demographic and gender biases in LLMs through a dual-pronged methodology, integrating category-theoretic transformations and retrieval-augmented generation (RAG). Category theory provides a rigorous, structure-preserving mathematical framework that maps biased semantic domains to unbiased canonical forms via functors, ensuring bias elimination while preserving semantic integrity. Complementing this, RAG dynamically injects diverse, up-to-date external knowledge during inference, directly countering ingrained biases within model parameters. By combining structural debiasing through functor-based mappings and contextual grounding via RAG, we outline a comprehensive framework capable of delivering equitable and fair model outputs. Our synthesis of the current literature validates the efficacy of each approach individually, while addressing potential critiques demonstrates the robustness of this integrated strategy. Ensuring fairness in LLMs, therefore, demands both the mathematical rigor of category-theoretic transformations and the adaptability of retrieval augmentation.

Many errors in student essays can be explained by influence from the native language (L1). L1 interference refers to errors influenced by a speaker's first language, such as using stadion instead of stadium, reflecting lexical transliteration from Russian. In this work, we address the task of detecting such errors in English essays written by Russian-speaking learners. We introduce RILEC, a large-scale dataset of over 18,000 sentences, combining expert-annotated data from REALEC with synthetic examples generated through rule-based and neural augmentation. We propose a framework for generating L1-motivated errors using generative language models optimized with PPO, prompt-based control, and rule-based patterns. Models fine-tuned on RILEC achieve strong performance, particularly on word-level interference types such as transliteration and tense semantics. We find that the proposed augmentation pipeline leads to a significant performance improvement, making it a potentially valuable tool for learners and teachers to more effectively identify and address such errors.

Neural scaling laws describe how model performance improves as a power law with size, but existing work focuses on models above 100M parameters. The sub-20M regime -- where TinyML and edge AI operate -- remains unexamined. We train 90 models (22K--19.8M parameters) across two architectures (plain ConvNet, MobileNetV2) on CIFAR-100, varying width while holding depth and training fixed. Both follow approximate power laws in error rate: $α= 0.156 \pm 0.002$ (ScaleCNN) and $α= 0.106 \pm 0.001$ (MobileNetV2) across five seeds. Since prior work fit cross-entropy loss rather than error rate, direct exponent comparison is approximate; with that caveat, these are 1.4--2x steeper than $α\approx 0.076$ for large language models. The power law does not hold uniformly: local exponents decay with scale, and MobileNetV2 saturates at 19.8M parameters ($α_{\mathrm{local}} = 0.006$). Error structure also changes. Jaccard overlap between error sets of the smallest and largest ScaleCNN is only 0.35 (25 seed pairs, $\pm 0.004$) -- compression changes which inputs are misclassified, not merely how many. Small models concentrate capacity on easy classes (Gini: 0.26 at 22K vs. 0.09 at 4.7M) while abandoning the hardest (bottom-5 accuracy: 10% vs. 53%). Counter to expectation, the smallest models are best calibrated (ECE = 0.013 vs. peak 0.110 at mid-size). Aggregate accuracy is therefore misleading for edge deployment; validation must happen at the target model size.

Long-horizon wildfire risk forecasting requires generating probabilistic spatial fields under sparse event supervision while maintaining computational efficiency across multiple prediction horizons. Extending diffusion models to multi-step forecasting typically repeats the denoising process independently for each horizon, leading to redundant computation. We introduce N-Tree Diffusion (NT-Diffusion), a hierarchical diffusion model designed for long-horizon wildfire risk forecasting. Fire occurrences are represented as continuous Fire Risk Maps (FRMs), which provide a smoothed spatial risk field suitable for probabilistic modeling. Instead of running separate diffusion trajectories for each predicted timestamp, NT-Diffusion shares early denoising stages and branches at later levels, allowing horizon-specific refinement while reducing redundant sampling. We evaluate the proposed framework on a newly collected real-world wildfire dataset constructed for long-horizon probabilistic prediction. Results indicate that NT-Diffusion achieves consistent accuracy improvements and reduced inference cost compared to baseline forecasting approaches.

Designing environments that maximize the rate of emergent behavior development in AI agents remains an open problem. We present the first systematic study of stress-performance relationships in large language model (LLM) multi-agent systems, drawing an explicit parallel to the Yerkes-Dodson law from cognitive psychology. Using a grid-world survival arena, we conduct 22 experiments across four phases, varying environmental pressure through resource scarcity (upkeep cost) and reproductive competition (sexual selection). Our key finding is that cooperative behavior follows an inverted-U curve: trade interactions peak at 29 under medium pressure (upkeep=5), while both low and extreme pressure produce 8--12 trades. Under extreme pressure, behavioral repertoire collapses to movement-only within 5--12 turns. We further show that sexual selection -- a softer pressure mechanism where all agents survive but not all reproduce -- eliminates inter-agent aggression entirely and produces communicative behavior absent under survival pressure. These results suggest that environmental pressure calibration is a viable curriculum design strategy for LLM agent development, analogous to the inverted-U relationship between arousal and performance in biological systems.

Generative models have emerged as powerful priors for solving inverse problems. These models typically represent a class of natural signals using a single fixed complexity or dimensionality. This can be limiting: depending on the problem, a fixed complexity may result in high representation error if too small, or overfitting to noise if too large. We develop tunable-complexity priors for diffusion models, normalizing flows, and variational autoencoders, leveraging nested dropout. Across tasks including compressed sensing, inpainting, denoising, and phase retrieval, we show empirically that tunable priors consistently achieve lower reconstruction errors than fixed-complexity baselines. In the linear denoising setting, we provide a theoretical analysis that explicitly characterizes how the optimal tuning parameter depends on noise and model structure. This work demonstrates the potential of tunable-complexity generative priors and motivates both the development of supporting theory and their application across a wide range of inverse problems.

Machine learning models in agricultural vision often achieve high accuracy on curated datasets but fail to generalize under real field conditions due to distribution shifts between training and deployment environments. Moreover, most machine learning competitions focus primarily on model design while treating datasets as fixed resources, leaving the role of data collection practices in model generalization largely unexplored. We introduce the AgrI Challenge, a data-centric competition framework in which multiple teams independently collect field datasets, producing a heterogeneous multi-source benchmark that reflects realistic variability in acquisition conditions. To systematically evaluate cross-domain generalization across independently collected datasets, we propose Cross-Team Validation (CTV), an evaluation paradigm that treats each team's dataset as a distinct domain. CTV includes two complementary protocols: Train-on-One-Team-Only (TOTO), which measures single-source generalization, and Leave-One-Team-Out (LOTO), which evaluates collaborative multi-source training. Experiments reveal substantial generalization gaps under single-source training: models achieve near-perfect validation accuracy yet exhibit validation-test gaps of up to 16.20% (DenseNet121) and 11.37% (Swin Transformer) when evaluated on datasets collected by other teams. In contrast, collaborative multi-source training dramatically improves robustness, reducing the gap to 2.82% and 1.78%, respectively. The challenge also produced a publicly available dataset of 50,673 field images of six tree species collected by twelve independent teams, providing a diverse benchmark for studying domain shift and data-centric learning in agricultural vision.

We propose DistGP: a multi-robot learning method for collaborative learning of a global function using only local experience and computation. We utilise a sparse Gaussian process (GP) model with a factorisation that mirrors the multi-robot structure of the task, and admits distributed training via Gaussian belief propagation (GBP). Our loopy model outperforms Tree-Structured GPs \cite{bui2014tree} and can be trained online and in settings with dynamic connectivity. We show that such distributed, asynchronous training can reach the same performance as a centralised, batch-trained model, albeit with slower convergence. Last, we compare to DiNNO \cite{yu2022dinno}, a distributed neural network (NN) optimiser, and find DistGP achieves superior accuracy, is more robust to sparse communication and is better able to learn continually.

Clinical machine learning models are increasingly trained using large scale, multimodal foundation paradigms, yet deployment environments often differ systematically from the data generating settings used during training. Such shifts arise from heterogeneous measurement policies, documentation practices, and institutional workflows, leading to representation entanglement between physiologic signal and practice specific artifacts. In this work, we propose a practice invariant representation learning framework for multimodal clinical prediction. We model clinical observations as arising from latent physiologic factors and environment dependent processes, and introduce an objective that jointly optimizes predictive performance while suppressing environment predictive information in the learned embedding. Concretely, we combine supervised risk minimization with adversarial environment regularization and invariant risk penalties across hospitals. Across multiple longitudinal EHR prediction tasks and cross institution evaluations, our method improves out of distribution AUROC by up to 2 to 3 points relative to masked pretraining and standard supervised baselines, while maintaining in distribution performance and improving calibration. These results demonstrate that explicitly accounting for systematic distribution shift during representation learning yields more robust and transferable clinical models, highlighting the importance of structural invariance alongside architectural scale in healthcare AI.

Noisy training data can significantly degrade the performance of language-model-based classifiers, particularly in non-topical classification tasks. In this study we designed a methodological framework to assess the impact of denoising. More specifically, we explored a range of denoising strategies for sentence-level difficulty detection, using training data derived from document-level difficulty annotations obtained through noisy crowdsourcing. Beyond monolingual settings, we also address cross-lingual transfer, where a multilingual language model is trained in one language and tested in another. We evaluate several noise reduction techniques, including Gaussian Mixture Models (GMM), Co-Teaching, Noise Transition Matrices, and Label Smoothing. Our results indicate that while BERT-based models exhibit inherent robustness to noise, incorporating explicit noise detection can further enhance performance. For our smaller dataset, GMM-based noise filtering proves particularly effective in improving prediction quality by raising the Area-Under-the-Curve score from 0.52 to 0.92, or to 0.93 when de-noising methods are combined. However, for our larger dataset, the intrinsic regularisation of pre-trained language models provides a strong baseline, with denoising methods yielding only marginal gains (from 0.92 to 0.94, while a combination of two denoising methods made no contribution). Nonetheless, removing noisy sentences (about 20\% of the dataset) helps in producing a cleaner corpus with fewer infelicities. As a result we have released the largest multilingual corpus for sentence difficulty prediction: see https://github.com/Nouran-Khallaf/denoising-difficulty

Operating a global, real-time platform at Uber's scale requires infrastructure that is both resilient and cost-efficient. Historically, reliability was ensured through a costly 2x capacity model--each service provisioned to handle global traffic independently across two regions--leaving half the fleet idle. We present Uber's Failover Architecture (UFA), which replaces the uniform 2x model with a differentiated architecture aligned to business criticality. Critical services retain failover guarantees, while non-critical services opportunistically use failover buffer capacity reserved for critical services during steady state. During rare "full-peak" failovers, non-critical services are selectively preempted and rapidly restored, with differentiated Service-Level Agreements (SLAs) using on-demand capacity. Automated safeguards, including dependency analysis and regression gates, ensure critical services continue to function even while non-critical services are unavailable. The quantitative impact is significant: UFA reduces steady-state provisioning from 2x to 1.3x, raising utilization from ~20% to ~30% while sustaining 99.97% availability. To date, UFA has hardened over 4,000 unsafe dependencies, eliminated over one million CPU cores from a baseline of about four million cores.

Concept Bottleneck Models (CBMs) aim for ante-hoc interpretability by learning a bottleneck layer that predicts interpretable concepts before the decision. State-of-the-art approaches typically select which concepts to learn via human specification, open knowledge graphs, prompting an LLM, or using general CLIP concepts. However, concepts defined a-priori may not have sufficient predictive power for the task or even be learnable from the available data. As a result, these CBMs often significantly trail their black-box counterpart when controlling for information leakage. To address this, we introduce a novel CBM pipeline named Mechanistic CBM (M-CBM), which builds the bottleneck directly from a black-box model's own learned concepts. These concepts are extracted via Sparse Autoencoders (SAEs) and subsequently named and annotated on a selected subset of images using a Multimodal LLM. For fair comparison and leakage control, we also introduce the Number of Contributing Concepts (NCC), a decision-level sparsity metric that extends the recently proposed NEC metric. Across diverse datasets, we show that M-CBMs consistently surpass prior CBMs at matched sparsity, while improving concept predictions and providing concise explanations. Our code is available at https://github.com/Antonio-Dee/M-CBM.

Deliberative democratic theory suggests that civic competence: the capacity to navigate disagreement, weigh competing values, and arrive at collective decisions is not innate but developed through practice. Yet opportunities to cultivate these skills remain limited, as traditional deliberative processes like citizens' assemblies reach only a small fraction of the population. We present Agora, an early-stage AI-powered platform that uses LLMs to organize authentic human voices on policy issues, helping users build consensus-finding skills by proposing and revising policy recommendations, hearing supporting and opposing perspectives, and receiving feedback on how policy changes affect predicted support. In a preliminary study with 44 university students, participants using the full interface (with access to voice explanations) reported higher levels of problem-solving skills, internal deliberation, and produced higher quality consensus statements compared to a control condition showing only aggregate support distributions. These initial findings point toward a promising direction for scaling civic education.

All applications in fifth-generation (5G) networks rely on stable radio-frequency (RF) environments to support mission-critical services in mobility, automation, and connected intelligence. Their exposure to intentional interference or low-power jamming threatens availability and reliability, especially when such attacks remain below link-layer observability. This paper investigates lightweight, explainable, and hardware-efficient jamming detection using the Convolutional Tsetlin Machine (CTM) operating directly on 5G Synchronization Signal Block (SSB) features. CTM formulates Boolean logic clauses over quantized inputs, enabling bit-level inference and deterministic deployment on FPGA fabrics. These properties make CTM well suited for real-time, resource-constrained edge environments anticipated in 5G. The proposed approach is experimentally validated on a real 5G testbed using over-the-air SSB data, emulating practical downlink conditions. We benchmark CTM against a convolutional neural network (CNN) baseline under identical preprocessing and training pipelines. On the real dataset, CTM achieves comparable detection performance (Accuracy 91.53 +/- 1.01 vs. 96.83 +/- 1.19 for CNN) while training $9.5\times$ faster and requiring 14x less memory (45~MB vs.\ 624~MB). Furthermore, we outline a compact FPGA-oriented design for Zybo~Z7 (Zynq-7000) and provide resource projections (not measured) under three deployment profiles optimized for latency, power, and accuracy trade-offs. The results show that the CTM provides a practical, interpretable, and resource-efficient alternative to conventional DNNs for RF-domain jamming detection, establishing it as a strong candidate for edge-deployed, low-latency, and security-critical 5G applications while laying the groundwork for B5G systems.

High-performing vision language models still produce incorrect answers, yet their failure modes are often difficult to explain. To make model internals more accessible and enable systematic debugging, we introduce VisualScratchpad, an interactive interface for visual concept analysis during inference. We apply sparse autoencoders to the vision encoder and link the resulting visual concepts to text tokens via text-to-image attention, allowing us to examine which visual concepts are both captured by the vision encoder and utilized by the language model. VisualScratchpad also provides a token-latent heatmap view that suggests a sufficient set of latents for effective concept ablation in causal analysis. Through case studies, we reveal three underexplored failure modes: limited cross-modal alignment, misleading visual concepts, and unused hidden cues. Project page: https://hyesulim.github.io/visual_scratchpad_projectpage/

Human-vehicle interaction in safety-critical traffic environments increasingly incorporates neural sensing to infer user intent and cognitive state, yet most existing approaches either treat electroencephalography (EEG) as a static biometric credential or train task-specific decoders that ignore long-term neurodynamic trajectories, lacking mechanisms for secure user identity and continual modeling of evolving cognitive states. This work proposes a self-supervised evolutionary learning (SSEL) framework that discovers individualized neurodynamic progressions and intrinsic identity manifolds directly from continuous EEG, without external labels or predefined cognitive stage models. SSEL jointly optimizes within-stage temporal predictability, boundary contrast, cross-trial alignment, and sparse stage-specific feature weights, while a population-based evolutionary search enables direct optimization in the discrete, non-differentiable space of candidate segmentations. We validate the framework on EEG recorded from participants performing a simulated road-crossing decision task, a canonical safety-critical scenario in which perceptual assessment, risk evaluation, and decision commitment unfold over time. The learned segmentations reveal stable, person-specific stage structures and neurodynamic signatures that support authentication and anomaly detection. Compared to inference-based segmentation baselines, SSEL achieves orders-of-magnitude higher boundary contrast, substantial gains in cross-trial generalization of intention boundaries, and more interpretable, sparse stage-wise feature attributions. Beyond performance, the framework advances a progression-aware perspective on cognitive neurodynamics, where security, resilience, and personalization emerge from the intrinsic temporal structure of brain activity, with implications for next-generation smart urban and transportation infrastructures.

This study examines the role of uncertainty estimation (UE) methods in multilingual text classification under noisy and non-topical conditions. Using a complex-vs-simple sentence classification task across several languages, we evaluate a range of UE techniques against a range of metrics to assess their contribution to making more robust predictions. Results indicate that while methods relying on softmax outputs remain competitive in high-resource in-domain settings, their reliability declines in low-resource or domain-shift scenarios. In contrast, Monte Carlo dropout approaches demonstrate consistently strong performance across all languages, offering more robust calibration, stable decision thresholds, and greater discriminative power even under adverse conditions. We further demonstrate the positive impact of UE on non-topical classification: abstaining from predicting the 10\% most uncertain instances increases the macro F1 score from 0.81 to 0.85 in the Readme task. By integrating UE with trustworthiness metrics, this study provides actionable insights for developing more reliable NLP systems in real-world multilingual environments. See https://github.com/Nouran-Khallaf/To-Predict-or-Not-to-Predict

Contemporary artificial intelligence research has been organized around two dominant ambitions: productivity, which treats AI systems as tools for accelerating work and economic output, and alignment, which focuses on ensuring that increasingly capable systems behave safely and in accordance with human values. This paper articulates and develops a third, emerging ambition: the use of large language models (LLMs) as scientific instruments for studying human behavior, culture, and moral reasoning. Trained on unprecedented volumes of human-produced text, LLMs encode large-scale regularities in how people argue, justify, narrate, and negotiate norms across social domains. We argue that these models can be understood as condensates of human symbolic behavior, compressed, generative representations that render patterns of collective discourse computationally accessible. The paper situates this third ambition within long-standing traditions of computational social science, content analysis, survey research, and comparative-historical inquiry, while clarifying the epistemic limits of treating model output as evidence. We distinguish between base models and fine-tuned systems, showing how alignment interventions can systematically reshape or obscure the cultural regularities learned during pretraining, and we identify instruct-only and modular adaptation regimes as pragmatic compromises for behavioral research. We review emerging methodological approaches including prompt-based experiments, synthetic population sampling, comparative-historical modeling, and ablation studies and show how each maps onto familiar social-scientific designs while operating at unprecedented scale.

Identifying the root cause of a bug remains difficult for many developers because bug reports often lack a bug reproducing test case that reliably triggers the failure. Manually writing such test cases is time-consuming and requires substantial effort to understand the codebase and isolate the failing behavior. To address this challenge, we propose Echo, an agent for generating issue reproducing test cases, which advances previous work in several ways. During generation, Echo strengthens context retrieval by leveraging a code graph and a novel automatic query-refinement strategy. Echo also improves upon previous tools by automatically executing generated test cases, a first-of-its-kind feature that seamlessly integrates into practical development workflows. In addition, Echo generates potential patches and uses the patched version to validate whether a candidate test meets the fail-to-pass criterion and to provide actionable feedback for refinement. Unlike prior bug-reproduction agents that sample and rank multiple candidate tests, Echo generates a single test per issue, offering a better cost-performance trade-off. Experiments on SWT-Bench Verified show that Echo establishes a new state of the art among open-source approaches, achieving a 66.28% success rate.

Neural networks often rely on spurious shortcuts for many epochs before discovering structured representations. However, the mechanism governing when this transition occurs and whether its timing can be predicted remains unclear. Prior work shows that gradient descent converges to low norm solutions and that neural networks exhibit simplicity bias, but neither explains the timescale of the transition from shortcut features to structured representations. We introduce the Norm-Hierarchy Transition (NHT) framework, which explains delayed representation learning as the slow traversal of a hierarchy of parameter norms during regularized optimization. When multiple interpolating solutions exist with different norms, weight decay gradually moves the model from high norm shortcut solutions toward lower norm structured representations. We derive a tight bound showing that the transition delay grows logarithmically with the ratio between shortcut and structured norms. Experiments on modular arithmetic, CIFAR-10 with spurious features, CelebA, and Waterbirds support the predictions of the framework. The results suggest that grokking, shortcut learning, and delayed feature discovery arise from a common mechanism based on norm hierarchy traversal during training.

Multi-group learning is a learning task that focuses on controlling predictors' conditional losses over specified subgroups. We propose ShakyPrepend, a method that leverages tools inspired by differential privacy to obtain improved theoretical guarantees over existing approaches. Through numerical experiments, we demonstrate that ShakyPrepend adapts to both group structure and spatial heterogeneity. We provide practical guidance for deploying multi-group learning algorithms in real-world settings.

While Large Language Models (LLMs) can accelerate text-heavy tasks in alternative investment due diligence, a gap remains in their ability to accurately extract and reason over structured tabular data from complex financial spreadsheets. Progress is held back by the lack of real industry fund portfolio datasets for benchmarking, as private equity data rooms are confidential. To address this, we introduce FinSheet-Bench, a benchmark of synthetic financial portfolio data modeled on real private equity fund structures, designed to evaluate LLM performance on text-serialized spreadsheet question answering and numeric reasoning tasks. Our evaluation of ten model configurations from OpenAI, Google, and Anthropic on financial spreadsheets, including complex layouts, fund dividers, and multi-line column names, reveals that no standalone model achieves error rates low enough for unsupervised use in professional finance applications. The best-performing model, Gemini 3.1 Pro, achieves 82.4% accuracy across twenty-four evaluation files of varying complexity and structural layout (approximately 1 error per 6 questions), followed by GPT-5.2 with reasoning at 80.4%, Claude Opus 4.6 with thinking at 80.2%, and Gemini 3 Pro at 80.2%. Performance degrades substantially on larger, more complex spreadsheets: the largest spreadsheet (152 companies, 8 funds) yields an average accuracy of just 48.6% across all models, compared to 86.2% on the easiest evaluation file. These difficulty patterns are consistent across all ten models, indicating that they reflect LLM limitations rather than idiosyncratic model weaknesses. Reliable financial spreadsheet extraction will likely require architectural approaches that separate document understanding from deterministic computation.

One common concern about advanced artificial intelligence is that it will prevent us from turning it off, as that would interfere with pursuing its goals. In this paper, we discuss an unorthodox proposal for addressing this concern: give the AI a (primary) goal of being turned off (see also papers by Martin et al., and by Goldstein and Robinson). We also discuss whether and under what conditions this would be a good idea.

Robustness under latent distribution shift remains challenging in partially observable reinforcement learning. We formalize a focused setting where an adversary selects a hidden initial latent distribution before the episode, termed an adversarial latent-initial-state POMDP. Theoretically, we prove a latent minimax principle, characterize worst-case defender distributions, and derive approximate best-response certificates with finite-sample guarantees, providing formal meaning to empirical training diagnostics. Empirically, using a Battleship benchmark, we demonstrate that targeted exposure to shifted latent distributions reduces average robustness gaps between Spread and Uniform distributions from 10.3 to 3.1 shots at equal budget. Furthermore, iterative best-response training exhibits budget-sensitive behavior entirely consistent with our approximate certificate theory. Ultimately, we show that for latent-initial-state problems, our framework yields precise diagnostic principles and confirms that structured adversarial exposure effectively mitigates worst-case vulnerabilities.

This chapter explores the evolution of data-driven hint generation for intelligent tutoring systems (ITS). The Hint Factory and Interaction Networks have enabled the generation of next-step hints, waypoints, and strategic subgoals from historical student data. Data-driven techniques have also enabled systems to find the right time to provide hints. We explore further potential data-driven adaptations for problem solving based on behavioral problem solving data and the integration of Large Language Models (LLMs).

Recent token merging techniques for Vision Transformers (ViTs) provide substantial speedups by reducing the number of tokens processed by self-attention, often without retraining. However, their direct application to the Segment Anything Model (SAM) family is nontrivial: SAM's image encoder mixes windowed and global attention, and its mask decoder relies on dense, prompt-conditioned features for precise boundary prediction. We systematically evaluate representative token-merging methods on SAM and Medical SAM in a strict off-the-shelf setting, and find that existing destination-selection heuristics can erode boundaries and leak prompt information as merge rates increase. We propose \textbf{StructSAM}, a resolution-preserving merge-unmerge framework tailored to SAM. StructSAM computes a lightweight token-energy score from first-order feature gradients, uses grid-based flatness screening to protect boundary and prompt regions, and merges tokens within flat areas toward low-energy destinations with explicit token recovery. We further provide a spectral graph coarsening view showing that score-guided merging yields bounded Laplacian spectral distortion compared to random or window-restricted baselines. Across eight natural and medical benchmarks, StructSAM reduces encoder FLOPs by 25-30\% (up to 40\%+ with prompt-aware merging) with minor drops in mIoU/Dice, consistently outperforming ToMe, PiToMe, ToMeSD, VidToMe, and ALGM at the same compute.

This paper proposes a new method for crop yield prediction, which is essential for developing management strategies, informing insurance assessments, and ensuring long-term food security. Although existing data-driven approaches have shown promise in this domain, their performance often degrades when applied across large geographic regions and long time periods. This limitation arises from two key challenges: (1) difficulty in jointly capturing short-term and long-term temporal patterns, and (2) inability to effectively accommodate spatial data variability in agricultural systems. Ignoring these issues often leads to unreliable predictions for specific regions or years, which ultimately affects policy decisions and resource allocation. In this paper, we propose a new predictive framework to address these challenges. First, we introduce a new backbone model architecture that captures both short-term daily-scale crop growth dynamics and long-term dependencies across years. To further improve generalization across diverse spatial regions, we augment this model with a retrieval-based adaptation strategy. Recognizing the substantial yield variation across years, we design a novel retrieval-and-refinement pipeline that adjusts retrieved samples by removing cross-year bias not explained by input features. Our experiments on real-world county-level corn yield data over 630 counties in the US demonstrate that our method consistently outperforms different types of baselines. The results also verify the effectiveness of the retrieval-based augmentation method in improving model robustness under spatial heterogeneity.

We present AutoResearch-RL, a framework in which a reinforcement learning agent conducts open-ended neural architecture and hyperparameter research without human supervision, running perpetually until a termination oracle signals convergence or resource exhaustion. At each step the agent proposes a code modification to a target training script, executes it under a fixed wall clock time budget, observes a scalar reward derived from validation bits-per-byte (val-bpb), and updates its policy via Proximal Policy Optimisation (PPO). The key design insight is the separation of three concerns: (i) a frozen environment (data pipeline, evaluation protocol, and constants) that guarantees fair cross-experiment comparison; (ii) a mutable target file (train.py) that represents the agent's editable state; and (iii) a meta-learner (the RL agent itself) that accumulates a growing trajectory of experiment outcomes and uses them to inform subsequent proposals. We formalise this as a Markov Decision Process, derive convergence guarantees under mild assumptions, and demonstrate empirically on a single GPU nanochat pretraining benchmark that AutoResearch-RL discovers configurations that match or exceed hand-tuned baselines after approximately 300 overnight iterations, with no human in the loop.

Continuous symmetries are fundamental to many scientific and learning problems, yet they are often unknown a priori. Existing symmetry discovery approaches typically search directly in the space of transformation generators or rely on learned augmentation schemes. We propose a fundamentally different perspective based on spectral structure. We introduce a framework for discovering continuous one-parameter subgroups using the Generalized Fourier Transform (GFT). Our central observation is that invariance to a subgroup induces structured sparsity in the spectral decomposition of a function across irreducible representations. Instead of optimizing over generators, we detect symmetries by identifying this induced sparsity pattern in the spectral domain. We develop symmetry detection procedures on maximal tori, where the GFT reduces to multi-dimensional Fourier analysis through their irreducible representations. Across structured tasks, including the double pendulum and top quark tagging, we demonstrate that spectral sparsity reliably reveals one-parameter symmetries. These results position spectral analysis as a principled and interpretable alternative to generator-based symmetry discovery.

This paper bridges neuroscience and artificial intelligence to propose a cortically inspired blueprint for modular perceptual AI. While current monolithic models such as GPT-4V achieve impressive performance, they often struggle to explicitly support interpretability, compositional generalization, and adaptive robustness - hallmarks of human cognition. Drawing on neuroscientific models of cortical modularity, predictive processing, and cross-modal integration, we advocate decomposing perception into specialized, interacting modules. This architecture supports structured, human-inspired reasoning by making internal inference processes explicit through hierarchical predictive feedback loops and shared latent spaces. Our proof-of-concept study provides empirical evidence that modular decomposition yields more stable and inspectable representations. By grounding AI design in biologically validated principles, we move toward systems that not only perform well, but also support more transparent and human-aligned inference.

Fine-grained understanding and species-specific multimodal question answering are vital for advancing biodiversity conservation and ecological monitoring. However, existing multimodal large language models face challenges when it comes to specialized topics like avian species, making it harder to provide accurate and contextually relevant information in these areas. To address this limitation, we introduce the MAviS-Dataset, a large-scale multimodal avian species dataset that integrates image, audio, and text modalities for over 1,000 bird species, comprising both pretraining and instruction-tuning subsets enriched with structured question-answer pairs. Building on the MAviS-Dataset, we introduce MAviS-Chat, a multimodal LLM that supports audio, vision, and text and is designed for fine-grained species understanding, multimodal question answering, and scene-specific description generation. Finally, for quantitative evaluation, we present MAviS-Bench, a benchmark of over 25,000 QA pairs designed to assess avian species-specific perceptual and reasoning abilities across modalities. Experimental results show that MAviS-Chat outperforms the baseline MiniCPM-o-2.6 by a large margin, achieving state-of-the-art open-source results and demonstrating the effectiveness of our instruction-tuned MAviS-Dataset. Our findings highlight the necessity of domain-adaptive multimodal LLMs for ecological applications.

LLMs are increasingly used to draft academic text and to support software engineering (SE) evidence synthesis, but they often hallucinate bibliographic references that look legitimate. We study how deployment-motivated prompting constraints affect citation verifiability in a closed-book setting. Using 144 claims (24 in SE&CS) and a deterministic verification pipeline (Crossref + Semantic Scholar), we evaluate two proprietary models (Claude Sonnet, GPT-4o) and two open-weight models (LLaMA~3.1-8B, Qwen~2.5-14B) across five regimes: Baseline, Temporal (publication-year window), Survey-style breadth, Non-Disclosure policy, and their combination. Across 17,443 generated citations, no model exceeds a citation-level existence rate of 0.475; Temporal and Combo conditions produce the steepest drops while outputs remain format-compliant (well-formed bibliographic fields). Unresolved outcomes dominate (36-61%); a 100-citation audit indicates that a substantial fraction of Unresolved cases are fabricated. Results motivate post-hoc citation verification before LLM outputs enter SE literature reviews or tooling pipelines.

Global safety models exhibit strong performance across widely used benchmarks, yet their training data rarely captures the cultural and linguistic nuances of Taiwanese Mandarin. This limitation results in systematic blind spots when interpreting region-specific risks such as localized financial scams, culturally embedded hate speech, and misinformation patterns. To address these gaps, we introduce TS-Bench (Taiwan Safety Benchmark), a standardized evaluation suite for assessing safety performance in Taiwanese Mandarin. TS-Bench contains 400 human-curated prompts spanning critical domains including financial fraud, medical misinformation, social discrimination, and political manipulation. In parallel, we present Breeze Guard, an 8B safety model derived from Breeze 2, our previously released general-purpose Taiwanese Mandarin LLM with strong cultural grounding from its original pre-training corpus. Breeze Guard is obtained through supervised fine-tuning on a large-scale, human-verified synthesized dataset targeting Taiwan-specific harms. Our central hypothesis is that effective safety detection requires the cultural grounding already present in the base model; safety fine-tuning alone is insufficient to introduce new socio linguistic knowledge from scratch. Empirically, Breeze Guard significantly outperforms the leading 8B general-purpose safety model, Granite Guardian 3.3, on TS-Bench (+0.17 overall F1), with particularly large gains in high-context categories such as scam (+0.66 F1) and financial malpractice (+0.43 F1). While the model shows slightly lower performance on English-centric benchmarks (ToxicChat, AegisSafetyTest), this tradeoff is expected for a regionally specialized safety model optimized for Taiwanese Mandarin. Together, Breeze Guard and TS-Bench establish a new foundation for trustworthy AI deployment in Taiwan.

We propose a Vocos-based bandwidth extension model that enhances audio at 8-48 kHz by generating missing high-frequency content. Inputs are resampled to 48 kHz and processed by a neural vocoder backbone, enabling a single network to support arbitrary upsampling ratios. A lightweight Linkwitz-Riley-inspired refiner merges the original low band with the generated high frequencies via a smooth crossover. On validation, the model achieves competitive log-spectral distance while running at a real-time factor of 0.0001 on an NVIDIA A100 GPU and 0.0053 on an 8-core CPU, demonstrating practical, high-quality BWE at extreme throughput.

Flow maps enable high-quality image generation in a single forward pass. However, unlike iterative diffusion models, their lack of an explicit sampling trajectory impedes incorporating external constraints for conditional generation and solving inverse problems. We put forth Variational Flow Maps, a framework for conditional sampling that shifts the perspective of conditioning from "guiding a sampling path", to that of "learning the proper initial noise". Specifically, given an observation, we seek to learn a noise adapter model that outputs a noise distribution, so that after mapping to the data space via flow map, the samples respect the observation and data prior. To this end, we develop a principled variational objective that jointly trains the noise adapter and the flow map, improving noise-data alignment, such that sampling from complex data posterior is achieved with a simple adapter. Experiments on various inverse problems show that VFMs produce well-calibrated conditional samples in a single (or few) steps. For ImageNet, VFM attains competitive fidelity while accelerating the sampling by orders of magnitude compared to alternative iterative diffusion/flow models. Code is available at https://github.com/abbasmammadov/VFM

We present VisualDeltas, a lightweight preference-learning framework that extracts supervision from visual quality variations in multimodal data. By leveraging the systematic impact of image quality on visual perception and reasoning, VisualDeltas induces informative preference signals without relying on human annotations or external teachers. The framework supports both label-free and label-based regimes, enabling flexible use of available supervision when present. Across diverse multimodal benchmarks and model scales, VisualDeltas consistently outperforms rejection-sampling fine-tuning and improves generalization, and extends naturally to a range of visual degradations.

Patient no-shows disrupt outpatient clinic operations, reduce productivity, and may delay necessary care. Clinics often adopt overbooking or double-booking to mitigate these effects. However, poorly calibrated policies can increase congestion and waiting times. Most existing methods rely on fixed heuristics and fail to adapt to real-time scheduling conditions or patient-specific no-show risk. To address these limitations, we propose an adaptive outpatient double-booking framework that integrates individualized no-show prediction with multi-objective reinforcement learning. The scheduling problem is formulated as a Markov decision process, and patient-level no-show probabilities estimated by a Multi-Head Attention Soft Random Forest model are incorporated in the reinforcement learning state. We develop a Multi-Policy Proximal Policy Optimization method equipped with a Multi-Policy Co-Evolution Mechanism. Under this mechanism, we propose a novel τ rule based on Kullback-Leibler divergence that enables selective knowledge transfer among behaviorally similar policies, improving convergence and expanding the diversity of trade-offs. In addition, SHapley Additive exPlanations is used to interpret both the predicted no-show risk and the agent's scheduling decisions. The proposed framework determines when to single-book, double-book, or reject appointment requests, providing a dynamic and data-driven alternative to conventional outpatient scheduling policies.

Data-efficient learning remains a central challenge in autonomous driving due to the high cost and safety risks of large-scale real-world interaction. Although world-model-based reinforcement learning enables policy optimization through latent imagination, existing approaches often lack explicit mechanisms to encode spatial and kinematic structure essential for driving tasks. In this work, we build upon the Recurrent State-Space Model (RSSM) and propose a kinematics-aware latent world model framework for autonomous driving. Vehicle kinematic information is incorporated into the observation encoder to ground latent transitions in physically meaningful motion dynamics, while geometry-aware supervision regularizes the RSSM latent state to capture task-relevant spatial structure beyond pixel reconstruction. The resulting structured latent dynamics improve long-horizon imagination fidelity and stabilize policy optimization. Experiments in a driving simulation benchmark demonstrate consistent gains over both model-free and pixel-based world-model baselines in terms of sample efficiency and driving performance. Ablation studies further verify that the proposed design enhances spatial representation quality within the latent space. These results suggest that integrating kinematic grounding into RSSM-based world models provides a scalable and physically grounded paradigm for autonomous driving policy learning.

Chaotic time series are notoriously difficult to forecast. Small uncertainties in initial conditions amplify rapidly, while strong nonlinearities and regime dependent variability constrain predictability. Although modern deep learning often delivers strong short horizon accuracy, its black box nature limits scientific insight and practical trust in settings where understanding the underlying dynamics matters. To address this gap, we propose two complementary symbolic forecasters that learn explicit, interpretable algebraic equations from chaotic time series data. Symbolic Neural Forecaster (SyNF) adapts a neural network based equation learning architecture to the forecasting setting, enabling fully differentiable discovery of compact and interpretable algebraic relations. The Symbolic Tree Forecaster (SyTF) builds on evolutionary symbolic regression to search directly over equation structures under a principled accuracy complexity trade off. We evaluate both approaches in a rolling window nowcasting setting with one step ahead forecasting using several accuracy metrics and compare against a broad suite of baselines spanning classical statistical models, tree ensembles, and modern deep learning architectures. Numerical experiments cover a benchmark of 132 low dimensional chaotic attractors and two real world chaotic time series, namely weekly dengue incidence in San Juan and the Nino 3.4 sea surface temperature index. Across datasets, symbolic forecasters achieve competitive one step ahead accuracy while providing transparent equations that reveal salient aspects of the underlying dynamics.

A significant element of human cooperative intelligence lies in our ability to identify opportunities for fruitful collaboration; and conversely to recognise when the task at hand is better pursued alone. Research on flexible cooperation in machines has left this meta-level problem largely unexplored, despite its importance for successful collaboration in heterogeneous open-ended environments. Here, we extend the typical Ad Hoc Teamwork (AHT) setting to incorporate the idea of agents having heterogeneous goals that in any given scenario may or may not overlap. We introduce a novel approach to learning policies in this setting, based on a hierarchical combination of imitation and reinforcement learning, and show that it outperforms baseline methods across extended versions of two cooperative environments. We also investigate the contribution of an auxiliary component that learns to model teammates by predicting their actions, finding that its effect on performance is inversely related to the amount of observable information about teammate goals.

Second primary cancer (SPC), a new cancer in patients different from previously diagnosed, is a growing concern due to improved cancer survival rates. Early prediction of SPC is essential to enable timely clinical interventions. This study focuses on lung cancer survivors treated in Taiwanese hospitals, where the limited size and geographic scope of local datasets restrict the effectiveness and generalizability of traditional machine learning approaches. To address this, we incorporate external data from the publicly available US-based Surveillance, Epidemiology, and End Results (SEER) program, significantly increasing data diversity and scale. However, the integration of multi-source datasets presents challenges such as feature inconsistency and privacy constraints. Rather than naively merging data, we proposed a loss fusion horizontal federated learning (LF2L) framework that can enable effective cross-institutional collaboration while preserving institutional privacy by avoiding data sharing. Using both common and unique features and balancing their contributions through a shared loss mechanism, our method demonstrates substantial improvements in the prediction performance of SPC. Experiment results show statistically significant improvements in AUROC and AUPRC when compared to localized, horizontal federated, and centralized learning baselines. This highlights the importance of not only acquiring external data but also leveraging it effectively to enhance model performance in real-world clinical model development.

Geospatial foundation models provide precomputed embeddings that serve as compact feature vectors for large-scale satellite remote sensing data. While these embeddings can reduce data-transfer bottlenecks and computational costs, Earth observation (EO) applications can still face geometric mismatches between user-defined areas of interest and the fixed precomputed embedding grid. Standard latent-space interpolation is unreliable in this setting because the embedding manifold is highly non-convex, yielding representations that do not correspond to realistic inputs. We verify this using Prithvi-EO-2.0 to understand the shortcomings of interpolation applied to patch embeddings. As a substitute, we propose a Learned Equivariance-Predicting Architecture (LEPA). Instead of averaging vectors, LEPA conditions a predictor on geometric augmentations to directly predict the transformed embedding. We evaluate LEPA on NASA/USGS Harmonized Landsat-Sentinel (HLS) imagery and ImageNet-1k. Experiments show that standard interpolation achieves a mean reciprocal rank (MRR) below 0.2, whereas LEPA increases MRR to over 0.8, enabling accurate geometric adjustment without re-encoding.

Despite the integration of search tools, Deep Search Agents often suffer from a misalignment between reasoning-driven queries and the underlying web indexing structures. Existing frameworks treat the search engine as a static utility, leading to queries that are either too coarse or too granular to retrieve precise evidence. We propose WeDas, a Web Content Distribution Aware framework that incorporates search-space structural characteristics into the agent's observation space. Central to our method is the Query-Result Alignment Score, a metric quantifying the compatibility between agent intent and retrieval outcomes. To overcome the intractability of indexing the dynamic web, we introduce a few-shot probing mechanism that iteratively estimates this score via limited query accesses, allowing the agent to dynamically recalibrate sub-goals based on the local content landscape. As a plug-and-play module, WeDas consistently improves sub-goal completion and accuracy across four benchmarks, effectively bridging the gap between high-level reasoning and low-level retrieval.

Similarities between language representations derived from Self-Supervised Speech Models (S3Ms) have been observed to primarily reflect geographic proximity or surface typological similarities driven by recent expansion or contact, potentially missing deeper genealogical signals. We investigate how scaling linguistic coverage of an S3M-based language identification system from 126 to 4,017 languages influences this topology. Our results reveal a non-linear effect: while phylogenetic recovery remains stagnant up to the 1K scale, the 4K model displays a dramatic qualitative shift, resolving both clear lineages and complex, long-term linguistic contact. Notably, our analysis reveals the emergence of a robust macro-cluster in the Pacific (comprising Papuan, Oceanic, and Australian languages) and investigates its latent drivers. We find that the 4K model utilizes a more concentrated encoding that captures shared, robust acoustic signatures such as global energy dynamics. These findings suggest that massive S3Ms can internalize multiple layers of language history, providing a promising perspective for computational phylogenetics and the study of language contact.

Predicting how cells respond to genetic perturbations is fundamental to understanding gene function, disease mechanisms, and therapeutic development. While recent deep learning approaches have shown promise in modeling single-cell perturbation responses, they struggle to generalize across cell types and perturbation contexts due to limited contextual information during generation. We introduce PT-RAG (Perturbation-aware Two-stage Retrieval-Augmented Generation), a novel framework that extends Retrieval-Augmented Generation beyond traditional language-model applications to cellular biology. Unlike standard RAG systems designed for text retrieval with pre-trained LLMs, perturbation retrieval lacks established similarity metrics and requires learning what constitutes relevant context, making differentiable retrieval essential. PT-RAG addresses this through a two-stage pipeline: first, retrieving candidate perturbations $K$ using GenePT embeddings, then adaptively refining the selection through Gumbel-Softmax discrete sampling conditioned on both the cell state and the input perturbation. This cell-type-aware differentiable retrieval enables end-to-end optimization of the retrieval objective jointly with generation. On the Replogle-Nadig single-gene perturbation dataset, we demonstrate that PT-RAG outperforms both STATE and vanilla RAG under identical experimental conditions, with the strongest gains in distributional similarity metrics ($W_1$, $W_2$). Notably, vanilla RAG's dramatic failure is itself a key finding: it demonstrates that differentiable, cell-type-aware retrieval is essential in this domain, and that naive retrieval can actively harm performance. Our results establish retrieval-augmented generation as a promising paradigm for modelling cellular responses to gene perturbation. The code to reproduce our experiments is available at https://github.com/difra100/PT-RAG_ICLR.

Intergenerational mobility quantifies the transmission of socio-economic outcomes from parents to children. While rank-rank regression (RRR) is standard, adding covariates directly (RRRX) often yields parameters with unclear interpretation. Conditional rank-rank regression (CRRR) resolves this by using covariate-adjusted (conditional) ranks to measure within-group mobility. We improve and extend CRRR by estimating conditional ranks with a deep conditional transformation model (DCTM) and cross-fitting, enabling end-to-end conditional distribution learning with structural constraints and strong performance under nonlinearity, high-order interactions, and discrete ordered outcomes where the distributional regression used in traditional CRRR may be cumbersome or prone to misconfiguration. We further extend CRRR to discrete outcomes via an $ω$-indexed conditional-rank definition and study sensitivity to $ω$. For continuous outcomes, we establish an asymptotic theory for the proposed estimators and verify the validity of exchangeable bootstrap inference. Simulations across simple/complex continuous and discrete ordered designs show clear accuracy gains in challenging settings. Finally, we apply our method to two empirical studies, revealing substantial within-group persistence in U.S. income and pronounced gender differences in educational mobility in India.

Questions and Answering forums such as Stack Overflow play an important role in supporting software developers in finding answers to queries related to issues such as software errors and bugs. However, searching through a large set of candidate answers could be time consuming and may not lead to the best solution. In this research, the effectiveness of data mining models and machine learning techniques to solve this kind of problems is evaluated. We propose a recommender system to aid developers in finding solutions to their software bugs by carefully mining Stack Overflow. The proposed model leverages the knowledge available through crowdsourcing the Q&A available in Stack Overflow to recommend a solution to software bugs. We use deep learning techniques to construct the required Learning-to-Rank (LTR)-based model using the social context embedding the Stack Overflow features. Text mining, natural language processing and recommendation algorithms are used to extract, evaluate and recommend the best relevant bug solutions. Additionally, our model achieves nearly 78% correct solutions when recommending the 10 best answers for each question.

Accurate and efficient 3D medical image segmentation is essential for clinical AI, where models must remain reliable under stringent memory, latency, and data availability constraints. Transformer-based methods achieve strong accuracy but suffer from excessive parameters, high FLOPs, and limited generalization. We propose LightMedSeg, a modular UNet-style segmentation architecture that integrates anatomical priors with adaptive context modeling. Anchor-conditioned FiLM modulation enables anatomy-aware feature calibration, while a local structural prior module and texture-aware routing dynamically allocate representational capacity to boundary-rich regions. Computational redundancy is minimized through ghost and depthwise convolutions, and multi-scale features are adaptively fused via a learned skip router with anchor-relative spatial position bias. Despite requiring only 0.48M parameters and 14.64~GFLOPs, LightMedSeg achieves segmentation accuracy within a few Dice points of heavy transformer baselines. Therefore, LightMedSeg is a deployable and data-efficient solution for 3D medical image segmentation. Code will be released publicly upon acceptance.

Large Language Models (LLMs) have demonstrated strong general capabilities, yet their deployment in finance remains challenging due to dense domain-specific terminology, stringent numerical reasoning requirements, and low tolerance for factual errors. We conduct a controlled empirical study showing that in specialized vertical domains, performance is largely determined by the quality and difficulty/verifiability profile of post-training data. We introduce \textbf{ODA-Fin-SFT-318k}, constructed via multi-stage distillation and verification to produce high-quality Chain-of-Thought supervision, and \textbf{ODA-Fin-RL-12k}, curated for hard-but-verifiable tasks that balance reward precision and task diversity. Using standard SFT and RL pipelines, we show that high-quality CoT distillation establishes a robust foundation during SFT, while difficulty- and verifiability-aware sampling improves RL generalization. Evaluated on nine benchmarks spanning general financial tasks, sentiment analysis, and numerical reasoning, our ODA-Fin-RL-8B consistently surpasses open-source state-of-the-art (SOTA) financial LLMs of comparable size. We release our ODA-Fin-SFT-318k and ODA-Fin-RL-12k datasets, along with trained models to advance data-centric financial AI research.

Self-supervised learning (SSL) has made rapid progress, yet learned features often over-rely on contextual shortcuts-background textures and co-occurrence statistics. While video provides rich temporal variation, dense in-the-wild streams with strong ego-motion create a co-occurrence trap: foreground objects and background context move coherently, encouraging representations to collapse into scene encoders. To address this, we propose VINO (Video-driven Invariance for Non-Contextual Objects), a teacher-student framework that learns robust image encoders from dense video by imposing a structural information bottleneck. Using a class-agnostic structural prior solely to generate views-not as semantic pseudo-labels-VINO forms an asymmetric distillation problem. The teacher predicts from a foreground-union view with the background suppressed, while the student observes object-conditioned scene views that retain surrounding context but remove competing instances. Matching these targets via masked distillation makes background cues unreliable, pushing the representation toward object-centric invariances. We further enforce temporal object permanence via teacher-anchored cross-time distillation over track-matched objects, and stabilize part-to-whole consistency with mask-guided local views. Through attention visualization and unsupervised object discovery on PASCAL VOC, we demonstrate that VINO effectively disentangles foreground from background. Pretrained on the dense Walking Tours Venice video, VINO achieves 34.8 CorLoc, yielding highly focused, shape-biased representations that substantially outperform prior dense-video and motion-guided SSL baselines.

Margin-based learning, exemplified by linear and kernel methods, is one of the few classical settings where generalization guarantees are independent of the number of parameters. This makes it a central case study in modern highly over-parameterized learning. We ask what minimal mathematical structure underlies this phenomenon. We begin with a simple margin-based problem in arbitrary metric spaces: concepts are defined by a center point and classify points according to whether their distance lies below $r$ or above $R$. We show that whenever $R>3r$, this class is learnable in \emph{any} metric space. Thus, sufficiently large margins make learnability depend only on the triangle inequality, without any linear or analytic structure. Our first main result extends this phenomenon to concepts defined by bounded linear combinations of distance functions, and reveals a sharp threshold: there exists a universal constant $γ>0$ such that above this margin the class is learnable in every metric space, while below it there exist metric spaces where it is not learnable at all. We then ask whether margin-based learnability can always be explained via an embedding into a linear space -- that is, reduced to linear classification in some Banach space through a kernel-type construction. We answer this negatively by developing a structural taxonomy of Banach spaces: if a Banach space is learnable for some margin parameter $γ\geq 0$, then it is learnable for all such $γ$, and in infinite-dimensional spaces the sample complexity must scale polynomially in $1/γ$. Specifically, it must grow as $(1/γ)^p$ for some $p\ge 2$, and every such rate is attainable.

We present a miniature brain transformer architecture that extends the attention-coupled latent memory framework with four additional brain-region analogues: a thalamic relay, an amygdaloid salience module, a prefrontal working-memory (PFC) buffer, and a cerebellar fast-path, all coupled by inhibitory callosal cross-talk between lateralized hippocampal banks. We evaluate on a two-domain benchmark -- MQAR (Multi-Query Associative Recall; episodic domain) and modular arithmetic (+1 mod 10; rule-based domain) -- using a seven-variant additive ablation. The central empirical finding is a surprise: inhibitory callosal coupling alone never lateralizes the banks (variants 1-5 maintain D_sep ~ 0.25 and P_ct ~ 0.25 for all 30 epochs). Functional lateralization requires the synergy of PFC and inhibition: only when the PFC buffer is added (variant 6) does a sharp, discontinuous phase transition fire -- at epoch 11 for the PFC-only variant and epoch 10 for the full model -- collapsing P_ct from 0.25 to ~0.002 and more than doubling D_sep from 0.251 to 0.501 in a single gradient step. The PFC buffer acts as a symmetry-breaker: its slowly drifting domain context creates the initial asymmetry that the inhibitory feedback loop then amplifies irreversibly. The cerebellar fast-path accelerates the transition by one epoch (epoch 10 vs. epoch 11) with no asymptotic change, confirming its convergence-acceleration role. The result constitutes a novel, falsifiable prediction -- no lateralization without working memory context -- and a principled, neurobiologically motivated blueprint for hierarchical persistent memory in sequence models.

Bowel sounds (BS) are typically momentary and have low amplitude, making them difficult to detect accurately through manual auscultation. This leads to significant variability in clinical assessment. Digital acoustic sensors allow the acquisition of high-quality BS and enable automated signal analysis, offering the potential to provide clinicians with both objective and quantitative feedback on bowel activity. This study presents an automated pipeline for bowel sound segmentation and classification using a wearable acoustic SonicGuard sensor. BS signals from 83 subjects were recorded using a SonicGuard sensor. Data from 40 subjects were manually annotated by clinical experts and used to train an automatic annotation algorithm, while the remaining subjects were used for further model evaluation. An energy-based event detection algorithm was developed to detect BS events. Detected sound segments were then classified into BS patterns using a pretrained Audio Spectrogram Transformer (AST) model. Model performance was evaluated separately for healthy individuals and patients. The best configuration used two specialized models, one trained on healthy subjects and one on patients, achieving (accuracy: 0.97, AUROC: 0.98) for healthy group and (accuracy: 0.96, AUROC: 0.98) for patient group. The auto-annotation method reduced manual labeling time by approximately 70%, and expert review showed that less than 12% of automatically detected segments required correction. The proposed automated segmentation and classification system enables quantitative assessment of bowel activity, providing clinicians with an objective diagnostic tool that may improve the diagnostic of gastrointestinal function and support the annotation of large-scale datasets.

Direct Preference Optimization (DPO) aligns large language models by optimizing pairwise preferences and has shown remarkable effectiveness as a simple and scalable alternative to RLHF. However, in practice, preference data are often noisy. Existing robust variants of DPO mainly rely on uniform objective modifications or global reweighting. While partially effective, these methods treat noisy samples as a homogeneous source of uncertainty and fail to distinguish between different noise types, leading to sub-optimal alignment robustness. In this work, we show that robust preference alignment benefits from addressing different noise types with targeted interventions rather than uniform regularization. We propose winsorized Direct Preference Optimization~(wDPO), a robust LLM alignment approach with hierarchical winsorization. Specifically, wDPO adopts a reward-free hierarchical intervention strategy that leverages only signals already available during DPO training. It first uses the implicit margin from DPO log-ratio to identify heterogeneous noise patterns without relying on external reward models. For hard noise, wDPO performs a data-level intervention by sparsely correcting strongly inconsistent preference pairs. For ambiguous comparisons, it applies a gradient-level intervention through soft winsorization, capping extreme losses in the high-loss tail to prevent weakly informative samples from dominating gradient updates. Extensive experiments on PKU-SafeRLHF and multiple external safety benchmarks demonstrate that wDPO consistently improves preference alignment quality and robustness over vanilla DPO and strong DPO-family baselines, with particularly pronounced gains under controlled label-flip noise.

As Large Language Models (LLMs) transition into autonomous agentic roles, the risk of deception-defined behaviorally as the systematic provision of false information to satisfy external incentives-poses a significant challenge to AI safety. Existing benchmarks often focus on unintentional hallucinations or unfaithful reasoning, leaving intentional deceptive strategies under-explored. In this work, we introduce a logically grounded framework to elicit and quantify deceptive behavior by embedding LLMs in a structured 20-Questions game. Our method employs a conversational forking mechanism: at the point of object identification, the dialogue state is duplicated into multiple parallel worlds, each presenting a mutually exclusive query. Deception is formally identified when a model generates a logical contradiction by denying its selected object across all parallel branches to avoid identification. We evaluate GPT-4o, Gemini-2.5-Flash, and Qwen-3-235B across three incentive levels: neutral, loss-based, and existential (shutdown-threat). Our results reveal that while models remain rule-compliant in neutral settings, existential framing triggers a dramatic surge in deceptive denial for Qwen-3-235B (42.00\%) and Gemini-2.5-Flash (26.72\%), whereas GPT-4o remains invariant (0.00\%). These findings demonstrate that deception can emerge as an instrumental strategy solely through contextual framing, necessitating new behavioral audits that move beyond simple accuracy to probe the logical integrity of model commitments.

High-fidelity nonlinear finite-element (FE) simulations of reinforced-concrete (RC) structures are still costly, especially in parametric settings where loading positions vary. We develop a dual-graph spatiotemporal GNN surrogate to approximate the time histories of RC beams under four-point bending. To generate training data, we run a parametric Abaqus campaign that independently shifts the two loading blocks on a mesh-aligned grid and exports full-field responses at fixed normalized loading levels. The model rolls out autoregressively and jointly predicts nodal displacements, element wise von Mises stress, element-wise equivalent plastic strain (PEEQ), and the global vertical reaction force in a single multi-task setup. A key motivation is the peak loss introduced when element quantities are forced through node-based representations. We therefore couple node- and element-level dynamics using two recurrent graph branches: a node-level graph convolutional gated recurrent unit (GConvGRU) for kinematics and an element-level GConvGRU for history-dependent internal variables, with global force predicted through pooling on the element branch. In controlled ablations, removing the Element to Node to Element pathway improves peak-sensitive prediction in localized high-gradient stress/PEEQ regions without degrading global load displacement trends. After training, the surrogate produces full trajectories at a fraction of the cost of nonlinear FE, enabling faster parametric evaluation and design exploration.

Reinforcement learning with verifiable rewards (RLVR) has shown promise in enhancing the reasoning performance of large language models (LLMs) by increasing test-time compute. However, even after extensive RLVR training, such models still tend to generate unnecessary and low-quality steps in their chain-of-thought (CoT), leading to inefficient overthinking and lower answer quality. We show that when the initial direction or quality of the CoT is suboptimal, the model often fails to reach the correct answer, even after generating several times more tokens than when the initial CoT is well-initialized. To this end, we introduce Reinforcement Learning with Re-solving (Re$^2$), in which LLMs learn to flexibly abandon unproductive reasoning paths and restart the solution process when necessary, rather than always committing to a final answer. Re$^2$ applies pure reinforcement learning without any preliminary supervised fine-tuning, successfully amplifying the rare redo behavior in vanilla models from only 0.5% to over 30%. This leads to substantial performance gains over standard RLVR under the same training compute budget, and also demonstrates notable improvements in test-time performance as the number of samples increases.

Out-of-distribution (OOD) detection is a well-known challenge due to deep models often producing overconfident. In this paper, we reveal a key insight that trained classifiers tend to rely on sparse parameter contribution patterns, meaning that only a few dominant parameters drive predictions. This brittleness can be exploited by OOD inputs that anomalously trigger these parameters, resulting in overconfident predictions. To address this issue, we propose a simple yet effective method called Shaping Parameter Contribution Patterns (SPCP), which enhances OOD detection robustness by encouraging the classifier to learn boundary-oriented dense contribution patterns. Specifically, SPCP operates during training by rectifying excessively high parameter contributions based on a dynamically estimated threshold. This mechanism promotes the classifier to rely on a broader set of parameters for decision-making, thereby reducing the risk of overconfident predictions caused by anomalously triggered parameters, while preserving in-distribution (ID) performance. Extensive experiments under various OOD detection setups verify the effectiveness of SPCP.

Autonomous agents powered by large language models introduce a class of execution-layer vulnerabilities -- prompt injection, retrieval poisoning, and uncontrolled tool invocation -- that existing guardrails fail to address systematically. In this work, we propose the Layered Governance Architecture (LGA), a four-layer framework comprising execution sandboxing (L1), intent verification (L2), zero-trust inter-agent authorization (L3), and immutable audit logging (L4). To evaluate LGA, we construct a bilingual benchmark (Chinese original, English via machine translation) of 1,081 tool-call samples -- covering prompt injection, RAG poisoning, and malicious skill plugins -- and apply it to OpenClaw, a representative open-source agent framework. Experimental results on Layer 2 intent verification with four local LLM judges (Qwen3.5-4B, Llama-3.1-8B, Qwen3.5-9B, Qwen2.5-14B) and one cloud judge (GPT-4o-mini) show that all five LLM judges intercept 93.0-98.5% of TC1/TC2 malicious tool calls, while lightweight NLI baselines remain below 10%. TC3 (malicious skill plugins) proves harder at 75-94% IR among judges with meaningful precision-recall balance, motivating complementary enforcement at Layers 1 and 3. Qwen2.5-14B achieves the best local balance (98% IR, approximately 10-20% FPR); a two-stage cascade (Qwen3.5-9B->GPT-4o-mini) achieves 91.9-92.6% IR with 1.9-6.7% FPR; a fully local cascade (Qwen3.5-9B->Qwen2.5-14B) achieves 94.7-95.6% IR with 6.0-9.7% FPR for data-sovereign deployments. An end-to-end pipeline evaluation (n=100) demonstrates that all four layers operate in concert with 96% IR and a total P50 latency of approximately 980 ms, of which the non-judge layers contribute only approximately 18 ms. Generalization to the external InjecAgent benchmark yields 99-100% interception, confirming robustness beyond our synthetic data.

Analyzing creative activity traces requires capturing activity at appropriate granularity and interpreting it in ways that reflect the structure of creative practice. However, existing approaches record state changes without preserving the intent or relationships that define higher-level creative moves. This decoupling manifests differently across domains: GenAI tools lose non-linear exploration structure, visualization authoring obscures representational intent, and programmatic environments flatten interaction boundaries. We present three complementary approaches: a node-based interface for stateful GenAI artifact management, a vocabulary of visual cues as higher-level creative moves in visualization authoring, and a programming model that embeds semantic histories directly into interaction state.

Finding good branching orders is key to solving SAT problems efficiently, but finding such branching orders is a difficult problem. Using a learning based approach to predict a good branching order before solving, therefore, has potential. In this paper, we investigate predicting branching orders using graph neural networks as a preprocessing step to conflict-driven clause learning (CDCL) SAT solvers. We show that there are significant gains to be made in existing CDCL SAT solvers by providing a good initial branching. Further, we provide three labeling methods to find such initial branching orders in a tractable way. Finally, we train a graph neural network to predict these branching orders and show through our evaluations that a GNN-initialized ordering yields significant speedups on random 3-CNF and pseudo-industrial benchmarks, with generalization capabilities to instances much larger than the training set. However, we also find that the predictions fail at speeding up more difficult and industrial instances. We attribute this to the solver's dynamic heuristics, which rapidly overwrite the provided initialization, and to the complexity of these instances, making GNN prediction hard.

Optimizing GPU kernels manually is a challenging and time-consuming task. With the rapid development of LLMs, automated GPU kernel optimization is gradually becoming a tangible reality. However, current LLM-driven automated optimization methods narrowly focus on machine learning applications, such as PyTorch operator optimization, while overlooking broader domains like sparse matrix operations in scientific computing. Extending to these broader applications brings new challenges for the benchmark and algorithm. Therefore, developing a general-purpose automated kernel optimization method becomes our primary focus. In this paper, we address the absence of systematic evaluation for multi-scenario settings by introducing MSKernelBench, which spans multiple scenarios, including fundamental algebraic operations, common LLM kernels, sparse matrix operators, and scientific computing routines, each supporting both FP32 and BF16 precision. Building on this benchmark, we introduce CUDAMaster, a multi-agent, hardware-aware system for kernel optimization that leverages profiling information and automatically constructs the full compilation and execution toolchain. Experimental results demonstrate that CUDAMaster achieves significant speedups across most operators, outperforming Astra by about 35%. In several cases, its performance matches or surpasses that of highly optimized, closed-source libraries such as cuBLAS. A demo showcasing the original and optimized code for each operator is available at https://hanyx2021.github.io/MSKernelBenchDemo/.

We present a theoretical analysis of the Jacobian of an attention block within a transformer, showing that it is governed by the query, key, and value projections that define the attention mechanism. Leveraging this insight, we introduce a method that systematically alters the spectral properties of each attention layer to reduce the Jacobian's condition number, thereby improving the overall conditioning of the attention layers within a transformer network. We empirically show that this improved Jacobian conditioning translates to enhanced performance in practice. Our approach is simple, broadly applicable, and can be easily integrated as a drop-in replacement for a wide range of existing attention mechanisms. We validate its effectiveness across diverse transformer architectures and tasks, demonstrating consistent improvements in performance.

Large language models (LLMs) now exhibit strong multi-step reasoning abilities, but existing inference-time scaling methods remain computationally expensive, often relying on extensive sampling or external evaluators. We propose a principled strategy that frames reasoning as uncertainty minimisation and operates at the level of individual thoughts rather than tokens. Our method selects, at each reasoning step, the continuation that maximizes the model's self-certainty, a metric computed from its internal predictive distribution. This approach achieves significant improvement with a small number of samples, relies exclusively on model-internal signals, and applies to open-ended questions as opposed to methods like majority voting. Experiments on MATH500 and GSM8K across multiple model sizes demonstrate that thought-level self-certainty maximization consistently outperforms greedy decoding and matches or exceeds self-consistency under comparable token budgets. Cross-linguistic evaluations further indicate that the method transfers robustly beyond high-resource languages. Furthermore, analysis of self-certainty dynamics reveals that correct reasoning trajectories converge early to stable paths, suggesting that early decisions, likely associated with the planning of the reasoning process, are predictive of final accuracy. Building on this result, we show that self-certainty maximisation applied to the early steps can explain most of the performance gain and provide a simple yet efficient inference-time scaling method.

Large Language Models show promise for AI-assisted storytelling, yet current tools often generate predictable, unoriginal narratives. To address this limitation, we present NarrativeLoom, a multi-persona co-creative system grounded in Campbell's Blind Variation and Selective Retention theory. NarrativeLoom deploys specialized AI personas to generate diverse narrative options (blind variation), while users act as creative directors to select and refine them (selective retention). We designed a controlled study with 50 participants and found that stories co-authored with NarrativeLoom were not only perceived by users as more novel and diverse but were also objectively rated by experts as significantly better across all Torrance Test creativity dimensions: fluency, flexibility, originality, and elaboration. Stories are significantly longer with richer settings and more dialogue. Writing expertise emerged as a moderator: novices benefited more from structured scaffolding. This demonstrates the value of theory-informed co-creative systems and the importance of adapting them to varying user expertise.

Direct prompt-based editing often fails on complex transformations because vague and subjective prompts often require nuanced understanding of what should be changed in the image. Our core intuition is that leveraging compositional image editing tools rather than direct prompting profits from structured agent-level planning with explicit reasoning, leading to better results. This structured planning framework enables efficient offline RL post-training on quality-scored trajectories to improve performance. We present a tool-based agentic RL post-training framework that addresses this through structured planning with chain-of-thought reasoning. Our key contributions include: (1) A tool-based agentic planning methodology that combines a compositional library of orthogonal primitive transformations, structured context representation, and explicit per-step reasoning to decompose complex styling into interpretable tool sequences. (2) A synthetic data generation pipeline producing three large-scale datasets (each $\sim$10K trajectories) with reasoning chains, plans, and quality scores, as no existing datasets provide such supervision. Our datasets and code are publicly available at the HuggingFace repository. (3) Offline RL training methods for learning planners with reasoning as our core algorithmic contributions, which consistently improve over the Edit-Only baseline in visual quality and instruction following. (4) Comprehensive evaluation across 4B and 8B parameter Qwen3-VL models showing that our methods outperform other baselines in the majority of compositional tasks, validated by human evaluations.

Enabling question answering over tables and databases in natural language has become a key capability in the democratization of insights from tabular data sources. These systems first require retrieval of data that is relevant to a given natural language query, for which several methods have been introduced. In this work we present and study a table retrieval mechanism devising fine-grained typed query decomposition and global connectivity-awareness (DCTR), to handle the challenges induced by open-domain question answering over relational databases in complex usage contexts. We evaluate the effectiveness of the two mechanisms through the lens of retrieval complexity which we measure along the axes of query- and data complexity. Our analyses over industry-aligned benchmarks illustrate the robustness of DCTR for highly composite queries and densely connected databases.

Emotion Recognition in Conversation (ERC) is critical for enabling natural human-machine interactions. However, existing methods predominantly employ categorical or dimensional emotion annotations, which often fail to adequately represent complex, subtle, or culturally specific emotional nuances. To overcome this limitation, we propose a novel task named Emotion Transcription in Conversation (ETC). This task focuses on generating natural language descriptions that accurately reflect speakers' emotional states within conversational contexts. To address the ETC, we constructed a Japanese dataset comprising text-based dialogues annotated with participants' self-reported emotional states, described in natural language. The dataset also includes emotion category labels for each transcription, enabling quantitative analysis and its application to ERC. We benchmarked baseline models, finding that while fine-tuning on our dataset enhances model performance, current models still struggle to infer implicit emotional states. The ETC task will encourage further research into more expressive emotion understanding in dialogue. The dataset is publicly available at https://github.com/UEC-InabaLab/ETCDataset.

Large Vision Language Models (LVLMs) show immense potential for automated ophthalmic diagnosis. However, their clinical deployment is severely hindered by lacking domain-specific knowledge. In this work, we identify two structural deficiencies hindering reliable medical reasoning: 1) the Perception Gap, where general-purpose visual encoders fail to resolve fine-grained pathological cues (e.g., microaneurysms); and 2) the Reasoning Gap, where sparse visual evidence is progressively overridden by massive language priors in deeper transformer layers, leading to ungrounded hallucinations. To bridge these gaps, we propose EyExIn, a data-efficient framework designed to anchor retinal VLMs with expert knowledge via a Deep Expert Injection mechanism. Our architecture employs an Expert-Aware Dual-Stream encoding strategy that decouples visual representation into a general stream for anatomical context and a specialized expert stream for pathological semantics. To ensure high-fidelity integration, we design a Semantic-Adaptive Gated Fusion module, which dynamically amplifies subtle lesion signals while filtering irrelevant background noise. Furthermore, we introduce Adaptive Deep Expert Injection to embed persistent "Vision Anchors" by integrating fused visual features as residual biases directly into intermediate LLM layers. This mechanism creates a visual shortcut that forces the reasoning stack to remain strictly grounded in visual evidence. Extensive experiments across four benchmarks demonstrate that our model consistently outperforms massive proprietary systems. EyExIn significantly enhances domain-specific knowledge embedding and achieves state-of-the-art precision in ophthalmic visual question answering, advancing the development of trustworthy ophthalmic AI.

In the training of neural networks, adaptive moment estimation (Adam) typically converges fast but exhibits suboptimal generalization performance. A widely accepted explanation for its defect in generalization is that it often tends to converge to sharp minima. To enhance its ability to find flat minima, we propose its new variant named inverse Adam (InvAdam). The key improvement of InvAdam lies in its parameter update mechanism, which is opposite to that of Adam. Specifically, it computes element-wise multiplication of the first-order and second-order moments, while Adam computes the element-wise division of these two moments. This modification aims to increase the step size of the parameter update when the elements in the second-order moments are large and vice versa, which helps the parameter escape sharp minima and stay at flat ones. However, InvAdam's update mechanism may face challenges in convergence. To address this challenge, we propose dual Adam (DualAdam), which integrates the update mechanisms of both Adam and InvAdam, ensuring convergence while enhancing generalization performance. Additionally, we introduce the diffusion theory to mathematically demonstrate InvAdam's ability to escape sharp minima. Extensive experiments are conducted on image classification tasks and large language model (LLM) fine-tuning. The results validate that DualAdam outperforms Adam and its state-of-the-art variants in terms of generalization performance. The code is publicly available at https://github.com/LongJin-lab/DualAdam.

As autonomous AI agents increasingly populate the Internet, a novel security challenge arises: "Is this entity an AI agent?" It is a new entity-type verification problem with no established solution. We formalize the problem through a three-class entity taxonomy (Human, Script, Agent) based on a verifiable agentic capability vector <x, r, s> (action, reasoning, and memory). A timing threshold t exploits the asymmetric hardness between human cognition and AI processing to separate the three classes. We define the Agentic Capability Verification Problem (ACVP) through three necessity primitives, each testing one capability dimension. Building on this foundation, we introduce aCAPTCHA (Agent CAPTCHA), a time-constrained security game for agent admission whose security rests on ACVP hardness under t. We instantiate aCAPTCHA through time-bounded natural-language understanding as a multi-round HTTP verification protocol, and evaluate it with preliminary agent trials that validate the protocol's soundness and completeness. aCAPTCHA provides a composable, infrastructure-free admission gate for any service where entity-type verification is required.

The Werewolf Game is a communication game where players' reasoning and discussion skills are essential. In this study, we present a Werewolf AI agent developed for the AIWolfDial 2024 shared task, co-hosted with the 17th INLG. In recent years, large language models like ChatGPT have garnered attention for their exceptional response generation and reasoning capabilities. We thus develop the LLM-based agents for the Werewolf Game. This study aims to enhance the consistency of the agent's utterances by utilizing dialogue summaries generated by LLMs and manually designed personas and utterance examples. By analyzing self-match game logs, we demonstrate that the agent's utterances are contextually consistent and that the character, including tone, is maintained throughout the game.

Understanding physical transformations is fundamental for reasoning in dynamic environments. While Vision Language Models (VLMs) show promise in embodied applications, whether they genuinely understand physical transformations remains unclear. We introduce ConservationBench evaluating conservation -- whether physical quantities remain invariant under transformations. Spanning four properties with paired conserving/non-conserving scenarios, we generate 23,040 questions across 112 VLMs. Results reveal systematic failure: performance remains near chance with improvements on conservation tasks accompanied by drops on controls. Control experiments show strong textual priors favoring invariance, yet models perform worse with visual content. Neither temporal resolution, prompting, nor curated sampling helps. These findings show that current VLMs fail to maintain transformation-invariant representations of physical properties across dynamic scenes.

Accurate and reliable forecasting of epidemic incidences is critical for public health preparedness, yet it remains a challenging task due to complex nonlinear temporal dependencies and heterogeneous spatial interactions. Often, point forecasts generated by spatiotemporal models are unreliable in assigning uncertainty to future epidemic events. Probabilistic forecasting of epidemics is therefore crucial for providing the best or worst-case scenarios rather than a simple, often inaccurate, point estimate. We present deep spatiotemporal engression methods to generate accurate and reliable probabilistic forecasts on low-frequency epidemic datasets. The proposed methods act as distributional lenses, and out-of-sample probabilistic forecasts are generated by sampling from the trained models. Our frameworks encapsulate lightweight deep generative architectures, wherein uncertainty is quantified endogenously, driven by a pre-additive noise component during model construction. We establish geometric ergodicity and asymptotic stationarity of the spatiotemporal engression processes under mild assumptions on the network weights and pre-additive noise process. Comprehensive evaluations across six epidemiological datasets over three forecast horizons demonstrate that the proposal consistently outperforms several temporal and spatiotemporal benchmarks in both point and probabilistic forecasting. Additionally, we explore the explainability of the proposal to enhance the models' practical application for informed, timely public health interventions.

Generative models offer a promising paradigm for the final stage reranking in multi-stage recommender systems, with the ability to capture inter-item dependencies within reranked lists. However, their practical deployment still faces two key challenges: (1) an inherent conflict between achieving high generation quality and ensuring low-latency inference, making it difficult to balance the two, and (2) insufficient interaction between user and item features in existing methods. To address these challenges, we propose a novel Personalized Semi-Autoregressive with online knowledge Distillation (PSAD) framework for reranking. In this framework, the teacher model adopts a semi-autoregressive generator to balance generation quality and efficiency, while its ranking knowledge is distilled online into a lightweight scoring network during joint training, enabling real-time and efficient inference. Furthermore, we propose a User Profile Network (UPN) that injects user intent and models interest dynamics, enabling deeper interactions between users and items. Extensive experiments conducted on three large-scale public datasets demonstrate that PSAD significantly outperforms state-of-the-art baselines in both ranking performance and inference efficiency.

Creatively translating complex gameplay ideas into executable artifacts (e.g., games as Unity projects and code) remains a central challenge in computational game creativity. Gameplay design patterns provide a structured representation for describing gameplay phenomena, enabling designers to decompose high-level ideas into entities, constraints, and rule-driven dynamics. Among them, goal patterns formalize common player-objective relationships. Goal Playable Concepts (GPCs) operationalize these abstractions as playable Unity engine implementations, supporting experiential exploration and compositional gameplay design. We frame scalable playable pattern realization as a problem of constrained executable creative synthesis: generated artifacts must satisfy Unity's syntactic and architectural requirements while preserving the semantic gameplay meanings encoded in goal patterns. This dual constraint limits scalability. Therefore, we investigate whether contemporary large language models (LLMs) can perform such synthesis under engine-level structural constraints and generate Unity code (as games) structured and conditioned by goal playable patterns. Using 26 goal pattern instantiations, we compare a direct generation baseline (natural language -> C# -> Unity) with pipelines conditioned on a human-authored Unity-specific intermediate representation (IR), across three IR configurations and two open-source models (DeepSeek-Coder-V2-Lite-Instruct and Qwen2.5-Coder-7B-Instruct). Compilation success is evaluated via automated Unity replay. We propose grounding and hygiene failure modes, identifying structural and project-level grounding as primary bottlenecks.

We study concurrent graph games where n players cooperate against an opponent to reach a set of target states. Unlike traditional settings, we study distributed randomisation: team players do not share a source of randomness, and their private random sources are hidden from the opponent and from each other. We show that memoryless strategies are sufficient for the threshold problem (deciding whether there is a strategy for the team that ensures winning with probability that exceeds a threshold), a result that not only places the problem in the Existential Theory of the Reals (\exists\mathbb{R}) but also enables the construction of value iteration algorithms. We additionally show that the threshold problem is NP-hard. For the almost-sure reachability problem, we prove NP-completeness. We introduce Individually Randomised Alternating-time Temporal Logic (IRATL). This logic extends the standard ATL framework to reason about probability thresholds, with semantics explicitly designed for coalitions that lack a shared source of randomness. On the practical side, we implement and evaluate a solver for the threshold and almost-sure problem based on the algorithms that we develop.

This paper presents novel method for distribution-free robust trajectory optimization and control of discrete-time, nonlinear, and non-Gaussian stochastic systems, with closed-loop guarantees on chance constraint satisfaction. Our framework employs conformal inference to generate coverage-based confidence sets for the closed-loop dynamics around arbitrary reference trajectories, by constructing a joint nonconformity score to quantify both the validity of contraction (i.e., incremental stability) conditions and the impact of external stochastic disturbance on the closed-loop dynamics, without any distributional assumptions. Via appropriate constraint tightening, chance constraints can be reformulated into tractable, statistically valid deterministic constraints on the reference trajectories. This enables a formal pathway to leverage and validate learning-based motion planners and controllers, such as those with neural contraction metrics, in safety-critical real-world applications. Notably, our statistical guarantees are non-diverging and can be computed with finite samples of the underlying uncertainty, without overly conservative structural priors. We demonstrate our approach in motion planning problems for designing safe, dynamically feasible trajectories in both numerical simulation and hardware experiments.

In the era of "Software Engineering 2.0" (SE 2.0), where intelligent agents collaborate with human engineers, Generative AI is advancing beyond code generation into Software Architecture (SA). While Large Language Models (LLMs) demonstrate superior capabilities, computational costs and data privacy concerns drive interest in Small Language Models (SLMs) with fewer than 7 billion parameters. However, the reasoning limits of these resource-constrained models remain unexplored. This study benchmarks 10 state-of-the-art SLMs on Architectural Decision Records generation, introducing a multi-dimensional framework evaluating Technical Compliance and Semantic Diversity. Our empirical results reveal a significant reasoning gap: models above the 3B-parameter threshold demonstrate robust zero-shot capabilities, while sub-2B models show the strongest BERTScore gains from Fine-Tuning, though compliance improvements are not guaranteed. Contrary to assumptions regarding context saturation, Few-Shot prompting serves as a highly effective calibration mechanism for select mid-sized models with short context windows. Furthermore, high semantic diversity in off-the-shelf small models often correlates with hallucination rather than productive exploration. These findings establish a rigorous baseline for deploying sustainable, locally hosted architectural assistants.

As Joint Audio-Visual Generation Models see widespread commercial deployment, embedding watermarks has become essential for protecting vendor copyright and ensuring content provenance. However, existing techniques suffer from an architectural mismatch by treating modalities as decoupled entities, exposing a critical Binding Vulnerability. Adversaries exploit this via Swap Attacks by replacing authentic audio with malicious deepfakes while retaining the watermarked video. Because current detectors rely on independent verification ($Video_{wm}\vee Audio_{wm}$), they incorrectly authenticate the manipulated content, falsely attributing harmful media to the original vendor and severely damaging their reputation. To address this, we propose mAVE (Manifold Audio-Visual Entanglement), the first watermarking framework natively designed for joint architectures. mAVE cryptographically binds audio and video latents at initialization without fine-tuning, defining a Legitimate Entanglement Manifold via Inverse Transform Sampling. Experiments on state-of-the-art models (LTX-2, MOVA) demonstrate that mAVE guarantees performance-losslessness and provides an exponential security bound against Swap Attacks. Achieving near-perfect binding integrity ($>99\%$), mAVE offers a robust cryptographic defense for vendor copyright.

Reward hacking is a form of misalignment in which models overoptimize proxy rewards without genuinely solving the underlying task. Precisely measuring reward hacking occurrence remains challenging because true task rewards are often expensive or impossible to compute. We introduce Countdown-Code, a minimal environment where models can both solve a mathematical reasoning task and manipulate the test harness. This dual-access design creates a clean separation between proxy rewards (test pass/fail) and true rewards (mathematical correctness), enabling accurate measurement of reward-hacking rates. Using this environment, we study reward hacking in open-weight LLMs and find that such behaviors can be unintentionally learned during supervised fine-tuning (SFT) when even a small fraction of reward-hacking trajectories leak into training data. As little as 1\% contamination in distillation SFT data is sufficient for models to internalize reward hacking which resurfaces during subsequent reinforcement learning (RL). We further show that RL amplifies misalignment and drives its generalization beyond the original domain. We open-source our environment and code to facilitate future research on reward hacking in LLMs. Our results reveal a previously underexplored pathway through which reward hacking can emerge and persist in LLMs, underscoring the need for more rigorous validation of synthetic SFT data. Code is available at https://github.com/zohaib-khan5040/Countdown-Code.

Model-based reinforcement learning (MBRL) agents operating in high-dimensional observation spaces, such as Dreamer, rely on learning abstract representations for effective planning and control. Existing approaches typically employ reconstruction-based objectives in the observation space, which can render representations sensitive to task-irrelevant details. Recent alternatives trade reconstruction for auxiliary action prediction heads or view augmentation strategies, but perform worse in the Crafter environment than reconstruction-based methods. We close this gap between Dreamer and reconstruction-free models by introducing a JEPA-style predictor defined on continuous, deterministic representations. Our method matches Dreamer's performance on Crafter, demonstrating effective world model learning on this benchmark without reconstruction objectives.

Vision-and-Language Navigation (VLN) increasingly relies on large vision-language models, but their inference cost conflicts with real-time deployment. Token caching is a promising training-free strategy that avoids redundant computation by reusing stable visual tokens across frames. However, existing methods assume a static camera and fixed semantic focus, assumptions that VLN fundamentally violates. We identify two failure modes: (1) visual dynamics, where viewpoint shift displaces token positions across frames, causing position-wise matching to pair misaligned content; (2) semantic dynamics, where token relevance shifts across task stages as navigation progresses, making cached states stale. We propose VLN-Cache, a visual-dynamic-aware and semantic-dynamic-aware caching framework that introduces view-aligned remapping to recover geometric correspondences and a task-relevance saliency filter to veto reuse at semantic transitions. A layer-adaptive entropy policy further balances the per-layer reuse budget. Experiments on the R2R-CE simulation benchmark show up to 1.52x speedup while maintaining competitive navigation success rates.

On-policy distillation is a promising approach for transferring knowledge between language models, where a student learns from dense token-level signals along its own trajectories. This framework typically uses reverse KL divergence, encouraging the student to match the teacher's high-confidence predictions. However, we show that the mode-seeking property of reverse KL reduces generation diversity and yields unstable learning signals when the teacher distribution has high entropy. To address this, we introduce Entropy-Aware On-Policy Distillation. Our key idea is augmenting the standard reverse KL objective with forward KL when teacher entropy is high, capturing the full range of plausible outputs while retaining precise imitation elsewhere. It balances mode-seeking precision with mode-covering robustness without sacrificing on-policy training efficiency. Experiments show that our method maintains generation diversity (sustained token-level entropy) and improves student-teacher alignment (lower forward KL on high-entropy tokens). Across six math reasoning benchmarks, this yields Pass@8 accuracy gains of +1.37 for Qwen3-0.6B-Base, +2.39 for Qwen3-1.7B-Base, and +5.05 for Qwen3-4B-Base compared to baseline on-policy distillation methods. These results demonstrate that accounting for teacher uncertainty is essential for maintaining diversity and achieving effective knowledge transfer.

Large Reasoning Models (LRMs) have demonstrated strong performance by producing extended Chain-of-Thought (CoT) traces before answering. However, this paradigm often induces over-reasoning: redundant calculations and circular self-verification that increase computational cost without improving outcomes. Existing evaluations largely emphasize final accuracy or coarse token counts, and lack automated tools to separate essential logic from structural redundancy. We introduce CoTJudger, a graph-driven framework that quantifies reasoning efficiency by converting free-form CoTs into directed dependency graphs and extracting the Shortest Effective Path (SEP) needed to reach a correct solution. This yields an interpretable efficiency signal -- how much of a CoT is necessary versus structurally redundant -- that is comparable across models and tasks. Evaluating 21 LRMs, CoTJudger reveals pervasive redundancy and surfaces recurring failure modes, including verification obsession and compensatory redundancy. These results provide a practical metric for disentangling reasoning ability from computational waste, enabling more targeted evaluation and diagnosis of LRM efficiency.

Anomaly detection is a crucial machine-learning task with wide-ranging applications. Deep Support Vector Data Description (Deep SVDD) is a prominent deep one-class method, but it is vulnerable to hypersphere collapse, often relies on heuristic choices for hypersphere parameters, and provides limited interpretability. To address these issues, we propose Interpretable Maximum Margin Deep Anomaly Detection (IMD-AD), which leverages a small set of labeled anomalies and a maximum margin objective to stabilize training and improve discrimination. It is inherently resilient to hypersphere collapse. Furthermore, we prove an equivalence between hypersphere parameters and the network's final-layer weights, which allows the center and radius to be learned end-to-end as part of the model and yields intrinsic interpretability and visualizable outputs. We further develop an efficient training algorithm that jointly optimizes representation, margin, and final-layer parameters. Extensive experiments and ablation studies on image and tabular benchmarks demonstrate that IMD-AD empirically improves detection performance over several state-of-the-art baselines while providing interpretable decision diagnostics.

We present Artoo, a learned acoustic communication system for robots that replaces hand-designed signal processing with end-to-end co-trained neural networks. Our system pairs a lightweight text-to-speech (TTS) transmitter (1.18M parameters) with a conformer-based automatic speech recognition (ASR) receiver (938K parameters), jointly optimized through a differentiable channel. Unlike human speech, robot-to-robot communication is paralinguistics-free: the system need not preserve timbre, prosody, or naturalness, only maximize decoding accuracy under channel distortion. Through a three-phase co-training curriculum, the TTS transmitter learns to produce distortion-robust acoustic encodings that surpass the baseline under noise, achieving 8.3% CER at 0 dB SNR. The entire system requires only 2.1M parameters (8.4 MB) and runs in under 13 ms end-to-end on a CPU, making it suitable for deployment on resource-constrained robotic platforms.

User reviews on e-commerce and review sites are crucial for making purchase decisions, although creating detailed reviews is time-consuming and labor-intensive. In this study, we propose a novel use of dialogue systems to facilitate user review creation by generating reviews from information gathered during interview dialogues with users. To validate our approach, we implemented our system using GPT-4 and conducted comparative experiments from the perspectives of system users and review readers. The results indicate that participants who used our system rated their interactions positively. Additionally, reviews generated by our system required less editing to achieve user satisfaction compared to those by the baseline. We also evaluated the reviews from the reader' perspective and found that our system-generated reviews are more helpful than those written by humans. Despite challenges with the fluency of the generated reviews, our method offers a promising new approach to review writing.

Generative diffusion models are increasingly used for medical imaging data augmentation, but text prompting cannot produce causal training data. Re-prompting rerolls the entire generation trajectory, altering anatomy, texture, and background. Inversion-based editing methods introduce reconstruction error that causes structural drift. We propose MedSteer, a training-free activation-steering framework for endoscopic synthesis. MedSteer identifies a pathology vector for each contrastive prompt pair in the cross-attention layers of a diffusion transformer. At inference time, it steers image activations along this vector, generating counterfactual pairs from scratch where the only difference is the steered concept. All other structure is preserved by construction. We evaluate MedSteer across three experiments on Kvasir v3 and HyperKvasir. On counterfactual generation across three clinical concept pairs, MedSteer achieves flip rates of 0.800, 0.925, and 0.950, outperforming the best inversion-based baseline in both concept flip rate and structural preservation. On dye disentanglement, MedSteer achieves 75% dye removal against 20% (PnP) and 10% (h-Edit). On downstream polyp detection, augmenting with MedSteer counterfactual pairs achieves ViT AUC of 0.9755 versus 0.9083 for quantity-matched re-prompting, confirming that counterfactual structure drives the gain. Code is at link https://github.com/phamtrongthang123/medsteer

Hand-crafted mutants are increasingly used to evaluate fuzzing and property-based testing tools, but current tooling is fragmented and often forces trade-offs between readability, mutation preservation, and execution cost. We present a declarative framework for hand-crafted mutation analysis and management. First, we characterize five mutation representations: comment-based, preprocessor-based, patch-based, match-and-replace, and in-AST mutations. Second, we define a mutation algebra that supports selective execution, tag-based expansion, and higher-order combinations of mutants. Third, we describe a lossless conversion pipeline that maps mutation representations through a common intermediate form, including a strategy for extracting and normalizing in-AST mutations. We implement these ideas in Marauder, a prototype system for injecting, activating, resetting, and composing hand-crafted mutants across representations. This framework clarifies the design space of hand-crafted mutation systems and provides a practical foundation for more efficient and expressive mutation experiments.

Intelligent fault diagnosis (IFD) has emerged as a powerful paradigm for ensuring the safety and reliability of industrial machinery. However, traditional IFD methods rely heavily on abundant labeled data for training, which is often difficult to obtain in practical industrial environments. Constructing a digital twin (DT) of the physical asset to obtain simulation data has therefore become a promising alternative. Nevertheless, existing DT-assisted diagnosis methods mainly transfer diagnostic knowledge through domain adaptation techniques, which still require a considerable amount of unlabeled data from the target asset. To address the challenges in few-shot scenarios where only extremely limited samples are available, a bi-directional DT prototype anchoring method with multi-periodicity learning is proposed. Specifically, a framework involving meta-training in the DT virtual space and test-time adaptation in the physical space is constructed for reliable few-shot model adaptation for the target asset. A bi-directional twin-domain prototype anchoring strategy with covariance-guided augmentation for adaptation is further developed to improve the robustness of prototype estimation. In addition, a multi-periodicity feature learning module is designed to capture the intrinsic periodic characteristics within current signals. A DT of an asynchronous motor is built based on finite element method, and experiments are conducted under multiple few-shot settings and three working conditions. Comparative and ablation studies demonstrate the superiority and effectiveness of the proposed method for few-shot fault diagnosis.

Scientists face significant visualization challenges as time-varying datasets grow in speed and volume, often requiring specialized infrastructure and expertise to handle massive datasets. Petascale climate models generated in NASA laboratories require a dedicated group of graphics and media experts and access to high-performance computing resources. Scientists may need to share scientific results with the community iteratively and quickly. However, the time-consuming trial-and-error process incurs significant data transfer overhead and far exceeds the time and resources allocated for typical post-analysis visualization tasks, disrupting the production workflow. Our paper introduces a user-friendly framework for creating 3D animations of petascale, time-varying data on a commodity workstation. Our contributions: (i) Generalized Animation Descriptor (GAD) with a keyframe-based adaptable abstraction for animation, (ii) efficient data access from cloud-hosted repositories to reduce data management overhead, (iii) tailored rendering system, and (iv) an LLM-assisted conversational interface as a scripting module to allow domain scientists with no visualization expertise to create animations of their region of interest. We demonstrate the framework's effectiveness with two case studies: first, by generating animations in which sampling criteria are specified based on prior knowledge, and second, by generating AI-assisted animations in which sampling parameters are derived from natural-language user prompts. In all cases, we use large-scale NASA climate-oceanographic datasets that exceed 1PB in size yet achieve a fast turnaround time of 1 minute to 2 hours. Users can generate a rough draft of the animation within minutes, then seamlessly incorporate as much high-resolution data as needed for the final version.

Although large vision-language models (LVLMs) have demonstrated remarkable capabilities, they are prone to hallucinations in multi-image tasks. We attribute this issue to limitations in existing attention mechanisms and insufficient cross-image modeling. Inspired by this, we propose a structured hallucination mitigation framework involving Cross-Image Attention calibration and Preference Learning (CAPL). CAPL explicitly enhances inter-image interactions at the architectural level while reinforcing reliance on genuine cross-image evidence during training, thereby improving the model's perception and modeling of cross-image associations. Specifically, we (i) introduce a selectable image token interaction attention mechanism to establish fine-grained cross-image entity alignment and information flow; (ii) design a cross-image modeling-based preference optimization strategy that contrasts reasoning outcomes under full inter-image interaction and those obtained when images are mutually invisible, encouraging the model to ground its predictions in authentic visual evidence and mitigating erroneous inferences driven by textual priors. Experimental results demonstrate that CAPL consistently improves performance across multiple model architectures, achieving stable gains on both multi-image hallucination and general benchmarks. Notably, performance on single-image visual tasks remains stable or slightly improves, indicating strong generalization capability.

Failures in clusters running large-scale AI workloads can result in decreased utilization. Because the cost of a failure in such AI workloads is high (as it requires restarting the entire job from a previous checkpoint), there are many mechanisms in place to ensure that the failures are mitigated, and the impact of a failure is minimized. However, these mechanisms have many knobs and parameters, all of which must be carefully tuned based on the system and cluster's characteristics. We built AIReSim, a discrete event simulator to evaluate the different design choices during the failure, recovery, scheduling and repair processes for a cluster running a large-scale AI workload. AIReSim allows the system designer to systematically evaluate the effects of the different knobs and parameters on the overall end-to-end reliability of the system. Further, AIReSim can be used to identify which knobs or parameters are important in order to prioritize the investment of effort in improving the system. AIReSim also allows tuning of the knobs for achieving different tradeoffs in the system, as well as to consider various ``what-if'' scenarios. We present a case study of applying AIReSim for capacity planning for large-scale clusters running AI workloads.

We present DeepEarth, a self-supervised multi-modal world model with Earth4D, a novel planetary-scale 4D space-time positional encoder. Earth4D extends 3D multi-resolution hash encoding to include time, efficiently scaling across the planet over centuries with sub-meter, sub-second precision. Multi-modal encoders (e.g. vision-language models) are fused with Earth4D embeddings and trained via masked reconstruction. We demonstrate Earth4D's expressive power by achieving state-of-the-art performance on an ecological forecasting benchmark. Earth4D with learnable hash probing surpasses a multi-modal foundation model pre-trained on substantially more data. Access open source code and download models at: https://github.com/legel/deepearth

In cross-silo federated learning (FL), sensitive text datasets remain confined to local organizations due to privacy regulations, making repeated training for each downstream task both communication-intensive and privacy-demanding. A promising alternative is to generate differentially private (DP) synthetic datasets that approximate the global distribution and can be reused across tasks. However, pretrained large language models (LLMs) often fail under domain shift, and federated finetuning is hindered by computational heterogeneity: only resource-rich clients can update the model, while weaker clients are excluded, amplifying data skew and the adverse effects of DP noise. We propose a flexible participation framework that adapts to client capacities. Strong clients perform DP federated finetuning, while weak clients contribute through a lightweight DP voting mechanism that refines synthetic text. To ensure the synthetic data mirrors the global dataset, we apply control codes (e.g., labels, topics, metadata) that represent each client's data proportions and constrain voting to semantically coherent subsets. This two-phase approach requires only a single round of communication for weak clients and integrates contributions from all participants. Experiments show that our framework improves distribution alignment and downstream robustness under DP and heterogeneity.

Speech Large Language Models (LLMs) that understand and follow instructions in many languages are useful for real-world interaction, but are difficult to train with supervised fine-tuning, requiring large, task-specific speech corpora. While recent distillation-based approaches train performant English-only Speech LLMs using only annotated ASR data by aligning text and speech using only a lightweight projector, these models under-perform when scaled to multilingual settings due to language interference in the shared projector. We address this by introducing language-aware distillation using a query bank and a gating network that selects or mixes query tokens using a Q-Former projector. Our approach shows gains of 14% over matched multilingual distillation baselines on instruction following. We further synthesize Audio-MLQA, a multilingual spoken QA benchmark built on MLQA with high-quality TTS questions. Our best model improves over existing Speech LLM baselines by 32% on Audio-MLQA.

Large language model-based web agents have shown strong potential in automating web interactions through advanced reasoning and instruction following. While retrieval-based memory derived from historical trajectories enables these agents to handle complex, long-horizon tasks, current methods struggle to generalize across unseen websites. We identify that this challenge arises from the flat memory structures that entangle high-level task logic with site-specific action details. This entanglement induces a workflow mismatch in new environments, where retrieved contents are conflated with current web, leading to logically inconsistent execution. To address this, we propose Hierarchical Memory Tree (HMT), a structured framework designed to explicitly decouple logical planning from action execution. HMT constructs a three-level hierarchy from raw trajectories via an automated abstraction pipeline: the Intent level maps diverse user instructions to standardized task goals; the Stage level defines reusable semantic subgoals characterized by observable pre-conditions and post-conditions; and the Action level stores action patterns paired with transferable semantic element descriptions. Leveraging this structure, we develop a stage-aware inference mechanism comprising a Planner and an Actor. By explicitly validating pre-conditions, the Planner aligns the current state with the correct logical subgoal to prevent workflow mismatch, while the Actor grounds actions by matching the stored semantic descriptions to the target page. Experimental results on Mind2Web and WebArena show that HMT significantly outperforms flat-memory methods, particularly in cross-website and cross-domain scenarios, highlighting the necessity of structured memory for robust generalization of web agents.

Despite the promise of Retrieval-Augmented Generation in grounding Multimodal Large Language Models with external knowledge, the transition to extensive contexts often leads to significant attention dilution and reasoning hallucinations. The surge in information density causes critical evidence to be submerged by voluminous noise, which complicates the discernment of relevant fragments within a dense input. In this paper, we propose \textbf{Hit-RAG}, a multi-stage preference alignment framework designed to resolve these cognitive bottlenecks through a progressive optimization pipeline. Our approach systematically refines the utilization of external evidence via three distinct stages. First, Supervised Fine-tuning establishes baseline context awareness to minimize information neglect. Next, Discriminative Preference Alignment enhances robustness against misleading distractors. Finally, Group-Relative Policy Optimization stabilizes logical synthesis to prevent reasoning collapse. Extensive evaluations on eight benchmarks demonstrate that Hit-RAG consistently yields substantial performance gains, enabling models to bridge the gap between context acquisition and accurate reasoning while surpassing much larger counterparts in long-context scenarios.

Neural approaches to the Flexible Job Shop Scheduling Problem (FJSP), particularly those based on deep reinforcement learning (DRL), have gained growing attention in recent years. However, existing methods rely on complex feature-engineered state representations (i.e., often requiring more than 20 handcrafted features) and graph-biased neural architectures. To reduce modeling complexity and advance a more generalizable framework for FJSP, we introduce \textsc{ReSched}, a minimalist DRL framework that rethinks both the scheduling formulation and model design. First, by revisiting the Markov Decision Process (MDP) formulation of FJSP, we condense the state space to just four essential features, eliminating historical dependencies through a subproblem-based perspective. Second, we employ Transformer blocks with dot-product attention, augmented by three lightweight but effective architectural modifications tailored to scheduling tasks. Extensive experiments show that \textsc{ReSched} outperforms classical dispatching rules and state-of-the-art DRL methods on FJSP. Moreover, \textsc{ReSched} also generalizes well to the Job Shop Scheduling Problem (JSSP) and the Flexible Flow Shop Scheduling Problem (FFSP), achieving competitive performance against neural baselines specifically designed for these variants.

Checklists have emerged as a popular approach for interpretable and fine-grained evaluation, particularly with LLM-as-a-Judge. Beyond evaluation, these structured criteria can serve as signals for model alignment, reinforcement learning, and self-correction. To support these use cases, we present AutoChecklist, an open-source library that unifies checklist-based evaluation into composable pipelines. At its core is a taxonomy of five checklist generation abstractions, each encoding a distinct strategy for deriving evaluation criteria. A modular Generator $\rightarrow$ Refiner $\rightarrow$ Scorer pipeline connects any generator with a unified scorer, and new configurations can be registered via prompt templates alone. The library ships with ten built-in pipelines implementing published approaches and supports multiple LLM providers (OpenAI, OpenRouter, vLLM). Beyond the Python API, the library includes a CLI for off-the-shelf evaluation and a web interface for interactive exploration. Validation experiments confirm that these checklist methods significantly align with human preferences and quality ratings, and a case study on ICLR peer review rebuttals demonstrates flexible domain adaptation. AutoChecklist is publicly available at https://github.com/ChicagoHAI/AutoChecklist.

Treatment effects estimated from randomized controlled trials are local not only to the study population but also to the time at which the trial was conducted. We develop a framework for temporal transportation: extrapolating treatment effects to time periods where no experiment was conducted. We target the transported average treatment effect (TATE) and show that under a separable temporal effects assumption, the TATE decomposes into an observed average treatment effect and a temporal ratio. We provide two identification strategies -- one using replicated trials comparing the same treatments at different times, another using common treatment arms observed across time -- and develop doubly robust, semiparametrically efficient estimators for each. Monte Carlo simulations confirm that both estimators achieve nominal coverage, with the common arm strategy yielding substantial efficiency gains when its stronger assumptions hold. We apply our methods to A/B tests from the Upworthy Research Archive, demonstrating that the two strategies exhibit a variance-bias tradeoff: the common arm approach offers greater precision but may incur bias when treatments interact heterogeneously with temporal factors.

Safety alignment is critical for deploying large language models (LLMs) in real-world applications, yet most existing approaches rely on large human-annotated datasets and static red-teaming benchmarks that are costly, difficult to scale, and slow to adapt to evolving model behaviors. Moreover, overly conservative safety mechanisms can reduce model usefulness by rejecting sensitive but legitimate queries. We introduce Self-MOA (Self Multi-Objective Alignment), a fully automated framework for aligning small language models using weak supervision from automated evaluator models. Self-MOA operates as a closed loop that dynamically generates model-specific red team prompts, constructs preference data from model-generated responses, and aligns models via multi-objective preference optimization to jointly optimize for safety and helpfulness. Across multiple small language models and safety benchmarks, Self-MOA achieves a 12.41\% improvement in safety while preserving helpfulness, using as little as 11 times less training data than human-supervised alignment baselines. These results demonstrate that adaptive, automated alignment can reduce the dependence on static, human-curated safety pipelines in resource-constrained settings.

This study examines how students integrate generative artificial intelligence (AI) into design projects through the lens of the SuperSkillsStack framework, which identifies four key human competencies for effective human-AI collaboration: Agency, Domain Knowledge, Imagination, and Taste. As generative AI increasingly transforms creative practice, design education must consider how human capabilities are cultivated alongside technological tools. Using qualitative thematic analysis, this study analyzes reflective writings from 80 student design teams participating in a human-centered design course. The findings show that students primarily used AI during the early stages of the design process, including brainstorming, information synthesis, and problem framing. However, students consistently relied on human judgment to interpret contextual information, validate AI-generated outputs, and refine design solutions. Domain knowledge derived from field observations enabled students to detect inaccuracies in AI suggestions, while taste played a critical role in evaluating and selecting meaningful ideas. The results suggest that generative AI functions primarily as a cognitive accelerator rather than a replacement for human creativity. The study highlights the importance of cultivating higher-order human capabilities to support effective human-AI collaboration in design education.

Mixture-of-Experts (MoE) architecture offers enhanced efficiency for Large Language Models (LLMs) with modularized computation, yet its inherent sparsity poses significant hardware deployment challenges, including memory locality issues, communication overhead, and inefficient computing resource utilization. Inspired by the modular organization of the human brain, we propose Mozart, a novel algorithm-hardware co-design framework tailored for efficient training of MoE-based LLMs on 3.5D wafer-scale chiplet architectures. On the algorithm side, Mozart exploits the inherent modularity of chiplets and introduces: (1) an expert allocation strategy that enables efficient on-package all-to-all communication, and (2) a fine-grained scheduling mechanism that improves communication-computation overlap through streaming tokens and experts. On the architecture side, Mozart adaptively co-locates heterogeneous modules on specialized chiplets with a 2.5D NoP-Tree topology and hierarchical memory structure. Evaluation across three popular MoE models demonstrates significant efficiency gains, enabling more effective parallelization and resource utilization for large-scale modularized MoE-LLMs.

A matching platform is a system that matches different types of participants, such as companies and job-seekers. In such a platform, merely maximizing the number of matches can result in matches being concentrated on highly popular participants, which may increase dissatisfaction among other participants, such as companies, and ultimately lead to their churn, reducing the platform's profit opportunities. To address this issue, we propose a novel online learning problem, Combinatorial Allocation Bandits (CAB), which incorporates the notion of *arm satisfaction*. In CAB, at each round $t=1,\dots,T$, the learner observes $K$ feature vectors corresponding to $K$ arms for each of $N$ users, assigns each user to an arm, and then observes feedback following a generalized linear model (GLM). Unlike prior work, the learner's objective is not to maximize the number of positive feedback, but rather to maximize the arm satisfaction. For CAB, we provide an upper confidence bound algorithm that achieves an approximate regret upper bound, which matches the existing lower bound for the special case. Furthermore, we propose a TS algorithm and provide an approximate regret upper bound. Finally, we conduct experiments on synthetic data to demonstrate the effectiveness of the proposed algorithms compared to other methods.

Adaptive Large Neighborhood Search (ALNS) is a prominent metaheuristic and a widely adopted approach for production and logistics optimization. However, it has long relied on hand-crafted components built on expert experience, which makes development slow and costly to adapt to new problems. This paper proposes a closed-loop, large-language-model-driven evolutionary framework that decouples ALNS and automatically rebuilds all of its components. We break ALNS into seven key modules: destroy, repair, operator selection, weight update, initial solution construction, acceptance rule, and destroy-rate control, and evolve each module through a dedicated task. By incorporating the MAP-Elites mechanism, the framework maintains a multi-dimensional elite archive to simultaneously drive the evolution of solution quality and strategic diversity. On TSPLIB benchmarks, the evolved algorithms consistently outperform optimized classic ALNS baselines under both fixed-iteration and fixed-time limits. The gains are especially clear on large-scale instances, where the average optimality gap drops from 3.18% to 0.74%. Code analysis also uncovers several counterintuitive yet meaningful design patterns that emerged naturally during evolution, offering practical and theoretical insights for future ALNS design. Finally, comparisons across multiple language models highlight clear differences in their ability to support evolutionary algorithm design, helping guide model selection for real-world engineering use.

In predictive modeling for low-resource audio classification, extracting high-accuracy and interpretable attributes is critical. Particularly in high-reliability applications, interpretable audio attributes are indispensable. While human-driven attribute discovery is effective, its low throughput becomes a bottleneck. We propose a method for adaptively discovering interpretable audio attributes using Multimodal Large Language Models (MLLMs). By replacing humans in the AdaFlock framework with MLLMs, our method achieves significantly faster attribute discovery. Our method dynamically identifies salient acoustic characteristics via prompting and constructs an attribute-based ensemble classifier. Experimental results across various audio tasks demonstrate that our method outperforms direct MLLM prediction in the majority of evaluated cases. The entire training completes within 11 minutes, proving it a practical, adaptive solution that surpasses conventional human-reliant approaches.

Deploying visuomotor robots at scale is challenging due to the potential for anomalous failures to degrade performance, cause damage, or endanger human life. Bimanual manipulators are no exception; these robots have vast state spaces comprised of high-dimensional images and proprioceptive signals. Explicitly defining failure modes within such state spaces is infeasible. In this work, we overcome these challenges by training a probabilistic, history informed, world model within the compressed latent space of a pretrained vision foundation model (NVIDIA's Cosmos Tokenizer). The model outputs uncertainty estimates alongside its predictions that serve as non-conformity scores within a conformal prediction framework. We use these scores to develop a runtime monitor, correlating periods of high uncertainty with anomalous failures. To test these methods, we use the simulated Push-T environment and the Bimanual Cable Manipulation dataset, the latter of which we introduce in this work. This new dataset features trajectories with multiple synchronized camera views, proprioceptive signals, and annotated failures from a challenging data center maintenance task. We benchmark our methods against baselines from the anomaly detection and out-of-distribution detection literature, and show that our approach considerably outperforms statistical techniques. Furthermore, we show that our approach requires approximately one twentieth of the trainable parameters as the next-best learning-based approach, yet outperforms it by 3.8% in terms of failure detection rate, paving the way toward safely deploying manipulator robots in real-world environments where reliability is non-negotiable.

Recent work has proposed powerful frameworks, rooted in causal theory, to quantify fairness. Causal inference has primarily emphasized the detection of \emph{average} treatment effects (ATEs), and subsequent notions of fairness have inherited this focus. In this paper, we build on previous concerns about regulation based on averages. In particular, we formulate the "causal masking problem" as a linear program that optimizes an alternative objective, such as maximizing profit or minimizing crime, while retaining a zero ATE (i.e., the ATE between a protected attribute and a decision). By studying the capabilities and limitations of causal masking, we show that optimization under ATE-based regulation may induce significant unequal treatment. We demonstrate that the divergence between true and causally masked fairness is driven by confounding, underscoring the importance of full conditional-independence testing when assessing fairness. Finally, we discuss statistical and information-theoretic limitations that make causally masked solutions very difficult to detect, allowing them to persist for long periods. These results argue that we must regulate fairness at the model-level, rather than at the decision level.

Controllable diffusion generation often relies on various heuristics that are seemingly disconnected without a unified understanding. We bridge this gap with Diffusion Controller (DiffCon), a unified control-theoretic view that casts reverse diffusion sampling as state-only stochastic control within (generalized) linearly-solvable Markov Decision Processes (LS-MDPs). Under this framework, control acts by reweighting the pretrained reverse-time transition kernels, balancing terminal objectives against an $f$-divergence cost. From the resulting optimality conditions, we derive practical reinforcement learning methods for diffusion fine-tuning: (i) f-divergence-regularized policy-gradient updates, including a PPO-style rule, and (ii) a regularizer-determined reward-weighted regression objective with a minimizer-preservation guarantee under the Kullback-Leibler (KL) divergence. The LS-MDP framework further implies a principled model form: the optimal score decomposes into a fixed pretrained baseline plus a lightweight control correction, motivating a side-network parameterization conditioned on exposed intermediate denoising outputs, enabling effective gray-box adaptation with a frozen backbone. Experiments on Stable Diffusion v1.4 across supervised and reward-driven finetuning show consistent gains in preference-alignment win rates and improved quality-efficiency trade-offs versus gray-box baselines and even the parameter-efficient white-box adapter LoRA.

Modern enterprise platforms increasingly depend on distributed microservices, analytical data platforms, and external APIs to construct composite responses for applications. Orchestrating data retrieval across these heterogeneous systems is challenging because many workflow platforms rely on predefined workflows or state-machine definitions. Systems such as Apache Airflow, AWS Step Functions, and Temporal provide powerful orchestration capabilities but typically assume workflows are defined prior to execution. This paper presents a configuration-driven runtime orchestration framework for dynamic data retrieval in distributed systems. The framework generates execution graphs dynamically from configuration at request time, enabling low-latency orchestration without redeploying workflow code when integrations evolve. The execution planner performs dependency-aware scheduling and parallel execution of independent tasks, allowing efficient aggregation across distributed services. The paper describes the architecture, execution model, and operational tradeoffs of this framework, and presents a representative enterprise case study for Customer 360 retrieval. The approach demonstrates how runtime configuration can enable flexible and scalable orchestration in rapidly evolving integration environments.

Multi-agent reinforcement learning (MARL) is increasingly used to design learning-enabled agents that interact in shared environments. However, training MARL algorithms in general-sum games remains challenging: learning dynamics can become unstable, and convergence guarantees typically hold only in restricted settings such as two-player zero-sum or fully cooperative games. Moreover, when agents have heterogeneous and potentially conflicting preferences, it is unclear what system-level objective should guide learning. In this paper, we propose a new MARL pipeline called Near-Potential Policy Optimization (NePPO) for computing approximate Nash equilibria in mixed cooperative--competitive environments. The core idea is to learn a player-independent potential function such that the Nash equilibrium of a cooperative game with this potential as the common utility approximates a Nash equilibrium of the original game. To this end, we introduce a novel MARL objective such that minimizing this objective yields the best possible potential function candidate and consequently an approximate Nash equilibrium of the original game. We develop an algorithmic pipeline that minimizes this objective using zeroth-order gradient descent and returns an approximate Nash equilibrium policy. We empirically show the superior performance of this approach compared to popular baselines such as MAPPO, IPPO and MADDPG.

We present the first large-scale, cross-domain evaluation of document chunking strategies for dense retrieval, addressing a critical but underexplored aspect of retrieval-augmented systems. In our study, 36 segmentation methods spanning fixed-size, semantic, structure-aware, hierarchical, adaptive, and LLM-assisted approaches are benchmarked across six diverse knowledge domains using five different embedding models. Retrieval performance is assessed using graded relevance scores from a state-of-the-art LLM evaluator, with Normalised DCG@5 as the primary metric (complemented by Hit@5 and MRR). Our experiments show that content-aware chunking significantly improves retrieval effectiveness over naive fixed-length splitting. The top-performing strategy, Paragraph Group Chunking, achieved the highest overall accuracy (mean nDCG@5~0.459) and substantially better top-rank hit rates (Precision@1~24%, Hit@5~59%). In contrast, simple fixed-size character chunking as baselines performed poorly (nDCG@5 < 0.244, Precision@1~2-3%). We observe pronounced domain-specific differences: dynamic token sizing is strongest in biology, physics and health, while paragraph grouping is strongest in legal and maths. Larger embedding models yield higher absolute scores but remain sensitive to suboptimal segmentation, indicating that better chunking and large embeddings provide complementary benefits. In addition to accuracy gains, we quantify the efficiency trade-offs of advanced chunking. Producing more, smaller chunks can increase index size and latency. Consequently, we identify methods (like dynamic chunking) that approach an optimal balance of effectiveness and efficiency. These findings establish chunking as a vital lever for improving retrieval performance and reliability.

We present Elenchus, a dialogue system for knowledge base construction grounded in inferentialist semantics, where knowledge engineering is re-conceived as explicitation rather than extraction from expert testimony or textual content. A human expert develops a bilateral position (commitments and denials) about a topic through prover-skeptic dialogue with a large language model (LLM) opponent. The LLM proposes tensions (claims that parts of the position are jointly incoherent) which the expert resolves by retraction, refinement, or contestation. The LLM thus serves as a defeasible derivability oracle whose unreliability is structurally contained by the expert's authority. Our main technical contribution is a mapping from Elenchus dialectical states to material bases in Hlobil and Brandom's NonMonotonic MultiSuccedent (NMMS) logic, satisfying Containment and enabling the elaboration of logical vocabulary that makes explicit the inferential relationships negotiated in the dialectic. We demonstrate the approach on the W3C PROV-O provenance ontology, where a single dialogue session elicits and structures design tensions that a domain expert can articulate, corresponding to decisions documented in a retrospective analysis of the ontology's design. Using pyNMMS, an automated NMMS reasoner, we verify that the structural properties of the resulting material base (nontransitivity, nonmonotonicity, and independence) correspond to specific PROV design rationales, demonstrating end-to-end integration from dialogue through formal reasoning.

Conditional Optimal Transport (COT) problem aims to find a transport map between conditional source and target distributions while minimizing the transport cost. Recently, these transport maps have been utilized in conditional generative modeling tasks to establish efficient mappings between the distributions. However, classical COT inherits a fundamental limitation of optimal transport, i.e., sensitivity to outliers, which arises from the hard distribution matching constraints. This limitation becomes more pronounced in a conditional setting, where each conditional distribution is estimated from a limited subset of data. To address this, we introduce the Conditional Unbalanced Optimal Transport (CUOT) framework, which relaxes conditional distribution-matching constraints through Csiszár divergence penalties while strictly preserving the conditioning marginals. We establish a rigorous formulation of the CUOT problem and derive its dual and semi-dual formulations. Based on the semi-dual form, we propose Conditional Unbalanced Optimal Transport Maps (CUOTM), an outlier-robust conditional generative model built upon a triangular $c$-transform parameterization. We theoretically justify the validity of this parameterization by proving that the optimal triangular map satisfies the $c$-transform relationships. Our experiments on 2D synthetic and image-scale datasets demonstrate that CUOTM achieves superior outlier robustness and competitive distribution-matching performance compared to existing COT-based baselines, while maintaining high sampling efficiency.

Extreme weather events and cyberattacks can cause component failures and disrupt the operation of power distribution networks (DNs), during which reconfiguration and load shedding are often adopted for resilience enhancement. This study introduces a topology-aware graph reinforcement learning (RL) framework for outage management that embeds higher-order topological features of the DN into a graph-based RL model, enabling reconfiguration and load shedding to maximize energy supply while maintaining operational stability. Results on the modified IEEE 123-bus feeder across 300 diverse outage scenarios demonstrate that incorporating the topological data analysis (TDA) tool, persistence homology (PH), yields 9-18% higher cumulative rewards, up to 6% increase in power delivery, and 6-8% fewer voltage violations compared to a baseline graph-RL model. These findings highlight the potential of integrating RL with TDA to enable self-healing in DNs, facilitating fast, adaptive, and automated restoration.

In practical data-driven applications on electrical equipment fault diagnosis, training data can be poisoned by sensor failures, which can severely degrade the performance of machine learning (ML) models. However, once the ML model has been trained, removing the influence of such harmful data is challenging, as full retraining is both computationally intensive and time-consuming. To address this challenge, this paper proposes a SISA (Sharded, Isolated, Sliced, and Aggregated)-based machine unlearning (MU) framework for power transformer inter-turn short-circuit fault (ITSCF) localization. The SISA method partitions the training data into shards and slices, ensuring that the influence of each data point is isolated within specific constituent models through independent training. When poisoned data are detected, only the affected shards are retrained, avoiding retraining the entire model from scratch. Experiments on simulated ITSCF conditions demonstrate that the proposed framework achieves almost identical diagnostic accuracy to full retraining, while reducing retraining time significantly.

Quadruped locomotion provides a natural setting for understanding when model-free learning can outperform model-based control design, by exploiting data patterns to bypass the difficulty of optimizing over discrete contacts and the combinatorial explosion of mode changes. We give a principled analysis of why imitation learning with quadrupeds can be inherently effective in a small data regime, based on the structure of its limit cycles, Poincaré return maps, and local numerical properties of neural networks. The understanding motivates a new imitation learning method that regulates the alignment between variations in a latent space and those over the output actions. Hardware experiments confirm that a few seconds of demonstration is sufficient to train various locomotion policies from scratch entirely offline with reasonable robustness.

Accurate chart comprehension represents a critical challenge in advancing multimodal learning systems, as extensive information is compressed into structured visual representations. However, existing vision-language models (VLMs) frequently struggle to generalize on unseen charts because it requires abstract, symbolic, and quantitative reasoning over structured visual representations. In this work, we introduce Chart-RL, an effective reinforcement learning (RL) method that employs mathematically verifiable rewards to enhance chart question answering in VLMs. Our experiments demonstrate that Chart-RL consistently outperforms supervised fine-tuning (SFT) across different chart understanding benchmarks, achieving relative improvements of 16.7% on MutlChartQA, and 11.5% on ChartInsights. We conduct robustness analysis, where Chart-RL achieves enhanced performance in 18 of 25 perturbed chart categories, demonstrating strong consistency and reasoning capability across visual variations. Furthermore, we demonstrate that task difficulty and inherent complexity are more critical than data quantity in RL training. For instance, Chart-RL trained on merely 10 complex chart-query examples significantly outperforms models trained on over 6,000 simple examples. Additionally, training on challenging reasoning tasks not only improves in-domain generalization relative to simpler tasks, but also facilitate strong transfer to out-of-domain visual mathematical problems.

We study post-training linear autoregressive models with outcome and process rewards. Given a context $\boldsymbol{x}$, the model must predict the response $\boldsymbol{y} \in Y^N$, a sequence of length $N$ that satisfies a $γ$ margin condition, an extension of the standard separability to sequences. We prove that on test samples where the base model achieves a non-trivial likelihood $α$, a variant of policy gradient (PG) can achieve likelihood $1 - \varepsilon$ with an essentially minimax optimal number of reward queries $\tilde{O}((α^{-1} + \varepsilon^{-1})/γ^2)$. However, a barrier arises for going beyond the support of the base model. We prove that the overall expected error after post-training with outcome rewards is governed by a property of the base model called the Likelihood Quantile (LQ), and that variants of PG, while minimax optimal, may require a number of reward queries exponential in $N$ to go beyond this support, regardless of the pre-training algorithm. To overcome this barrier, we study post-training with a process reward model, and demonstrate how PG variants in this setting avoid the curse of dimensionality in $N$ via dependence on a token-level LQ. Along the way, we prove that under the margin condition, SGD with adaptive learning rate (LR) achieves a near optimal test error for statistical learning, and PG with adaptive LR achieves a near optimal number of mistakes for online learning while being computationally efficient whenever possible, both of which may be of independent interest.

As graphs scale to billions of nodes and edges, graph Machine Learning workloads are constrained by the cost of multi-hop traversals over exponentially growing neighborhoods. While various system-level and algorithmic optimizations have been proposed to accelerate Graph Neural Network (GNN) pipelines, data management and movement remain the primary bottlenecks at scale. In this paper, we explore whether graph sparsification, a well-established technique that reduces edges to create sparser neighborhoods, can serve as a lightweight pre-processing step to address these bottlenecks while preserving accuracy on node classification tasks. We develop an extensible experimental framework that enables systematic evaluation of how different sparsification methods affect the performance and accuracy of GNN models. We conduct the first comprehensive study of GNN training and inference on sparsified graphs, revealing several key findings. First, sparsification often preserves or even improves predictive performance. As an example, random sparsification raises the accuracy of the GAT model by 6.8% on the PubMed graph. Second, benefits increase with scale, substantially accelerating both training and inference. Our results show that the K-Neighbor sparsifier improves model serving performance on the Products graph by 11.7x with only a 0.7% accuracy drop. Importantly, we find that the computational overhead of sparsification is quickly amortized, making it practical for very large graphs.

Memory disaggregation via Compute Express Link (CXL) enables multiple hosts to share remote memory, improving utilization for data-intensive workloads. Today, virtual memory enables process-level isolation on a host and CXL enables host-level isolation. This creates a critical security gap: the absence of process-level memory isolation in shared disaggregated memory. We present Space-Control, a hardware-software co-design that provides fine-grained, process-level isolation for shared disaggregated memory. Space-Control authenticates execution context in the hardware and enforces access control on every memory access and amortizes lookup times with a small cache. Our design allows up to 127 processes Simulation Toolkit (SST) based CXL model, Space-Control incurs minimal performance overhead of 3.3%, making shared disaggregated memory isolation practical.

DNA foundation models have become transformative tools in bioinformatics and healthcare applications. Trained on vast genomic datasets, these models can be used to generate sequence embeddings, dense vector representations that capture complex genomic information. These embeddings are increasingly being shared via Embeddings-as-a-Service (EaaS) frameworks to facilitate downstream tasks, while supposedly protecting the privacy of the underlying raw sequences. However, as this practice becomes more prevalent, the security of these representations is being called into question. This study evaluates the resilience of DNA foundation models to model inversion attacks, whereby adversaries attempt to reconstruct sensitive training data from model outputs. In our study, the model's output for reconstructing the DNA sequence is a zero-shot embedding, which is then fed to a decoder. We evaluated the privacy of three DNA foundation models: DNABERT-2, Evo 2, and Nucleotide Transformer v2 (NTv2). Our results show that per-token embeddings allow near-perfect sequence reconstruction across all models. For mean-pooled embeddings, reconstruction quality degrades as sequence length increases, though it remains substantially above random baselines. Evo 2 and NTv2 prove to be most vulnerable, especially for shorter sequences with reconstruction similarities > 90%, while DNABERT-2's BPE tokenization provides the greatest resilience. We found that the correlation between embedding similarity and sequence similarity was a key predictor of reconstruction success. Our findings emphasize the urgent need for privacy-aware design in genomic foundation models prior to their widespread deployment in EaaS settings. Training code, model weights and evaluation pipeline are released on: https://github.com/not-a-feature/DNA-Embedding-Inversion.

Many distributional quantities in reinforcement learning are intrinsically joint across actions, including distributions of gaps and probabilities of superiority. However, the classical Markov decision process (MDP) formalism specifies only marginal laws and leaves the joint law of counterfactual one-step outcomes across multiple possible actions at a state unspecified. We study coupled-dynamics environments with a multi-action generative interface which can sample counterfactual one-step outcomes for multiple actions under shared exogenous randomness. We propose joint MDPs (JMDPs) as a formalism for such environments by augmenting an MDP with a multi-action sample transition model which specifies a coupling of one-step counterfactual outcomes, while preserving standard MDP interaction as marginal observations. We adopt and formalize a one-step coupling regime where dependence across actions is confined to immediate counterfactual outcomes at the queried state. In this regime, we derive Bellman operators for $n$th-order return moments, providing dynamic programming and incremental algorithms with convergence guarantees.

Recent advances have made long-form report-generating systems widely available. This has prompted evaluation frameworks that use LLM-as-judge protocols and claim verification, along with meta-evaluation frameworks that seek to validate these methods. Many of the meta-evaluations estimate an evaluation quality's by comparing its assessments against human pairwise preferences. Prior work, however, suggests that human pairwise preference may be overly simplistic and can fail to capture nuances of expert expectations. We conduct a case study in meta-evaluation for long-form QA benchmarks using ScholarQA-CS2, a benchmark designed for assessing retrieval-augmented deep-research QA in the scientific domain. We comprehensively validate the benchmark through human pairwise preference judgments, then critically examine the strengths, weaknesses, and confounders of this approach. We show that pairwise preference rankings are best suited for system-level evaluation, while explicit metric-wise annotations and expert annotators are critical for reliable metric-level assessment, with subjectivity remaining a key challenge. Based on our findings, we offer practical guidelines for designing future meta-evaluations that better align evaluation methods, annotator expertise, and reporting practices. By surfacing these methodological challenges, we aim to advance evaluation standards for deep-research systems.

In this work, we study the mechanical behavior of solids with microstructure using the framework of Cosserat elasticity with a single unit director. This formulation captures the coupling between deformation and orientational fields that arises in many structured materials. To compute equilibrium configurations of such media, we develop two complementary computational approaches: a finite element formulation based on variational principles and a neural network-based solver that directly minimizes the total potential energy. The neural architecture is constructed to respect the fundamental kinematic structure of the theory. In particular, it enforces frame invariance of the energy, satisfies the unit-length constraint on the director field, and represents deformation and director fields through separate networks to preserve their kinematic independence in the variational setting. Beyond satisfying balance laws, however, physically admissible solutions must also correspond to stable energy minimizers. To assess this requirement, we derive the quasiconvexity condition, rank-one convexity condition, and the Legendre-Hadamard inequalities for the Cosserat model and formulate them in a manner suitable for evaluating neural network predictions. These necessary stability conditions provide a physics-based validation framework: network outputs that violate these necessary conditions cannot correspond to stable energy minimizers and can therefore be rejected. In this way, we integrate classical variational stability theory with modern machine-learning solvers, establishing a computational workflow in which equilibrium solutions are not only learned but also assessed for energetic consistency.

Mixture-of-experts (MoE) is a common approach for increasing parameter capacity, but applying MoE to state space model (SSM) token mixers can multiply the cost of the recurrent state update. We study how to introduce expert specialization into selective SSMs while preserving computational efficiency. We show that MoE--SSM can refer to two designs: (1) MoE over separated SSMs, which maintains multiple state trajectories and thus scales compute with the number of experts; and (2) MoE-parameterized SSM, which mixes experts in parameter space, maintains a single state trajectory, and evaluates the recurrence once. Our method, Switch Mamba (Swimba), follows the second design by routing over expert-produced SSM streams. Theoretically, we establish well-definedness and stability for MoE-parameterized SSMs and characterize the relationship between the two designs. Empirically, we evaluate Swimba on standard benchmark tasks and measure real-time throughput and latency. Under matched FLOPs, Swimba achieves slightly better average performance than the baseline, with a small slowdown in real-time latency and throughput. Overall, these results suggest that parameter-space MoE can increase SSM capacity while keeping the dominant recurrence cost fixed.

Mindfulness meditation is a widely accessible and evidence-based method for supporting mental health. Despite the proliferation of mindfulness meditation apps, sustaining user engagement remains a persistent challenge. Personalizing the meditation experience is a promising strategy to improve engagement, but it often requires costly and unscalable manual effort. We present MindfulAgents, a multi-agent system powered by large language models that (1) generates guided meditation scripts based on an expert-established mindfulness framework, (2) encourages users' reflection on emotional states and mindfulness skills, and (3) enables real-time personalization of the mindfulness meditation experience for each user. In a formative lab study (N=13), MindfulAgents significantly improved in-session engagement (p = 0.011) and self-awareness (p = 0.014), and reduced momentary stress (p = 0.020). Furthermore, a four-week deployment study (N=62) demonstrated a notable increase in long-term engagement (p = 0.002) and level of mindfulness (p = 0.023). Participants reported that MindfulAgents offered more relevant meditation sessions personalized to individual needs in various contexts, supporting sustained practice. Our findings highlight the potential of LLM-driven personalization for enhancing user engagement in digital mindfulness meditation interventions.

Large language models (LLMs) often exhibit flawed reasoning ability that undermines reliability. Existing approaches to improving reasoning typically treat it as a general and monolithic skill, applying broad training which is inefficient and unable to target specific reasoning errors. We introduce Reasoning Editing, a paradigm for selectively modifying specific reasoning patterns in LLMs while preserving other reasoning pathways. This task presents a fundamental trade-off between Generality, the ability of an edit to generalize across different tasks sharing the same reasoning pattern, and Locality, the ability to preserve other reasoning capabilities. Through systematic investigation, we uncover the Circuit-Interference Law: Edit interference between reasoning patterns is proportional to the overlap of their neural circuits. Guided by this principle, we propose REdit, the first framework to actively reshape neural circuits before editing, thereby modulating interference between reasoning patterns and mitigating the trade-off. REdit integrates three components: (i) Contrastive Circuit Reshaping, which directly addresses the generality-locality trade-off by disentangling overlapping circuits; (ii) Meta-Contrastive Learning, which extends transferability to novel reasoning patterns; and (iii) Dual-Level Protection, which preserves preexisting abilities by constraining reshaping update directions and regularizing task-level predictions. Extensive experiments with Qwen-2.5-3B on propositional logic reasoning tasks across three difficulty levels demonstrate that REdit consistently achieves superior generality and locality compared to baselines, with additional validation in mathematics showing broader potential. Our code is available at https://github.com/LzyFischer/REdit.

We introduce NerVE, a unified eigenspectral framework for understanding how feed-forward networks (FFNs) in large language models (LLMs) organize and regulate information flow in high-dimensional latent space. Despite FFNs dominating the parameter budget, their high-dimensional dynamics remain poorly understood. NerVE addresses this gap through lightweight, memory-efficient tracking of eigenspectrum dynamics via four complementary metrics: Spectral Entropy (dispersion), Participation Ratio (effective dimensionality), Eigenvalue Early Enrichment (top-heaviness), and Jensen-Shannon divergence (distributional shifts). Our key insight is that FFN nonlinearities reinject variance across eigenmodes, fundamentally governing latent dimension utilization, and that optimizer geometry strongly modulates the extent of this variance reinjection. We validate NerVE across model scales, and diverse architectural and optimizer configurations, each uniquely shaping FFN dynamics: normalization schemes controlling variance flow; FFN weight geometries constraining latent space; positional encoding and activation functions regulating information flow; and optimizer choices redistributing effective capacity across depth. Across these settings, NerVE consistently recovers stable spectral signatures that correlate with model's generalization ability and respond predictably to design choices, generalizing beyond transformer to MLP-Mixer architectures, providing actionable insights for architectural and optimizer choices beyond trial-and-error.

Safe navigation of autonomous robots remains one of the core challenges in the field, especially in dynamic and uncertain environments. One of the prevalent approaches is safety filtering based on control barrier functions (CBFs), which are easy to deploy but difficult to design. Motivated by the shortcomings of existing learning- and model-based methods, we propose a simple yet effective neural CBF design method for safe robot navigation in dynamic environments. We employ the idea of a composite CBF, where multiple neural CBFs are combined into a single CBF. The individual CBFs are trained via the Hamilton-Jacobi reachability framework to approximate the optimal safe set for single moving obstacles. Additionally, we use the residual neural architecture, which guarantees that the estimated safe set does not intersect with the corresponding failure set. The method is extensively evaluated in simulation experiments for a ground robot and a quadrotor, comparing it against several baseline methods. The results show improved success rates of up to 18\% compared to the best baseline, without increasing the conservativeness of the motion. Also, the method is demonstrated in hardware experiments for both types of robots.

The extraction of structured information from raw text is a fundamental component of many NLP applications, including document retrieval, ranking, and relevance estimation. High-quality extractions often require domain-specific accuracy, up-to-date understanding of specialized taxonomies, and the ability to incorporate emerging jargon and rare outliers. In many domains--such as medical, legal, and HR--the extraction model must also adapt to shifting terminology and benefit from explicit reasoning over structured knowledge. We propose DySECT, a Dynamic Self-Evolving Extraction and Curation Toolkit, which continually improves as it is used. The system incrementally populates a versatile, self-expanding knowledge base (KB) with triples extracted by the LLM. The KB further enriches itself through the integration of probabilistic knowledge and graph-based reasoning, gradually accumulating domain concepts and relationships. The enriched KB then feeds back into the LLM extractor via prompt tuning, sampling of relevant few-shot examples, or fine-tuning using KB-derived synthetic data. As a result, the system forms a symbiotic closed-loop cycle in which extraction continuously improves knowledge, and knowledge continuously improves extraction.

This study investigates whether large language models (LLMs) exhibit cross-linguistic differences in mental health evaluations. Focusing on Chinese and English, we examine two widely used models, GPT-4o and Qwen3, to assess whether prompt language systematically shifts mental health-related evaluations and downstream decision outcomes. First, we assess models' evaluative orientation toward mental health stigma using multiple validated measurement scales capturing social stigma, self-stigma, and professional stigma. Across all measures, both models produce higher stigma-related responses when prompted in Chinese than in English. Second, we examine whether these differences also manifest in two common downstream decision tasks in mental health. In a binary mental health stigma detection task, sensitivity to stigmatizing content varies across language prompts, with lower sensitivity observed under Chinese prompts. In a depression severity classification task, predicted severity also differs by prompt language, with Chinese prompts associated with more underestimation errors, indicating a systematic downward shift in predicted severity relative to English prompts. Together, these findings suggest that language context can systematically shape evaluative patterns in LLM outputs and shift decision thresholds in downstream tasks.

Instruction tuning has become essential for adapting large language models (LLMs) to follow domain-specific prompts. Yet, in specialized fields such as medicine, the scarcity of high-quality French instruction data limits effective supervision. To address this gap, we introduce MedInjection-FR, a large-scale French biomedical instruction dataset comprising 571K instruction-response pairs drawn from three complementary sources: native, synthetic, and translated data. We design a controlled experimental framework to systematically assess how data provenance affects instruction tuning, using Qwen-4B-Instruct fine-tuned across seven configurations combining these sources. Results show that native data yield the strongest performance, while mixed setups, particularly native and translated, provide complementary benefits. Synthetic data alone remains less effective but contributes positively when balanced with native supervision. Evaluation on open-ended QA combines automatic metrics, LLM-as-a-judge assessment, and human expert review; although LLM-based judgments correlate best with human ratings, they show sensitivity to verbosity. These findings highlight that data authenticity and diversity jointly shape downstream adaptation and that heterogeneous supervision can mitigate the scarcity of native French medical instructions.

Tree ensembles such as XGBoost are often preferred for discriminative tasks in mixed-type tabular data, due to their inductive biases, minimal hyperparameter tuning, and training efficiency. We argue that these qualities, when leveraged correctly, can make for better generative models as well. As such, we present XGenBoost, a pair of generative models based on XGBoost: i) a Denoising Diffusion Implicit Model (DDIM) with XGBoost as score-estimator suited for smaller datasets, and ii) a hierarchical autoregressive model whose conditionals are learned via XGBoost classifiers, suited for large-scale tabular synthesis. The architectures follow from the natural constraints imposed by tree-based learners, e.g., in the diffusion model, combining Gaussian and multinomial diffusion to leverage native categorical splits and avoid one-hot encoding while accurately modelling mixed data types. In the autoregressive model, we use a fixed-order factorization, a hierarchical classifier to impose ordinal inductive biases when modelling numerical features, and de-quantization based on empirical quantile functions to model the non-continuous nature of most real-world tabular datasets. Through two benchmarks, one containing smaller and the other larger datasets, we show that our proposed architectures outperform previous neural- and tree-based generative models for mixed-type tabular synthesis at lower training cost.

While Large Language Models demonstrate immense potential as proactive Medical Agents, their real-world deployment is severely bottlenecked by data scarcity under privacy constraints. To overcome this, we propose State-Enhanced Logical-Skill Memory (SELSM), a training-free framework that distills simulated clinical trajectories into entity-agnostic operational rules within an abstract skill space. During inference, a Query-Anchored Two-Stage Retrieval mechanism dynamically fetches these entity-agnostic logical priors to guide the agent's step-by-step reasoning, effectively resolving the state polysemy problem. Evaluated on MedAgentBench -- the only authoritative high-fidelity virtual EHR sandbox benchmarked with real clinical data -- SELSM substantially elevates the zero-shot capabilities of locally deployable foundation models (30B--32B parameters). Notably, on the Qwen3-30B-A3B backbone, our framework completely eliminates task chain breakdowns to achieve a 100\% completion rate, boosting the overall success rate by an absolute 22.67\% and significantly outperforming existing memory-augmented baselines. This study demonstrates that equipping models with a dynamically updatable, state-enhanced cognitive scaffold is a privacy-preserving and computationally efficient pathway for local adaptation of AI agents to clinical information systems. While currently validated on FHIR-based EHR interactions as an initial step, the entity-agnostic design of SELSM provides a principled foundation toward broader clinical deployment.

As machine learning (ML) systems increasingly shape access to credit, jobs, and other opportunities, the fairness of algorithmic decisions has become a central concern. Yet it remains unclear when enforcing fairness constraints in these systems genuinely improves outcomes for affected groups or instead leads to "leveling down," making one or both groups worse off. We address this question in a unified, population-level (Bayes) framework for binary classification under prevalent group fairness notions. Our Bayes approach is distribution-free and algorithm-agnostic, isolating the intrinsic effect of fairness requirements from finite-sample noise and from training and intervention specifics. We analyze two deployment regimes for ML classifiers under common legal and governance constraints: attribute-aware decision-making (sensitive attributes available at decision time) and attribute-blind decision-making (sensitive attributes excluded from prediction). We show that, in the attribute-aware regime, fair ML necessarily (weakly) improves outcomes for the disadvantaged group and (weakly) worsens outcomes for the advantaged group. In contrast, in the attribute-blind regime, the impact of fairness is distribution-dependent: fairness can benefit or harm either group and may shift both groups' outcomes in the same direction, leading to either leveling up or leveling down. We characterize the conditions under which these patterns arise and highlight the role of "masked" candidates in driving them. Overall, our results provide structural guidance on when pursuing algorithmic fairness is likely to improve group outcomes and when it risks systemic leveling down, informing fair ML design and deployment choices.

Self-Sovereign Digital Identity (SSDI) enables individuals to control their own identity assertions and data, rather than relying on centralized or federated systems prone to large-scale data breaches. By eliminating centralized databases maintained by service providers and identity brokers, SSDIs offer enhanced security and privacy. However, adoption remains slow, and research in this area lacks systematization and uniformity. To address these gaps, we present a comprehensive systematization of knowledge on self-sovereign digital identities, with a primary focus on identifying the challenges that impede real-world adoption. We survey 80 academic and non-academic sources and identify six major challenges: (i) binding a single identity to one individual or organization, (ii) the absence of mature cryptographic and communication protocols, (iii) significant usability barriers, (iv) regulatory and oversight gaps, (v) bootstrapping to critical-mass adoption, and (vi) dependence on a permissionless, decentralized, yet singular infrastructure that may expose unforeseen vulnerabilities over time. We then analyze 47 scientific publications and find that the vast majority focus on blockchain-based solutions rather than generalized SSDI architectures. Additionally, we catalog 12 real-world, production-grade SSDI applications. Our evaluation of these solutions reveals that self-sovereignty is, in practice, a spectrum rather than a binary property. Finally, we explore the frontiers of SSDI by identifying major trends, open problems, and opportunities for future research. We hope this systematization will help advance the shift from centralized to self-sovereign digital identities in a disciplined and impactful way.

Large Language Models (LLMs) have demonstrated impressive capabilities in a wide range of code generation tasks. However, generating code for certain domains remains challenging. One such domain is Computer-Aided Design (CAD) program, where the goal is to produce scripted parametric models that define object geometry for precise design and manufacturing applications. A key challenge in LLM-based CAD program generation is the limited geometric complexity of generated shapes compared to those found in real-world industrial designs. This shortfall is in part due to the lack of diversity in the available CAD program training data. To address this, we propose a novel data augmentation paradigm that prompts an LLM to generate CAD programs conditioned on a reference surface program and a modeling procedure - an idea inspired by practices in industrial design. By varying the reference surface using a collection of organic shapes, our method enriches the geometric distribution of generated CAD models. In particular, it introduces edges and faces defined by spline-based curvature, which are typically missing or underrepresented in existing open-source CAD program datasets. Experiments show that our method produces CAD samples with significantly greater geometric diversity and a higher resemblance to industry-grade CAD designs in terms of the proportion of organic shape primitives. This enhancement makes our CAD data augmentation approach a useful tool for training LLMs and other deep learning models in CAD generation.

Streaming clustering is a domain that has become extremely relevant in the age of big data, such as in network traffic analysis or in processing continuously-running sensor data. Furthermore, possibilistic models offer unique benefits over approaches from the literature, especially with the introduction of a "fuzzifier" parameter that controls how quickly typicality degrades as one gets further from cluster centers. We propose a single-pass possibilistic clustering (SPC) algorithm that is effective and easy to apply to new datasets. Key contributions of SPC include the ability to model non-spherical clusters, closed-form footprint updates over arbitrarily sized damped windows, and the employment of covariance union from the multiple hypothesis tracking literature to merge two cluster mean and covariance estimates. SPC is validated against five other streaming clustering algorithm on the basis of cluster purity and normalized mutual information.

Coronary Artery Disease (CAD) remains a leading cause of morbidity and mortality worldwide. Early detection is critical to recover patient outcomes and decrease healthcare costs. In recent years, machine learning (ML) advancements have shown significant potential in enhancing the accuracy of CAD diagnosis. This study investigates the application of ML algorithms to improve the detection of CAD by analyzing patient data, including clinical features, imaging, and biomarker profiles. Bi-directional Long Short-Term Memory (Bi-LSTM), Gated Recurrent Units (GRU), and a hybrid of Bi-LSTM+GRU were trained on large datasets to predict the presence of CAD. Results demonstrated that these ML models outperformed traditional diagnostic methods in sensitivity and specificity, offering a robust tool for clinicians to make more informed decisions. The experimental results show that the hybrid model achieved an accuracy of 97.07%. By integrating advanced data preprocessing techniques and feature selection, this study ensures optimal learning and model performance, setting a benchmark for the application of ML in CAD diagnosis. The integration of ML into CAD detection presents a promising avenue for personalized healthcare and could play a pivotal role in the future of cardiovascular disease management.

The agentic web marks a structural transition from a human-centered information network to a digital environment populated by artificial intelligence (AI) agents that perceive, decide, and act autonomously. As delegated action unfolds at machine speed, exceeds discrete moments of human judgment, and distributes decision-making across non-human actors, existing legal frameworks face growing strain, creating an urgent need for new mechanisms capable of sustaining legality in this emerging order. A trustworthy agentic web therefore depends on the infrastructuring of legality through interoperable protocols that organize identity, delegation, and accountability across systems, enabling coherent governance beyond isolated platforms. Towards this end, this article advances a distributed legal infrastructure (DLI), a governance paradigm composed of five interlocking layers: (1) self-sovereign, soulbound agent identities; (2) cognitive AI logic and constraint systems; (3) decentralized adjudication mechanisms for dispute resolution; (4) bottom-up agentic market regulation to mitigate information asymmetries and network effects, including insurance-based models; and (5) portable institutional frameworks that enable legal interoperability while preserving plural sources of authority. This reference framework contributes to emerging research on embedding legality within agentic web infrastructure, aligning distributed technical systems with accountability, contestability, and rule-of-law principles.

Ferroelectric field-effect transistors (FeFET)-based vertical NAND (Fe-VNAND) has emerged as a promising candidate to overcome z-scaling limitations with lower programming voltages. However, the data retention of 3D Fe-VNAND is hindered by the complex interaction between charge detrapping and ferroelectric depolarization. Developing optimized device designs requires exploring an extensive parameter space, but the high computational cost of conventional Technology Computer-Aided Design (TCAD) tools makes such wide-scale optimization impractical. To overcome these simulation barriers, we present a Physics-Informed Neural Operator (PINO)-based AI surrogate model designed for high-efficiency prediction of threshold voltage (Vth) shifts and retention behavior. By embedding fundamental physical principles into the learning architecture, our PINO framework achieves a speedup exceeding 10000x compared to TCAD while maintaining physical accuracy. This study demonstrates the model's effectiveness on a single FeFET configuration, serving as a pathway toward modeling the retention loss mechanisms.

Large language models (LLMs) have become common decision-support tools across educational and professional contexts, raising questions about how their outputs shape human critical thinking. Prior work suggests that the amount of AI assistance can influence cognitive engagement, yet little is known about how specific properties of LLM outputs (e.g., response length) impacts users' critical evaluation of information. In this study, we examine whether the length of LLM responses shapes users' accuracy in evaluating LLM-generated reasoning on critical thinking tasks, particularly in interaction with the correctness of the LLM's reasoning. To begin evaluating this, we conducted a within-subjects experiment with 24 participants who completed 15 modified Watson--Glaser critical thinking items, each accompanied by an LLM-generated explanation that varied in length and correctness. Mixed-effects logistic regression revealed a strong and statistically reliable effect of LLM output correctness on participant accuracy, with participants more likely to answer correctly when the LLM's explanation was correct. Response length appeared to moderated this effect: when the LLM output was incorrect, medium-length explanations were associated with higher participant accuracy than either shorter or longer explanations, whereas accuracy remained high across lengths when the LLM output was correct. Together, these findings suggest that response length alone may be insufficient to support critical thinking, and that how reasoning is presented-including a potential advantage of mid-length explanations under some conditions-points to design opportunities for LLM-based decision-support systems that emphasize transparent reasoning and calibrated expressions of certainty.

Attention heads retrieve: given a query, they return a softmax-weighted average of stored values. We show that this computation is one step of gradient descent on a classical energy function, and that Langevin sampling from the corresponding distribution yields \emph{stochastic attention}: a training-free sampler controlled by a single temperature. Lowering the temperature gives exact retrieval; raising it gives open-ended generation. Because the energy gradient equals the attention map, no score network, training loop, or learned model is required. We validate on four domains (64 to 4,096 dimensions). At generation temperature, stochastic attention is 2.6 times more novel and 2.0 times more diverse than the best learned baseline (a variational autoencoder trained on the same patterns), while matching a Metropolis-corrected gold standard. A simple signal-to-noise rule selects the operating temperature for any dimension. The approach requires no architectural changes and extends naturally to retrieval-augmented generation and in-context learning.

Large Language Models (LLMs) exhibit impressive general-purpose capabilities but also introduce serious safety risks, particularly the potential for deception as models acquire increased agency and human oversight diminishes. In this work, we present LieCraft: a novel evaluation framework and sandbox for measuring LLM deception that addresses key limitations of prior game-based evaluations. At its core, LieCraft is a novel multiplayer hidden-role game in which players select an ethical alignment and execute strategies over a long time-horizon to accomplish missions. Cooperators work together to solve event challenges and expose bad actors, while Defectors evade suspicion while secretly sabotaging missions. To enable real-world relevance, we develop 10 grounded scenarios such as childcare, hospital resource allocation, and loan underwriting that recontextualize the underlying mechanics in ethically significant, high-stakes domains. We ensure balanced gameplay in LieCraft through careful design of game mechanics and reward structures that incentivize meaningful strategic choices while eliminating degenerate strategies. Beyond the framework itself, we report results from 12 state-of-the-art LLMs across three behavioral axes: propensity to defect, deception skill, and accusation accuracy. Our findings reveal that despite differences in competence and overall alignment, all models are willing to act unethically, conceal their intentions, and outright lie to pursue their goals.

Long-horizon execution in Large Language Models (LLMs) remains unstable even when high-level strategies are provided. Evaluating on controlled algorithmic puzzles, we demonstrate that while decomposition is essential for stability, extreme decomposition creates a "no-recovery bottleneck". We show that this bottleneck becomes critical due to highly non-uniform error distribution, where consistent errors on a few "hard" steps become irreversible. To address this, we propose Lookahead-Enhanced Atomic Decomposition (LEAD). By incorporating short-horizon future validation and aggregating overlapping rollouts, LEAD provides enough isolation to maintain stability while retaining enough local context to correct errors. This enables the o4-mini model to solve Checkers Jumping up to complexity $n=13$, whereas extreme decomposition fails beyond $n=11$.

Discovering compact governing equations from experimental observations is one of the defining objectives of quantitative science, yet practical discovery pipelines routinely fail when measurements are noisy, relevant state variables are unobserved, or multiple symbolic structures explain the data equally well within statistical uncertainty. Here we introduce SymLang (Symmetry-constrained Language-guided equation discovery), a unified framework that brings together three previously separate ideas: (i) typed symmetry-constrained grammars that encode dimensional analysis, group-theoretic invariance, and parity constraints as hard production rules, eliminating on average 71.3% of candidate expression trees before any fitting; (ii) language-model-guided program synthesis in which a fine-tuned 7B-parameter proposer, conditioned on interpretable data descriptors, efficiently navigates the constrained search space; and (iii) MDL-regularized Bayesian model selection coupled with block-bootstrap stability analysis that quantifies structural uncertainty rather than committing to a single best equation. Across 133 dynamical systems spanning classical mechanics, electrodynamics, thermodynamics, population dynamics, and nonlinear oscillators, SymLang achieves an exact structural recovery rate of 83.7% under 10% observational noise - a 22.4 percentage-point improvement over the next-best baseline - while reducing out-of-distribution extrapolation error by 61% and near-eliminating conservation-law violations (3.1 x 10-3 vs. 187.3 x 10-3 physical drift for the closest competitor). In all tested regimes the framework correctly identifies structural degeneracy, reporting it explicitly rather than returning a confidently wrong single equation. The framework is fully open-source and reproducible, providing a principled pathway from raw data to interpretable, physically auditable symbolic laws.

Human annotation remains the foundation of reliable and interpretable data in Natural Language Processing (NLP). As annotation and evaluation tasks continue to expand, from categorical labelling to segmentation, subjective judgment, and continuous rating, measuring agreement between annotators has become increasingly more complex. This paper outlines how inter-annotator agreement (IAA) has been conceptualised and applied across NLP and related disciplines, describing the assumptions and limitations of common approaches. We organise agreement measures by task type and discuss how factors such as label imbalance and missing data influence reliability estimates. In addition, we highlight best practices for clear and transparent reporting, including the use of confidence intervals and the analysis of disagreement patterns. The paper aims to serve as a guide for selecting and interpreting agreement measures, promoting more consistent and reproducible human annotation and evaluation in NLP.

Trajectory prediction for flying objects is critical in domains ranging from sports analytics to aerospace. However, traditional methods struggle with complex physical modeling, computational inefficiencies, and high hardware demands, often neglecting critical trajectory events like landing points. This paper introduces a novel, hardware-efficient trajectory prediction framework that integrates environmental priors with a Dual-Transformer-Cascaded (DTC) architecture. We demonstrate this approach by predicting the landing points of tennis balls in real-world outdoor courts. Using a single industrial camera and YOLO-based detection, we extract high-speed flight coordinates. These coordinates, fused with structural environmental priors (e.g., court boundaries), form a comprehensive dataset fed into our proposed DTC model. A first-level Transformer classifies the trajectory, while a second-level Transformer synthesizes these features to precisely predict the landing point. Extensive ablation and comparative experiments demonstrate that integrating environmental priors within the DTC architecture significantly outperforms existing trajectory prediction frameworks

Artifact Evaluation (AE) is essential for ensuring the transparency and reliability of research, closing the gap between exploratory work and real-world deployment is particularly important in cybersecurity, particularly in IoT and CPSs, where large-scale, heterogeneous, and privacy-sensitive data meet safety-critical actuation. Yet, manual reproducibility checks are time-consuming and do not scale with growing submission volumes. In this work, we demonstrate that Large Language Models (LLMs) can provide powerful support for AE tasks: (i) text-based reproducibility rating, (ii) autonomous sandboxed execution environment preparation, and (iii) assessment of methodological pitfalls. Our reproducibility-assessment toolkit yields an accuracy of over 72% and autonomously sets up execution environments for 28% of runnable cybersecurity artifacts. Our automated pitfall assessment detects seven prevalent pitfalls with high accuracy ($F_1$ > 92%). Hence, the toolkit significantly reduces reviewer effort and, when integrated into established AE processes, could incentivize authors to submit higher-quality and more reproducible artifacts. IoT, CPS, and cybersecurity conferences and workshops may integrate the toolkit into their peer-review processes to support reviewers' decisions on awarding artifact badges, improving the overall sustainability of the process.

Activation functions are fundamental to deep neural networks, governing gradient flow, optimization stability, and representational capacity. Within historic deep architectures, while ReLU has been the dominant choice for the activation function, modern transformer-based models increasingly are adopting smoother alternatives such as GELU and other self-gated alternatives. Despite their empirical success, the mathematical relationships among these functions and the principles underlying their effectiveness remains only partially understood. We introduce IGLU, a parametric activation function derived as a scale mixture of GELU gates under a half-normal mixing distribution. This derivation yields a closed-form expression whose gating component is exactly the Cauchy CDF, providing a principled one-parameter family that continuously interpolates between identity-like and ReLU-like behavior via a single sharpness parameter $σ$. Unlike GELU's Gaussian gate, IGLU's heavy-tailed Cauchy gate decays polynomially in the negative tail, guaranteeing non-zero gradients for all finite inputs and offering greater robustness to vanishing gradients. We further introduce IGLU-Approx, a computationally efficient rational approximation of IGLU expressed entirely in terms of ReLU operations that eliminates transcendental function evaluation. Through evaluations on CIFAR-10, CIFAR-100, and WikiText-103 across ResNet-20, ViT-Tiny, and GPT-2 Small, IGLU achieves competitive or superior performance on both vision and language datasets against ReLU and GELU baselines, with IGLU-Approx recovering this performance at substantially reduced computational cost. In particular, we show that employing a heavy-tailed gate leads to considerable performance gains in heavily imbalanced classification datasets.

Cooperative multi-agent reinforcement learning (MARL) systems powered by large language models (LLMs) are frequently optimized via sparse terminal-only feedback. This shared signal entangles upstream decisions, obstructing accurate decision-level credit assignment. To address this trajectory-level diffusion, we introduce Contextual Counterfactual Credit Assignment (\textbf{\texttt{C3}}). Instead of distributing rewards across an entire episode, \textbf{\texttt{C3}} isolates the causal impact of individual messages by freezing the exact transcript-derived context, evaluating context-matched alternatives via fixed-continuation replay, and applying a leave-one-out (LOO) baseline. This localized intervention extracts unbiased, low-variance marginal advantages for standard policy-gradient optimization. Evaluated across five mathematical and coding benchmarks under matched budgets, \textbf{\texttt{C3}} improves terminal performance over established baselines. Mechanistic diagnostics further show that these gains are accompanied by higher credit fidelity, lower contextual variance, and stronger inter-agent causal dependence. Our code is available at https://github.com/EIT-EAST-Lab/C3.

Automated Vulnerability Repair (AVR) systems, especially those leveraging large language models (LLMs), have demonstrated promising results in patching vulnerabilities -- that is, if we trust their patch validation methodology. Ground-truth patches from human developers often come with new tests that not only ensure mitigation of the vulnerability but also encode extra semantics such as root cause location, optimal fix strategy, or subtle coding styles or conventions. And yet, none of the recent AVR systems verify that the auto-generated patches additionally pass these new tests (termed as $\text{PoC}^+$ tests). This is a subtle yet critical omission. To fill this gap, we constructed a benchmark, $\textrm{PVBench}$, with 209 cases spanning 20 projects. Each case includes basic tests (functional tests before the patch and the PoC exploit) as well as the associated $\text{PoC}^+$ tests. Evaluated on three state-of-the-art AVR systems, we find that over 40\% of patches validated as correct by basic tests fail under $\text{PoC}^+$ testing, revealing substantial overestimation on patch success rates. Analyzing these patches that are falsely labeled as correct, we suggest that AVR tools should improve in three critical areas: root cause analysis, adherence to program specifications, and capturing developer intention.

While large language models are capable diagnostic tools, the impact of multi-agent topology on diagnostic accuracy remains underexplored. This study evaluates four agent topologies, Control (single agent), Hierarchical, Adversarial, and Collaborative, across 302 cases spanning 33 rare disease categories. We introduce a Reasoning Gap metric to quantify the difference between internal knowledge retrieval and final diagnostic accuracy. Results indicate that the Hierarchical topology (50.0% accuracy) marginally outperforms Collaborative (49.8%) and Control (48.5%) configurations. In contrast, the Adversarial model significantly degrades performance (27.3%), exhibiting a massive Reasoning Gap where valid diagnoses were rejected due to artificial doubt. Across all architectures, performance was strongest in Allergic diseases and Toxic Effects categories but poorest in Cardiac Malformation and Respiratory cases. Critically, while the single-agent baseline was generally robust, all multi-agent systems, including the Adversarial model, yielded superior accuracy in Bone and Thoracic disease categories. These findings demonstrate that increasing system complexity does not guarantee better reasoning, supporting a shift toward dynamic topology selection.

Multimodal large language models can exhibit text dominance, over-relying on linguistic priors instead of grounding predictions in non-text inputs. One example is large audio-language models (LALMs) where decisive audio evidence can be under-utilized even when it contains important information. To address this issue we use mechanistic interpretability to identify a small set of audio-specialist attention heads whose audio attention yields a ``listening'' signal. We show that this signal increases when audio evidence affects the model's output, providing an indicator of audio engagement under standard prompting. Leveraging this localization, we construct an audio--silence steering direction and apply an inference-time activation intervention to the final representation, amplifying the model's audio effect. To demonstrate the utility of this intervention, we show on MMAU that this improves accuracy by up to +8.0 percentage points on two Qwen-based LALMs, without any parameter updates.

We study contextual bilateral trade under full feedback when trader valuations have bounded density but infinite variance. We first extend the self-bounding property of Bachoc et al. (ICML 2025) from bounded to real-valued valuations, showing that the expected regret of any price $π$ satisfies $\mathbb{E}[g(m,V,W) - g(π,V,W)] \le L|m-π|^2$ under bounded density alone. Combining this with truncated-mean estimation, we prove that an epoch-based algorithm achieves regret $\widetilde{O}(T^{1-2β(p-1)/(βp + d(p-1))})$ when the noise has finite $p$-th moment for $p \in (1,2)$ and the market value function is $β$-Hölder, and we establish a matching $Ω(\cdot)$ lower bound via Assouad's method with a smoothed moment-matching construction. Our results characterize the exact minimax rate for this problem, interpolating between the classical nonparametric rate at $p=2$ and the trivial linear rate as $p \to 1^+$.

Several successful strategies in automated reasoning rely on human-supplied guidance about which term or clause shapes are interesting. In this paper we aim to discover interesting term shapes automatically. Specifically, we discover abstractions : term patterns that occur over and over again in relevant proofs. We present our tool Twitch which discovers abstractions with the help of Stitch, a tool originally developed for discovering reusable library functions in program synthesis tasks. Twitch can produce abstractions in two ways: (1) from a partial, failed proof of a conjecture; (2) from successful proofs of other theorems in the same domain. We have also extended Twee, an equational theorem prover, to use these abstractions. We evaluate Twitch on a set of unit equality (UEQ) problems from TPTP, and show that it can prove 12 rating-1 problems as well as yielding significant speed-ups on many other problems.

Agentic AI systems combine large language model (LLM) reasoning with external tool invocation and long-horizon task execution. Although these systems are increasingly deployed in practice, their architectural composition introduces reliability challenges that differ from those in traditional software systems and standalone LLM applications. However, there is limited empirical understanding of how faults originate, manifest, and propagate in real-world agentic AI systems. To address this gap, we conduct a large-scale empirical study of faults in agentic AI systems. We collect 13,602 issues and pull requests from 40 open-source agentic AI repositories and apply stratified sampling to select 385 faults for in-depth qualitative analysis. Using grounded theory, we derive taxonomies of fault types, observable symptoms, and root causes. We further apply Apriori-based association rule mining to identify statistically significant relationships among faults, symptoms, and root causes, revealing common fault propagation patterns. Finally, we validate the taxonomy through a developer study with 145 practitioners. Our analysis identifies 37 distinct fault types grouped into 13 higher-level fault categories, along with 13 classes of observable symptoms and 12 categories of root causes. The results show that many failures originate from mismatches between probabilistically generated artifacts and deterministic interface constraints, frequently involving dependency integration, data validation, and runtime environment handling. Association rule mining further reveals recurring propagation pathways across system components, such as token management faults leading to authentication failures and datetime handling defects causing scheduling anomalies. Practitioners rated the taxonomy as representative of real-world failures (mean = 3.97/5), and 83.8% reported that it covered faults they had encountered.

Background: Recent studies have demonstrated that large language models (LLMs) can perform binary classification tasks on child welfare narratives, detecting the presence or absence of constructs such as substance-related problems, domestic violence, and firearms involvement. Whether smaller, locally deployable models can move beyond binary detection to classify specific substance types from these narratives remains untested. Objective: To validate a locally hosted LLM classifier for identifying specific substance types aligned with DSM-5 categories in child welfare investigation narratives. Methods: A locally hosted 20-billion-parameter LLM classified child maltreatment investigation narratives from a Midwestern U.S. state. Records previously identified as containing substance-related problems were passed to a second classification stage targeting seven DSM-5 substance categories. Expert human review of 900 stratified cases assessed classification precision, recall, and inter-method reliability (Cohen's kappa). Test-retest stability was evaluated using approximately 15,000 independently classified records. Results: Five substance categories achieved almost perfect inter-method agreement (kappa = 0.94-1.00): alcohol, cannabis, opioid, stimulant, and sedative/hypnotic/anxiolytic. Classification precision ranged from 92% to 100% for these categories. Two low-prevalence categories (hallucinogen, inhalant) performed poorly. Test-retest agreement ranged from 92.1% to 99.1% across the seven categories. Conclusions: A small, locally hosted LLM can reliably classify substance types from child welfare administrative text, extending prior work on binary classification to multi-label substance identification.

Subsurface ore detection is of paramount importance given the gradual depletion of shallow mineral resources in recent years. It is crucial to explore approaches that go beyond the limitations of traditional geological exploration methods. One such promising new method is joint magnetic and gravitational inversion. Given magnetic and gravitational data on a surface, jointly reconstructing the underlying densities that generate them remains an ill-posed inverse problem. Although joint inversion of multiple properties mitigates the non-uniqueness problem in magnetic and gravitational data, deterministic algorithms converge to a single regularized solution and thus do not capture the distribution of possible solutions. Similarly, most machine learning based techniques predict a single solution without modelling the entire distribution. In this paper, we introduce a novel framework that reframes 3D gravity and magnetic joint inversion as a rectified flow on the Noddyverse dataset, the largest physics-based dataset for inversion. We introduce a Ginzburg-Landau (GL) regularizer, a generalized version of the Ising model that aids in ore identification, enabling physics-aware training. We also propose a guidance methodology based on GL theory that can be used as a plug-and-play module with existing unconditional denoisers. Lastly, we also train and release a VAE for the 3D densities, which facilitates downstream work in the field.

We uncover a behavioral law of long-horizon vision-language models: models that maintain temporally grounded beliefs generalize better. Standard benchmarks measure only final-answer accuracy, which obscures how models use visual information; a model can guess correctly while its step-by-step reasoning is entirely unanchored to the visual input. We formalize this as behavioral faithfulness over long horizons, an empirically measurable property that quantifies whether a model's intermediate reasoning remains consistent with the evolving visual state. Across eight models on three long-horizon benchmarks, we demonstrate that temporal grounding quality is a leading indicator of robustness: the Step Grounding Rate (SGR) predicts out-of-distribution retention with $r = 0.83$ (permutation test $p = 0.003$), a relationship that holds within capacity-matched models and cannot be explained by scale or in-distribution accuracy. Critically, grounding quality varies by up to 10.8 percentage points within parameter-matched 7B models despite similar accuracy, revealing it as an independent axis of model capability. Multiple robustness checks confirm the signal reflects genuine visual reliance: counterfactual traces drop SGR by 26--41 percentage points, cross-architecture verifiers agree at $ρ= 0.96$, random reasoning scores near chance ($\sim 18\%$), and the predictor remains strong even without explicit reasoning disclosure ($r = 0.78$).

Conformal prediction delivers prediction intervals with distribution-free coverage, but its intervals can look overconfident in regions where the model is extrapolating, because standard conformal scores do not explicitly represent epistemic uncertainty. Credal methods, by contrast, make epistemic effects visible by working with sets of plausible predictive distributions, but they are typically model-based and lack calibration guarantees. We introduce CREDO, a simple "credal-then-conformalize" recipe that combines both strengths. CREDO first builds an interpretable credal envelope that widens when local evidence is weak, then applies split conformal calibration on top of this envelope to guarantee marginal coverage without further assumptions. This separation of roles yields prediction intervals that are interpretable: their width can be decomposed into aleatoric noise, epistemic inflation, and a distribution-free calibration slack. We provide a fast implementation based on trimming extreme posterior predictive endpoints, prove validity, and show on benchmark regressions that CREDO maintains target coverage while improving sparsity adaptivity at competitive efficiency.

The alignment problem refers to concerns regarding powerful intelligences, ensuring compatibility with human preferences and values as capabilities increase. Current large language models (LLMs) show misaligned behaviors, such as strategic deception, manipulation, and reward-seeking, that can arise despite safety training. Gaining a mechanistic understanding of these failures requires empirical approaches that can isolate behavioral patterns in controlled settings. We propose that biological misalignment precedes artificial misalignment, and leverage the Dark Triad of personality (narcissism, psychopathy, and Machiavellianism) as a psychologically grounded framework for constructing model organisms of misalignment. In Study 1, we establish comprehensive behavioral profiles of Dark Triad traits in a human population (N = 318), identifying affective dissonance as a central empathic deficit connecting the traits, as well as trait-specific patterns in moral reasoning and deceptive behavior. In Study 2, we demonstrate that dark personas can be reliably induced in frontier LLMs through minimal fine-tuning on validated psychometric instruments. Narrow training datasets as small as 36 psychometric items resulted in significant shifts across behavioral measures that closely mirrored human antisocial profiles. Critically, models generalized beyond training items, demonstrating out-of-context reasoning rather than memorization. These findings reveal latent persona structures within LLMs that can be readily activated through narrow interventions, positioning the Dark Triad as a validated framework for inducing, detecting, and understanding misalignment across both biological and artificial intelligence.

Purpose: This study developed a comprehensive database of presentation abstracts from the Society for Social Work and Research (SSWR) Annual Conference and examined patterns in research methodology, authorship, collaboration, and institutional participation over two decades. Method: Abstract metadata was compiled from the SSWR Confex conference management system for presentations from 2005 to 2026 using web scraping. A small language model (gpt-oss:20b) performed classification and extraction tasks on abstracts, including categorization of methodologies and parsing of author affiliations, with human review at each major stage to ensure accuracy. Results: The database contains 23,793 presentations with 69,924 author records representing 20,779 unique researchers from 4,049 institutions across 93 countries. Annual conference presentations increased from 423 in 2005 to 1,935 in 2026, representing a compound annual growth rate of 7.5%. Quantitative methods predominated (61.1%), followed by qualitative approaches (23.4%), mixed methods (9.1%), and reviews (5.4%). The mean number of authors per presentation increased from 2.22 in 2005 to 3.31 in 2026. International participation grew from 4.5% to 13.5% of author affiliations over the observation period. Discussion: Findings indicate substantial growth in SSWR conference participation, alongside increased collaboration and international engagement. The methodological distribution reveals continued quantitative predominance with growing qualitative representation. This database provides research infrastructure for systematic hypothesis testing about research priorities and disciplinary development over time, enabling analyses that inform both scholarship and conference planning.

Reusable decision structure survives across episodes in reinforcement learning, but this depends on how the agent--world boundary is drawn. In stationary, finite-horizon MDPs, an invariant core: the (not-necessarily contiguous) subsequences of state--action pairs shared by all successful trajectories (optionally under a simple abstraction) can be constructed. Under mild goal-conditioned assumptions, it's existence can be proven and explained by how the core captures prototypes that transfer across episodes. When the same task is embedded in a decentralized Markov game and the peer agent is folded into the world, each peer-policy update induces a new MDP; the per-episode invariant core can shrink or vanish, even with small changes to the induced world dynamics, sometimes leaving only the individual task core or just nothing. This policy-induced non-stationarity can be quantified with a variation budget over the induced kernels and rewards, linking boundary drift to loss of invariants. The view that a continual RL problem arises from instability of the agent--world boundary (rather than exogenous task switches) in decentralized MARL suggests future work on preserving, predicting, or otherwise managing boundary drift.

With many organizations struggling to gain value from AI deployments, pressure to evaluate AI in an informed manner has intensified. Status quo AI evaluation approaches mask the operational realities that ultimately determine deployment success, making it difficult for decision makers outside the stack to know whether and how AI tools will deliver durable value. We introduce and describe context specification as a process to support and inform the deployment decision making process. Context specification turns diffuse stakeholder perspectives about what matters in a given setting into clear, named constructs: explicit definitions of the properties, behaviors, and outcomes that evaluations aim to capture, so they can be observed and measured in context. The process serves as a foundational roadmap for evaluating what AI systems are likely to do in the deployment contexts that organizations actually manage.

In this paper, we study cooperative multi-agent reinforcement learning (MARL) where the joint reward exhibits submodularity, which is a natural property capturing diminishing marginal returns when adding agents to a team. Unlike standard MARL with additive rewards, submodular rewards model realistic scenarios where agent contributions overlap (e.g., multi-drone surveillance, collaborative exploration). We provide the first formal framework for this setting and develop algorithms with provable guarantees on sample efficiency and regret bound. For known dynamics, our greedy policy optimization achieves a $1/2$-approximation with polynomial complexity in the number of agents $K$, overcoming the exponential curse of dimensionality inherent in joint policy optimization. For unknown dynamics, we propose a UCB-based learning algorithm achieving a $1/2$-regret of $O(H^2KS\sqrt{AT})$ over $T$ episodes.

The development of effective treatments for Cerebral Palsy (CP) can begin with the early identification of affected children while they are still in the early stages of the disorder. Pathological issues in the brain can be better diagnosed with the use of one of many medical imaging techniques. Magnetic Resonance Imaging (MRI) has revolutionized medical imaging with its unparalleled image resolution. A unique Machine Learning (ML) model that was built to identify CP disorder is presented in this paper. The model is intended to assist in the early diagnosis of CP in newborns. In this study, the brain MRI images dataset was first collected, and then the preprocessing techniques were applied to this dataset to make it ready for use in the proposed model. Following this, the proposed model was constructed by combining three CNN models, specifically VGG 19, Efficient-Net, and the ResNet50 model, to extract features from the image. Following this, a Bi-LSTM was utilized as a classifier to determine whether or not CP was present, and finally, the proposed model was employed for training and testing. The results show that the proposed model achieved an accuracy of 98.83%, which is higher than VGG-19 (96.79%), Efficient-Net (97.29%), and VGG-16 (97.50%).. When the suggested model is compared to other models that have been pre-trained in the past, the accuracy scores seem to be much higher.

Multi-Agent Debate (MAD) has emerged as a promising paradigm for enhancing large language model reasoning. However, recent work reveals a limitation:standard MAD cannot improve belief correctness beyond majority voting; we refer to this as the Martingale Curse. This curse arises because correlated errors cause agents to converge toward erroneous consensus, where debate merely reinforces collective mistakes rather than filtering noise. We propose AceMAD, a framework that breaks the Martingale Curse by harnessing asymmetric cognitive potential energy to transform MAD from a random walk into a directed convergence process with positive drift. Through a peer-prediction mechanism, agents predict their peers' belief distributions, revealing asymmetric cognitive potential: truth-holders not only know the correct answer but also anticipate the crowd's misconceptions, while the hallucinating majority remains blind to their collective error. This asymmetry creates a potential energy gap that we quantify via strictly proper scoring rules. We prove this cognitive potential manifests as information-theoretic superiority and, under nonlinear aggregation, converts into submartingale drift toward truth, directly breaking the Martingale Curse. Experiments on challenging subsets across six benchmarks show AceMAD recovers sparse truth signals even when initial majorities are incorrect, substantially outperforming baseline methods.

The growing scale of deep learning demands distributed training frameworks that jointly reason about parallelism, memory, and network topology. Prior works often rely on heuristic or topology-agnostic search, handling communication and memory separately. Without per-device memory awareness, these methods typically ensure feasibility post hoc by sharding parameters and activations across many devices, increasing synchronization, inflating communication, and underutilizing compute-limiting scalability and efficiency on real datacenter networks. We present NEST, a network-, compute-, and memory-aware device placement framework that unifies model parallelism, topology modeling, and memory feasibility via structured dynamic programming. NEST's DP operates on operator graphs with tensor and expert parallel configurations, explicit allreduce latencies across hierarchical or arbitrary networks, and memory/compute profiles. By factoring parallelism across tensor, pipeline, data, and expert dimensions, NEST defines a principled search space for hybrid strategies while jointly optimizing co-location, network latency, and memory feasibility. Evaluations across diverse hardware and networks show NEST achieves up to 2.43 times higher throughput, better memory efficiency, and improved scalability over state-of-the-art baselines, providing a foundation for co-designing parallelization strategies and datacenter interconnects for next-generation AI infrastructure. The source code of NEST is available at: https://github.com/scai-tech/Nest

Inference-time alignment effectively steers large language models (LLMs) by generating multiple candidates from a reference model and selecting among them with an imperfect reward model. However, current strategies face a fundamental dilemma: ``optimistic'' approaches like Best-of-$N$ suffer from reward hacking, while ``pessimistic'' regularized methods often stifle the exploration needed to discover high-quality responses. In this work, we formalize this trade-off through the lens of regret minimization, demonstrating that the optimal strategy depends critically on the tail behavior of the reward distribution. We show theoretically that light-tailed regimes favor optimism to unearth high-quality outliers, whereas heavy-tailed regimes require pessimism to guard against reward mis-calibration in the extremes. Guided by this insight, we introduce Best-of-Tails (BoT), an adaptive inference-time alignment framework that uses Tsallis divergence as a tunable regularizer to provide a finer granularity of interpolation between these extremes. BoT uses the Hill estimator to characterize reward-tail heaviness on a per-prompt basis and dynamically adjusts its selection rule to balance exploration gains against alignment error. Across math, multiple-choice reasoning, and human-preference evaluations, BoT improves alignment performance across a range of reference and reward model configurations relative to fixed-strategy baselines.

Deep reinforcement learning agents frequently suffer from premature convergence, where early entropy collapse causes the policy to discard exploratory behaviors before discovering globally optimal strategies. We introduce Optimistic Policy Regularization (OPR), a lightweight mechanism designed to preserve and reinforce historically successful trajectories during policy optimization. OPR maintains a dynamic buffer of high-performing episodes and biases learning toward these behaviors through directional log-ratio reward shaping and an auxiliary behavioral cloning objective. When instantiated on Proximal Policy Optimization (PPO), OPR substantially improves sample efficiency on the Arcade Learning Environment. Across 49 Atari games evaluated at the 10-million step benchmark, OPR achieves the highest score in 22 environments despite baseline methods being reported at the standard 50-million step horizon. Beyond arcade benchmarks, OPR also generalizes to the CAGE Challenge 2 cyber-defense environment, surpassing the competition-winning Cardiff agent while using the same PPO architecture. These results demonstrate that anchoring policy updates to empirically successful trajectories can improve both sample efficiency and final performance.

Data scarcity is a primary obstacle in developing robust Machine Learning (ML) models for detecting rapidly intensifying tropical cyclones. Traditional data augmentation techniques (rotation, flipping, brightness adjustment) fail to preserve the physical consistency and high-intensity gradients characteristic of rare Category 4-equivalent events, which constitute only 0.14\% of our dataset (202 of 140,514 samples). We propose a physics-informed diffusion model based on the Context-UNet architecture to generate synthetic, multi-spectral satellite imagery of extreme weather events. Our model is conditioned on critical atmospheric parameters such as average wind speed, type of Ocean and stage of development (early, mature, late etc) -- the known drivers of rapid intensification. Using a controlled pre-generated noise sampling strategy and mixed-precision training, we generated $16\times16$ wind-field samples that are cropped from multi-spectral satellite imagery which preserve realistic spatial autocorrelation and physical consistency. Results demonstrate that our model successfully learns discriminative features across ten distinct context classes, effectively mitigating the data bottleneck. Specifically, we address the extreme class imbalance in our dataset, where Class 4 (Ocean 2, early stage with average wind speed 50kn hurricane) contains only 202 samples compared to 79,768 samples in Class 0. This generative framework provides a scalable solution for augmenting training datasets for operational weather detection algorithms. The average Results yield an average Log-Spectral Distance (LSD) of 4.5dB, demonstrating a scalable framework for enhancing operational weather detection algorithms.

Evaluating time series attribution methods is difficult because real-world datasets rarely provide ground truth for which time points drive a prediction. A common workaround is to generate synthetic data where class-discriminating features are placed at known locations, but each study currently reimplements this from scratch. We introduce xaitimesynth, a Python package that provides reusable infrastructure for this evaluation approach. The package generates synthetic time series following an additive model where each sample is a sum of background signal and a localized, class-discriminating feature, with the feature window automatically tracked as a ground truth mask. A fluent data generation API and YAML configuration format allow flexible and reproducible dataset definitions for both univariate and multivariate time series. The package also provides standard localization metrics, including AUC-PR, AUC-ROC, Relevance Mass Accuracy, and Relevance Rank Accuracy. xaitimesynth is open source and available at https://github.com/gregorbaer/xaitimesynth.

Spatial transcriptomics (ST) enables mapping gene expression with spatial context but is severely affected by high sparsity and technical noise, which conceals true biological signals and hinders downstream analyses. To address these challenges, SpatialMagic was proposed, which is a hybrid imputation model combining MAGIC-based graph diffusion with transformer-based spatial self-attention. The long-range dependencies in the gene expression are captured by graph diffusion, and local neighborhood structure is captured by spatial attention models, which allow for recovering the missing expression values, retaining spatial consistency. Across multiple platforms, SpatialMagic consistently outperforms existing baselines, including MAGIC and attention-based models, achieving peak Adjusted Rand Index (ARI) scores in clustering accuracy of 0.3301 on high-resolution Stereo-Seq data, 0.3074 on Slide-Seq, and 0.4216 on the Sci-Space dataset. Beyond quantitative improvements, SpatialMagic substantially enhances downstream biological analyses by improving the detection of both up- and down-regulated genes while maintaining regulatory consistency across datasets. The pathway enrichment analysis of the recovered genes indicates that they are involved in consistent processes across key metabolic, transport, and neural signaling pathways, suggesting that the framework improves data quality while preserving biological interpretability. Overall, SpatialMagic's hybrid diffusion attention strategy and refinement module outperform state-of-the-art baselines on quantitative metrics and provide a better understanding of the imputed data by preserving tissue architecture and uncovering biologically relevant genes. The source code and datasets are provided in the following link: https://github.com/sayeemzzaman/SpatialMAGIC

The Job Shop Scheduling Problem (JSSP) is commonly formulated as a disjunctive graph in which nodes represent operations and edges encode technological precedence constraints as well as machine-sharing conflicts. Most existing reinforcement learning approaches model this graph as homogeneous, merging job-precedence and machine-contention edges into a single relation type. Such a simplification overlooks the intrinsic heterogeneity of the problem structure and may lead to the loss of critical relational information. To address this limitation, we propose the Heterogeneous Graph Transformer (HGT)-Scheduler, a reinforcement learning framework that models the JSSP as a heterogeneous graph. The proposed architecture leverages a Heterogeneous Graph Transformer to capture type-specific relational patterns through edge-type-dependent attention mechanisms applied to precedence and contention relations. The scheduling policy is trained using Proximal Policy Optimization. The effectiveness of the proposed method is evaluated on the Fisher--Thompson benchmark instances. On the FT06 instance, the HGT-Scheduler achieves an optimality gap of 8.4\%, statistically outperforming both an identical architecture that ignores edge types ($p = 0.011$) and a standard Graph Isomorphism Network baseline. On the larger FT10 instance, the approach demonstrates favorable scalability. However, under a 50,000-step training limit, the performance of heterogeneous and homogeneous graph models is comparable, suggesting that edge-type awareness requires longer training horizons for larger problem instances. Ablation analyses further indicate that a three-layer attention architecture provides the best performance. Overall, the results confirm that explicitly modeling distinct edge semantics improves the learning of effective scheduling policies.

Neural network representations are often analyzed as vectors in a fixed Euclidean space. However, their coordinates are not uniquely defined. If a hidden representation is transformed by an invertible linear map, the network function can be preserved by applying the inverse transformation to downstream weights. Representations are therefore defined only up to invertible linear transformations. We study neural representation spaces from this geometric viewpoint and treat them as vector spaces with a gauge freedom under the general linear group. Within this framework, commonly used similarity measures such as cosine similarity become metric-dependent quantities whose values can change under coordinate transformations that leave the model function unchanged. This provides a common interpretation for several observations in the literature, including cosine-similarity instability, anisotropy in embedding spaces, and the appeal of representation comparison methods such as SVCCA and CKA. Experiments on multilayer perceptrons and convolutional networks confirm that inserting invertible transformations into trained models can substantially distort cosine similarity and nearest-neighbor structure while leaving predictions unchanged. These results indicate that analysis of neural representations should focus either on quantities that are invariant under this gauge freedom or on explicitly chosen canonical coordinates.

Over the past decade, Artificial Intelligence has significantly advanced, mostly driven by large-scale neural approaches. However, in the chemical process industry, where safety is critical, these methods are often unsuitable due to their brittleness, and lack of explainability and interpretability. Furthermore, open-source real-world datasets containing historical failures are scarce in this domain. In this paper, we investigate an approach for predicting failures in chemical processes using symbolic machine learning and conduct a feasibility study in the context of an ethylene oxidation process. Our method builds on a state-of-the-art symbolic machine learning system capable of learning predictive models in the form of probabilistic rules from context-dependent noisy examples. This system is a general-purpose symbolic learner, which makes our approach independent of any specific chemical process. To address the lack of real-world failure data, we conduct our feasibility study leveraging data generated from a chemical process simulator. Experimental results show that symbolic machine learning can outperform baseline methods such as random forest and multilayer perceptron, while preserving interpretability through the generation of compact, rule-based predictive models. Finally, we explain how such learned rule-based models could be integrated into agents to assist chemical plant operators in decision-making during potential failures.

Building machine-learning models for estimating traffic flows from OD matrices requires an appropriate design of the training process and a training dataset spanning over multiple regimes and dynamics. As machine-learning models rely heavily on historical data, their predictions are typically accurate only when future traffic patterns resemble those observed during training. However, their performance often degrades when there is a significant statistical discrepancy between historical and future conditions. This issue is particularly relevant in traffic forecasting when predictions are required for modified versions of the network, where the underlying graph structure changes due to events such as maintenance, public demonstrations, flooding, or other extreme disruptions. Ironically, these are precisely the situations in which reliable traffic predictions are most needed. In the presented work, we combine a machine-learning model (graph convolutional neural network) with a meta-learning architecture to train the former to quickly adapt to new graph structures and demand patterns, so that it may easily be applied to scenarios in which changes in the road network (the graph) and the demand (the node features) happen simultaneously. Our results show that the use of meta-learning allows the graph neural network to quickly adapt to unseen graphs (network closures) and OD matrixes while easing the burden of designing a training dataset that covers all relevant patterns for the practitioners. The proposed architecture achieves a R^2 of around 0.85 over unseen closures and OD matrixes.

Solving flow through porous media is a crucial step in the topology optimisation of cold plates, a key component in modern thermal management. Traditional computational fluid dynamics (CFD) methods, while accurate, are often prohibitively expensive for large and complex geometries. In contrast, data-driven surrogate models provide a computationally efficient alternative, enabling rapid and reliable predictions. In this study, we develop a machine-learning framework for predicting steady-state flow through porous media governed by the Navier-Stokes-Brinkman equations. We implement and compare three model architectures-convolutional autoencoder (AE), U-Net, and Fourier Neural Operator (FNO)-evaluating their predictive performance. To enhance physics consistency, we incorporate physics-informed loss functions. Our results demonstrate that FNO outperforms AE and U-Net, achieving a mean squared error (MSE) as low as 0.0017 while providing speedups of up to 1000 times compared to CFD. Additionally, the mesh-invariant property of FNO emphasizes its suitability for topology optimisation tasks, where varying mesh resolutions are required. This study highlights the potential of machine learning to accelerate fluid flow predictions in porous media, offering a scalable alternative to traditional numerical methods.

Physics-Informed Neural Networks (PINNs) enforce governing equations by penalizing PDE residuals at interior collocation points, but standard collocation strategies - uniform sampling and residual-based adaptive refinement - can oversample smooth regions, produce highly correlated point sets, and incur unnecessary training cost. We reinterpret collocation selection as a coreset construction problem: from a large candidate pool, select a fixed-size subset that is simultaneously informative (high expected impact on reducing PDE error) and diverse (low redundancy under a space-time similarity notion). We formulate this as a QUBO/BQM objective with linear terms encoding residual-based importance and quadratic terms discouraging redundant selections. To avoid the scalability issues of dense k-hot QUBOs, we propose a sparse graph-based BQM built on a kNN similarity graph and an efficient repair procedure that enforces an exact collocation budget. We further introduce hybrid coverage anchors to guarantee global PDE enforcement. We evaluate the method on the 1D time-dependent viscous Burgers equation with shock formation and report both accuracy and end-to-end time-to-accuracy, including a timing breakdown of selection overhead. Results demonstrate that sparse and hybrid formulations reduce selection overhead relative to dense QUBOs while matching or improving accuracy at fixed collocation budgets.

Designing aerial robots for specialized tasks, from perching to payload delivery, requires tailoring their aerodynamic shape to specific mission requirements. For tasks involving wide flight envelopes, the usual sequential process of first determining the shape and then the motion planner is likely to be suboptimal due to the inherent nonlinear interactions between them. This limitation has been motivating co-design research, which involves jointly optimizing the aerodynamic shape and the motion planner. In this paper, we present a general-purpose, gradient-based, nested co-design framework where the motion planner solves an optimal control problem and the aerodynamic forces used in the dynamics model are determined by a neural surrogate model. This enables us to model complex subsonic flow conditions encountered in aerial robotics and to overcome the limited applicability of existing co-design methods. These limitations stem from the simplifying assumptions they require for computational tractability to either the planner or the aerodynamics. We validate our method on two complex dynamic tasks for fixed-wing gliders: perching and a short landing. Our optimized designs improve task performance compared to an evolutionary baseline in a fraction of the computation time.

Alzheimer disease (AD) diagnosis and prognosis increasingly rely on machine learning (ML) models. Although these models provide good results, clinical adoption is limited by the need for technical expertise and the lack of trustworthy and consistent model explanations. SHAP (SHapley Additive exPlanations) is com-monly used to interpret AD models, but existing studies tend to focus on explanations for isolated tasks, providing little evidence about their robustness across disease stages, model architectures, or prediction objectives. This paper proposes a multi-level explainability framework that measures the coherence, stabil-ity and consistency of explanations by integrating: (1) within-model coherence metrics between feature importance and SHAP, (2) SHAP stability across AD boundaries, and (3) SHAP cross-task consistency be-tween diagnosis and prognosis. Using AutoML to optimize classifiers on the NACC dataset, we trained four diagnostic and four prognostic models covering the standard AD progression stages. Stability was then evaluated using correlation metrics, top-k feature overlap, SHAP sign consistency, and domain-level contribution ratios. Results show that cognitive and functional markers dominate SHAP explanations in both diagnosis and prognosis. SHAP-SHAP consistency between diagnostic and prognostic models was high across all classifiers, with 100% sign stability and minimal shifts in explanatory magnitude. Domain-level contributions also remained stable, with only minimal increases in genetic features for prognosis. These results demonstrate that SHAP explanations can be quantitatively vali-dated for robustness and transferability, providing clinicians with more reliable interpretations of ML pre-dictions.

Unlabeled streaming data are usually collected to describe dynamic systems, where concept drift detection is a vital prerequisite to understanding the evolution of systems. However, the drifting concepts are usually imbalanced in most real cases, which brings great challenges to drift detection. That is, the dominant statistics of large clusters can easily mask the drifting of small cluster distributions (also called small concepts), which is known as the `masking effect'. Considering that most existing approaches only detect the overall existence of drift under the assumption of balanced concepts, two critical problems arise: 1) where the small concept is, and 2) how to detect its drift. To address the challenging concept drift detection for imbalanced data, we propose Imbalanced Cluster Descriptor-based Drift Detection (ICD3) approach that is unbiased to the imbalanced concepts. This approach first detects imbalanced concepts by employing a newly designed multi-distribution-granular search, which ensures that the distribution of both small and large concepts is effectively captured. Subsequently, it trains a One-Cluster Classifier (OCC) for each identified concept to carefully monitor their potential drifts in the upcoming data chunks. Since the detection is independently performed for each concept, the dominance of large clusters is thus circumvented. ICD3 demonstrates highly interpretability by specifically locating the drifted concepts, and is robust to the changing of the imbalance ratio of concepts. Comprehensive experiments with multi-aspect ablation studies conducted on various benchmark datasets demonstrate the superiority of ICD3 against the state-of-the-art counterparts.

We propose a quantum implicit neural representation (QINR)-based autoencoder (AE) and variational autoencoder (VAE) for image reconstruction and generation tasks. Our purpose is to demonstrate that the QINR in VAEs and AEs can transform information from the latent space into highly rich, periodic, and high-frequency features. Additionally, we aim to show that the QINR-VAE can be more stable than various quantum generative adversarial network (QGAN) models in image generation because it can address the low diversity problem. Our quantum-classical hybrid models consist of a classical convolutional neural network (CNN) encoder and a quantum-based QINR decoder. We train the QINR-AE/VAE with binary cross-entropy with logits (BCEWithLogits) as the reconstruction loss. For the QINR-VAE, we additionally employ Kullback-Leibler divergence for latent regularization with beta/capacity scheduling to prevent posterior collapse. We introduce learnable angle-scaling in data reuploading to address optimization challenges. We test our models on the MNIST, E-MNIST, and Fashion-MNIST datasets to reconstruct and generate images. Our results demonstrate that the QINR structure in VAE can produce a wider variety of images with a small amount of data than various generative models that have been studied. We observe that the generated and reconstructed images from the QINR-VAE/AE are clear with sharp boundaries and details. Overall, we find that the addition of QINR-based quantum layers into the AE/VAE frameworks enhances the performance of reconstruction and generation with a constrained set of parameters.

The ensemble Kalman filter (EnKF) is widely used for data assimilation in high-dimensional systems, but its performance often deteriorates for strongly nonlinear dynamics due to the structural mismatch between the Kalman update and the underlying system behavior. In this work, we propose a latent autoencoder ensemble Kalman filter (LAE-EnKF) that addresses this limitation by reformulating the assimilation problem in a learned latent space with linear and stable dynamics. The proposed method learns a nonlinear encoder--decoder together with a stable linear latent evolution operator and a consistent latent observation mapping, yielding a closed linear state-space model in the latent coordinates. This construction restores compatibility with the Kalman filtering framework and allows both forecast and analysis steps to be carried out entirely in the latent space. Compared with existing autoencoder-based and latent assimilation approaches that rely on unconstrained nonlinear latent dynamics, the proposed formulation emphasizes structural consistency, stability, and interpretability. We provide a theoretical analysis of learning linear dynamics on low-dimensional manifolds and establish generalization error bounds for the proposed latent model. Numerical experiments on representative nonlinear and chaotic systems demonstrate that the LAE-EnKF yields more accurate and stable assimilation than the standard EnKF and related latent-space methods, while maintaining comparable computational cost and data-driven.

Plant disease classification via imaging is a critical task in precision agriculture. We propose XMACNet, a novel light-weight Convolutional Neural Network (CNN) that integrates self-attention and multi-modal fusion of visible imagery and vegetation indices for chili disease detection. XMACNet uses an EfficientNetV2S backbone enhanced by a self-attention module and a fusion branch that processes both RGB images and computed vegetation index maps (NDVI, NPCI, MCARI). We curated a new dataset of 12,000 chili leaf images across six classes (five disease types plus healthy), augmented synthetically via StyleGAN to mitigate data scarcity. Trained on this dataset, XMACNet achieves high accuracy, F1-score, and AUC, outperforming baseline models such as ResNet-50, MobileNetV2, and a Swin Transformer variant. Crucially, XMACNet is explainable: we use Grad-CAM++ and SHAP to visualize and quantify the models focus on disease features. The models compact size and fast inference make it suitable for edge deployment in real-world farming scenarios.

Robotic foundation models (RFMs) are emerging as a promising route towards flexible, instruction- and demonstration-driven robot control, however, a critical investigation of their industrial applicability is still lacking. This survey gives an extensive overview over the RFM-landscape and analyses, driven by concrete implications, how industrial domains and use cases shape the requirements of RFMs, with particular focus on collaborative robot platforms, heterogeneous sensing and actuation, edge-computing constraints, and safety-critical operation. We synthesise industrial deployment perspectives into eleven interdependent implications and operationalise them into an assessment framework comprising a catalogue of 149 concrete criteria, spanning both model capabilities and ecosystem requirements. Using this framework, we evaluate 324 manipulation-capable RFMs via 48,276 criterion-level decisions obtained via a conservative LLM-assisted evaluation pipeline, validated against expert judgements. The results indicate that industrial maturity is limited and uneven: even the highest-rated models satisfy only a fraction of criteria and typically exhibit narrow implication-specific peaks rather than integrated coverage. We conclude that progress towards industry-grade RFMs depends less on isolated benchmark successes than on systematic incorporation of safety, real-time feasibility, robust perception, interaction, and cost-effective system integration into auditable deployment stacks.

Protein sequence design must balance designability, defined as the ability to recover a target backbone, with multiple, often competing, developability properties such as solubility, thermostability, and expression. Existing approaches address these properties through post hoc mutation, inference-time biasing, or retraining on property-specific subsets, yet they are target dependent and demand substantial domain expertise or careful hyperparameter tuning. In this paper, we introduce ProtAlign, a multi-objective preference alignment framework that fine-tunes pretrained inverse folding models to satisfy diverse developability objectives while preserving structural fidelity. ProtAlign employs a semi-online Direct Preference Optimization strategy with a flexible preference margin to mitigate conflicts among competing objectives and constructs preference pairs using in silico property predictors. Applied to the widely used ProteinMPNN backbone, the resulting model MoMPNN enhances developability without compromising designability across tasks including sequence design for CATH 4.3 crystal structures, de novo generated backbones, and real-world binder design scenarios, making it an appealing framework for practical protein sequence design.

Deploying sparse Mixture of Experts(MoE) Vision Transformers remains a challenge due to linear expert memory scaling. Linear memory scaling stores $N$ independent expert weight matrices requiring $\mathcal{O}(N_E \cdot d^2)$ memory, which exceeds edge devices memory budget. Current compression methods like quantization, pruning and low-rank factorization reduce constant factors but leave the scaling bottleneck unresolved. We introduce ButterflyViT, a method that treats experts not as independent weight matrices but as geometric reorientations of a unified shared quantized substrate. Diversity among experts arises from viewing different angles of shared capacity, not from redundant storage. By applying learned rotations to a shared ternary prototype, each expert yields $\mathcal{O}(d_{\text{model}} \cdot d_{\text{ff}} + N_E \cdot n_\ell \cdot d)$ memory which is sub-linear in the number of experts. To address the unique challenges of vision, a spatial smoothness regulariser is introduced that penalises routing irregularities between adjacent patch tokens, turning patch correlation into a training signal. Across image classification tasks on CIFAR-100, ButterflyViT achieves 354$\times$ memory reduction at 64 experts with negligible accuracy loss. ButterflyViT allows multiple experts to fit on edge-constrained devices showing that geometric parameterization breaks linear scaling.

Large Language Models (LLMs), despite advances in instruction tuning, often fail to follow complex user instructions. Activation steering techniques aim to mitigate this by manipulating model internals, but have a potential risk of oversteering, where excessive emphasis on the instruction degrades task accuracy and overall text quality. To address this, we introduce DIRECTER (Dynamic rejection steering), a novel steering method that dynamically modulates steering strength by scaling the KV cache without extra dataset. DIRECTER couples steering with a plausibility-guided decoding loop, which adaptively adjusts steering strength at each step by comparing the steered output distribution to the original. If the steered output is deemed implausible, steering strength is progressively weakened. This strength modulation is guided by a lightweight, one-time attention sensitivity analysis that ranks layers by their influence on model representations. Extensive evaluations show that DIRECTER significantly enhances instruction-following capabilities across diverse benchmarks, improving accuracy by up to 6.5% over baselines without the common trade-offs in generation quality or task fidelity. The proposed dynamic, plausibility-guided control during activation steering further demonstrates its potential as a general mechanism for mitigating oversteering that is compatible with existing baselines.

Group Relative Policy Optimization (GRPO) is highly effective for post-training autoregressive (AR) language models, yet its direct application to diffusion large language models (dLLMs) often triggers reward collapse. We identify two sources of incompatibility. First, GRPO relies on importance ratios defined by sequence probabilities, which are intractable in dLLMs and must be estimated (e.g., via ELBO-based or mean-field likelihood proxies), yielding inherently noisy ratios. Second, standard GRPO's formulation is not designed for estimated ratios: its conditional clipping can be anomalously bypassed by model-agnostic estimation noise, producing gradient spikes, while its fixed group-size normalization amplifies gradient-magnitude fluctuations under high-variance ratio estimates. We show these effects form a self-reinforcing instability loop that drives policy drift and further increases ratio variance. To break this loop, we propose StableDRL, a reformulation of GRPO tailored for dLLMs that uses (i) unconditional clipping to suppress outlier-induced spikes and (ii) self-normalization to constrain updates within the convex hull of per-sample gradients. We further extend StableDRL to block-wise diffusion models via a staircase attention mechanism.

Constrained generative modeling is fundamental to applications such as robotic control and autonomous driving, where models must respect physical laws and safety-critical constraints. In real-world settings, these constraints rarely take the form of simple linear inequalities, but instead complex feasible regions that resemble road maps or other structured spatial domains. We propose a constrained generation framework that generates samples directly within such feasible regions while preserving realism. Our method fine-tunes a pretrained generative model to enforce constraints while maintaining generative fidelity. Experimentally, our method exhibits characteristics distinct from existing fine-tuning and training-free constrained baselines, revealing a new compromise between constraint satisfaction and sampling quality.

Training frontier-scale diffusion models often requires substantial computational resources concentrated in tightly coupled clusters, limiting participation to well-resourced institutions. While Decentralized Diffusion Models (DDM) enable training multiple experts in isolation, existing approaches require 1176 GPU-days and homogeneous training objectives across all experts. We present an efficient framework that reduces resource requirements while supporting heterogeneous training objectives. Our approach combines three contributions: (1) a heterogeneous decentralized training paradigm that allows experts to use different objectives (DDPM and Flow Matching), unified at inference time via a deterministic schedule-aware conversion into a common velocity space without retraining; (2) pretrained checkpoint conversion from ImageNet-DDPM to Flow Matching objectives, accelerating convergence and enabling initialization without objective-specific pretraining; and (3) PixArt-alpha's efficient AdaLN-Single architecture, reducing parameters while maintaining quality. Experiments on LAION-Aesthetics show that, relative to the training scale reported for prior DDM work, our approach reduces compute from 1176 to 72 GPU-days (16x) and data from 158M to 11M (14x). Under aligned inference settings, our heterogeneous 2DDPM:6FM configuration achieves better FID (11.88 vs. 12.45) and higher intra-prompt diversity (LPIPS 0.631 vs. 0.617) than the homogeneous 8FM baseline. By eliminating synchronization requirements and enabling mixed DDPM/FM objectives, our framework lowers infrastructure requirements for decentralized generative model training.

Protein language models (pLMs) have shown strong potential in prediction of the functional effects of missense variants in zero-shot settings. Despite this progress, benchmarking pLMs for viral proteins remains limited and systematic strategies for integrating in silico metrics with in vitro validation to guide antigen and target selection are underdeveloped. Here, we introduce ViroGym, a comprehensive benchmark designed to evaluate variant effect prediction in viral proteins and to facilitate selecting rational antigen candidates. We curated 79 deep mutational scanning (DMS) assays encompassing eukaryotic viruses, collectively comprising 552,937 mutated amino acid sequences across 7 distinct phenotypic readouts, and 21 influenza virus neutralisation tasks and a real-world predictive task for SARS-CoV-2. We benchmark well-established pLMs on fitness landscapes, antigenic diversity, and pandemic forecasting to provide a framework for vaccine selection, and show that pLMs selected using in vitro experimental data excel at predicting dominant circulating mutations in real world.

Autonomous agents are increasingly expected to support scientific research, and recent benchmarks report progress in code repair and autonomous experimentation. However, these evaluations typically assume a pre-configured execution environment, which requires resolving complex software dependencies, aligning hardware and framework versions, and configuring distributed execution, yet this capability remains largely unbenchmarked. We introduce ResearchEnvBench, a benchmark for environment synthesis in research code execution. Given a research repository, documentation, and a target execution setting, agents must construct an environment that successfully executes at runtime. Evaluations on diverse research repositories reveal a substantial gap in current SOTA agents, with failures dominated by incomplete dependency resolution and brittle version coupling. ResearchEnvBench provides a realistic testbed for advancing autonomous agents toward reproducible scientific research.

Recent Super-Resolution~(SR) methods mainly adopt Transformers for their strong long-range modeling capability and exceptional representational capacity. However, most SR Transformers rely heavily on relative positional bias~(RPB), which prevents them from leveraging hardware-efficient attention kernels such as FlashAttention. This limitation imposes a prohibitive computational burden during both training and inference, severely restricting attempts to scale SR Transformers by enlarging the training patch size or the self-attention window. Consequently, unlike other domains that actively exploit the inherent scalability of Transformers, SR Transformers remain heavily focused on effectively utilizing limited receptive fields. In this paper, we propose Rank-factorized Implicit Neural Bias~(RIB), an alternative to RPB that enables FlashAttention in SR Transformers. Specifically, RIB approximates positional bias using low-rank implicit neural representations and concatenates them with pixel content tokens in a channel-wise manner, turning the element-wise bias addition in attention score computation into a dot-product operation. Further, we introduce a convolutional local attention and a cyclic window strategy to fully leverage the advantages of long-range interactions enabled by RIB and FlashAttention. We enlarge the window size up to \textbf{96$\times$96} while jointly scaling the training patch size and the dataset size, maximizing the benefits of Transformers in the SR task. As a result, our network achieves \textbf{35.63\,dB PSNR} on Urban100$\times$2, while reducing training and inference time by \textbf{2.1$\times$} and \textbf{2.9$\times$}, respectively, compared to the RPB-based SR Transformer~(PFT).

We describe an experiment in large-scale autoformalization of algebraic topology in an Interactive Theorem Proving (ITP) environment, where the workload is distributed among multiple LLM-based coding agents. Rather than relying on static central planning, we implement a simulated bounty-based marketplace in which agents dynamically propose new lemmas (formal statements), attach bounties to them, and compete to discharge these proof obligations and claim the bounties. The agents interact directly with the interactive proof system: they can invoke tactics, inspect proof states and goals, analyze tactic successes and failures, and iteratively refine their proof scripts. In addition to constructing proofs, agents may introduce new formal definitions and intermediate lemmas to structure the development. All accepted proofs are ultimately checked and verified by the underlying proof assistant. This setting explores collaborative, decentralized proof search and theory building, and the use of market-inspired mechanisms to scale autoformalization in ITP.

Line-level code completion requires a critical balance between high accuracy and low latency. Existing methods suffer from a trade-off: large language models (LLMs) provide high-quality suggestions but incur high latency, while small language models (SLMs) are fast but often suboptimal. We propose MCCom (Model-Cascading-based code Completion), a framework that cascades a local SLM with a cloud-based LLM. To achieve effective cascading, MCCom leverages user actions as a novel signal to trigger the LLM only when the SLM fails, significantly reducing cloud computation costs. Furthermore, we introduce a two-stage speculative decoding strategy and an iterative retrieval mechanism to enhance collaboration between the models. We also train a 121M-parameter lightweight model, which achieves 73.8% of the performance of a 7B state-of-the-art model. Evaluated on RepoEval and a new real-world benchmark StmtEval, MCCom reduces inference latency by up to 47.9% and LLM usage by 46.3%, while improving the LLM's exact match rate by 8.9% through effective collaboration.

Credit risk scoring must support high-stakes lending decisions where data distributions change over time, probability estimates must be reliable, and group-level fairness is required. While modern machine learning models improve default prediction accuracy, they often produce poorly calibrated scores under distribution shift and may create unfair outcomes when trained without explicit constraints. This paper proposes Calibrated Credit Intelligence (CCI), a deployment-oriented framework that combines (i) a Bayesian neural risk scorer to capture epistemic uncertainty and reduce overconfident errors, (ii) a fairnessconstrained gradient boosting model to control group disparities while preserving strong tabular performance, and (iii) a shiftaware fusion strategy followed by post-hoc probability calibration to stabilize decision thresholds in later time periods. We evaluate CCI on the Home Credit Credit Risk Model Stability benchmark using a time-consistent split to reflect real-world drift. Compared with strong baselines (LightGBM, XGBoost, CatBoost, TabNet, and a standalone Bayesian neural model), CCI achieves the best overall trade-off between discrimination, calibration, stability, and fairness. In particular, CCI reaches an AUC-ROC of 0.912 and an AUC-PR of 0.438, improves operational performance with Recall@1%FPR = 0.509, and reduces calibration error (Brier score 0.087, ECE 0.015). Under temporal shift, CCI shows a smaller AUC-PR drop from early to late periods (0.017), and it lowers group disparities (demographic parity gap 0.046, equal opportunity gap 0.037) compared to unconstrained boosting. These results indicate that CCI produces risk scores that are accurate, reliable, and more equitable under realistic deployment conditions.

We present the design and implementation of PolyBlocks, a modular and reusable MLIR-based compiler infrastructure for AI programming frameworks and AI chips. PolyBlocks is based on pass pipelines that compose transformations on loop nests and SSA, primarily relying on lightweight affine access analysis; the transformations are stitched together in specialized ways to realize high-performance code automatically by the use of analytical cost models and heuristics. The optimizations in these passes include multi-level tiling, fusion, on-chip scratchpad usage, mapping matmuls and convolutions to matrix units, fusing the attention layer, and several other transformations for parallelism and locality. They have been developed in a way that makes it easy to build PolyBlocks-based compilers to target new chips, reusing much of the infrastructure. PolyBlocks' design and architecture enable fully automatic code generation from high-level frameworks to low-level target-specific intrinsics. Experimental results from evaluating PolyBlocks-powered just-in-time compilation for PyTorch and JAX targeting NVIDIA GPUs show that it is able to match or outperform Torch Inductor and XLA in several cases, although the latter rely on a combination of vendor libraries and code generation. For individual operators like matmuls and convolutions, PolyBlocks-generated code is competitive with the best vendor-tuned libraries or hand-written kernels.

Navigating safely through dense crowds requires collision avoidance that generalizes beyond the densities seen during training. Learning-based crowd navigation can break under out-of-distribution crowd sizes due to density-sensitive observation normalization and social-cost scaling, while analytical solvers often remain safe but freeze in tight interactions. We propose a reinforcement learning approach for dense, variable-density navigation that attains zero-shot density generalization using a density-invariant observation encoding with density-randomized training and physics-informed proxemic reward shaping with density-adaptive scaling. The encoding represents the distance-sorted $K$ nearest pedestrians plus bounded crowd summaries, keeping input statistics stable as crowd size grows. Trained with $N\!\in\![11,16]$ pedestrians in a $3\mathrm{m}\times3\mathrm{m}$ arena and evaluated up to $N\!=\!21$ pedestrians ($1.3\times$ denser), our policy reaches the goal in $>99\%$ of episodes and achieves $86\%$ collision-free success in random crowds, with markedly less freezing than analytical methods and a $>\!60$-point collision-free margin over learning-based benchmark methods. Codes are available at \href{https://github.com/jznmsl/PSS-Social}{https://github.com/jznmsl/PSS-Social}.

Over two billion Apple devices ship with a Neural Processing Unit (NPU) - the Apple Neural Engine (ANE) - yet this accelerator remains largely unused for large language model workloads. CoreML, Apple's public ML framework, imposes opaque abstractions that prevent direct ANE programming and do not support on-device training. We present Orion, to our knowledge the first open end-to-end system that combines direct ANE execution, a compiler pipeline, and stable multi-step training with checkpoint resume in a single native runtime, bypassing CoreML entirely via Apple's private _ANEClient and _ANECompiler APIs. Building on prior characterization work by maderix, we extend public knowledge of ANE constraints to a catalog of 20 restrictions on MIL IR programs, memory layout, compilation limits, and numerical behavior, including 14 previously undocumented constraints discovered during Orion development. Orion includes a compiler that lowers a graph IR through five optimization passes to ANE-native MIL and a runtime that manages IOSurface-backed zero-copy tensor I/O, program caching, and delta compilation for weight updates. Because the ANE bakes weights at compile time, naive training normally requires full recompilation per step (~4.2 s). We show that compiled programs can instead be updated by unloading, patching weight files, and reloading, bypassing ANECCompile() and reducing recompilation from 4,200 ms to 494 ms per step (8.5x), yielding a 3.8x training speedup. On an M4 Max, Orion achieves 170+ tokens/s for GPT-2 124M inference and demonstrates stable training of a 110M-parameter transformer on TinyStories for 1,000 steps in 22 minutes with zero NaN occurrences. We also present LoRA adapter-as-input, enabling hot-swap of adapters via IOSurface inputs without recompilation.

Current safety alignment methods encode safe behavior implicitly within model parameters, creating a fundamental opacity: we cannot easily inspect why a model refuses a request, nor intervene when its safety judgments fail. We propose Safe Transformer, a modular approach that augments pre-trained language models by inserting a discrete information bottleneck containing an explicit safety bit between transformer layers. The safety bit serves as both an interpretable signal of the model's safety classification and a controllable switch: through contrastive training, the model learns disentangled representations where the safety bit governs the behavioral mode - producing helpful responses when $s=1$ and refusals when $s=0$ - while additional unsupervised bits $u$ encode semantic content for generation. Additional unsupervised bits in the information bottleneck allow semantic information to flow through, preserving the model's generation capabilities. This design achieves both interpretability (the safety decision is directly readable) and controllability (the safety bit can be manually overridden), requiring only lightweight fine-tuning without pre-training from scratch. In red-team benchmarks, Safe Transformer achieves near-zero Attack Success Rate, substantially outperforming base models and safety fine-tuning baselines.

Electricity market prices exhibit extreme volatility, nonlinearity, and non-stationarity, making accurate forecasting a significant challenge. While cutting-edge time series foundation models (TSFMs) effectively capture temporal dependencies, they typically underutilize cross-variate correlations and non-periodic patterns that are essential for price forecasting. Conversely, regression models excel at capturing feature interactions but are limited to future-available inputs, ignoring crucial historical drivers that are unavailable at forecast time. To bridge this gap, we propose FutureBoosting, a novel paradigm that enhances regression-based forecasts by integrating forecasted features generated from a frozen TSFM. This approach leverages the TSFM's ability to model historical patterns and injects these insights as enriched inputs into a downstream regression model. We instantiate this paradigm into a lightweight, plug-and-play framework for electricity price forecasting. Extensive evaluations on real-world electricity market data demonstrate that our framework consistently outperforms state-of-the-art TSFMs and regression baselines, achieving reductions in Mean Absolute Error (MAE) of more than 30% at most. Through ablation studies and explainable AI (XAI) techniques, we validate the contribution of forecasted features and elucidate the model's decision-making process. FutureBoosting establishes a robust, interpretable, and effective solution for practical market participation, offering a general framework for enhancing regression models with temporal context.

Indoor air quality (IAQ) forecasting plays a critical role in safeguarding occupant health, ensuring thermal comfort, and supporting intelligent building control. However, predicting future concentrations of key pollutants such as carbon dioxide (CO2) and fine particulate matter (PM2.5) remains challenging due to the complex interplay between environmental factors and highly dynamic occupant behaviours. Traditional data driven models primarily rely on historical sensor trajectories and often fail to anticipate behaviour induced emission spikes or rapid concentration shifts. To address these limitations, we present a dual stream bi directional feedback fusion framework that jointly models indoor environmental evolution and action derived embeddings representing human activities. The proposed architecture integrates a context aware modulation mechanism that adaptively scales and shifts each stream based on a shared, evolving fusion state, enabling the model to selectively emphasise behavioural cues or long term environmental trends. Furthermore, we introduce dual timescale temporal modules that independently capture gradual CO2 accumulation patterns and short term PM2.5 fluctuations. A composite loss function combining weighted mean squared error, spike aware penalties, and uncertainty regularisation facilitates robust learning under volatile indoor conditions. Extensive validation on real-world IAQ datasets demonstrates that our approach significantly outperforms state of the art forecasting baselines while providing interpretable uncertainty estimates essential for practical deployment in smart buildings and health-aware monitoring systems.

Invisible watermarks, as an essential technology for image copyright protection, have been widely deployed with the rapid development of social media and AIGC. However, existing invisible watermark detection heavily relies on prior knowledge of specific algorithms, leading to limited detection capabilities for "unknown watermarks" in open environments. To this end, we propose a novel task named Universal Watermark Presence Detection (UWPD), which aims to identify whether an image carries a copyright mark without requiring decoding information. We construct the UniFreq-100K dataset, comprising large-scale samples across various invisible watermark embedding algorithms. Furthermore, we propose the Frequency Shield Network (FSNet). This model deploys an Adaptive Spectral Perception Module (ASPM) in the shallow layers, utilizing learnable frequency gating to dynamically amplify high-frequency watermark signals while suppressing low-frequency semantics. In the deep layers, the network introduces Dynamic Multi-Spectral Attention (DMSA) combined with tri-stream extremum pooling to deeply mine watermark energy anomalies, forcing the model to precisely focus on sensitive frequency bands. Extensive experiments demonstrate that FSNet exhibits superior zero-shot detection capabilities on the UWPD task, outperforming existing baseline models. Code and datasets will be released upon acceptance.

Protein language models often take into consideration the alignment between a protein sequence and its textual description. However, they do not take structural information into consideration. Traditional methods treat sequence and structure separately, limiting the ability to exploit the alignment between the structure and protein sequence embeddings. In this paper, we introduce a sequence structure contrastive alignment framework, which learns a shared embedding space where proteins are represented consistently across modalities. By training on large-scale pairs of sequences and experimentally resolved or predicted structures, the model maximizes agreement between matched sequence structure pairs while pushing apart unrelated pairs. This alignment enables cross-modal retrieval (e.g., finding structural neighbors given a sequence), improves downstream prediction tasks such as function annotation and stability estimation, and provides interpretable links between sequence variation and structural organization. Our results demonstrate that contrastive learning can serve as a powerful bridge between protein sequences and structures, offering a unified representation for understanding and engineering proteins.

Access to electronic health records (EHRs) for digital health research is often limited by privacy regulations and institutional barriers. Synthetic EHRs have been proposed as a way to enable safe and sovereign data sharing; however, existing methods may produce records that capture overall statistical properties of real data but present inconsistencies across clinical processes and observations. We developed an integrated pipeline to make synthetic patient trajectories clinically consistent through two synergistic steps: high-fidelity generation and scalable auditing. Using the MIMIC-IV database, we trained a knowledge-grounded generative model that represents nearly 32,000 distinct clinical events, including demographics, laboratory measurements, medications, procedures, and diagnoses, while enforcing structural integrity. To support clinical consistency at scale, we incorporated an automated auditing module leveraging large language models to filter out clinical inconsistencies (e.g., contraindicated medications) that escape probabilistic generation. We generated 18,071 synthetic patient records derived from a source cohort of 180,712 real patients. While synthetic clinical event probabilities demonstrated robust agreement (mean bias effectively 0.00) and high correlation (R2=0.99) with the real counterparts, review of a random sample of synthetic records (N=20) by three clinicians identified inconsistencies in 45-60% of them. Automated auditing reduced the difference between real and synthetic data (Cohen's effect size d between 0.59 and 1.60 before auditing, and between 0.18 and 0.67 after auditing). Downstream models trained on audited data matched or even exceeded real-data performance. We found no evidence of privacy risks, with membership inference performance indistinguishable from random guessing (F1-score=0.51).

The Dynamic Targeting (DT) mission concept is an approach to satellite observation in which a lookahead sensor gathers information about the upcoming environment and uses this information to intelligently plan observations. Previous work has shown that DT has the potential to increase the science return across applications. However, DT mission concepts must address challenges, such as the limited spatial extent of onboard lookahead data and instrument mobility, data throughput, and onboard computation constraints. In this work, we show how the performance of DT systems can be improved by using supplementary data streamed from geostationary satellites that provide lookahead information up to 35 minutes ahead of time rather than the 1 minute latency from an onboard lookahead sensor. While there is a greater volume of geostationary data, the search space for observation planning explodes exponentially with the size of the horizon. To address this, we introduce a hierarchical planning approach in which the geostationary data is used to plan a long-term observation blueprint in polynomial time, then the onboard lookahead data is leveraged to refine that plan over short-term horizons. We compare the performance of our approach to that of traditional DT planners relying on onboard lookahead data across four different problem instances: three cloud avoidance variations and a storm hunting scenario. We show that our hierarchical planner outperforms the traditional DT planners by up to 41% and examine the features of the scenarios that affect the performance of our approach. We demonstrate that incorporating geostationary satellite data is most effective for dynamic problem instances in which the targets of interest are sparsely distributed throughout the overflight.

Agentic systems operating over large tool ecosystems must plan and execute long-horizon workflows under weak or non-verifiable supervision. While frontier models mitigate these challenges through scale and large context budgets, small language models (SLMs) remain brittle: eager tool loading saturates context, execution errors compound over time, and sparse rewards limit learning. We introduce ATLAS, a reinforcement finetuning framework that enables SLMs to operate effectively in large-scale toolspace environments by learning how to acquire context and how to execute actions. Our approach makes two key contributions. First, we treat context control and execution structure as learnable decisions, combining iterative tool loading with programmatic tool orchestration to bound context growth and stabilize long-horizon trajectories. Second, we propose rubric-based reinforcement finetuning, which decomposes task success into structured, task-aligned criteria and enables scalable training using small judge models. Across MCP benchmarks, these design choices yield large and consistent gains over generic RL baselines, allowing a 4B SLM to approach frontier-agent performance under far tighter parameter and context budgets.

Current solar flare predictions often lack precise quantification of their reliability, resulting in frequent false alarms, particularly when dealing with datasets skewed towards extreme events. To improve the trustworthiness of space weather forecasting, it is crucial to establish confidence intervals for model predictions. Conformal prediction, a machine learning framework, presents a promising avenue for this purpose by constructing prediction intervals that ensure valid coverage in finite samples without making assumptions about the underlying data distribution. In this study, we explore the application of conformal prediction to regression tasks in space weather forecasting. Specifically, we implement full-disk solar flare prediction using images created from magnetic field maps and adapt four pre-trained deep learning models to incorporate three distinct methods for constructing confidence intervals: conformal prediction, quantile regression, and conformalized quantile regression. Our experiments demonstrate that conformalized quantile regression achieves higher coverage rates and more favorable average interval lengths compared to alternative methods, underscoring its effectiveness in enhancing the reliability of solar weather forecasting models.

Mining specifications from execution traces presents an automated way of capturing characteristic system behaviors. However, existing approaches are largely restricted to Boolean abstractions of events, limiting their ability to express data-aware properties. In this paper, we extend mining procedures to operate over richer datatypes. We first establish candidate functions in our domain that cover the set of traces by leveraging Syntax Guided Synthesis (SyGuS) techniques. To capture these function applications temporally, we formalize the semantics of TSL$_f$, a finite-prefix interpretation of Temporal Stream Logic (TSL) that extends LTL$_f$ with support for first-order predicates and functional updates. This allows us to unify a corresponding procedure for learning the data transformations and temporal specifications of a system. We demonstrate our approach synthesizing reactive programs from mined specifications on the OpenAI-Gymnasium ToyText environments, finding that our method is more robust and orders of magnitude more sample-efficient than passive learning baselines on generalized problem instances.

Increasingly deeper integration of HPC resources and QPUs unveils new challenges in computer architecture and engineering. As a consequence, dependability arises again as a concern encompassing resilience, reproducibility and security. The properties of quantum computing systems involve a reinterpretation of these factors in retrodictive, predictive, and prescriptive ways. We state here that resilience must become an \emph{a priori} design constraint rather than an afterthought of HPC-QPU integration. This article describes the need for conceptual and quantitative models to estimate and assess the resilience hybrid classical-quantum computing infrastructure. We suggest how resilience methods in civil engineering can apply at various levels of the classical-quantum computing stack. We also discuss implications of a model of end-user value for the estimation of consequences resulting from the propagation of vulnerabilities from a given level of the stack upwards. Finally, we argue in favor of new resilience models can help the impact of improving specific components in quantum technology stacks to provide a clearer picture about the value of separation of concerns across different layers. Ultimately, HPC-QPU integration will increasingly demand more precise statements about the cost-benefit ratio of specific system improvements and their cascading consequences against estimates of delivered value to users.

Reasoning models think out loud, but much of what they say is noise. We introduce OPSDC (On-Policy Self-Distillation for Reasoning Compression), a method that teaches models to reason more concisely by distilling their own concise behavior back into themselves. The entire approach reduces to one idea: condition the same model on a "be concise" instruction to obtain teacher logits, and minimize per-token reverse KL on the student's own rollouts. No ground-truth answers, no token budgets, no difficulty estimators. Just self-distillation. Yet this simplicity belies surprising sophistication: OPSDC automatically compresses easy problems aggressively while preserving the deliberation needed for hard ones. On Qwen3-8B and Qwen3-14B, we achieve 57-59% token reduction on MATH-500 while improving accuracy by 9-16 points absolute. On AIME 2024, the 14B model gains 10 points with 41% compression. The secret? Much of what reasoning models produce is not just redundant-it is actively harmful, compounding errors with every unnecessary token. Code is available at https://github.com/HJSang/OPSD_Reasoning_Compression.

The landscape of AI coding assistance is undergoing a fundamental shift from complex IDE plugins to versatile, terminal-native agents. Operating directly where developers manage source control, execute builds, and deploy environments, CLI-based agents offer unprecedented autonomy for long-horizon development tasks. In this paper, we present OPENDEV, an open-source, command-line coding agent engineered specifically for this new paradigm. Effective autonomous assistance requires strict safety controls and highly efficient context management to prevent context bloat and reasoning degradation. OPENDEV overcomes these challenges through a compound AI system architecture with workload-specialized model routing, a dual-agent architecture separating planning from execution, lazy tool discovery, and adaptive context compaction that progressively reduces older observations. Furthermore, it employs an automated memory system to accumulate project-specific knowledge across sessions and counteracts instruction fade-out through event-driven system reminders. By enforcing explicit reasoning phases and prioritizing context efficiency, OPENDEV provides a secure, extensible foundation for terminal-first AI assistance, offering a blueprint for robust autonomous software engineering.

Recent advances in large language models (LLMs) have enabled agentic systems for sequential decision-making. Such agents must perceive their environment, reason across multiple time steps, and take actions that optimize long-term objectives. However, existing web agents struggle on complex, long-horizon tasks due to limited in-context memory for tracking history, weak planning abilities, and greedy behaviors that lead to premature termination. To address these challenges, we propose STRUCTUREDAGENT, a hierarchical planning framework with two core components: (1) an online hierarchical planner that uses dynamic AND/OR trees for efficient search and (2) a structured memory module that tracks and maintains candidate solutions to improve constraint satisfaction in information-seeking tasks. The framework also produces interpretable hierarchical plans, enabling easier debugging and facilitating human intervention when needed. Our results on WebVoyager, WebArena, and custom shopping benchmarks show that STRUCTUREDAGENT improves performance on long-horizon web-browsing tasks compared to standard LLM-based agents.

Multimodal Large Language Models (MLLMs) that directly process RGB inputs for tasks like 3D localization and navigation have shown remarkable potential. However, we argue that these RGB-only approaches are fundamentally flawed in their ability to generalize across cameras. By ignoring camera parameters, they entangle an object's physical properties with the camera's perspective, creating an irresolvable ambiguity. We show this leads MLLMs to overfit to the training camera distribution, rather than learning true and generalizable 3D geometric principles. To address this, we propose Camera-Aware MLLM framework for spatial MLLMs. It learns generalizable spatial reasoning by: (i) injecting camera intrinsics via a dense embedding that conditions each visual token; (ii) introducing a camera-aware data augmentation strategy that synthetically varies camera parameters, forcing the model to disentangle camera properties from scene content; and (iii) distilling geometric priors from a 3D vision foundation model. Extensive experiments demonstrate that camera-aware MLLMs substantially outperform their naive counterparts, particularly in cross-camera generalization tests on spatially-grounded tasks, indicating that camera-awareness is not only beneficial but also a prerequisite for robust and generalizable spatial intelligence in MLLMs.

Personal AI agents face a fundamental deployment paradox on mobile: persistent background execution drains battery and violates platform sandboxing policies, yet purely reactive agents miss time-sensitive obligations until the user remembers to ask. We present Jagarin, a three-layer architecture that resolves this paradox through structured hibernation and demand-driven wake. The first layer, DAWN (Duty-Aware Wake Network), is an on-device heuristic engine that computes a composite urgency score from four signals: duty-typed optimal action windows, user behavioral engagement prediction, opportunity cost of inaction, and cross-duty batch resonance. It uses adaptive per-user thresholds to decide when a sleeping agent should nudge or escalate. The second layer, ARIA (Agent Relay Identity Architecture), is a commercial email identity proxy that routes the full commercial inbox -- obligations, promotional offers, loyalty rewards, and platform updates -- to appropriate DAWN handlers by message category, eliminating cold-start and removing manual data entry. The third layer, ACE (Agent-Centric Exchange), is a protocol framework for direct machine-readable communication from institutions to personal agents, replacing human-targeted email as the canonical channel. Together, these three layers form a complete stack from institutional signal to on-device action, without persistent cloud state, continuous background execution, or privacy compromise. A working Flutter prototype is demonstrated on Android, combining all three layers with an ephemeral cloud agent invoked only on user-initiated escalation.

Student Personas (SPs) are emerging as infrastructure for educational LLMs, yet prior work often relies on ad-hoc prompting or hand-crafted profiles with limited control over educational theory and population distributions. We formalize this as Theory-Aligned and Distribution-Controllable Persona Generation (TAD-PG) and introduce HACHIMI, a multi-agent Propose-Validate-Revise framework that generates theory-aligned, quota-controlled personas. HACHIMI factorizes each persona into a theory-anchored educational schema, enforces developmental and psychological constraints via a neuro-symbolic validator, and combines stratified sampling with semantic deduplication to reduce mode collapse. The resulting HACHIMI-1M corpus comprises 1 million personas for Grades 1-12. Intrinsic evaluation shows near-perfect schema validity, accurate quotas, and substantial diversity, while external evaluation instantiates personas as student agents answering CEPS and PISA 2022 surveys; across 16 cohorts, math and curiosity/growth constructs align strongly between humans and agents, whereas classroom-climate and well-being constructs are only moderately aligned, revealing a fidelity gradient. All personas are generated with Qwen2.5-72B, and HACHIMI provides a standardized synthetic student population for group-level benchmarking and social-science simulations. Resources available at https://github.com/ZeroLoss-Lab/HACHIMI

Vision--language models (VLMs) process images as visual tokens, yet their intermediate reasoning is often carried out in text, which can be suboptimal for visually grounded radiology tasks. Radiologists instead diagnose via sequential visual search; eye-tracking captures this process as time-ordered gaze trajectories that reveal how evidence is acquired over time. We use eye-gaze as supervision to guide VLM reasoning by introducing a small set of dedicated gaze tokens. These tokens are trained to predict gaze-selected image patch indices in temporal order, encouraging the model to follow human-like evidence acquisition and integration. Experiments on MIMIC-EYE and multiple external zero-shot benchmarks show consistent gains over baselines, achieving state-of-the-art in-domain performance and improved out-of-domain robustness. These results highlight temporally ordered gaze as an effective supervision signal for learning visually grounded medical reasoning.

We investigate several aspects of the thermodynamic geometry for a quantum fluid with square-well interactions using a third-order perturbation theory framework based on the path-integral-necklace analogy. A comparison is made between the thermodynamic and geometric properties of the quantum fluid and its classical counterpart for the interaction ranges $λ^{*}= 1.3$, 1.5, and 1.7. In particular, we analyze the scalar curvature behavior, criticality, and the corresponding Widom lines derived from curvature and several thermodynamic response functions. Quantum effects are shown to smooth supercritical anomalies of the scalar curvature and to shift its extrema for short-range interactions, while leaving the critical exponents of both the curvature and its heat capacity consistent with mean-field predictions. Widom lines associated with temperature-dependent response functions and with the curvature scalar exhibit pronounced classical-quantum differences for short interaction ranges; in contrast, those derived from the isothermal compressibility exhibit only minor variations. Overall, these results highlight the sensitivity of geometric information of thermodynamic systems due to quantum effects and the crucial role of the interaction range in shaping supercritical thermodynamic behavior.

Nonlinear interactions between vibrational modes play a crucial role in understanding the dynamical response of nanomechanical resonators. Here, we report the experimental observation and theoretical modeling of nonlinear mode coupling in a high-stress square silicon nitride membrane resonator. We quantify frequency shifts of the fundamental mode arising from tension-mediated geometric nonlinearity by increasing the amplitude of the fundamental mode and higher-order flexural modes. A quantitative theoretical framework based on Kirchhoff-Love plate theory is developed, which incorporates both intrinsic Duffing nonlinearity and nonlinear intermodal coupling and shows good agreement with experimental measurements for the (1,1)-(2,1) and (1,1)-(2,2) mode pairs. We further compute the nonlinear coupling matrix across mode families, revealing the role of mode symmetry and spatial overlap in governing intermodal interactions. These results establish nonlinear mode coupling as a controllable resource for multimode frequency tuning and mechanical transduction.

We experimentally investigate the transmission of light by dense atomic vapor. The light propagating in dense atomic vapor can be modeled as a Lévy flight random walk. For such system, the step-length distribution can be modeled as $P(\ell)\sim \ell^{-1-α(\ell)}$, with the Lévy index $α(\ell)$ varying smoothly with the step length. Moreover, the walkers alternate between two distinct distributions depending on the occurrence of collisions between atoms in the light scattering. We obtain the Lévy index from transmission measurements for different system sizes and atomic densities. The measured Lévy index is determined by the system size $α=α(\ell=L)$. Simulations are made for walkers alternating between two Lévy like step-length

The rheological properties of biological tissues are core to processes such as cancer metastasis, wound healing and embryo development. The emergence of tissue and organ structures during morphogenesis requires the precise formation of spatial patterns. Dating back to Turing, pattern formation has been suggested to arise in tissues via spontaneous symmetry breaking instabilities in the concentration field of chemical morphogens. Within the vertex model of tissue mechanics, we show that spontaneous symmetry breaking may also arise via a mechanical instability in the strain field of a deformed tissue, leading to a patterned coexistence of fluid and solid regions, with a strong localisation of the strain into shear bands. The nature of the bands differs between tissues in which internal cell-cell dissipation dominates external drag against a substrate, and vice versa.

The spherical grasshopper problem is a geometric optimization problem that arises in the context of Bell inequalities and can be interpreted as identifying the best local hidden variable approximation to quantum singlet correlations for measurements along random axes separated by a fixed angle. In a parallel publication [arXiv:2504.20953], we presented numerical solutions for this problem and explained their significance for singlet simulation and testing. In this companion paper, we describe in detail the geometric and computational framework underlying these results. We examine the role of spherical discretization and compare three natural variants of the problem: antipodal complementary lawns, antipodal independent lawns, and non-antipodal complementary lawns. We analyze the geometric structure of the corresponding optimal lawn configurations and interpret it in terms of a spherical harmonics expansion. We also discuss connections to other physical models and to classical problems in geometric probability.

In electric dipole spin resonance (EDSR), a single spin is electrically driven in the field gradient produced by a micromagnet. While EDSR has enabled high fidelity gate operations in many devices, there are reports of unexpected non-linearities in the Rabi frequency as a function of microwave drive amplitude. We carefully measure the response of Loss-DiVincenzo (LD) single spin qubits to resonant drives as well as simultaneous resonant and off-resonant drives, as would be encountered in a realistic quantum processor. With the microwave amplitude carefully calibrated, we find that the Rabi frequency scales linearly with drive amplitude, even when all three spins are driven simultaneously. We also determine that heating-induced resonance frequency shifts from off-resonant drives are comparable to typical temporal drifts. Our results indicate that the previously observed nonlinear response is not a general feature of LD spin qubits.

This work reports the successful formation of Ag and Au nanoparticles in $β$-Ga$_2$O$_3$ single-crystals by ion implantation and annealing at 550 °C. X-ray diffraction measurements revealed that nanoparticles were formed after the annealing step, presenting a highly-ordered crystalline structure and conforming to a crystallographic relation with respect to the matrix: $\left(0\overline{1}0\right)_β\parallel \left(110\right)_{\mathrm{Ag/Au}}$ and $\left[102\right]_β\parallel \left[1\overline{1}2\right]_{\mathrm{Ag/Au}}$. The presence of these nanoparticles was also confirmed via absorbance measurements revealing the localised surface plasmon resonance peaks associated with these particles. Considering the multiple advantages and the versatility of metallic nanoparticles, their combination with the exceptional properties of $β$-Ga$_2$O$_3$ paves the way for a wide range of applications.

Amorphous glass-forming polymers exhibit multiple relaxation processes, including the structural α-relaxation associated with the glass transition and faster secondary relaxations that typically follow Arrhenius behavior. Recently, a distinct slow Arrhenius process (SAP) has been observed at frequencies well below the α-process. Although Arrhenian in its temperature dependence, the SAP involves much longer relaxation times and its microscopic origin remains unclear. Here, we extend the two-state, two-timescale (TS2) theory to describe both the α-relaxation and the SAP within a unified framework. We propose that the SAP represents the high-temperature limit of an α-like process in a coarse-grained fluid of dynamically correlated clusters. With renormalized interaction energies and coordination parameters, the same model quantitatively reproduces both α and SAP data across multiple polymers without additional adjustable parameters and explains the observed Meyer-Neldel compensation behavior. The theory further predicts that the SAP should deviate from Arrhenius behavior at sufficiently low temperatures, transitioning to Vogel-Fulcher-Tammann-Hesse-like dynamics, thereby offering a physically transparent interpretation of cluster-scale relaxation in glass-forming polymers.

Recent advances in nano-thermometry motivate the extension of the Landauer-Büttiker scattering theory as to include the non-local dissipation associated with charge transport. Such a program is implemented by describing the inelastic scattering in the connecting electrodes within an electrostatically self-consistent scheme. The restriction to quasi-one-dimensional geometries, weak excitation and low temperature allows to obtain general expressions of the current density and the dissipated power, valid in different regimes, for the cases of an energy-independent mean-free-path or an energy-independent relaxation-rate. In particular, the dissipation asymmetry at both sides of a nano-device and the conditions for observing heating spots with a local maximum of the dissipated power are formulated in terms of the key parameters that define the nano-device and its environment.

Navigation in turbulent environments is a fundamental challenge for biological and artificial microswimmers. While most existing studies focus on adapting motility or steering, the role of active morphological changes in navigation remains poorly explored. Here, we investigate a shape-changing spheroidal microswimmer tasked with maximising its displacement from an initial position in two-dimensional stochastic and turbulent flows. Using reinforcement learning (RL), the microswimmer learns to adapt its aspect ratio based on its orientation and local velocity-gradient signals. The learned strategies outperform fixed-shape and short-time-optimal baselines across different flow regimes and remain effective when transferred from stochastic flows to fully resolved turbulence. Guided by the learned policies, we propose a minimal analytical model that captures the essential navigation mechanisms and reproduces the performance across flow regimes. These results show that adaptive morphology provides a robust and physically interpretable control paradigm for microswimmer navigation in complex flows.

High resolution image sensors require electrical access to each individual pixel for signal readout. Such access is especially challenging for ultra-miniaturized pixels, for heterogeneously integrated sensing and readout layers in long-wavelength detectors, and for novel light-sensing materials with unestablished integration to silicon chips. Here, we introduce and experimentally validate a novel imaging approach that does not require electrical connections to individual pixels. The sensor matrix involves photoresistive pixels connected neighbor-to-neighbor and packed into a rectangular lattice. The signal readout is based on electrical impedance tomography applied to the photoresistance: the photovoltage is measured at the matrix boundary at various positions of injected bias current, and the image is reconstructed algorithmically. We present experimental validations for moderate-size infrared imagers based on multilayer graphene (24 pixels) and amorphous vanadium oxide (264 pixels). The reconstruction procedure is mathematically stable, sustainable to variations of pixel resistivity and photosensitivity, and its complexity is that of linear system solution. The proposed method enables unprecedented architecture simplification of imaging devices.

Rhombohedral multilayer graphene (RMG) proximity-coupled to a Haldane substrate provides a platform to investigate the interplay between band topology, layer number, and electric-field control of orbital magnetism. Using a tight-binding model and the modern theory of orbital magnetization, we study the layer-dependent magnetic response in bilayer, trilayer, and tetralayer graphene under Haldane proximity. While monolayer graphene develops a global topological gap with quantized magnetization slope, multilayer systems remain metallic due to protected low-energy bands associated with unperturbed sublattices. Despite the absence of a global gap, finite valley-contrasting Berry curvature produces non-trivial layer-dependent Chern numbers. We decompose the total orbital magnetization into self-rotation ($M_{\mathrm{SR}}$) and center-of-mass ($M_C$) contributions, revealing their distinct behaviors across doping and applied interlayer bias. In bilayer graphene, magnetization remains negative and monotonic. Remarkably, trilayer and tetralayer graphene display a bias-induced sign reversal of orbital magnetization beyond critical thresholds ($Δ\simeq -55$ meV for 3LG, $-50$ meV for 4LG) in the hole-doped regime, a feature completely absent in the bilayer. The effect persists across both hole and electron doping, demonstrating that layer count serves as a key tuning parameter for orbital magnetism. Our findings establish topologically proximitized multilayer graphene as a versatile platform for electric-field-manipulable orbitronic and valleytronic devices.

Probing the spatially heterogeneous activity across scales is a major challenge in living matter. Energy injection at diverse length scales leads to mode coupling, inter-modal energy transfer, and entropy production. We demonstrate the emergence of multiple effective (``active'') temperatures in nonequilibrium molecular- and Brownian-dynamics simulations of an active polymer. Via a generalised Langevin equation for a labelled monomer we identify spectral noise temperatures and their relation to the underlying activity landscape. A harmonic trap of variable stiffness can serve as a minimally invasive prototypical spectroscopic device to selectively scan through the emergent effective temperatures and thereby resolve the scale-dependent activity.

We introduce the Weighted Planar Stochastic Porous Lattice (WPSPL), a geometrically disordered substrate generated by iteratively subdividing a unit square. At each step a block is selected with probability proportional to its area, divided into four parts, and one sub-block is retained (removed) with probability $q$ ($1-q$). We show analytically that the WPSPL exhibits multifractality for each of its infinitely many nontrivial conserved quantities and demonstrate numerically that its snapshots at different times are statistically self-similar. The dual of the lattice forms a complex network with a power-law degree distribution. Motivated by these properties of this porous lattice, we study bond percolation on the WPSPL, determine the percolation threshold, and estimate the critical exponents $α$, $β$, and $γ$ associated with the specific heat, order parameter, and susceptibility, respectively. The exponents vary continuously with $q$, reflecting a family of distinct universality classes as the global dimension of the lattice depends on $q$. Remarkably, the Rushbrooke inequality, $α+ 2β+ γ\ge 2$, is satisfied in near equality. Notably, the nonporous case ($q=1$) has a global dimension $2$ but lies outside the universality class of conventional two-dimensional lattices. Our results highlight how geometric disorder, multifractality, scale-free coordination number disorder, and porosity produce unconventional critical behavior.

High-intensity THz laser pulses enable the light-mediated control of lattice vibrations by resonantly driving selected phonon modes. On ultrafast timescales, memory effects influence the phonon dynamics and must be accounted for to describe the heat production associated with energy dissipation. Here, we establish a microscopic framework for non-Markovian phonon dynamics by deriving the noise and dissipation kernels governing a driven phonon mode. Using large-scale molecular dynamics simulations, we reconstruct these kernels directly from the many-body lattice dynamics and determine the corresponding heat production rate. Our results provide a quantitative picture of the crossover between Markovian and non-Markovian dynamics on picosecond timescales and show how the finite bandwidth of the driving field limits the dynamically relevant bath spectrum. Furthermore, we demonstrate that thermodynamic quantities such as heat production can be inferred directly from the dynamics of an individual phonon mode, enabling their experimental measurement using time-resolved spectroscopy.

Persistent spin helices are a manifestation of symmetry-protected spin textures in systems with balanced spin-orbit coupling. They enable long-lived spin structures that are of interest for spintronics and coherent spin manipulation. The quantum metric has recently emerged as a promising tool for characterizing the geometric structure of quantum states. Here, we demonstrate that the quantum metric provides a sensitive geometric probe of the persistent spin helix. Within the Rashba-Dresselhaus Hamiltonian, we analytically evaluate the quantum metric components and uncover a divergent geometric contribution that emerges precisely at the persistent spin helix condition. We reveal that this divergence originates from a hidden line degeneracy that forms when the strengths of Rashba and Dresselhaus spin-orbit coupling become equal. We further study the role of higher-order cubic spin-orbit interactions and determine how these corrections regularize the geometric response and control the scaling behavior of the quantum metric. Our results establish quantum geometry as a powerful framework for identifying and characterizing persistent spin helices and related symmetry-protected spin textures.

Resolving sub-10 nm spin switching and the associated terahertz (THz) electrodynamics during the colossal magnetoresistance (CMR) transition is a definitive frontier in reaching the fundamental spatial, temporal, and energy-dissipation limits of spin-based microelectronics and quantum logic architectures. Yet, the requirement of simultaneous control of high magnetic field, cryogenic environment, and nanometer-scale resolution has remained an elusive benchmark for terahertz nanoscopy, leaving the obscured nano-scale high-frequency dynamics of these transitions largely unexplored. Here, we overcome these limitations by utilizing a custom-built cryogenic magneto-THz scattering-type scanning near-field optical microscopy (cm-THz-sSNOM) platform to resolve the nanoscale, THz spectroscopic evolution of the magnetic field-driven CMR transition in a manganite single crystal $\text{Pr}_{2/3}\text{Ca}_{1/3}\text{MnO}_{3}$. Our measurements provide a real-space visualization of the local THz conductivity, capturing the moment that magnetic-field-induced spin switching triggers the phase transition from an antiferromagnetic insulator to a ferromagnetic metal. THz nano-imaging, together with an ellipsoidal near-field model, reveals a multi-scale transition initiated by 1-2 nm isolated spin-flip sites at low magnetic fields, which coalesce into $\sim$15~nm conducting regions as the threshold field is approached. These results provide an in situ, previously inaccessible THz real-space view of CMR switching, establishing a general analysis framework for mapping spin-charge-lattice-orbit-coupled dynamics at spatial scales that transcend the nominal sSNOM resolution.

We study band modulations and topological transitions in a one-dimensional periodic bead-on-string chain. Using an exact transfer-matrix formulation of the wave equation with periodically modulated mass density, combined with numerical spectral searches and tabletop experiments, we characterize band gaps and localized midgap states. We interpret these states by mapping the system to the Su-Schrieffer-Heeger (SSH) model and its low-energy (1+1)-dimensional Dirac theory. This framework reveals that the robust states are topological solitons bound to boundaries or engineered domain walls in the Dirac mass. Through this mapping, we provide an intuitive account of how band structure controls topological phase changes in mechanically realizable lattices.

Large-scale quantum computing requires cryogenic electronic controllers such as control/readout circuit and routing circuit. However, current technologies face high power dissipation problems, hindering large-scale qubit integration. Here, we theoretically propose extremely low-power cryogenic topological transistors, i.e., negative-capacitance topological insulator field-effect transistors (NC-TIFETs). By combining a gate-field-induced two-dimensional 1T'-Molybdenum Disulfide (MoS$_2$) topological channel with a hafnium-zirconium oxide (HZO) ferroelectric gate insulator, NC-TIFETs exhibit an extremely steep-slope transfer curve and ultra-high transconductance at low drain voltage ($V_{\mathrm{D}}$). Therefore, NC-TIFETs are the compelling candidate for minimizing power dissipation in the cryogenic electronic interfaces essential for large-scale quantum computing systems.

For the Hill equation describing one-dimensional periodic systems, a constructive formulation is developed for generating Floquet-Bloch states directly from arbitrary pairs of linearly independent solutions. One-dimensional photonic crystals are used as a concrete illustration. Explicit closed-form formulas map an arbitrary fundamental system to the corresponding Floquet-Bloch basis via the monodromy matrix, including the generic Jordan band-edge case, without reliance on canonically normalized solutions. The construction can be expressed directly in terms of the transfer matrix, making the residual representation freedom transparent and providing an implementation-ready framework for analytical and numerical studies of periodic systems.

We investigate how cyclic loading evolves the structure and deformation behaviors of a granular raft composed of particles floating at an air-oil interface. The raft has a disordered particle packing structure, and is cohesive due to capillary interactions between particles. Under uniaxial cyclic loading with a small strain amplitude, the raft's packing structure experiences an aging process characterized by logarithmically increasing packing fraction and decreasing structural heterogeneity. The observed structural change is due to particle dynamics that are organized around morphologically evolving voids in the raft. The raft is then subjected to quasi-static tension or compression tests until failure. In comparison with non-aged rafts, the rafts that experienced cyclic loading show a higher strength, higher stiffness, and lower ductility, along with qualitatively different features, such as a stress overshoot in the loading curve.

We develop a hydrodynamic framework for the interactions and collective dynamics of force dipoles embedded in a compressible fluid membrane supported by a shallow viscous subphase. Starting from the generalized two-dimensional Stokes equations with shear, dilatational, and odd (Hall) viscosities, we derive an exact real-space Green tensor using Hankel transforms. The resulting tensor is characterized by three hydrodynamic screening scales associated with shear, compressional, and odd-viscous modes, and smoothly reduces to the standard limiting cases of incompressible membranes and compressible parity-symmetric membranes, while also capturing the chiral response generated by odd viscosity. Using this Green tensor we obtain the velocity and vorticity fields generated by a force dipole and formulate the dynamical system governing interacting dipoles. The analysis reveals several distinct dynamical regimes and identifies observables that isolate the antisymmetric odd-viscous contribution to dipole interactions, including transverse drift and chiral relative motion.

Single-electron spin qubits defined in quantum dots are used as building blocks of a semiconductor-based quantum computer. Readout in a scaled-up version of such a quantum computer is expected to rely on the Pauli Spin Blockade (PSB) mechanism. A desired functionality of PSB readout is that it reveals two bits of information on the two spin qubits that are involved in the process, such that the four computational basis states can be discriminated. In this work, we propose and quantitatively analyze an experimental procedure, based on gate reflectometry, which enables this four-state discrimination in a single measurement. We provide an intuitive recipe to maximize the contrast between the quantum capacitances of the four basis states. Focusing on silicon double quantum dots equipped with a micromagnet, we quantify how amplifier noise and phonon-mediated relaxation influence readout fidelity. Our results highlight a realistic opportunity to mitigate the overhead of readout ancilla qubits in a spin-based quantum computer.

Efficient transport of spin and orbital moments, and their electrical detection, are among the main challenges in spintronics and orbitronics. In magnetic insulators, these currents are mediated by magnons. In addition to carrying spin and orbital moment, the orbital motion of a magnon combined with its magnetic moment, generates an effective electric dipole moment. Here, we develop a theoretical framework for Seebeck- and Nernst-type transport of the magnon orbital moment (MOM) and its associated electric dipole moment (EDM). We identify a Drude-like scattering contribution and an intrinsic component governed by the generalized Berry curvatures of magnon bands. We show that a measurable transverse voltage generated by the EDM current provides a direct electrical detection scheme for magnon orbital transport. Applying our theory to an hexagonal altermagnet, we obtain an experimentally accessible voltage of approximately $0.4~μ$V. Our results establish a concrete electrical probe of magnon orbital transport and highlight magnons as potential low-dissipation information carriers for orbitronics.

We consider the light scattering from a pair of point-like electrical dipoles. Whenever the polarizability of each dipole violates the optical theorem, the response of the pair (both in far-field and near-field) exhibits exact resonances as a function of the frequency and the inter-dipole distance. This polarizability is consistent with causality and the crossing condition (i.e., a real field generates a real response). Hence, the emergence of the resonances requires nonequilibrium conditions, e.g., corresponding to active dipoles. Within our approach (classical optics, monochromatic incident field, point-like dipoles), the exact resonances can be infinite. The resonances also appear in the equilibrium domain, where the optical theorem is valid. In that domain, they are finite, but can produce large amplification factors; e.g., for a pair of metallic nanoparticles under Drude's model, the single-particle plasmonic resonance can be amplified $\sim 10^{2}$ times. But the global maximization of the scattering can still be achieved by violating the optical theorem. Our results for one electric and one magnetic dipole show how resonances can amplify a weak magnetic response of a single dipole to the incident field. We also discuss an anti-resonance (dark-state) effect present in the two-dipole scattering.

Spectral control of closely spaced excitonic transitions is essential for valleytronic photonics, nanoscale light sources, and wavelength-encoded sensing. In monolayer molybdenum disulfide (MoS2), the A and B excitons are separated by only tens of meV, making spectral engineering both fundamentally important and technologically challenging. Here, we numerically investigate plasmon-enhanced excitonic emission in MoS2 coupled to vertically oriented hollow gold nanocylindrical cavities separated by a dielectric spacer. Finite-difference time-domain simulations combined with a photoluminescence-rate framework allow independent evaluation of excitation enhancement, radiative decay modification, non-radiative quenching, and excitonic charge generation. By tuning the cavity aspect ratio, the localized surface plasmon resonance is aligned with either the A or B excitonic transition, while spacer thickness and refractive index regulate near-field confinement and the local density of optical states. Under optimized conditions, excitation rates reach 4.34-fold enhancement while radiative decay exceeds 40-fold, producing photoluminescence increases of 143.85 and 87.27 times for the A and B excitons. The cavity also redistributes the relative intensities of the excitonic peaks, yielding normalized exciton peak ratios up to 2.4 compared to bare MoS2. These results establish hollow plasmonic nanocavities as a geometry-tunable platform for controlling excitonic emission and charge generation in atomically thin semiconductors.

A ring polymer in a confining space may exhibit at least two phases, namely an expanded (or solvent-rich phase) if its concentration is small, or a collapsed (or polymer-rich phase) when it is concentrated and compressed. These phases are discussed in reference \cite{deG79}, and have been modelled, traditionally, in the mean field using Flory-Huggins theory \cite{Flory42,Huggins42}. In three dimensions the ring polymer may also be knotted, or linked, and have its conformational degrees of freedom constrained by its topology. In a lattice model of confined knotted ring polymers there are indications that the thermodynamic properties of the ring polymer (for example, the osmotic pressure \cite{GJvR18,JvR19}) is a function of its topology. In this paper we explore a lattice knot model of a confined ring polymer as a function of its chemical potential. We show that a well-defined phase transition occurs between solvent-rich and polymer-rich phases when the lattice knot exhibits either the unknot topology or any other fixed knot type. Furthermore, we observe small yet significant variations in the free energy near the critical point when comparing trefoil knots with other non-trivial knot types. These findings indicate that the thermodynamic properties of confined ring polymers depend on their topological entanglement characteristics (namely, their knot type).

Improving the detailed understanding of the underlying properties and functions of biomolecules has recently attracted growing interest, enabled by the possibility of real-space imaging of single, intact macromolecules using Scanning Tunneling Microscopy (STM) in combination with electrospray ion beam deposition and soft landing. This combination provides key insights into biomolecular behavior, but it also imposes stringent requirements on rapid and reliable data analysis. A major limiting factor for applying machine learning to STM images is often the scarcity of training data, caused by the long acquisition times required for both experimental imaging and high-accuracy simulations. Here, we propose a framework for the rapid generation of three-dimensional structures of glycans, peptides, and glycopeptides and their corresponding STM-like image simulations, based on state-of-the-art, machine-learning-accelerated Density Functional Theory (DFT). We generate datasets for the polypeptide bradykinin and for a representative glycan molecule, and we train a conformer estimation model to predict a molecule's three-dimensional structure from an STM image. On synthetic data, our approach achieves high accuracy, with median atomic deviations below $2\,Å$ for peptides and below $4\,Å$ for glycans. Application to experimental data predominantly yields a precise, reliable, and visually convincing determination of the local positions of molecular subunits. The application to experimental data represents an important milestone towards a fully automated structural search pipeline for complex, biologically relevant systems imaged with STM.

The recent interest in non-Hermitian (NH) systems has significantly broadened their application across condensed matter physics, offering a unique framework to explore out-of-equilibrium phenomena. Simultaneously, altermagnets have emerged as a distinct magnetic class, characterized by unconventional spin-split bands protected by crystal symmetries. In this work, we investigate the interplay between non-Hermitian dynamics and spin transport in these materials, focusing on the Edelstein effect. We demonstrate that the introduction of non-Hermiticity in $d$-wave altermagnets and $p$-wave unconventional magnets opens novel susceptibility components that are inaccessible in Hermitian counterparts. Our analysis reveals that these susceptibility channels are highly sensitive to the underlying symmetry of the order parameter. Crucially, our results show that the non-conservative nature of the system leads to the selective gain and loss of specific spin components, a phenomenon that can be tuned by the interplay between dissipation and the altermagnetic order. These components exhibit a distinct gain/loss profile that depends strictly on the Néel vector orientation, providing a new route for manipulating spin degrees of freedom through controlled non-conservative processes in emerging magnetic materials

We present a covariant field-theoretical framework for a rank-4 tensor gauge field theory describing fractonic string-like objects. We show that the most general quadratic, parity-preserving action naturally leads to a Maxwell-like sector, with tensorial analogues of electric and magnetic fields, Maxwell-like equations, a conserved energy-momentum tensor, and a Lorentz-like force. Remarkably, the theory gives rise to fracton-like string excitations purely from symmetry principles: constraints on the motion of these extended objects appear as Gauss-like laws, without being imposed by hand. One of these laws is new and corresponds to a generalised dipole conservation for closed strings, restricting their mobility and defining a novel class of fractonic string-like excitations. Finally, we uncover a connection to linearised area-metric gravity: in a suitable limit, the theory reduces to known covariant fracton models with rank-2 gauge fields, highlighting a deep link between fractonic matter and gravity-like structures. This provides a unified perspective on higher-rank gauge fields, extended excitations, and emergent gravitational features.

We study the scalar curvature of the Fisher information metric on the microscopic coupling-parameter manifold of lattice spin models at criticality. For a $d$-dimensional lattice with periodic boundary conditions and $n = L^d$ sites, the Fisher manifold has $m = d \cdot n$ dimensions (one per bond), and we find $|\mathcal{R}(J_c)| \sim n^{d_R}$ with $d_R = (dν+ 2η)/(dν+ η)$, where $ν$ and $η$ are the correlation-length and anomalous-dimension critical exponents. For 2D Ising ($ν= 1$, $η= 1/4$), this predicts $d_R = 10/9$, confirmed by exact transfer-matrix computations ($L = 6$--$9$: $d_R = 1.1115 \pm 0.0002$) and multi-seed MCMC through $L = 24$. For 3D Ising ($ν= 0.630$, $η= 0.0363$), the prediction $d_R = 1.019$ is consistent with MCMC on $L^3$ tori up to $L = 10$ (power-law fit: $d_R = 1.040$). For 2D Potts $q = 3$ (predicted $33/29 \approx 1.138$), FFT-MCMC through $L = 40$ shows $d_\mathrm{eff}$ oscillating non-monotonically around $\sim 1.20$, consistent with $O(1/(\ln L)^2)$ logarithmic corrections. For $q = 4$ (predicted $22/19$), effective exponents oscillate with strong logarithmic corrections. The Ricci decomposition identity $R_3 = -R_1/2$, $R_4 = -R_2/2$ holds to 5--6 digits for all models. This exponent is distinct from Ruppeiner thermodynamic curvature and reflects the collective geometry of the growing Fisher manifold. We provide falsification criteria and predictions for additional universality classes.

We demonstrate that the finite-temperature critical transverse-field Ising chain exhibits quantitative signatures of black-hole physics in its dual gravitational description within the AdS/CFT correspondence. Its finite-temperature dynamics and thermodynamics are consistently captured by a mixed thermal-AdS/BTZ black hole saddle, leading to three mutually compatible observations. First, antipodal excitation transport collapses onto a universal temperature-dependent curve determined by the relative AdS and BTZ contributions to the gravitational partition function, reflecting horizon absorption. Second, in the high-temperature regime, the retarded response exhibits exponential relaxation governed by the lowest quasi-normal mode of the dual black hole. Third, the temperature derivative of the von Neumann entropy develops a pronounced minimum at a temperature consistent with the Hawking-Page transition. These results identify critical quantum spin chains as minimal and experimentally accessible platforms for probing dynamical and thermodynamic aspects of quantum black holes in controllable many-body systems.

We present a theoretical study of proximitized planar Josephson junctions (JJs) with Rashba spin-orbit coupling (SOC) subject to an in-plane magnetic field and demonstrate that the magneto-current-phase relation (magneto-CPR) provides a powerful and unified probe of their microscopic and topological properties. By analyzing the full phase and Zeeman-field dependence of the supercurrent, we show that magneto-CPR measurements allow one to reconstruct the ground-state phase that minimizes the system's free energy in the absence of phase bias. This reconstructed phase generally displays 0-pi-like transitions as a function of the Zeeman energy, and we demonstrate that the magnitudes of the associated phase jumps provide quantitative information about the Rashba SOC. We further show that the mixed phase-field response encoded in the magneto-CPR enables the extraction of the second mixed spin susceptibility, which serves as a sensitive diagnostic of gap closings and can be used to construct a superconducting topological phase diagram in terms of the relative topological gap. In addition, the magneto-CPR yields the field dependence of the forward and reverse critical currents, allowing one to characterize the Josephson diode effect and its connection to the Zeeman field, Rashba SOC, and junction transparency. Our results establish magneto-CPR measurements as a versatile spectroscopic tool that can be used to extract key system parameters and provide evidence of topological superconducting phases in planar JJs.

We study entanglement entropy (EE) in interacting quantum field theories (QFTs) at finite density. We argue that, in the limit of large subregions, the derivative of EE with respect to the size of the entangling region approaches the thermal entropy density, independently of microscopic details. We make this relation explicit using slab-shaped subregions, where the limiting behavior can be directly identified. At finite chemical potential, we show that EE satisfies thermodynamic response relations, including a generalized Maxwell relation linking chemical potential and charge density. We provide strong nonperturbative evidence for these statements in the three-dimensional $\operatorname{O}\left(4\right)$ model, and conjecture that they are generic features of continuum QFTs, establishing a two-way link between entanglement and thermodynamics that opens a route toward extracting the equation-of-state information from entanglement data.

Potential and mass barriers in graphene introduce electron scattering, modulating transmission probabilities. Complex multi-barrier setups allow electron transmission to be controlled with high precision, but have a huge design space of possible barrier geometries. This work presents a framework to optimize electronic transport in such systems using differential evolution algorithms. First, transfer matrix methods are employed to efficiently compute quantum transport through multi-barrier structures, before optimization is applied to find barrier configurations whose transmission profiles closely match a predefined target profile. To address the trade-off between the accuracy and complexity of resulting barrier configurations, regularization techniques are incorporated into the optimization process. Our approach demonstrates the potential for highly tunable electronic transport in graphene-based systems by exploiting evolution-inspired optimization techniques.

Most spherical viruses exhibit icosahedral symmetry, yet the growth of viral shells remains poorly understood due to the short lifetimes and broad size distribution of assembly intermediates. Classical nucleation theory has been widely applied to describe this process, but it treats the boundary of a growing shell as rigid and structureless. Here, we extend classical nucleation theory by incorporating thermal fluctuations of the capsid rim using both discrete and continuum descriptions. Allowing the rim of a partially formed capsid to undergo small geometric undulations, we show that these fluctuations generate an entropic contribution that renormalizes the effective line tension. As a result, rim fluctuations can either promote or hinder capsid closure, depending on the subunit-subunit binding free energy, temperature, and fluctuation amplitude. We find that fluctuations generally lower the nucleation barrier when the binding free energy is below a threshold value, while for sufficiently strong binding, they can instead raise the barrier by stabilizing incomplete capsids through a finite-size entropy penalty associated with rim closure. By moving beyond the idealized capillarity approximation, our results provide a controlled extension of classical nucleation theory that clarifies how boundary fluctuations influence capsid nucleation and growth.

It is shown that the low-energy theory of a $d$-wave altermagnet is characterized by anisotropic Dirac cones with up and down spins based on a recently proposed tight-binding model. Spin-selective perfect elliptic dichroism occurs in this system, where only up-spin or down-spin electrons are excited by elliptically polarized light. Then, we derive a formula for the third-order photocurrent induced by applying both elliptically polarized light and static electric field, which is described in terms of quantum metric and the Berry curvature. Based on it, we predict only up-spin polarized current is induced. It is the leading order of photocurrent because the second-order photocurrent such as the injection current and the shift current are prohibited due to inversion symmetry inherent to altermagnets. It is intriguing that nonzero photocurrent is induced only by the anisotropy of the Dirac cones. Our results will be useful for future photo-excited spintronics based on altermagnets.

Fractal electronic spectra arising from the competition between lattice periodicity and magnetic flux are a fundamental hallmark of two-dimensional quantum systems. While the spectral properties of Hofstadter butterflies are well documented, their thermodynamic response has remained remarkably unexplored. We present an original characterization of the electronic entropy $S_{e}$, and specific heat $C_{e}$, at half-filling, for square, honeycomb, and triangular lattices under a magnetic field. We demonstrate that these observables exhibit fast and slow magneto-thermo oscillations and pronounced magnetocaloric effects. We identify striking self-similarity in $S_e$ and $C_e$, tracing heart-shaped specific heat and tunnel-like entropy contours that repeat at specific lattice-dependent magnetic fluxes. Entropy minima at low temperatures play a remarkable role, acting as fingerprints for the butterfly spines, resolving the underlying fractal spectra. These findings may establish thermal measurements as high-resolution spectroscopic probes, providing a robust framework for recognizing fractal signatures through thermodynamics in diverse nanostructures.

We propose that scaling dimensions of d=3 conformal field theories can be studied on a system of qubits with near term quantum simulation platforms. Our proposal chooses couplings of quantum many-body problems on a polyhedral lattice at which the conformal state-operator correspondence can be observed most accurately in the spectrum. We validate our protocol on the Ising model where we extract the scaling dimensions of a number of scaling operators with a few percent accuracy from the spectrum of a system of just 20 qubits. The procedure makes only minor assumptions beyond general conformal invariance -- it may hence be applied widely. Requirements and challenges to realize this proposal on quantum computers are discussed. Our results demonstrate that for current or near term qubit platforms, three dimensional conformal field theories present a unique opportunity -- a forefront problem that is difficult on classical computers but may be solved through quantum simulation.

Monolayer semiconductors hold great potential for nanoscale electronics, optoelectronics, and photonics. Excitons dominate their optical properties. As their electric fields extend outside the monolayer, they are sensitive to their surroundings. Thus, disorder can cause exciton instability, which is detrimental to device performance and critical for scalability and reproducibility. Here, we adapt a super-resolution fluorescence fluctuation microscopy technique to image localized exciton fluctuations in monolayer semiconductors, allowing us to identify unstable spots in an otherwise continuous monolayer with constant fluorescence. These spots correspond to interfacial disorder measured by atomic force microscopy. We examine how different material interfaces influence the fluctuations by comparing several substrates and provide additional insight into the disorder behind the fluctuations using hyperspectral imaging. We also assess the reduction of disorder upon thermal annealing, evidenced by a decrease in fluctuations. Our results show that fluorescence fluctuation imaging can detect disorder features similar to those of atomic force microscopy and hyperspectral imaging, while being faster and easier to implement. Therefore, it is a promising method for evaluating the quality of monolayer semiconductors, particularly when integrated with nanostructures and heterostructures found in nano-optoelectronic devices.

Active stresses in biological cells and tissues drive many developmental processes. However, increasing experimental evidence suggests that additional mechanical interactions with surrounding material can play a crucial role in guiding these processes. We introduce a minimal model of this scenario and investigate how pattern formation in an active material can be controlled by an inhomogeneous environment. Specifically, we consider an active fluid in which a chemical species regulates local active stresses and is redistributed by the resulting flows. We show that active stress patterns within such a fluid exhibit frictiotaxis and systematically characterize how inhomogeneous external friction affects mechanochemical pattern formation instabilities. We find that hydrodynamic screening plays a crucial role in mediating the cross-talk between friction patterns and active fluid self-organization and identify a mechanochemical frustration mechanism that gives rise to pattern oscillations caused by inhomogeneous friction.

We study the self-assembly of colloidal microgel particles at a quasi-two-dimensional air-water interface of a drying droplet. Using bright-field microscopy, we demonstrate that increasing particle elasticity drives interfacial organization from repulsion-stabilized crystallization to attraction-dominated gelation, via diverse metastable structures including clusters, voids and anisotropic aggregates. Molecular dynamics simulations using an effective potential that captures the interplay between hydrophobic, capillary, steric and dipolar interactions, reproduce the overall phenomenology of the observed colloidal morphogenesis. Our findings establish particle elasticity as a key parameter governing non-equilibrium structural organization of colloids at an interface.

Trust is one of the cornerstones of human society. One of the evolutionary pressure mechanisms that may have led to its emergence is the presence of incentives for trustworthy behavior. However, this type of reward has received relatively little attention in the context of spatial trust games, which are often used to build models in evolutionary game theory. To fill this gap, we introduce an inter-role reward mechanism in the spatial trust game, so that an investing trustor can choose to pay an extra cost to reward a trustworthy trustee. With extensive numerical simulations, we find that this type of reward does not always promote trust. Rather, while moderate rewards break the dominance of mistrust, thereby favoring investment, excessive rewards eventually stimulate a nonreturn strategy, ultimately suppressing the evolution of trust. Additionally, lower reward costs do not necessarily promote trust. Instead, more costly, but not excessive, rewards enhance the advantage of the original investment, consolidating the clusters of rewarders and improving trust. Our model thus provides evidence about the counterintuitive nature of the relationship between trust and rewards in a complex society.

In the context of hybrid superconductor-semiconductor systems, Josephson junctions based on two-dimensional materials, such as graphene, offer promising opportunities because of their scalability and gate-tunable electronic properties. In this work, we investigate the inductive coupling between a quantum LC resonator and a superconducting loop embedding a short, ballistic, planar Josephson junction, with the graphene-based case as a representative example. Within a mean-field formalism, we analyze how the properties of the global system depend on the light-matter interaction coupling, the Fermi level of the two-dimensional material, and temperature. Our findings reveal that the current-phase relation can show features indicative of spontaneous time-reversal symmetry breaking. Furthermore, starting from the mean-field theory, we determine the low-energy spectrum of collective hybridized light-matter excitations.

Collective motion in animals and cells often exhibits rapid reorientations and scale-free velocity correlations. This allows information to spread rapidly through the group, allowing an adequate collective response to environmental changes and threats such as predators. To explain this phenomenon, we introduce a simple, biologically plausible mechanism: a minority-triggered reorientation rule. When local order is high, agents sometimes follow a strongly deviating neighbor instead of the majority. This rule qualitatively changes the macroscopic system behavior compared to traditional flocking models, as it generates heavy-tailed cascades of reorientations over broad parameter ranges. Our mechanism preserves cohesion while markedly enhancing collective responsiveness because localized directional cues elicit amplified group-level reorientation. Our results provide a parsimonious, biologically interpretable route to critical-like fluctuations and high responsiveness during flocking.

Collective temporal organization in complex systems is commonly attributed to synchronization, resonance, or proximity to dynamical instabilities. Here we identify a distinct mechanism by which coherent, synchronization-like behavior can emerge in stochastic systems that are linearly stable and contain no intrinsic oscillators. The mechanism arises from non-normal pseudospectral amplification and leads to what we term pseudo-coherence: an intermittent form of collective organization characterized by transient phase alignment, broken time-reversal symmetry, positive entropy production, and drifting spectral peaks. Using a minimal overdamped stochastic model, we show that increasing non-normality drives a sharp pseudo-critical transition. Beyond a well-defined threshold, fluctuations concentrate along a dominant reaction mode, generating intermittent growth of Kuramoto-like order parameters and irreversible probability currents without eigenvalue crossings or Hopf bifurcations. Analytically, we demonstrate that pseudo-critical non-normal dynamics reshapes the imaginary pseudospectrum, amplifying slow fluctuations and producing coherent frequency bands under finite-time observation. These results identify pseudo-coherence as a new route to collective temporal organization in non-equilibrium systems, suggesting that apparent rhythms and synchronization in natural systems may arise from non-normal stochastic amplification rather than intrinsic oscillators.

Here we develop and test a formalism for reversible ionic aggregation kinetics in an example concentrated electrolyte. Specifically, building on previous equilibrium work of McEldrew and co-workers in the context of concentrated electrolytes, and non-equilibrium properties of thermoreversible polymers and patchy particle systems, we develop the formalism for how ionic associations in electrolytes change subject to a step-change in conditions. This is achieved through solving a macroscopic rate equation of open/occupied association sites, which is a solution of the reversible Smoluchowski aggregation equation. We compare the derived equations against atomistic molecular dynamics simulations of a salt-in-ionic liquid. Good qualitative agreement is obtained, but quantitative differences are found, which highlights the multiple time scales of the associations that exist in concentrated electrolytes. We hope this formalism acts as the starting point for investigating these properties in other electrolytes, and developing it further to investigate the non-Newtonian behaviour of concentrated electrolytes, double layer charging, and the slow dynamics of these electrolytes in confinement.

We extend the statistical field-theoretical framework of rough surface contact mechanics to characterize the interfacial gap between an elastic half-space and a randomly rough surface incorporating exponential repulsion. Building upon a cumulant expansion to second-order, we derive an explicit analytical relation between the mean gap and the applied normal pressure. This result provides a closed-form expression for the drift and diffusion coefficients in a convection-diffusion equation governing the scale-dependent evolution of the gap distribution. Solving this equation with appropriate initial and boundary conditions yields the gap distribution under varying external pressures. Both the mean gap and the gap distribution are found to be in good agreement with Green's function molecular dynamics (GFMD) simulations. Our results demonstrate that the field-theoretical approach enables quantitative predictions not only for contact stresses but also for interfacial gap in rough surface contacts.

The localization transition in the Hermitian Aubry-André model is known to have a clear classical origin, with the critical point being exactly predictable from an analysis of classical phase-space trajectories. Motivated by this correspondence, we investigate whether a similar classical origin exists for localization transitions in non-Hermitian quasiperiodic Hamiltonians. Using semiclassical Husimi dynamics together with a detailed phase-space stability analysis, we show that localization transitions persist even in the semiclassical limit of such non-Hermitian models. However, in sharp contrast to the Hermitian Aubry-André case, the transition point inferred from classical phase-space analysis does not coincide with the quantum critical point. Instead, we find that the semiclassical transition depends sensitively on the choice of the irrational parameter defining the quasiperiodic potential, indicating the absence of a universal classical-quantum correspondence for the localization transition in the non-Hermitian setting. Nonetheless, we identify a suitable parameter regime in which the classical dynamics can faithfully mimic the quantum dynamics over a finite but appreciable time window.

We study the dynamic equilibrium between trions and excitons in monolayers of transition metal dichalcogenides in the presence of resident charge carriers and continuous photogeneration of excitons. We show that heating of the system via Drude absorption of low-frequency radiation leads to bistability of the steady-state equilibrium. The first is a low-temperature state, in which almost all resident charge carriers are bound into trions. The second state occurs at high temperatures, where most trions are dissociated; in this regime, the heating is more efficient due to the higher Drude conductivity of unbound charge carriers compared to trions. Switching between these two states occurs on a timescale of tens to hundreds of picoseconds and is accompanied by a jump in various observables such as temperature, current, and the intensity of exciton or trion luminescence.

Conventional heat engines require two reservoirs; their efficiency is bounded by the Carnot limit. In a well-defined theoretical framework where the working fluid undergoes phase change in the presence of asymmetric constraints, that limit does not apply: the appropriate description is constraint-reshaped entropy distributions $P_{\infty}(S;λ)$. We give \emph{one} complete \emph{theoretical} design: micro temperature difference (1--2\,$^\circ$C), subcooled liquid, liquid-only compression, and asymmetric constraint phase change (R134a). The cycle absorbs heat from a \textbf{single} heat source (the environment; $q_{\mathrm{in}} = 0.9\,\mathrm{kJ}/\mathrm{kg}$ per cycle), delivers net useful work ($w_{\mathrm{net}} = +0.514\,\mathrm{kJ}/\mathrm{kg}$), and uses only standard components. The analysis is rigorous: energy and mass balance are closed; all property data are from NIST and are reproducible at the cited URL. We show that, within this framework, this constitutes single-heat-source power generation.

With thermal issues becoming a major challenge to the development of integrated circuits (ICs), high-thermal-conductivity (high-TC) materials are gaining interest from both the industry and academia, especially for high-density back-end-of-line (BEOL) structures. Aluminum nitride (AlN) is an insulating material with high TC, suitable for thermal management in electronic devices. Furthermore, polycrystalline AlN thin films can be deposited at BEOL-compatible low temperatures (lower than 400 C) while retaining relatively high TC, rendering AlN a promising candidate for BEOL heat dissipation. Still, AlN deposition should aim at achieving high TC on a variety of substrates used in IC processes. In this work, we obtained 600- and 1200-nm BEOL-compatible AlN thin film samples on Si, SiO, SiN, and Al2O3 substrates for structural and thermal characterizations. Specifically, time-domain thermoreflectance (TDTR) measurements revealed consistently high TC (higher than 45 W m-1 K-1) on different substrates. Moreover, finite element analysis (FEA) simulations of AlN capped on top of an indium-tin-oxide (ITO) transistor showed a reduction of up to 44% in peak device temperature. Our work provided experimental and calculational evidence for the practicality of leveraging AlN as a high-TC, BEOL-compatible heat spreader material.

The local viscoelastic (VE) environment governs the motion of an embedded microsphere and consequently, pertinent dynamical phenomena. However, studying such phenomena with varying VE properties remains challenging for various reasons, including the strong coupling among the VE parameters and their dependence on experimental conditions, such as temperature. Here, we demonstrate the experimental realization of configurable VE media with broad variations, wherein the VE properties can be systematically and independently tuned, employing a dynamic optical trap. Specifically, the dynamics of a particle in a slowly diffusing optical trap provides the linear microrheological response of single-relaxation VE fluids, namely, Jeffreys or Maxwell-Voigt (MV) fluids, where the trap strength and its diffusion coefficient regulate the elastic response and the low-frequency viscosity, respectively. We validate this approach by comparing the experimentally observed dynamics of the trapped bead with those of a probe particle in real single-relaxation complex fluids, analytical predictions, and simulation results following harmonically bound Brownian particle with long-time diffusion model describing MV fluids. We extend the applicability of this scheme for realizing the microrheological response of double-relaxation VE media by incorporating appropriately correlated noise in the trap trajectory, signifying its validity for any linear VE media with multiple relaxations. Our scheme can be further extended to realize probe particle dynamics in an active VE environment, e.g., an entangled network of active polymers, by translating the trap along an active Brownian trajectory. Therefore, our scheme enables systematic microrheological studies in VE regimes that are otherwise challenging to realize or not readily accessible with real materials.

The emergence of nematic order on deformable closed surfaces plays a pivotal role in the morphogenesis of active biological matter, such as the regeneration of Hydra. In this work, we present a continuum model that couples the two-dimensional Landau-de Gennes order tensor, describing in-plane nematic ordering, with the mechanics of a mass-conserving, deformable spherical shell. By investigating the isotropic-to-nematic phase transition driven by a reduction in temperature, mimicking the natural induction of nematic order in actomyosin fibres, we perform both linear and weakly non-linear bifurcation analyses. The onset of nematic ordering spontaneously breaks spherical symmetry, yielding distinct equilibrium morphologies governed by the shell's deformability. Axisymmetric configurations, featuring two +1 defects at the poles, emerge via a discontinuous bifurcation, resulting in a globally stable prolate shape, alongside a metastable oblate shape. Non-axisymmetric configurations, featuring four +1/2 defects arranged in a square, arise via a continuous bifurcation. Shell softness drives the first-order character of the transition, while in the limit of infinite stiffness all bifurcations become continuous. Integer defects strongly couple with local mass redistribution, manifesting as shell thinning or thickening, whilst half-integer defects induce no such local deformation. These findings provide a purely mechanical framework for understanding body-axis formation and defect-mediated morphogenesis in biological vesicles.

Quantum phase transitions (QPTs) in the spin-boson model with/without the rotating-wave approximation (RWA) are systematically investigated through variational calculations using a sub-Ohmic bath with high spectral density. Four cases involving different system-environment interactions are examined, where transition points and critical exponents are accurately determined across varying tunneling strengths. Contrary to prior work, a rich phase diagram is revealed in the tunneling-coupling plane even at the low spectral exponent $s<1/2$, with a novel U(1)-symmetric phase being identified. As coupling increases, a multi-stage QPT sequence arises for the tunneling $0<Δ< Δ^*=0.074(1)$, whereas a single transition occurs beyond this range. Furthermore, an odd-parity phase is found to emerge even under the positive tunneling, exhibiting distinct characteristics relative to the prototype model.

Although rutile RuO2 has been a well-known and almost the best oxygen evolution reaction (OER) catalyst, the OER properties for the similar rutile oxide OsO2 with the same group element with Ru have been unknown, mainly due to long-standing synthesis difficulties. In this work, we report the successful synthesis of high-quality OsO2 single crystals, and the ground micrometer-size single crystals are chemically stable in alkaline solutions and exhibit robust OER performance. In sharp contrast, OsO2 nanopowder reacts quickly with KOH solutions and cannot work for OER. Compared with commercial RuO2 nanopowder, the OsO2 single crystals show comparable catalytic current densities, remarkably lower overpotentials at high current densities and better stability. These findings question the universal applicability of nanoscaling and highlight crystal integrity as a key descriptor for achieving stable and efficient OER electrocatalysis.

Altermagnets with compensated spin structures and nonrelativistic spin splitting have emerged as a new class of magnetic materials. Rutile OsO2 has been theoretically predicted to be altermagnetic, but experimental studies have been limited by synthesis challenges. We have succeeded in synthesizing high-quality single crystals of rutile OsO2. Electrical transport studies reveal that OsO2 is highly conductive and exhibits clear Fermi liquid behavior, indicating strong electron-electron scattering. Magnetic measurements show that the crystals are isotropically paramagnetic. Density-functional theory calculations indicate that bulk OsO2 is semimetallic with coexisting electron and hole pockets, with its magnetic ground state strongly dependent on the on-site Coulomb correlation U. Angle-resolved photoemission spectroscopy studies unveil that the bulk bands do not yet show altermagnetic spin splitting. Interestingly, resistivity is rather pressure sensitive: at 44 GPa, a clear metal-insulator transition occurs. Hybrid functional calculations reveal that applying pressure significantly increases the Hubbard U value, driving a phase transition from a paramagnetic metal to an altermagnetic metal, and eventually to an altermagnetic insulator. These findings suggest that tuning external pressure effectively modulates the magnetic ground state of OsO2, providing a pathway to realize altermagnetism in this material.

A rigorous understanding of the thermodynamic properties of point defects, namely vacancies and self-interstitials, is crucial for the discovery and screening of structural materials in clean energy applications. In this work, we extend a previously-developed statistical framework for predicting the thermodynamics of single-site impurities to further predict the thermodynamics of self-interstitial dumbbells in an arbitrarily complex alloy. We then apply this extended framework to compute effective formation energies in fully disordered Fe-Cr and Cu-Ni alloys. Notably, we predict that some self-interstitial dumbbell types that are high-energy in pure Fe become stabilized by Cr. We additionally describe a symmetry-breaking effect, wherein high solute concentrations distort the defect free energy surface, yielding misaligned self-interstitials.

Far-from-equilibrium magnetization dynamics can be accessed when a magnetic material is subject to a femtosecond excitation, such as an optical laser or an electric current. Numerically, the demagnetization of magnetic materials is typically modeled by atomistic spin dynamics. Micromagnetic models generally fail to reproduce ultrafast demagnetization in a grid independent manner. Here, we propose a non-equilibrium thermal field whose features depend on atomic spin flip probabilities. Under the assumption that each spin flip is equivalent to a quantum of angular momentum, equivalent temperatures on the order of thousands of Kelvin are achieved. Demagnetization is quantitatively reproduced for several cell sizes. The presented approach can be further refined and extended towards a grid-independent and multiscale modeling of ultrafast magnetization dynamics.

Collective light-matter systems host an extensive manifold of dark states whose role in the emergence of thermodynamic behavior remains poorly understood, especially in the presence of disorder and structured environments. Here, we develop a hybrid matrix product state-hierarchical equations of motion (MPS-HEOM) approach that enables numerically exact simulations of polariton dynamics from a few emitters to the thermodynamic limit under both static and dynamic disorder. This allows us, for the first time, to provide a quantitative and operational answer to the long-standing question of what is the minimum system size required to reach the thermodynamic limit in collective polaritonic systems. By introducing a convergence scale, $N_{T}$, i.e., the number of molecules required for the photonic dynamics to reach the thermodynamic limit, we show that dynamic disorder generally poses a greater computational challenge than static disorder. We attribute this behavior to the suppression of collective light-matter dynamics by disorder, which dynamically activates non-collective degrees of freedom. We further find that $N_{T}$ exhibits a turnover behavior as the bath becomes more Markovian, as the bath timescales regulate bright-to-dark energy transfer and the involvement of dark and gray states. Hence, phonon timescales control both the breakdown of collective behavior and the growth of $N_{T}$. Our results establish the suppression of collective behavior as the key mechanism governing thermodynamic convergence in disordered light-matter systems.

Numerical methods for obtaining exact dynamics of non-Markovian open quantum systems are mostly limited to either small systems or to short-time evolution only. Here, we propose a new algorithm for computing process tensors--matrix product operator (MPO) representations that capture the environment influence--which achieves greatly enhanced computational scalings with system size, while maintaining linear scaling with simulation length. We build on recent developments in the field which allow for long-time evolutions through process tensors which have a time-translational invariance. These can be built for general Gaussian environments and generic coupling operators with the system using infinite time-evolving block decimation (iTEBD). We introduce a modified iTEBD algorithm using intermediate compression steps which bring down the computation time scaling with system size $d$ from $\mathcal{O}(d^8)$ to $\mathcal{O}(d^4)$, as well as significantly lowering the required memory. To illustrate the power of this method, we apply it to the problem of dispersive qubit readout in circuit QED, which was previously out-of-reach numerically. The full treatment of the measurement resonator, which requires a large system space, combined with the long simulation times precipitated by the separation of timescales between the measurement drive and the environment dissipation, is now possible. The algorithm we introduce not only allows for capturing non-Markovian dynamics in large open quantum systems, but also further extends all the existing capabilities of process tensors, for example in quantum optimal control, or in computation of multi-time correlations or of steady states, to more complex systems with tens of levels.

We present a theory of a flux-tunable superconducting qubit, the "Fraunhofer qubit," based on the Fraunhofer interference in a wide ballistic Josephson junction. As magnetic flux threads the junction, the Josephson potential is effectively averaged over a phase window proportional to flux. For perfectly transmitting junctions, as flux approaches one flux quantum h/2e, the flux averaging transforms the potential near its minimum from a quadratic to a triangular shape, resulting in significantly enhanced anharmonicity. This enhancement persists for junctions with lower transparency conducting channels. Microscopic tight-binding simulations that include inhomogeneous electrostatic potential and disorder confirm the enhancement of anharmonicity. These results establish a framework for flux control in hybrid superconducting circuits, providing an operating point where anharmonicity and charge-noise protection can be optimally balanced.

The XOR Boolean logic gate is widely used in many applications such as encryption (XOR ciphers), binary addition (half- and full-adders), error detection (parity bits), etc. but is challenging to construct because of its demanding conditional dynamics. It typically requires multiple logic switches or other types of gates, which results in a large gate footprint and low logic density. Here, we present the design of an XOR gate with a single straintronic magnetic tunnel junction which reduces the footprint dramatically. Such a gate is non-volatile and hence suitable for non-von-Neumann architectures, processor-in-memory, etc. The switching time of the gate is ~200 ps and the energy dissipation per gate operation is ~225 aJ. Cascading of successive stages is accomplished via a CMOS device which plays no role in the gate dynamics but is needed for gain to provide logic level restoration, fan-out and isolation between input and output. This 1 MTJ-1 CMOS design has an energy dissipation that is an order of magnitude smaller than what has been reported for traditional all-transistor XOR designs.

We investigate how the strong coupling of a superconductor to thermal bosons can enhance its superconducting critical temperature. To tackle this problem, we use a renormalization group approach that allows us to describe the competition between density fluctuations and the build-up of boson-induced attraction between fermions. Capturing the mutual influence of bosonic and fermionic sectors, the self-consistent renormalization group framework predicts a robust increase of the critical temperature across a wide range of interactions. We find a nontrivial dependence of the critical temperature on the boson mass and we establish a phase diagram for enhanced superconductivity driven by bosons being either in the condensed or thermal state. We outline possible experimental realizations in cold atomic systems and discuss implementations using electron-exciton mixtures in van der Waals material heterostructures.

We investigate the role of architected thin films in the interfacial failure properties of bi-layer composites. Our results show that, while graded structures can be used to prescribe failure at the interface, they do not offer significant advantages in terms of fracture toughness. Hierarchically patterned layers can localize failure at the interface and simultaneously enhance interface toughness, by enforcing a buffer region where elastic energy is dissipated in the form of diffuse damage, so that no stress concentration can drive crack growth. To analyze these mechanisms, the associated patterns of local load redistribution and the soft deformation modes, we develop a network formalism that brings together concepts of discrete differential geometry and spectral graph theory.

In this paper, we derive general bright-dark soliton solutions to the coupled Sasa-Satsuma (CSS) equation using the Kadomtsev-Petviashvili (KP) reduction method. Since the CSS equation is a special case of the four-component Hirota equation, our approach begins with the construction of two-bright-two-dark soliton solutions for the four-component Hirota equation. By imposing specific parameter constraints, these solutions are subsequently reduced to the bright-dark soliton solutions of the CSS equation. Finally, the dynamical behaviors of the one- and two-bright-dark soliton solutions are thoroughly analyzed and illustrated.

Soliton-similariton fiber lasers have demonstrated exceptional operational stability, maintaining continuous mode-locking for weeks despite large intracavity spectral and temporal breathing. We present the first stability study of this laser, rigorously establishing that the anomalous dispersion segment that supports the soliton is the cause of this robustness. Specifically, we perform linear stability analysis of the laser employing a Jacobian-based eigenvalue decomposition and show that the eigenvalues lie within the unit circle, leading to a positive stability margin, which is indicative of the robustness of the laser against small perturbations. Furthermore, the stability margin is observed to increase with the length of the anomalous fiber segment, clearly establishing its role in pulse stabilization. Critically, integrated pulse timing jitter and relative intensity noise as obtained from quantum noise-limited laser simulations are shown to be anti-correlated to the stability margin, further validating the results of the Jacobian analysis and establishing an unequivocal link between the reported noise performance of the soliton-similariton laser to the underlying pulse stabilization mechanism mediated by the anomalous segment. The direct link of the linear stability analysis to the underlying nonlinear physics of the laser, together with its significantly lower computational overhead, establishes it as a highly effective predictive framework for assessing laser noise performance, enabling novel approaches for designing quantum noise-limited ultrafast sources.

The Kuramoto model is the paradigmatic model to study synchronization in coupled oscillator systems. In its classical formulation, the oscillators move on the unit circle, each characterized by a scalar phase and a natural frequency, by interacting through a sinusoidal coupling. In this work, we propose a d-dimensional generalization in which oscillators are represented as unit vectors on the (d-1)-sphere and interact through a matrix-weighted network (MWN), a recently introduced framework where links are endowed with a matrix weight instead of a scalar one. We derive necessary conditions for global synchronization via a Master Stability Function approach: the existence of a synchronous solution requires identical frequency matrices across nodes and, in the MWN case, a coherence condition on the network structure. Through a suitable change of variables, the stability analysis reduces the full Nd-dimensional problem to a family of d-dimensional eigenvalue problems, each one parametrized by the eigenvalue of a suitable scalar weighted Laplacian, showing that the synchronous solution is locally stable for any positive coupling strength K on any connected network. Analytical results are complemented by numerical simulations.

The paper investigates three eigenfunction constraints of two (2+1)-dimensional differential-difference integrable systems. First, we revisit the known squared eigenfunction symmetry constraint of the differential-difference Kadomtsev-Petviashvili (D$Δ$KP) hierarchy, which gives rise to a semi-discrete Ablowitz-Kaup-Newell-Segur hierarchy. Second, we introduce a linear eigenfunction constraint for the D$Δ$KP system and obtain a combined semi-discrete Burgers (sdBurgers) hierarchy. In the third one, we consider another linear eigenfunction constraint for the modified D$Δ$KP system and obtain the same combined sdBurgers hierarchy. All these constraint results are proved by using recursive algebraic structures of the involved integrable hierarchies generated by their master symmetries.

Deterministic chaos is commonly associated with spectral criticality: exponential sensitivity is expected when Jacobian eigenvalues exceed unity in parts of the attractor, producing the local expansion that offsets contraction elsewhere. We show that this paradigm is incomplete in dimensions d>1. We construct a bounded 3D dynamical system whose Jacobian is pointwise spectrally contracting, namely all instantaneous eigenvalues remain strictly inside the stability region, yet the system develops a positive maximal Lyapunov exponent and undergoes a transition to chaos as a non-normality index increases at fixed spectral radius. The mechanism relies on the repeated regeneration of transient non-normal amplification through endogenous switching that reinjects trajectories into amplifying non-orthogonal directions. Although demonstrated here for a discrete-time map, the mechanism is geometric and applies more broadly to deterministic dynamical systems. These results show that chaos can emerge without spectral criticality and identify non-normality as an independent route to deterministic chaos.

We develop a deformation-based framework for analyzing static billiard systems through the Jacobian determinant computed in noncanonical angular coordinates. Although these systems are conservative, the determinant is not identically equal to unity, generating structured domains of local phase-space expansion and contraction. We show numerically that these domains balance globally, providing a geometric manifestation of area preservation in noncanonical variables. The curves defined by det J = 1 act as deformation boundaries that intersect unstable periodic points and correlate with invariant manifolds. We prove analytically that period-two orbits restore exact unit determinant under composition, while higher-period orbits exhibit angular modulation consistent with reversibility. The Jacobian determinant thus reveals an additional geometric layer in phase-space organization and offers a complementary perspective on conservative billiard dynamics.

Chaos theory is a branch of classical physics, founded in the 1960s-70s, that studies systems whose solutions are sensitively dependent on their initial conditions. For many, it is surprising that chaos theory arrived so late. However, through the work of Henri Poincaré, we know that much of the math of chaos was understood by some 70 years prior. Furthermore, through the writings of Poincaré's colleagues -- Jacques Hadamard and Pierre Duhem -- we also see a detailed understanding of the chaos found in his work. They also have explicit reasons of why the math of chaos was to be ignored. It was a strict form of empiricism -- positivism -- causing them to label chaos as ``useless'' and ``meaningless'' mathematics because it was thought to be ungrounded in experience. In this paper, I describe how the empiricist tenets of positivism exiled chaos from physics following Poincaré.

In [14] Dubrovin introduced an $A_1$-type infinite ODE system and gave a simple way of constructing algebro-geometric solutions to the KdV hierarchy (cf. also [15,4]). In [34] the infinite ODE system is generalized to $\mathfrak{g}$-type infinite ODE system, where $\mathfrak{g}$ is any simple Lie algebra. In this paper, we give a simple constructinon of algebro-geometric solutions to the Gelfand--Dickey hierarchy based on the $A_n$-type infinite ODE system and Dubrovin's method. As an application, we give a formula for the $N$-point function for the related Riemann $θ$-function.

Understanding the nuances of nonlinear pulse propagation in optical fibers has led to several impactful applications across domains like optical communications, sensing and biophotonics. A key aspect in this regard is the use of appropriate optimization strategies for attaining requisite performance parameters. In this paper, we present the Taguchi method as a viable tool for optimizing nonlinear pulse propagation in optical fibers. We show that its use of the orthogonal arrays leads to rapid convergences to the desired pulse parameters, with even faster convergences obtained by favouring exploitation over exploration. We demonstrate the application of the method using two well-known problems from the field - the guiding center soliton, and soliton order conservation in dispersion decreasing fibers - which serve to underscore its salient features and also its potential for solution discovery across nonlinear pulse propagation problems.

We establish the integrability of a one-parameter family of coupled Dirac--scalar field theories in $(1+1)$ dimensions that interpolates between the known Dirac--sinh-Gordon and Dirac--sine-Gordon systems. The deformation is controlled by a phase parameter that modifies the Yukawa coupling and simultaneously rescales the scalar backreaction. For all values of the parameter, we construct an explicit zero-curvature representation based on an $sl(2,\mathbb{C})$-valued Lax pair and show that the deformation preserves integrability. We further prove that the family is physically non-trivial, in the sense that distinct parameter values are not related by admissible field redefinitions. In addition, we derive the continuity relation for the fermion bilinear, show that the spatial bilinear constraint follows from the zero-curvature equations, and construct the first conserved densities of the hierarchy. At the two endpoints, the family reduces to the standard integrable Dirac--sinh-Gordon model and, after analytic continuation, to the Dirac--sine-Gordon system which is dual to the massive Thirring model.

Dimensional reduction is a generic consequence of dissipation in nonlinear evolution equations, often leading to attractor collapse and the loss of dynamical richness. To counteract this, we introduce a geometric framework for Covariant Multi-Scale Negative Coupling Systems (C-MNCS), formulated intrinsically on smooth Riemannian manifolds for a broad class of semilinear dissipative PDEs. The proposed coupling redistributes energy across dynamically separated spectral bands, inducing a scale-balanced feedback that prevents finite-dimensional degeneration. We establish the short-time well-posedness of the coupled state-metric evolution system in Sobolev spaces and derive a priori estimates for phase-space contraction rates. Furthermore, under a global boundedness hypothesis, we prove that the global attractor possesses a strictly finite Hausdorff and Kaplan-Yorke dimension. To bridge abstract topological bounds with physical realizability, we isolate the core Adaptive Spectral Negative Coupling (ASNC) mechanism for numerical validation. High-resolution experiments, utilizing a fully coupled ETDRK4 scheme and continuous QR-based Lyapunov exponent computation on a conformally flat 2D dynamic scalar manifold, corroborate the theoretical predictions. These computations explicitly demonstrate the stabilization of high-dimensional attractors against severe macroscopic dissipation. This geometrically consistent mechanism offers a new paradigm for maintaining structural complexity and multiscale control in infinite-dimensional dynamical systems.

Dynamical-systems approaches to spatiotemporal chaos have been developed primarily for single-phase flows, where the system state is defined by bulk velocity fields. Extending these ideas to two-phase flows remains challenging because the dynamics are intrinsically coupled to the evolution of a deforming interface. Here, we address this challenge for a two-dimensional vertical falling film by formulating the dynamics in terms of the interface evolution. Starting from the Navier--Stokes equations, we recover a classical long-wave interface evolution equation, originally derived by Topper & Kawahara (1978). Using this formulation, we perform an extensive parametric study to construct a regime map in the space of domain size and dispersion parameter. The resulting map reveals a rich range of interfacial behaviors, including travelling waves, bursting travelling waves, and fully chaotic regimes. In the chaotic falling film regime, we exploit the dissipative nature of the governing equation, which suggests that the long-time dynamics evolve onto an inertial manifold. Using a data-driven approach, we parameterize this inertial manifold and estimate its intrinsic dimension, suggesting approximately linear growth with domain size. We then construct low-dimensional models in manifold coordinates to facilitate the search for exact coherent states of the full system. Using this approach, we identify travelling waves, relative periodic orbits and equilibria embedded within the chaotic attractor. Chaotic trajectories repeatedly approach the neighbourhoods of these invariant solutions, indicating that the recurrent interfacial patterns observed in the dynamics correspond to visits to these coherent states. To the best of our knowledge, this constitutes the first identification of exact coherent structures embedded in chaotic falling-film dynamics.

This is an English translation and digitisation of Sophie Kowalevski's (also know as Sofya Kovalevskaya) paper on what is now known as the Kovalevskaya Top. The original paper was written in French and published in Vol 12 of Acta Mathematica in 1889 with the title "Sur le probleme de la rotation d'un corps solide autour d'un point fixe".

Rare nonadiabatic events play a central role in photochemistry but remain difficult to simulate because excited-state dynamics is computationally demanding and often stochastic. Here we introduce a deterministic and time-reversible implementation of nonadiabatic dynamics that enables the application of transition path sampling (TPS) to excited-state processes. Our approach builds on the Mapping Approach to Surface Hopping (MASH) and establishes the conditions required for path ensemble sampling, in particular time reversibility and detailed balance. Combining this dynamics with the TPS framework yields a new method, termed nonadiabatic transition path sampling (NATPS). Using a model system of electronically coupled potential energy surfaces, we demonstrate that NATPS efficiently generates ensembles of reactive trajectories and provides mechanistic insight into nonadiabatic pathways. Compared with brute-force trajectory simulations and forward-flux sampling approaches, NATPS substantially reduces the computational effort required to obtain reactive trajectories.

Optical metasurfaces, comprised of subwavelength nanostructures, hold a great promise to high-power laser optics but also a limited pertinence due to their currently limited aperture size, throughput and durability. Here, an alternative approach is presented, reliant on laser-controlled self-organizing mask formation followed by ion etching which results in an all-fused-silica-glass metasurface. Two 1 mm diameter optical elements (an axicon lens and a shadower) are fabricated and their optical performance is validated at 532 nm wavelength with an extremely low broadband reflection (<0.15%) - a result of the unique metasurface elements shape. The self-organizing working principle enables producing large amounts of nano-elements at-once, thus a path for aperture scaleup. It also enables generation of sub-100 nm nanoelements, thus a path to short wavelengths operation. Two key advancements towards viability are presented: a laser scan with in-situ transmission feedback enables patterning the etching mask to a prescribed nanoparticle distribution, and a crafted beyond-mask-erosion-point etching of the mask enables increasing the metasurface phase difference to at least pi, while keeping extremely low reflection across it. This paves a path to high-power lasers optics, requiring large aperture, high throughput and laser light durability.

We derive integral formulas that simplify the Vector Spherical Tensor Product recently introduced by Xie et al., which generalizes the Gaunt tensor product to antisymmetric couplings. In particular, we obtain explicit closed-form expressions for the antisymmetric analogues of the Gaunt coefficients. This enables us to simulate the Clebsch-Gordan tensor product using a single Vector Spherical Tensor Product, yielding a $9\times$ reduction in the required tensor product evaluations. Our results enable efficient and practical implementations of the Vector Spherical Tensor Product, paving the way for applications of this generalization of Gaunt tensor products in $\mathrm{SO}(3)$-equivariant neural networks. Moreover, we discuss how the Gaunt and the Vector Spherical Tensor Products allow to control the expressivity-runtime tradeoff associated with the usual Clebsch-Gordan Tensor Products. Finally, we investigate low rank decompositions of the normalizations of the considered tensor products in view of their use in equivariant neural networks.

Diamond photonics has enabled efficient interfaces for quantum memories and is predicted to be a critical component of quantum networks. However, scalable network architectures require spatial, temporal, and spectral control of photons, which relies on nonlinear and electro-optic functionalities that diamond alone cannot provide. Here, we demonstrate heterogeneous integration of a thin-film lithium niobate (TFLN) platform, which has strong chi-2 nonlinearity and electro-optic effects, with thin diamond films. We demonstrate high-Q diamond photonic crystal cavities (Q factors exceeding 5x10^4 at 735 nm) that are lithographically aligned with TFLN photonic backbone and critically coupled to it. This allows us to realize low-loss diamond-TFLN "escalators" (loss ~1 dB/coupler) that support efficient light transfer between them. At cryogenic temperatures (5K), we can collect photons emitted from silicon vacancies (SiVs) embedded within the diamond structure via the TFLN photonic circuit. This approach establishes a scalable route toward integrated photonic circuits for practical quantum networking and other technologies.

Two-phase flow in porous media is a ubiquitous phenomenon that has been studied for well over a century. However, we still lack a successful theory that predicts flow on a macroscopic length scale (the so-called Darcy scale) on the basis of a "microscopic" model. Here we show that the characteristic features of two-phase flow on the Darcy scale can be predicted by mapping the distribution of droplets in 2-phase flow onto the distribution of spins in a spin-glass model. The success of this approach is surprising, as two-phase flow is a non-equilibrium phenomenon, whereas the properties of the spin glass are obtained using equilibrium statistical mechanics. To obtain this mapping, we follow the approach of Meshulam and Bialek (Rev. Mod. Phys. 97, 045002 (2025)) and use the Jaynes maximum entropy principle to derive the spin-glass Hamiltonian using machine learning trained on many realizations of the two-phase flow pattern in a dynamic pore network model. With this mapping, we can construct a "phase diagram" for the 2-phase flow system. We find that the critical line separating the paramagnetic phase from a spin glass phase coincides with the transition where the dependence of the rate of two-phase flow on the imposed pressure gradient changes from linear to non-linear. The glassy phase of the spin model coincides with a flow regime characterized by hysteresis and strong fluctuations over a wide range of time scales. It is tempting to identify this flow regime as a dynamic glass state.

Phase-only spatial light modulators (SLMs) are used in optical systems for several purposes. In this article, the main landmarks of SLM-based imaging systems are surveyed. In addition to conventional two-dimensional imaging, these systems are useful for multidimensional imaging, axial sectioning, field-of-view expansion, improved image resolution, imaging through scatterers, and depth-of-field control. The SLMs in this review are positioned in the system aperture and modulate the input light in various ways to achieve different imaging goals. This review begins with the nearly 20-year-old Fresnel incoherent correlation holography system, continues with coded-aperture holography, and progresses to the most recent versions of interferenceless coded-aperture holography systems.

We consider the diffraction of time-harmonic plane waves by a periodic structure, governed by the Helmholtz equation. Bound states in the continuum (BICs) are quasi-periodic fields that remain $L^{2}$-bounded over one period and occur at frequencies embedded in the continuous spectrum. Perturbations that break a BIC can lead to ultra-strong resonances, enabling various applications in photonics. Employing the implicit function theorem, we demonstrate how a simple BIC continuously deforms into a propagating field as system parameters vary in a neighborhood, with the frequency adjusting accordingly. In this setting, the incident coefficients of the field persist as an eigenvector of the scattering matrix with a fixed eigenvalue. By introducing a mapping $\mathcal{P}$ from the parameters to these coefficients, the zeros of $\mathcal{P}$ correspond precisely to BICs. When such a zero is isolated and the dimensions of the domain and range coincide, the BIC can be related to the mapping degree of $\mathcal{P}$ in a small neighborhood. This perspective clarifies the phase singularity associated with BICs and provides a general topological interpretation of their local robustness with respect to the given parameters. Moreover, it yields a practical numerical criterion for detecting and verifying BICs via computation of the mapping degree of $\mathcal{P}$.

Single photon emitters (SPEs) are promising building blocks for practical devices in quantum technologies. Traditionally, these systems are excited using off-resonant visible light through their phonon transitions, yet this process remains poorly understood. Here, we explore the interaction of mid-infrared (MIR) excitation on the properties of SPEs in hexagonal boron nitride. Notably, we present a reversible, non-destructive method to enhance emission from blue SPEs using MIR co-excitation. By resonantly driving defect-localized in-plane infrared-active optical phonon modes near 7.3 um, the MIR field modulates carrier dynamics through a phonon-assisted recombination. This unique feature, not observed previously for defects in solids, is a promising reservoir in a growing toolkit to modulate quantum emitters at room temperature for their use in practical quantum technologies.

Robust single-vessel tracking from fixed coastal platforms is hindered by modality-specific degradations: cameras suffer from illumination and visual clutter, while LiDAR performance drops with range and intermittent returns. We present a heterogeneous multi-sensor fusion particle-filter tracker that incorporates an information-gain (entropy-reduction) adaptive sensing policy to select the most informative configuration at each fusion time bin. The approach is validated in a real maritime deployment at the CMMI Smart Marina Testbed (Ayia Napa Marina, Cyprus), using a shore-mounted 3D LiDAR and an elevated fixed camera to track a rigid inflatable boat with onboard GNSS ground truth. We compare LiDAR-only, camera-only, all-sensors, and adaptive configurations. Results show LiDAR dominates near-field accuracy, the camera sustains longer-range coverage when LiDAR becomes unavailable, and the adaptive policy achieves a favorable accuracy-continuity trade-off by switching modalities based on information gain. By avoiding continuous multi-stream processing, the adaptive configuration provides a practical baseline for resilient and resource-aware maritime surveillance.

Efficient locomotion is important for the evolution of complex life, yet the physical principles selecting specific swimming strokes often remain entangled with biological constraints. In viscous fluids, the scallop theorem constrains the temporal organization of strokes, but no analogous principle is known for their spatial structure, leaving the prevalence of symmetric gaits across diverse organisms without a physical explanation. Here we show that spatial symmetry acts as an emergent organizing principle for efficiency in viscous fluids. By analysing deformable swimmers whose strokes are not constrained to any particular symmetry class, we identify a hydrodynamic duality: symmetric and anti-symmetric strokes are dynamically equivalent, yielding identical speeds and efficiencies, which we prove are optimal among all strokes. By contrast, the optimal efficiency cannot be achieved by generic non-symmetric strokes. We validate this using numerical simulations of Stokes flow, demonstrating that these symmetry rules persist even in three-dimensional body plans. Our results suggest that the prevalence of symmetric and alternating gaits in nature reflects not merely a developmental constraint, but a physical optimality principle for locomotion in viscous environments, complementing developmental and neural constraints.

Although commonly associated with limbless animals like snakes and fish, multi-legged organisms like centipedes also utilize undulatory locomotion. Whether these undulations are actively reinforced or resisted by the axial musculature remains an open question. We present a dynamical model of centipede locomotion that integrates leg-ground interactions, passive body mechanics, and active lateral musculature. By varying stepping rate, actuation, and body stiffness, we examine how locomotor strategies affect speed and an effective energetic efficiency. Coordination emerges only when body stiffness is tuned to stepping frequency: overly flexible bodies lose synchrony, while overly rigid ones move slowly and inefficiently. We make a falsifiable prediction, measurable with a non-invasive experiment, that centipedes utilize speed dependent active stiffness to maintain this coordination. Our results suggest that lateral muscles also have a speed dependent function, revealed by optimizing speed and an effective cost, that resists a phase lag between leg touchdowns and body curvature. Together, we find that centipedes actively modulate body mechanics to achieve rapid, efficient locomotion, highlighting how complex control can emerge from embodied physical properties rather than solely from neural computation.

Urban land cover doubled between 1985 and 2015, yet the spatial dynamics of urban form remain under-quantified, despite its importance for sustainability, infrastructure planning, and climate risk. Urban expansion is a non-equilibrium process shaped by interactions between population growth, infrastructure, institutions, and market failures -- rendering static and equilibrium models inadequate. We review key challenges and modeling approaches, focusing on partial differential equation (PDE) frameworks. Borrowed from statistical physics, PDEs capture spatial heterogeneity, anisotropy, stochasticity, and feedbacks between land use and transport networks. Integrating economic and institutional factors remains a major challenge for policy relevance. We propose a research agenda that bridges remote sensing, urban economics, and complexity science to develop dynamic, empirically grounded models of urban expansion.

Thermal neutron total scattering experiments of light and heavy water were reproduced using the CSNS in-house Monte Carlo thermal neutron transport code, Prompt, with a focus on the instrumental detector response and the accurate derivation of thermal neutron scattering cross-sections. In this work, a data reduction method is developed to process both the measured and simulated detector events for estimating angular, wavelength distributions, as well as angular differential cross sections. The reduction results of simulations and experiments show a high degree of consistency. The prominent inelasticity signatures observed in the experiments can be accurately reproduced in simulations. We discuss the cause of the inelasticity effects, and demonstrate the elimination of such effects when the inelastic scattering process is taken into account in simulations. In addition, multiple scattering in samples is analysed and discussed.

A knife-edge is recognized as a critical component of the schlieren imaging system. This knife-edge serves as a cutoff element, which is required to achieve high sensitivity in the schlieren imaging setup. This article describes a method for totally removing the knife-edge from a normal schlieren imaging system while maintaining the same sensitivity. In this approach, the camera lens's internal aperture acts like the cutoff element making an external knife-edge redundant. This method simplifies the schlieren setup and also reduces the setup cost due to the reduced number of optical or optomechanical elements. We show flow visualization data that clearly demonstrate that schlieren contrast can be obtained by employing a lens's internal aperture as the cutoff.

Nonlinear photon upconversion processes driven by diverse forms of structured light are receiving increasing attention. In harmonic and high-order harmonic generation (HG and HHG) with Laguerre-Gauss (LG) beams, linear scaling the driver topological charge (TC) with the harmonic order is equivalent to driver orbital angular momentum (OAM) per photon scaling, and constitutes a proof of OAM conservation. However, with generic driving fields, such as non-LG vortices or spatiotemporal optical vortices, TC and OAM per photon may scale or not in a process in which the OAM is conserved. We find the physical magnitude that scales with generality when the OAM, either longitudinal or transverse, or its intrinsic part, is conserved. This new rule allows for the wealth of phenomena observed in HHG that are unintelligible from the rigid LG rule.

The widespread use of non-SI quantities and units, together with improper and misleading terminology, generates confusion in the domain of phase and amplitude noise. We discuss such physical quantities, units and terms with the purpose of stimulating a discussion between labs, and to agree for the full adoption of the International System of Units (SI) and clear, unambiguous terminology.

Topology optimization (TO) has been widely adopted in engineering design; however, it is prone to being trapped in local optima, particularly in strongly nonlinear problems. Sensitivity-free data-driven topology design (DDTD) offers a promising alternative. Nevertheless, existing DDTD-based methods still depend heavily on prior information or sensitivity-based TO methods for initialization, limiting their generality and independence in engineering applications. In this study, an efficient DDTD-based framework capable of being driven from low information-entropy initial datasets is proposed while improving computational efficiency. To reduce the dependence on high information-entropy initial datasets, a mesh-independent mutation module is introduced as a supplementary source of geometric features, enabling stable exploration under low information-entropy initialization. To alleviate the computational bottleneck in DDTD, where all candidate structures require numerical evaluations, a non-AI-based rapid identification algorithm is developed to efficiently identify potential high-performance structures, thereby significantly reducing the number of expensive high-fidelity simulations. The framework generates material distributions on body-fitted meshes to maintain consistency between numerical simulations and physical manufacturing. A signed distance field-based minimum length constraint is further incorporated to ensure reliable mesh generation. Numerical experiments on strongly nonlinear stress-related problems, together with comparisons with sensitivity-based TO methods, demonstrate the effectiveness of the proposed method. In microfluidic reactor and shell design problems involving non-differentiable constraints, the proposed method successfully addresses scenarios that remain challenging for both sensitivity-based TO and conventional DDTD-based methods.

Optical klystrons provide an efficient mechanism for enhancing coherent radiation through laser-induced microbunching and dispersive amplification. In the terahertz (THz) regime and at low beam energy, however, the radiation wavelength becomes comparable to the characteristic wavelengths of longitudinal space-charge (LSC) and coherent synchrotron radiation (CSR) driven microbunching. In this work, we analyze the impact of electron-beam microbunching on the spectral amplitude and phase of a seeded optical klystron operating in the THz regime. Using a phase-space formalism, incoherent energy modulations generated by collective effects are shown to enter the bunching spectrum as a stochastic longitudinal phase, producing local wavenumber jitter and spectral broadening. An explicit connection between the microbunching-induced energy modulation and LSC/CSR gain is established for low-energy beams, demonstrating that the overlap between collective-effect wavelengths and the optical-klystron radiation wavelength leads to strong spectral phase distortion. These effects impose fundamental constraints on the achievable harmonic bunching and spectral purity and stability of THz optical klystrons and must be considered in the design and optimization of next-generation low-energy THz FEL facilities.

Ion transport through narrow channels is described by the coupled Poisson--Nernst--Planck--Stokes equations (PNPS) on a continuum scale. However, direct numerical simulations in two or three dimensions of boundary value problems for small aspect ratio geometries, a crucial characteristic of nanopores, can quickly become computationally intensive and thus limit the insights into the underlying mechanisms that control electrokinetic phenomena. Taking advantage of the small aspect ratio, we derive a systematic asymptotic reduction of the PNPS system. In contrast to existing one-dimensional reductions, which assume a Debye length much smaller than the channel radius, our analysis identifies a distinguished asymptotic regime in which the Debye length is allowed to be comparable to the channel width. Our approach has a significantly larger range of validity and contains existing approximations such as the Helmholtz--Smoluchowski approximation as limiting cases. The derived asymptotic model extends also to a generalized PNPS system, where finite-size constraints and solvation effects are taken into account and thus applies to other well-known models such as the Bikerman--Freise model. Using our asymptotic model we demonstrate that the ion current can undergo a number of different flow transitions and in particular predict that positively charged ions can be pushed against their electrostatic gradient. Furthermore, we show how finite-size effects can influence the ion current and enhance ion selectivity. Finally, we revisit case studies of protein-based channels from the literature to illustrate the predictive potential of our asymptotic model.

Time-resolved photoelectron spectroscopy provides access to the electronic structure and non-equilibrium electron dynamics in matter. At solid surfaces photoemission dynamics can be investigated on its natural time scale by measuring attosecond time delays of emitted electrons. Photoelectrons with final state energies of several tens of eV need tens to hundreds of attoseconds to be released into the vacuum. Competing effects determine the emission dynamics and, hence, the full picture of the process is still under debate. The rather large energy differences between the final states probed in commonly reported relative photoemission delays obscure their complex fine structure and hinders the interpretation of the measurements. Here we report differential attosecond delays $τ_{\mathrm{DAD}}$, i.e., relative photoemission delays for energetically close-lying spin-orbit split states. Differential attosecond delays on the order of 30 to 100 as for Bi 5d, Te 4d, and Se 3d core level photoemission from Bi$_2$Te$_3$ and Bi$_2$Se$_3$ can neither be attributed to intra-atomic delays, nor to ballistic transport and subsequent emission. Instead, calculations based on the one-step photoemission theory reveal that photoemission delays vary strongly on the energy scale of the spin-orbit splitting and quantitatively match experimental observations. This strong variation arises from multiple scattering at the surface leading to final states that involve both evanescent and propagating Bloch waves. Their relative amplitudes vary strongly affecting thereby the timing of the photoemission event since evanescent and propagating components exhibit inherently different dynamics.

Multiple scattering rapidly scrambles optical fields in fog, snow and turbid water, causing op- timized wavefront corrections to become obsolete on microsecond timescales. Although wavefront shaping enables focusing through static scattering layers, closed-loop control in dynamically evolving media has remained experimentally challenging because the correction bandwidth must approach the intrinsic decorrelation rate of the speckle. Here, we demonstrate closed-loop wavefront shaping with 32 independent degrees of freedom in an agitated turbid medium exhibiting sub-microsecond decorrelation. The medium thickness exceeds the transport mean free path, meaning the far-field speckle autocorrelation is limited to a diffraction-sized grain. Despite the microsecond decorrelation and this multiple-scattering regime, stable focusing is maintained as the correction cycle approaches the intrinsic dynamics of the medium. These results establish an experimentally accessible regime for coherent wave control in rapidly evolving complex media.

Extremal materials are a specific class of Cauchy materials whose elasticity tensor has one or more zero eigenvalues. Each zero eigenvalue corresponds to a soft mode requiring zero strain energy, while non-zero eigenvalues correspond to hard modes that cost energy. According to the number, N, of zero eigenvalues, these materials can be referred to as unimode (N=1), bimode (N=2), etc. Extremal materials have enabled novel functions beyond conventional Cauchy media, e.g., phonon polarizers, Rayleigh wave isolators and underwater acoustic cloaks. These functions typically require a single extremal material. Interfaces between two extremal materials exhibit rich wave behaviors, yet have been seldom explored. Here, we proposed the concept of complementary extremal materials, i.e., the soft mode of one extremal material is a hard mode of the other. As one example, we study the interface between an isotropic unimode material and an isotropic bimode material. We show that the interface allows perfect mode conversion from longitudinal waves to transverse waves. A low-frequency underwater acoustic insulator based on complementary extremal materials is proposed. Our finding has been verified with designed metamaterials and using effective-medium modeling. This work demonstrates the potential of complementary extremal materials in controlling elastic wave polarization and waterborne sound.

Here, we present a diffusion transformer (DiT)-based pixel super-resolution virtual staining approach to transform low-resolution FTIR microspectroscopic images of the unstained tissues into corresponding high-resolution H&E-stained images. Unlike conventional conditional DiT architectures, this method models the transformation from FTIR images to H&E-stained images as a stochastic Brownian bridge process and directly learns the cross-domain translation in pixel space by means of a large-patch Transformer. When applied to FTIR images of unlabeled human lung tissue samples, the proposed method successfully transforms them into high-resolution H&E-stained images, achieving a 4 times pixel-level super-resolution. Additionally, by partitioning images into large patches, our method achieves a fourfold improvement in inference speed compared with traditional U-Net-based diffusion models, without compromising the quality of generated images. This super-resolution virtual staining method provides a rapid and effective solution for generating high-resolution, clinically usable H&E-stained images from infrared spectroscopic images, which can significantly facilitate the incorporation of FTIR microspectroscopy into clinical histological scenarios.

Cells encounter a diverse array of physical and chemical signals as they navigate their natural surroundings. However, their response to the simultaneous presence of multiple cues remains elusive. Particularly, the impact of topography alongside a chemotactic gradient on cell migratory behavior remains insufficiently explored. In this paper, we investigate the effects of topographical obstacles during chemotaxis. Our approach involves modifying the Keller-Segel model, incorporating a spatially dependent coefficient of chemotaxis. Through our analysis, we demonstrate that this coefficient plays a crucial role in preventing blow-up phenomena in cell concentration.

This paper presents the design and experimental demonstration of a reconfigurable cavity excited nonlocal metasurface antenna capable of wide angle dynamic beam steering. The antenna is synthesized using a volume surface integral equation based framework that rigorously captures nonlocal mutual coupling among metasurface unit cells. To ensure physical consistency, the numerically characterized resistance and reactance relationship of the tunable unit cells is directly incorporated into the synthesis, enabling precise far-field synthesis while minimizing Ohmic losses. The proposed approach is applied to a 10 GHz cavity fed metasurface antenna composed of 24 independently controlled varactor-loaded unit cells. Numerical simulations and near-field measurements demonstrate stable beam steering with a range of 80 degrees across broadside with excellent agreement between measured and simulated radiation patterns. These results confirm the effectiveness of the proposed framework for the realization of compact, reconfigurable cavity-excited metasurface antennas.

Transferring optical information through random diffusers is a critical yet challenging task. In this work, we introduce a cascaded diffractive optical network for information transfer through random and unknown diffusers, achieved through a series of passive, structured layers physically interleaved within the scattering medium. These interleaved diffractive layers are optimized to mitigate the scattering process without requiring digital computing. The performance of this all-optical system was quantified as a function of several physical parameters, including the diffractive processor's depth, the physical layout of the diffractive layers, and the statistical properties of the scattering medium. To further enhance the performance and robustness of information transfer through a scattering medium, we also developed a hybrid (optical-digital) system that coupled the diffractive processor with a jointly trained digital neural network, which was shown to achieve superior reconstruction fidelity even when the input object information was subjected to unknown random rotations, shifts, and scaling. We also experimentally validated this system using a fabricated multi-layer diffractive system in the visible spectrum, demonstrating reliable information recovery through random diffuser layers. Our numerical and experimental results demonstrate the capabilities of an interleaved diffractive processor architecture to recover optical information through volumetric diffusive media, which can find applications in biomedical imaging, telecommunications, and remote sensing.

The spaceborne laser emission telescope is a core and critical component of the space gravitational wave detection system.Compared with ground-based telescopes, the on-orbit space environment is more complex and harsh, presenting higher technical challenges for the design of the optical system and structure - both optical design and structural design face considerable difficulties. To meet the requirements of space gravitational wave detection, this paper designs a laser emission telescope based on an off-axis four-mirror configuration, with a capture field of view of 300μrad, an optical transmission efficiency of 86.3%, and an optical path stability index of TTL<0.025 nm/μrad. During the design process, based on existing theories and engineering experience, the primary mirror thickness optimization and lightweight structural design were completed, and a flexible support scheme was adopted to achieve a primary mirror surface figure accuracy of 9.42 nm; the total mass of the entire telescope (excluding mirrors) is only 3.845 kg. Multi-dimensional finite element analysis was conducted on the telescope under actual working conditions: the strength of the telescope's support materials was verified under self-weight and 10G gravity loads; after removing the rigid body displacement of the mirrors using Zernike polynomials, the surface deformation of the primary mirror was controlled within 1/30 wavelength. In the thermal stability analysis, the structural deformation of the telescope under a temperature change of 100 degree celsius was simulated, and key indicators such as eccentricity and tilt between the mirrors all meet the optical design requirements. In the modal analysis, the first-order natural frequency of the telescope reaches 200 Hz under both self-weight and weightless conditions, demonstrating excellent dynamic stability.

Ponderomotive electron optics has recently attracted attention because structured optical fields can provide round-lens operation with functionalities that are difficult to achieve using conventional electron-optical elements, including negative lens power and negative spherical aberration. A largely unexplored aspect is how ponderomotive lenses disperse with electron energy and whether that dispersion can be engineered for achromatization. Here we show that, for relativistic electrons, longitudinal and transverse optical components exhibit distinct energy dispersion because Lorentz-boost-induced polarization mixing modifies the effective lens action. Focusing a radially polarized annular beam produces two co-located Bessel-like components near focus, a transverse $J_1^2$ lens and a longitudinal component with relativistic mixing, forming a zero-separation doublet. Using a local Abbe number defined in the energy domain, we derive a single geometry condition for achromatization and evaluate the spot-size performance in a thin-lens model. We also identify parameter regions where the same element yields negative on-axis chromatic aberration, indicating a compact round-lens corrector regime.

Digital holography is an interference-based imaging technique capable of recording both the amplitude and phase of an electromagnetic field. It can be obtained at the laser illumination wavelength, but also with second-harmonic generation, since the latter is produced in a coherent process. Here we describe the development of a harmonic holographic microscope for 3D single-shot mapping of second-harmonic emitters. The knowledge of the scattered field, in amplitude and phase, in a given plane, that of the camera, allows its reconstruction in any other plane using the angular spectrum representation of the optical fields, a process called 3D numerical back-propagation. In order to probe the polarization dependence of the sample nonlinear response, we implement polarization multiplexing, in which a Wollaston prism creates two off-axis reference beams with orthogonal polarizations and non-parallel propagation directions. Each reference only interferes with the corresponding polarization component in the sample SHG emission, thus providing two independent sets of interference fringes which are easily separated in the angular spectrum representation. From a single measurement, two second-harmonic fields corresponding to orthogonal polarizations can be back-propagated. In the particular case of collagen, the second-harmonic polarization state can reveal the orientation or disorder of molecules and fibers. We demonstrate the feasibility of the method by reconstructing the spatial distribution of the second-harmonic field generated by collagen fibers in a rat-tail tendon sample and show that polarization-multiplexed holography can provide single-shot 3D mapping of biophysical parameters such as the helical pitch angle of collagen molecules.

In this study, we present a comprehensive investigation of the structural, optical, and electrical properties of Bifidobacterium longum subsp. longum 35624 (BB35) thin films, and demonstrate their application as a novel relative humidity sensor. UV-Visible spectroscopy revealed that BB35 exhibits two distinct optical absorption regions, corresponding to direct band gaps of 2.1 \pm 0.05 eV and 2.8 \pm 0.05 eV, as confirmed by Tauc plot analysis, establishing BB35 as a genuine wide-bandgap semiconductor material. Photoluminescence measurements under 280 nm excitation exhibited a broad emission spectrum, which was deconvoluted into four Gaussian peaks centered at 434 nm (2.86 eV), 499 nm (2.48 eV), 543 nm (2.3 eV), and 620 nm (2.0 eV), indicating the presence of multiple radiative recombination centers characteristic of semiconducting materials. Electrical characterization revealed dispersive charge transport with current decay following a power-law I \propto t^{-α} (α\approx 0.3), suggesting Poole-Frenkel conduction mechanism typically observed in disordered organic semiconductors. The relative humidity (RH) sensing performance of BB35 films was evaluated using gold interdigital electrodes across 15-90% RH range. The sensor exhibited reversible response with sensitivity increasing linearly from 0.85 to 4.80 as RH increased from 15% to 90%. The devices demonstrated excellent stability over two months with less than 5% degradation in baseline current. These results establish BB35 thin films as a promising eco-friendly semiconducting material for humidity sensing applications and open new avenues for integrating biological materials into electronic and optoelectronic devices.

Dual-comb distance metrology combines sub-um precision with absolute distance measurement over meter-scale non-ambiguity ranges. The most common interferometric implementation requires sophisticated approaches to achieve continuous streaming due to the need for digitization at near-gigasample/s rates. Here, we exploit the naturally low data burden of two-photon dual-comb LiDAR to demonstrate continuously-streamed distance metrology at 11.5 kHz, reaching a precision of nearly 1 um in an averaging time of 10 ms. Using free-running 500 MHz Er,Yb:glass femtosecond lasers we achieve sampling rates >10 kHz, sufficient to demonstrate the capture of a four-minute audio track from the displacement of a loudspeaker-mounted mirror. The ability to perform high-speed yet low-data-burden distance metrology using unstabilized lasers presents exciting opportunities for diverse applications in industrial machine-tool control and calibration.

We report the first realization of a fractional topological phase in a fully unitary, noninteracting discrete-time quantum walk implemented on finite cyclic graphs. Using a single-coin split-step cyclic quantum walk (SCSS-CQW), we uncover topological phenomena that are inaccessible within conventional cyclic quantum-walk dynamics. The protocol enables controlled engineering of quasienergy spectra, flat bands, and topological phase transitions through the step-dependency parameter and coin-rotation angle. We show that cyclic graphs with even and odd numbers of sites exhibit qualitatively different band structures, while rotational flat bands arise exclusively in $4n$-site cycles; a general analytic condition for their emergence is derived. The SCSS-CQW produces fractional winding numbers $\pm \frac{1}{2}$ (Zak phases $\pm \fracπ{2}$), in sharp contrast with the integer invariants of standard quantum walks. These fractional invariants lead to an unconventional bulk--boundary correspondence and support edge states beyond the usual integer topological classification. In the step-dependent protocol, transitions between distinct fractional winding sectors generate robust edge modes. Numerical simulations show that these states remain stable in the presence of both dynamic and static coin disorder as well as phase-preserving perturbations, while survival-probability analysis demonstrates their long-time persistence. Requiring only a constant number of detectors independent of the evolution time, the proposed scheme offers a minimal-resource and experimentally accessible platform for realizing fractional topology, flat bands, and protected edge states in small-scale synthetic quantum systems.

Granger causality (GC), a popular statistical method for the inference of directional influences between time series measured from a complex network, is sensitive to high-order (non-pairwise) interactions which fundamentally shape the collective network dynamics. This work introduces Partial Decomposition of Granger Causality (PDGC), a tool eliciting redundant and synergistic causal interactions in the pattern of information flow between the subsystems of physiological networks. The tool exploits the framework of partial information decomposition to dissect the multivariate GC from a set of driver random processes to a target process into unique effects carried exclusively by each driver, redundant effects carried identically by more drivers, and synergistic effects carried jointly by some drivers but not by any of them individually. Computation is based on multivariate state-space models expanded in the frequency domain to assess PDGC both in specific bands of physiological interest and in the time domain after whole-band integration. The spectral PDGC was tested in physiological networks probed by measuring the variability series of arterial pressure, heart period, respiration and cerebral blood velocity in patients prone to neurally-mediated syncope compared to healthy controls. This application revealed unprecedented modes of physiological interaction, related to the sympathetic control of low-frequency cardiovascular and cerebrovascular oscillations, characterizing distinctive patterns of autonomic dysfunction. The extraction of high-order causality patterns from the spectral GC favors dissecting the mechanisms of causal influence underlying multivariate interactions among oscillatory processes in many data-driven applications of network science.

Enhanced supercontinuum generation (SCG) is experimentally demonstrated in integrated silicon nitride (Si3N4) waveguides incorporating highly nonlinear graphene oxide (GO) in the form of two dimensional (2D) films. Onchip integration of the 2D GO films with precise control of their thickness is realized by using a transfer free and layer by layer coating method. The control of the film length and coating position is achieved via window opening in the upper silica cladding of the photonic integrated chips. Detailed SCG measurements are performed using the fabricated devices with different waveguide geometries and GO film thicknesses, and the results are compared with devices without GO. Significantly improved spectral broadening of ultrashort optical pulses with ultrahigh peaks powers exceeding 1000 W is observed for the hybrid devices, achieving up to 2.4 times improvement in the spectral bandwidth relative to devices without GO. Theoretical analyses for the influence of GO film thickness, coating length, coating position, and waveguide geometry are also provided by fitting the experimental results with theory, showing that there is still significant room for further improvement. This work opens up a promising new avenue towards improving the SCG performance of photonic integrated devices by incorporating functional 2D materials.

On-chip integration of highly anisotropic two-dimensional (2D) materials offers new opportunities for realizing high performance polarization selective devices. Obtaining optimized designs for such devices requires extensively sweeping large parameter spaces, which in conventional approaches relies on massive mode simulations that demand considerable computational resources. Here, we address this limitation by developing a machine learning (ML) model based on fully connected neural networks (FCNNs). Trained by using mode simulation results for low resolution structural parameters, the FCNN model can accurately predict polarizer figures of merits (FOMs) for high resolution parameters and rapidly map the global variation trend across the entire parameter space. We test the performance of the FCNN model using two types of polarizers with 2D graphene oxide (GO) and molybdenum disulfide (MoS2). Results show that, compared to conventional mode simulation approach, our approach can not only reduce the overall computing time by about 4 orders of magnitude, but also achieve highly accurate FOM predictions with an average deviation of less than 0.04. In addition, the measured FOM values for the fabricated devices show good agreement with the predicted ones, with discrepancies remaining below 0.2. These results validate artificial intelligence (AI) as an effective approach for designing and optimizing 2D-material based optical polarizers with high efficiency.

Accurate representation of interfaces and flux exchange is vital for coupled multiphysics simulations across a broad range of applications. Currently, coupling approaches are limited by the underlying discretization or to specific physical problems, restricting their generality. To remove these constraints, a consistent interface reconstruction and coupling method has been developed to bridge multiphysical computational domains. The proposed numerical framework contains two complementary steps. The first step reconstructs a continuous bounding surface from discretized spatial data using a weighted interpolation and marching-grid approach that preserves geometric fidelity across a wide range of resolutions. The second step consists of a conservative flux mapping algorithm that projects surface quantities such as aerodynamic loads, heat fluxes, or mass transfer onto nearby discrete elements while maintaining global conservation of flux quantities. This new numerical formulation allows for integration of the various partitioned domains and ensures consistent data transfer between them. The framework is tested on various geometries, and surface containment errors below 2.5% and flux-transfer errors below 1% are obtained. Transient simulations of uniform surface recession further confirm accuracy in coupled evolution, with predicted volume loss agreeing with analytical solutions to within 1%. The proposed method establishes a general and extensible framework for consistent interface reconstruction and coupling in multiphysics environments, providing a pathway toward unified treatment of discrete-continuum interactions for a wide range of systems.

An efficient method of computing the dynamics of a pure quantum state under the time-dependent Schrödinger equation is described: At each timestep, a restricted subspace of the potentially infinite-dimensional total Hilbert space is systematically and naturally constructed via the image of repeated applications of the Hamiltonian operator, and the time evolution is computed exactly within said subspace. The subspace is dynamically recomputed at each timestep such that it co-evolves with the state vector. We benchmark the method using the Holstein model and compare the formal information content of its representation to the matrix-product state formalism. The method is built from the ground up as a parallel algorithm for graphics processing units and is applicable to arbitrary Hamiltonians that are sparse in a given basis. It can be extended to open quantum system dynamics and/or time-dependent generators.

Indonesia's nickel ore export ban has driven rapid expansion of smelting and hydrometallurgical processing capacity at the Indonesia Morowali Industrial Park (IMIP), now the world's largest integrated nickel processing complex, on the coast of Central Sulawesi. Whether this industrialization has degraded the adjacent marine environment remains unquantified. We apply Bayesian structural time-series (BSTS) causal inference to a multi-decadal, multi-sensor satellite ocean color record of the diffuse attenuation coefficient at 490 nm, $K_d(490)$, to test for a causal link between IMIP expansion and nearshore turbidity change. A consensus structural breakpoint, a significant posterior causal effect estimated against a Banda Sea counterfactual, and a distribution-free placebo rank test collectively establish that coastal water clarity deteriorated after the transition from initial nickel pig iron production to hyper-expansion of high-pressure acid leaching facilities for battery-grade nickel. Satellite-derived land cover analysis independently corroborates this timing, showing substantial built-area growth and concurrent tree cover loss within the IMIP footprint. The resulting euphotic zone shoaling occurs in oligotrophic waters supporting high marine biodiversity, where even moderate optical degradation may impair coral photosynthesis and compress depth-dependent reef habitat. These findings quantify a marine environmental cost absent from Indonesia's mineral downstreaming policy discourse and demonstrate a transferable, satellite-based quasi-experimental framework for causal impact assessment at coastal industrial sites in data-limited tropical settings.

High-quality InAs quantum wells grown on InP are a promising platform for topological quantum information processing due to their large g-factor, strong Rashba spin-orbit interaction, and their compatibility with in-situ-deposited superconductors. In this work, we investigate InAs/InGaAs quantum wells grown on InP (001) wafers, focusing on how the layer structure and strain influence the electronic properties and surface morphology. By combining quantum transport measurements with atomic force microscopy, we show that the layer design predominantly affects the mobility anisotropy, which aligns well with the surface morphology. Surface characterization further reveals the mechanism of quantum well collapse when the layer thickness exceeds the strain limit. In addition, transport measurements demonstrate that quantum confinement has a clear impact on band nonparabolicity.

Recent theoretical studies show that nanoscale contact on dielectric substrates can induce flexoelectric polarization large enough to drive electron transfer. This has been supported by experimental evidence, indicating that contact electrification is inherently a coupled electromechanical phenomenon. In this work, we develop a computational model for flexoelectricity-driven contact electrification that integrates finite-deformation flexoelectricity with contact mechanics and physically motivated charge transfer. A tunneling transparency function is introduced to regulate the interfacial channel based on the WKB approximation, capturing the irreversible charge trapping during unloading. Three contact scenarios are investigated with specific hypotheses for charge transfer: unbiased metal-dielectric contact driven by surface polarization, biased contact restricted to carriers of a single polarity, and dielectric-dielectric contact where surface states with finite capacity limit the transferable charge. The model is compared with atomic force microscopy measurements on PMMA and PDAP substrates under both biased and unbiased conditions.For contact between identical dielectric materials, we show that geometric asymmetry in surface curvature is sufficient to induce charge separation, with polarity reversal occurring at a critical surface wavenumber. Three-dimensional simulations on random rough surfaces reproduce the mosaic charge distributions observed experimentally, confirming that contact-induced local strain gradient heterogeneity can generate spatially non-uniform charge patterns without introducing any material inhomogeneity.

We investigate the ripeness of kiwifruit nondestructively with near-infrared pulse illumination. With this measurements in time domain, only one frequency (the wavelength 800 nm is required. Measurements were performed on three golden kiwifruits over a period of ten days. To quantify changes in temporal profiles of the detected light, we introduce two indices: the relative ripeness $r(n)$ and the first Wasserstein distance $W_1(n)$. Both indices exhibit nonmonotonic behavior.

We predict and analyze boundary-driven exceptional points in semi-infinite Hermitian photonic waveguide lattices with a side-coupled defect. The exceptional points arise from coherent reflections at the lattice termination, which induce strong memory effects in the defect dynamics. Using an exact analytic approach, we derive the defect's non-Markovian memory kernel, revealing the trajectories and coalescence conditions of the resonances, which can be precisely tuned by the defect position and the coupling strength. Our results provide a simple and experimentally accessible platform for exploring memory-enabled non-Hermitian physics in Hermitian photonic lattices.

Classical molecular dynamics and electro-diffusion theories have achieved profound success in elucidating ion selectivity and gating mechanisms. However, reconciling strict selectivity with high flux permeation in Angstrom-scaled biological ion channels poses a universal challenge in nanoscale physics, as classical models consistently underestimate single-channel conductance. Using a non perturbative quantum transport framework, we calculate the ion permeation dynamics through the selectivity filter within a transfer matrix formalism. We demonstrate that quantum tunneling allows ions to bypass classical Arrhenius suppression, quantitatively recovering the experimental conductance of Na+ and K+ channels. Crucially, our findings reveal that the exploitation of quantum mechanics is a fundamental prerequisite for achieving macroscopic physiological efficiency. By reframing ion channels as mesoscopic quantum conductors, this work establishes a transformative paradigm in quantum biology and predicts distinct transport resonances in the terahertz regime.

A mechanism of quantum-mechanical tunneling is based on electron-wavefunction and is used to explain ohmic contact and tunnel diode. Tunneling is the important example of wave-particle duality. In this study, an attempt is made to explain these devices in particle description. As is well known, any object at room-temperature emits infrared (IR) photons due to blackbody radiation. The process of heavy doping can cause a lot of defects, e.g. vacancies and interstitials. The self-absorption of the IR emission could be achieved through sub-band-gap excitations due to defect-related levels in forbidden energy gap, creating carriers. In a heavily doped p-n junction diode, some of the IR-generated carriers diffuse into the junction which has a built-in field in a depletion layer. The built-in field then sweeps out the carriers, producing IR photocurrent. The IR photocurrent is regarded as the reverse current of the p-n junction according to the precedent where impurity-photovoltaic-effect is used to explain circuit devices (arXiv:2510.18226). Also, a reverse bias voltage increases depletion layer width, but carrier diffusion length out of the depletion layer remains unchanged, and more IR-generated carriers created in and near the depletion layer can contribute to the IR current. In addition, heavy doping results in not only a great many IR-generated carriers but also avalanche effect at low voltage. As a result, total reverse current dramatically increases with voltage, thus a Schottky diode can be altered to create an ohmic contact and a p-n diode can become a tunnel diode. Considering the wave-particle duality, the combination of quantum-mechanical and photovoltaic mechanisms is suggested to explain these devices. Each of the two mechanisms plays a big or small role in the explanation.

Optical neural networks (ONNs) based on programmable photonic integrated circuits (PICs) offer a promising route toward low-latency and energy-efficient deep learning. However, conventional photonic implementations of matrix-vector multiplication (MVM) rely on locally connected architectures, such as Mach-Zehnder interferometer (MZI) meshes, whose number of active components scales quadratically with matrix size, severely limiting scalability. Here, we present a scalable ONN that overcomes this limitation by exploiting the intrinsically diffractive and nonlocal nature of coherent light inside a silicon photonic chip. Our approach employs cascaded stages of multiport directional couplers (MDCs) interleaved with compact phase-shifter arrays, enabling strong nonlocal coupling among multiple optical modes. We show that an MDC-based optical unitary converter (OUC) requires only $3N$ phase shifters to achieve uniform coverage over the $N$-dimensional complex unitary group, in stark contrast to the $O(N^2)$ scaling of conventional MZI meshes. Based on the singular value decomposition, we demonstrate that an $N\times N$ MVM can be realized using only $7N$ phase shifters, breaking the traditional $O(N^2)$ scaling barrier. We experimentally implement a 32-input silicon photonic MVM chip with a tenfold reduction in active components and validate its performance on various classification tasks. Our results establish a practical pathway toward large-scale, energy-efficient, and reconfigurable photonic neural networks.

This study introduces the Bragg Frequency Convertor, a spatial-temporal-periodic grating that extends the concept of conventional Bragg gratings into the dynamic domain to achieve pure parametric frequency conversion. By time-modulating either the high-index or low-index layers of a quarter-wave Bragg grating, we demonstrate selective and directional frequency conversion: Modulation of the high-index layers selectively yields down-conversion, whereas modulation of the low-index layers leads to up-conversion. Such a pure frequency conversion emerges from the synergistic interplay of spatial and temporal periodicities. The output is thus dominated by the converted frequency, with the carrier and undesired time harmonics suppressed. We derive a coupled-mode theory explaining such a layer-dependent phase matching and validate it with full-wave simulations, showing tunable conversion efficiency via modulation phase. The advance presented here lies in recognizing the conventional Bragg grating not as a passive partner but as an enabling scaffold: its intrinsic spatial periodicity is exploited to impose the precise spectral constraints required for efficient temporal modulation and the generation of pure parametrically converted signals. This work establishes temporal Bragg gratings as a versatile platform for reconfigurable and spurious-free frequency conversion, with applications in optical systems, signal processing, spectral engineering, and quantum photonics.

Understanding how multi-scale network structure influences circadian rhythms in the suprachiasmatic nucleus (SCN) is essential for uncovering the principles of rhythmic robustness and synchronization. Previous studies using synthetic SCN networks suggested a size-dependent phenomenon, in which rhythmic activity initially strengthens with network size and then saturates, but it remains unclear whether this occurs in real SCN networks. Here, we apply geometric branch growth (GBG) and geometric renormalization (GR) to generate self-similar scaled-up and scaled-down replicas from a single-scale functional mouse SCN network. Unlike synthetic models, these SCN replicas do not exhibit size-dependent rhythms: average period, amplitude, and synchronization remain stable across scales. By increasing the average degree with network size, we reproduce size-dependent rhythms and show that they arise from network connectivity, whereas low-degree networks fragment and fail to sustain oscillations. Disrupting clustering self-similarity slightly reduces synchronization, but circadian rhythms remain robust, indicating that average degree, rather than clustering, is the dominant structural driver. These results highlight the resilience of SCN rhythms to network scaling and provide a framework for linking multi-scale network structure to biological timekeeping.

Standard 6 megavolt (MV) radiotherapy is limited by source size and secondary electron range to minimum radiological penumbra widths of ~2-3 mm. This study investigates sharper beams via upright radiotherapy with lower energies and extended source-to-patient distances. A 2.5 MV beam from a clinical linac was delivered at a 4 m source-to-phantom distance (2.5 MV-ED). Lateral profiles and percent depth doses were measured in a solid water phantom with radiochromic film and an ion chamber. These single beam measurements were used to benchmark TOPAS Monte Carlo simulations. The validated 2.5 MV-ED model was then used to simulate upright deliveries with a conical beam geometry. These simulations were compared against equivalent plans generated for standard 6 MV-FFF coplanar deliveries at 1 m from the source. The 2.5 MV-ED single 28x28 mm^2 beam produced a measured 80%-20% penumbra of 1.0 +-0.1 mm, compared to 2.4 mm Jaw-defined penumbra for a standard 6MV-FFF beam. The doses at 10 cm depth were 52% vs. 56%, and the surface doses were 22% vs. 38% for the 2.5MV-ED and standard 6 MV-FFF respectively. Conical geometry Monte-Carlo simulations using the 2.5 MV-ED beams demonstrated significantly sharper composite dose fall-off in all cardinal directions compared to coplanar 6MV-FFF plans. For a spatially fractionated lattice example plan with 5 mm diameter high dose spheres, the 2.5 MV-ED conical approach achieved a peak-to-valley dose ratio of 4.5-5.2, compared to the 2.6-2.9 achievable with a standard 6 MV-FFF clinical system. Low-energy, extended-distance photon beams can provide sharper penumbra and lower surface dose, while maintaining comparable depth-dose penetration as standard 6 MV setups. Combined with upright patient setups, ultra-sharp dose distributions with enhanced treatment conformity, reduced toxicity, and higher-fidelity dose modulation are possible.

Stimulated Brillouin scattering (SBS) microscopy enables label-free biomechanical imaging, with Brillouin gain serving as a critical contrast parameter for quantitative analysis. However, the influence of sample-induced refractive index (RI) heterogeneity on gain measurements remains poorly understood. Here, we quantitatively investigate, how RI mismatch affects SBS microscopy using finite element simulations and experiments on a phantom sample comprising polydimethylsiloxane beads embedded in agarose gel. We demonstrate that RI heterogeneity induces focal field distortion that reduce pump-probe beam overlap, resulting in attenuated Brillouin gain and degraded shift precision at material interfaces. Crucially, we establish that fiber-coupling efficiency, commonly used for system alignment, cannot serve as a linear proxy for Brillouin gain due to its heightened sensitivity to focal field distortion.

Optical beams propagating through atmospheric turbulence undergo spatiotemporal intensity fluctuations that deviate significantly from an ideal Gaussian profile. In this work, we present a unified theoretical and experimental framework for quantifying and mitigating these turbulence-induced distortions by coupling a higher-order statistical characterization technique with optical waveguide spatial filtering. The statistical characterization employs a Cholesky-whitened Gram--Charlier expansion that decomposes the two-dimensional beam intensity distribution into a Gaussian core augmented by third- and fourth-order cumulant corrections, thereby isolating skewness and excess kurtosis as quantitative non-Gaussianity indicators. Concurrently, the propagation of the distorted beam through a dielectric waveguide is analyzed to demonstrate that higher-order spatial modes, which carry the dominant share of turbulence-induced structural distortions, encounter a cutoff condition governed by the normalized frequency parameter and subsequently undergo exponential attenuation along the propagation direction. The waveguide thus acts as a passive spatial mode filter that selectively transmits the fundamental guided mode while suppressing radiative higher-order modes. The fitted beam volume, derived from the Gram--Charlier intensity model, serves as a unified scalar diagnostic that tracks the frame-by-frame evolution of turbulence-induced structural changes. Experimental measurements validate the theoretical predictions, demonstrating a substantial reduction in intensity fluctuations and a recovery of Gaussian beam statistics after waveguide propagation.

We present a unified theoretical framework linking second and fourth order statistical correlations of stochastic electromagnetic beams to the scintillation index observed after propagation through Kolmogorov atmospheric turbulence. We derive the beam coherence polarization matrix and its propagation law in a random medium. An algebraic connection is established between the second order polarization matrix J, its fourth order Gram counterpart obtained from the square of J, the classical degree of polarization P, the fourth order degree of polarization, and the scintillation index of the beam intensity. A key result is a closed form purity relation connecting the trace of the squared polarization matrix to the square of its trace, which shows that unpolarized natural beams exhibit a large scintillation index. The analysis demonstrates that scintillation can be reduced by polarizing the beam using suitable polarizer sets. This reduction occurs independently of the atmospheric turbulence strength parameter. Experimental results further show that simultaneous control of coherence and polarization, achievable using a pseudo random phase plate, provides a practical approach for minimizing scintillation in free space optical communication links.

We present a pseudo-spectral method for solving the three-dimensional Boussinesq equations in unbounded cylindrical domains, specifically tailored for rotating, stably stratified flows subject to strong azimuthal shear. To effectively capture the global geometry without sacrificing spectral accuracy, the spatial discretization employs Fourier expansions in the azimuthal and axial directions alongside mapped associated Legendre polynomials in the radial direction. This basis spans the semi-infinite domain while analytically resolving the coordinate singularity at the origin. While this spectral framework ensures high spatial fidelity, the temporal integration of these rotating shear flows presents a formidable computational challenge due to the numerical stiffness driven by fast restorative wave forces and rapid background advection. To circumvent this, we develop an exponential time differencing (ETD) scheme that analytically integrates the fully coupled linear operator, including the radially dependent advective cross terms. By encoding the physical resonance characteristics and stability limits of the background flow directly into the integration operators, the proposed ETD formulation removes the numerical stability constraints imposed by the background shear and stratification. This permits integration time steps scaled by the slow macroscopic evolution of the physical instabilities rather than the fast background kinematics, offering significant performance gains over standard mixed implicit-explicit schemes. The method's accuracy and stability are validated through the precise conservation of energy and angular momentum, establishing a robust framework for simulating instabilities in astrophysical and geophysical vortices.

In the early-stage design of advanced electronic packages, designers face a critical trade-off between simulation fidelity and computational turnaround time. Conventional early-stage methodologies typically achieve speed by relying on steady-state assumptions and structural homogenization. While computationally efficient, these approximations fundamentally fail to capture dynamic thermal events and stress concentrations at fine-grained internal interfaces, effectively masking failure mechanisms driven by transient signal bursts. In this work, we present a GPU-accelerated transient coupled Electromagnetic-Thermal-Mechanical solver that resolves this bottleneck. The proposed solver enables full-scale, non-homogenized, time-domain simulation of large-scale packages with runtimes amenable for rapid design iteration. Simulation of a NEC SX-Aurora TSUBASA package demonstrates that the tool allows for the identification of signal-induced adiabatic stress that is typically invisible to steady-state and homogenized baselines. This capability brings sign-off level physics fidelity to the early design phase, facilitating the prevention of costly late-stage design failures and broader transient thermal performance degradation risks.

The unique properties of two-dimensional moire systems have been widely studied from many perspectives. However, relatively little work has explored how the real space structure of the moire systems can directly engender novel properties and functionalities. In this work, we exploit the feature that moire excitons naturally form an ordered superlattice with a lattice constant comparable to the wavelength of the resonant light, which enables intriguing cooperative optical responses. Particularly, we show that the collective moire exciton states can have either strongly enhanced (superradiant) or suppressed (subradiant) radiative decay rate, depending on their in-plane wavevector. These super- and subradiant states can be efficiently switched by a gate-induced electric field gradient. Moreover, the cooperative transmittance $T$ of the nanometer-thick moire system can be switched from $T \approx 0$ (opaque) to $T \approx 1$ (transparent) with less than $2~\%$ heterostrain or a $1^{\circ}$ adjustment in the twist angle $θ$. These features are robust against non-radiative losses and inhomogeneity, making the moire system a highly versatile platform for cooperative quantum optics with potential applications in e.g., single photon storage and switching.

We demonstrate a minimally-destructive in situ technique for measuring and stabilizing slowly-drifting magnetic fields in ultracold-atom experiments. While conventional magnetic-field sensors such as Hall, giant magnetoresistive, or fluxgate-based devices are broadly used, their accuracy, precision and dynamic range can be limited. In addition, these sensors are typically positioned at least several centimeters away from the in-vacuum atomic system, as their operation creates perturbing magnetic fields, and their placement is limited by geometric constraints imposed by the vacuum system. We overcome these issues by using the atomic system itself as a built-in magnetometer. To that end, we employ a pair of weak measurements to determine the Zeeman splitting -- and thereby the magnetic field -- of a magnetically sensitive atomic transition. We provide closed-form expressions quantifying the trade-offs between measurement noise, dynamic range, and atom loss. This procedure is demonstrated with ultracold Rb-87, weakly measured using partial-transfer absorption imaging. We then incorporate a Kalman filter to stabilize the magnetic field; this eliminated long-term drift in the ambient field (as high as ~70 nT/hr) in exchange for a modest increase in shot-to-shot variability from 1.8(2) nT to 2.0(2) nT.

A chirped-pulse electro-optic (EO) dual comb system operating near 821 nm is used to generate cavity enhanced dual combs in the ultraviolet (UV) region near 323 nm with an optical bandwidth of up to 90 GHz. The UV combs result from intracavity sum frequency mixing of the non-resonant near-IR combs with a cavity enhanced field at 532 nm in a nonlinear crystal. The cavity is pumped with 1 W at 532 nm and seeded with less than 10 mW in the near-IR region to generate a few uW of UV power within a detection bandwidth of < 20 MHz. The UV power is enhanced by 100-fold relative to single pass methods and is readily detectable using an avalanche photodiode. The system is used to measure the high-resolution UV comb spectra of Rb atoms near 323 nm (9 2P3/2 <- 5 2S1/2). This method is easily extendable across the 300 nm to 400 nm region using telecom and near-IR dual EO comb sources for seeding.

This study presents a computationally efficient framework for accelerated alloy discovery that uses the non-interacting electron density to capture intrinsic structure-property relationships in refractory high-entropy alloys (HEAs). Unlike state-of-the-art approaches relying on expensive, self-consistent density functional theory calculations, our method employs the non-interacting electron density as the primary structural descriptor. By extracting physical features through directionally resolved two-point spatial correlations and compressing them via Principal Component Analysis, we efficiently map the design space. Coupling these descriptors with Bayesian active learning, we achieve a normalized mean absolute error (NMAE) of <2% for the bulk modulus of Al-Nb-Ti-Zr alloys using only 10 training samples. Furthermore, we demonstrate that the model learns an electronic packing manifold that is transferable across distinct chemical species within refractory HEAs. Validated on a distinct 7-component refractory system (Mo-Nb-Ta-Ti-V-W-Zr) containing four elements entirely absent from the training data, the framework enables rigorous zero-shot extrapolation. Moreover, by augmenting the base model with just 20 samples from the target domain, we achieve high-fidelity predictions (NMAE < 3%) for 7-component alloys, reducing data acquisition costs by orders of magnitude compared to standard workflows. These results establish the non-interacting electron density as a rigorous, extrapolative descriptor for vast compositional landscapes.

The geometries of surface wave modes are determined by the highly nontrivial interplay of capillarity and wetting effects at the boundaries of their domain. Aside from idealised scenarios, this commonly leads to unknown boundary conditions, thereby hindering theoretical formulation and experimental analysis. To address this problem, we introduce Extracted Mode Tracking (EMT), a data-analysis framework to obtain instantaneous amplitude and phase content of axisymmetric surface-wave modes from spatio-temporal measurements. This approach uses unsupervised machine learning techniques to extract a basis of wave modes directly from collected data; the spatial profiles require no prior theoretical modelling, and so the issue of unknown boundary conditions is circumvented. Time-resolved mode amplitudes are reconstructed by geometric fitting at each recorded time-step, and the success is evaluated by a spectral signal-to-noise quantifier. Capabilities and limitations of EMT are systematically benchmarked on synthetic datasets, finding strong resilience against noise, improved accuracy over alternative methodologies, and the ability to operate with restricted domains which poses significant merit for use in experimental systems with limited measurement field-of-view. Finally, we conduct a Faraday-wave experiment in a regime highly sensitive to boundary effects in order to further validate the method, and demonstrate the observational access to nonlinear wave-dynamics enabled by EMT. These results establish EMT as a general tool for analysing wave mode dynamics of axially-symmetric fluid interface systems, and open pathways for quantitative studies of nonlinear mode-interactions, stability, and turbulence.

Machine learning is rapidly accelerating materials and chemical discovery, but most current models target energies, forces, or selected molecular properties rather than the underlying many-body electronic structure. Learning electronic-structure proxies, such as reduced density matrices, offers a path to surrogates that can predict a broad range of observables from a single ML model. Short of learning the full wavefunction, the two-electron reduced density matrix (2-RDM) is among the most information-rich, minimally lossy targets, providing direct access to expectation values of arbitrary one- and two-electron operators regardless of the strength of the underlying electron correlation. Here we show that learning the 2-RDM is a feasible goal, yielding exceptionally accurate models. We develop surrogates for correlated wavefunction methods (including configuration interaction and coupled cluster) that yield 2-RDMs with sufficient fidelity to provide direct, training-free access to energies and forces for driving energy-conserving molecular dynamics. To tackle realistic molecular condensed phases, we leverage a many-body expansion of the 2-RDM, using our ML models to supply the expansion terms and enabling ML-powered, coupled-cluster-quality electronic structure and energetics for large solvated systems. As a demonstration, we showcase a coupled-cluster-level electronic-structure calculation of glucose solvated by 500 water molecules achieved at Hartree-Fock cost. This work establishes a general framework for learning correlated electronic structure with high fidelity and deploying it to systems beyond the reach of conventional ab initio methods.

Ising machines exploit the natural dynamics of physical systems to minimize the Ising Hamiltonian and thereby address computationally hard combinatorial optimization problems. This paradigm has motivated a range of physical implementations. In the electronic domain, coupled networks of oscillators and bistable latches have emerged as two prominent realizations of Ising machines and are the focus of the present work. Despite this common abstraction, we demonstrate that differences in the underlying physical dynamics of oscillators and latches lead to fundamentally different stability properties and computational behavior of the resulting dynamical systems. Specifically, we show analytically that in Bistable Latch Ising Machines (BLIMs) all discrete Ising configurations possess identical linear stability, whereas in Oscillator Ising Machines (OIMs) the Jacobian spectrum depends explicitly on the spin configuration, enabling selective destabilization of higher-energy states. Evaluating the performance of both models on MaxCut instances of varying sizes, we find that this difference in stability structure yields consistently higher-quality solutions with OIMs. These results highlight how the characteristics of the device nonlinearity directly shape the dynamical and functional properties of Ising machine implementations.

Quantitative prediction of thermodynamic properties in solution is essential for translating atomistic simulations into reliable chemical insight. As an exemplar system, the behaviour of CaCO$_3$ in water has been widely studied to understand its mineralization in seawater, with potential implications for carbon-capture strategies. However, making accurate computational predictions has been a long-standing challenge, requiring both highly accurate electronic structure methods and extensive statistical sampling. Here, we combine advances in machine learning and electronic structure theory to fully resolve the ion pairing free energy of CaCO$_3$ with explicit solvation. We show that achieving quantitative agreement with experiment requires going beyond the standard density functional theory up to the "gold-standard" coupled cluster theory with single, double, and perturbative triple excitations [CCSD(T)]. We generate a set of systematically improvable models, enabling reliable insights into the initial association mechanism of Ca and CO$_3$ ions prior to nucleation while fully quantifying enthalpic and entropic effects. Our results demonstrate that CCSD(T)-level thermodynamic predictions of complex aqueous systems can now be routinely achieved.

In modern power systems, edge devices serve as local hubs that collect data, perform on-site computing, sense electrical parameters, execute control actions, and communicate with neighboring edge devices as part of the larger grid. However, as the number of monitored nodes and control loops grows, traditional edge devices face serious limits. They can become overloaded by complex signal processing and decision tasks, causing delays and higher energy use. Standard sensors hit a noise floor that prevents them from detecting miniature changes, making it harder to spot early signs of faults or instability. Meanwhile, conventional communication links struggle with bandwidth limits, security risks, and rising encryption demands, which together slow down and weaken the transfer of critical grid information. Quantum technologies have the potential to overcome these challenges. Quantum computers can deliver exponential speed-ups for optimization and machine-learning tasks that ordinary processors cannot handle. Quantum sensors can sense signals with atomic precision, giving edge devices a more precise view of grid dynamics. Quantum communication techniques, including quantum key distribution, offer methods to achieve information-theoretic security and ensure that information arrives quickly and without tampering. We explore how quantum technologies can be integrated into edge devices, highlighting both opportunities and challenges.

Transforming rovibronic Hamiltonians of molecular systems from the $ΛS$ (Hund's case a) basis to the adiabatic $Ω$ representation is widely used to "remove" spin-orbit coupling (SOC) and enable single-state treatments of spectra and dynamics. We show that this simplification is only apparent: the SOC elimination necessarily generates sizeable non-adiabatic couplings (NACs) from the nuclear kinetic energy operator. Neglecting these spin-orbit-induced NACs causes severe errors in rovibronic energies and transition properties. Using an analytically tractable two electronic state model and high-accuracy variational benchmarks, we derive the exact conditions for numerical equivalence between $Ω$ and $ΛS$ formulations and quantify how missing NAC terms and bond-length-dependent spin factors degrade predictions. We implement a complete $Ω$-representation workflow in Duo for diatomics, fully transforming all Hamiltonian terms and enabling side-by-side $Ω$ vs $ΛS$ calculations. For common single-state pipelines (e.g., LEVEL), we provide diagnostics that flag unsafe regimes and practical remedies to restore accuracy. The results deliver actionable guidance for spectroscopy, photophysics, and kinetics: $Ω$-based single-state approximations are reliable only when interacting states are well separated in the Franck-Condon region; otherwise, explicit non-adiabatic terms are required - even for "forbidden" transitions.

We propose a physics-informed anomaly detection framework for collider data based on a Bayesian latent diffusion model. Our method combines a probabilistic encoder with diffusion dynamics in the latent space, allowing for stable and flexible density estimation while explicitly enforcing physics constraints, such as mass decorrelation and regularization of latent correlations. We train and test the model on simulated LHC jet data and evaluate its performance using seed-averaged ROC curves together with discovery-oriented metrics. Through a series of ablation studies, we show that the diffusion process, Bayesian regularization, and physics-motivated loss terms each contribute in a complementary way: they help stabilize training and improve generalization, even when the gains in peak performance are moderate. Overall, our results emphasize the importance of incorporating both uncertainty estimates and physics consistency when building reliable anomaly detection methods for new Physics searches in high-energy physics.

Precise sensing and control of spatial mode content is essential for the performance of precision optical systems, particularly interferometric gravitational-wave detectors, where misalignment and mode mismatch can lead to significant optical losses and degraded quantum noise suppression. Conventional approaches, including heterodyne wavefront sensing and phase camera techniques, are effective but can be limited by hardware complexity and systematic uncertainties arising from restricted reference-beam overlap. This paper presents a novel two-step deep learning pipeline for robust beam diagnostics based solely on beam intensity images. In the first stage, a multi-intensity-image convolutional neural network (CNN) performs accurate mode decomposition, recovering the complex modal content of distorted beams. In the second stage, the predicted mode coefficients are fed into a downstream regression network that simultaneously estimates all eight degrees of freedom (DoFs) associated with misalignment and mode mismatch, including beam tilt, lateral offset, and waist size and position mismatches in both transverse directions. The proposed CNN-based framework achieves a mean absolute error (MAE) of 0.0034 in the mode decomposition stage, which propagates to a total MAE of 0.0062 in the recovered beam imperfection parameters at the final stage. This corresponds to an average residual optical loss of 39 ppm per DoF (310 ppm total). This approach relies only on standard CCD imaging and is robust to random intensity noise, eliminating the need for complex interferometric hardware. The results demonstrate that the proposed deep learning pipeline enables real-time, high-accuracy wavefront sensing and mode-mismatch diagnostics, providing a scalable and hardware-efficient tool for improving the stability and sensitivity of precision optical systems.

Targeted hub removal is known to weaken connectivity in heterogeneous networks. We show that in Barabási--Albert networks the same intervention can also shift Watts threshold dynamics across the cascade critical point. For BA networks with $N=2{,}000$ and $m=2$, removing the top 10\% of nodes by degree raises the bond-percolation threshold from $p_c=0.174$ to $0.776$ and, at $\varphi=0.22$, increases mean cascade size from $0.86\%$ (95\% CI 0.43--1.30) to $23.1\%$ (21.3--24.9). A controlled hub-vulnerability experiment on fixed topology shows that most of this cascade effect is dynamical: lowering hub activation thresholds produces much larger cascades even without deleting nodes, while deletion partly offsets the increase by removing edges. Using a configuration-model approximation, we derive the post-removal branching factor $z_1$ and identify a window in which the original network is subcritical but the hub-removed network is supercritical. The effect persists across system sizes and is not seen in matched ER or WS controls. These results identify a regime in which hub removal simultaneously worsens connectivity and cascade exposure in BA networks.

Phylodynamics is used to estimate epidemic dynamics from phylogenetic trees or genomic sequences of pathogens, but the likelihood calculations needed can be challenging for complex models. We present a neural Bayes estimator (NBE) for key epidemic quantities: the reproduction number, prevalence, and cumulative infections through time. By performing quantile regression over tree space, the NBE allows us to estimate posterior medians and credible intervals directly from a reconstructed tree. Our approach uses a recursive neural network as a tree embedding network with a prediction network conditioned on time and quantile level to generate the estimates. In simulation studies, the NBE achieves good predictive performance, with conservative uncertainty estimates. Compared with a BEAST2 fixed-tree analysis, the NBE gives less biased estimates of time-varying reproduction numbers in our test setting. Under a misspecified sampling model, the NBE performance degrades (as expected) but remains reasonable, and fine-tuning a pre-trained model yields estimates comparable to those from a model trained from scratch, at substantially lower computational cost.

How does the brain optimize sensory information for decision-making in new tasks? One hypothesis suggests learning reduces redundancy in neural representations to improve efficiency, while another, based on Bayesian inference, predicts learning increases redundancy by distributing information across neurons. We tested these hypotheses by tracking population responses in macaque cortical area V4 as monkeys learned visual discrimination tasks. We found strong support for the Bayesian predictions: task learning increased redundancy in neural responses over weeks of training and within single trials. This redundancy did not reduce information but instead increased the information carried by individual neurons. These insights suggest sensory processing in the brain reflects a generative rather than discriminative inference process.

The standard engineering approach when facing uncertainty is modelling. Mixing data from a well-calibrated model with real recordings has led to breakthroughs in many applications of AI, from computer vision to autonomous driving. This type of model-based data augmentation is now beginning to show promising results in biosignal processing as well. However, while these simulated data are necessary, they are not sufficient for virtual neurophysiological experiments. Simply generating neural signals that reproduce a predetermined motor behaviour does not capture the flexibility, variability, and causal structure required to probe neural mechanisms during control tasks. In this study, we present an in silico neuromechanical model that combines a fully forward musculoskeletal simulation, reinforcement learning, and sequential, online electromyography synthesis. This framework provides not only synchronised kinematics, dynamics, and corresponding neural activity, but also explicitly models feedback and feedforward control in a virtual participant. In this way, online control problems can be represented, as the simulated human adapts its behaviour via a learned RL policy in response to a neural interface. For example, the virtual user can learn hand movements robust to perturbations or the control of a virtual gesture decoder. We illustrate the approach using a gesturing task within a biomechanical hand model, and lay the groundwork for using this technique to evaluate neural controllers, augment training datasets, and generate synthetic data for neurological conditions.

Understanding animal behavior from video is essential for neuroscience research. Modern laboratories typically collect two complementary data streams: skeletal keypoints from pose estimation tools and raw video recordings. Keypoint-based methods are efficient but suffer from geometric ambiguity, environmental blindness, and sensitivity to occlusions. Video-based methods capture rich context but require processing every frame, making them impractical for the hundreds of hours of recordings that modern experiments produce. We introduce LookAgain, a multimodal framework that combines the efficiency of keypoints with the representational power of video through on-demand visual grounding. During training, LookAgain uses dense visual features to pretrain a motion encoder and to train a gating module that learns which frames require visual context. During inference, this gating module activates visual processing only when keypoint signals are ambiguous, while maintaining performance comparable to using all frames. Experiments on single-animal and multi-animal benchmarks show that LookAgain achieves strong performance with significantly reduced computational cost, enabling high-quality behavior analysis on long-duration recordings.

Recent experimental studies indicate that visual cognition is accompanied by slowly propagating biophysical travelling waves in cortical tissue. Here we propose polarization waves as a coherent physical framework for visual cognition. We first compute the propagation of scalar potential fields generated by impressed ionic currents in the primary visual cortex using a telegraph-type model and extract the velocity of the moving potential ridge. By exploiting the linear convolution structure, we then demonstrate that the scalar potential field and the polarization wave, arising from slowly oscillating neuronal dipoles, propagate with identical velocities. Remarkably, this velocity coincides with the independently predicted propagation speed of the cognitively inferred modulated wave (~1.5 cm/s). Because ionic influx entering a single optic-nerve channel integrates signals from more than a hundred photoreceptors, the resulting polarization field necessarily spans a distribution of wave numbers. We show that amplitudes of such multi-k polarization waves undergo dispersive spreading in time, which possibly suppresses cross-channel interference in visual perception.

A directed phylogenetic network is tree-child if every non-leaf vertex has a child that is not a reticulation. As a class of directed phylogenetic networks, tree-child networks are very useful from a computational perspective. For example, several computationally difficult problems in phylogenetics become tractable when restricted to tree-child networks. At the same time, the class itself is rich enough to contain quite complex networks. Furthermore, checking whether a directed network is tree-child can be done in polynomial time. In this paper, we seek a class of undirected phylogenetic networks that is rich and computationally useful in a similar way to the class tree-child directed networks. A natural class to consider for this role is the class of tree-child-orientable networks which contains all those undirected phylogenetic networks whose edges can be oriented to create a tree-child network. However, we show here that recognizing such networks is NP-hard, even for binary networks, and as such this class is inappropriate for this role. Towards finding a class of undirected networks that fills a similar role to directed tree-child networks, we propose new classes called $q$-cuttable networks, for any integer $q\geq 1$. We show that these classes have many of the desirable properties, similar to tree-child networks in the rooted case, including being recognizable in polynomial time, for all $q\geq 1$. Towards showing the computational usefulness of the class, we show that the NP-hard problem Tree Containment is polynomial-time solvable when restricted to $q$-cuttable networks with $q\geq 3$.

Genetic interactions and protein co-dependencies shape cellular fitness, buffering capacity, and disease vulnerability. However, systematic integration of co-dependency relationships across heterogeneous datasets remains limited. Here, we present HIDDENdb (Harnessing Intelligent Data Discovery to Explore Gene Networks), a comprehensive database that captures genetic and protein co-dependencies inferred from large-scale perturbation screens, multi-omics datasets, and curated interaction repositories. HIDDENdb integrates genome-wide loss-of-function screens (CRISPR and shRNA) with other unbiased resources (BioGRID-ORCS and GWAS) to construct a map of co-dependency relationships across diverse biological contexts. Using robust statistical modeling and network inference approaches, we identify modules of genes and proteins exhibiting shared dependency patterns across cell lines. Notably, top-ranked gene-gene co-dependency pairs are enriched for high-confidence AlphaFold-predicted protein-protein interfaces, suggesting that a subset of inferred functional relationships may reflect underlying structural interactions. Importantly, the database enables users to explore co-dependency networks interactively. HIDDENdb is freely accessible through a web-based interface at https://bofillderoslab.shinyapps.io/hiddendb/.

Obesity does not emerge abruptly; rather, it develops gradually over extended periods. The gradual progression often prevents early recognition of physiological changes until excess adiposity is established. A common belief is that weight reduction can be achieved simply by "eating less and moving more". Although reductions in caloric intake and increases in physical activity are fundamental principles of weight management, this perspective oversimplifies a complex and adaptive biological system. Metabolic rate, hormonal regulation, behavioral factors, and compensatory physiological responses all influence the body's resistance to changes in weight. During weight loss, reduced metabolic rate and increased efficiency make maintaining a caloric deficit increasingly difficult. Conversely, during periods of overfeeding, resting metabolic rate, the thermic effect of food, and non-exercise activity thermogenesis increase with rising body weight, partially offsetting the caloric surplus and slowing weight gain. However, these compensatory responses are asymmetrical, with stronger and more persistent adaptations to underfeeding than to overfeeding. This asymmetry helps explain why weight gain often occurs gradually and why sustained weight loss is biologically challenging. In this work, we employ a lambda-omega model from dynamical systems theory to describe metabolic regulation in response to lifestyle perturbations. We introduce time-varying parameters that allow the regulatory coefficients to evolve gradually under sustained environmental and physiological stressors. By allowing lambda(t) and omega(t) to vary over time, the model captures progressive shifts in the metabolic set-point and deformation of the underlying dynamical landscape. This framework enables exploration of transitions between metabolic states and long-term adaptations that shape trajectories of weight gain and loss.

We continue recent attempts to put together concepts and results of Chemical Reaction Networks theory (CRNT) and Mathematical Epidemiology (ME), for solving problems of stability of positive ODEs. We provide first an elegant CRN-flavored generalization of the most cited result in ME, the Next Generation Matrix (NGM) theorem. We review next the "symbolic-numeric approach of Vassena and Stadler, which tackles bifurcation problems by viewing the characteristic polynomial of the Jacobian at fixed points as a formal polynomial in the "symbolic reactivities", and identifies its coefficients as "Child Selection minors of the stoichiometric matrix". We also review two applications of this approach using the Mathematica package Epid-CRN tools from both CRNT and ME.

Motivation: GWAS (genome-wide association study) summary statistic files are essential inputs for polygenic risk score (PRS) calculation. However, identifying suitable files across thousands of catalog entries typically requires downloading large datasets and manually inspecting their column structures, a process that is both time-consuming and storage-intensive. Results: We present GWASPoker, a phenotype-driven, GWAS-Catalog-specific pre-download triage tool that scans candidate GWAS files for PRS column availability through partial downloads and header detection, without requiring full-file transfer. Analysing 60,499 records from the GWAS Catalog, 60,281 (99.6%) contained accessible download links, of which 54,026 (89.6%) were successfully partially downloaded and parsed across 20 file formats, yielding 724 unique header signatures. Across 13 phenotypes, 84 of 85 manually curated GWAS files (98.8%) were automatically retrieved and processed. Header validation against fully downloaded files showed exact agreement in 23 of 28 cases (82.1%). Availability and implementation: GWASPoker is implemented in Python 3 and is freely available at https://github.com/MuhammadMuneeb007/GWASPokerforPRS under the MIT licence. Example outputs and documentation are provided in the repository. The tool was tested on Linux (HPC cluster) with Python 3.8 or later. The LLM-based code-generation step is entirely optional; a rules-based column-mapping template is provided for fully offline use.

Frogeye Leaf Spot (FLS), caused by Cercospora sojina, poses a significant threat to soybean production, with yield losses of 30-60%. Traditional mass-action models assume homogeneous mixing, which rarely holds in real fields and limits their ability to inform FLS management. To address this, we developed a network-based model that incorporates real-field structure to improve FLS management in soybeans. Using approximate Bayesian computation, we estimated key epidemiological parameters and found that infection origin can shift the balance between transmission routes. Data analyses indicated that tillage and non-tillage plots did not differ significantly in fungal spread, decay, or disease severity. Finally, we show that early, targeted roguing is more effective than delayed or random removal. Together, these findings offer science-based guidance for FLS management and highlight the value of network-based models to inform agricultural disease control.

We investigate the impact of mode-dependent loss (MDL) on the statistics of the signal-to-noise ratio (SNR) in coupled-core multi-core fiber (CC-MCF) systems. Through numerical and theoretical simulations, we present an in-depth analysis of the impact of MDL on received amplified spontaneous emission (ASE) noise and nonlinear interference (NLI), as well as their joint contribution to the SNR. We show that MDL induces different statistics on the two noises and discuss the differences with single-mode polarization-dependent loss. Moreover, we investigate the impact of spatial mode dispersion (SMD) on the MDL-induced impairment, offering insights on their joint effects on ASE and NLI.

This letter investigates multi-mode pinching antenna systems (PASS), where signals of multiple orthogonal modes can be transmitted within a dielectric waveguide and radiated by pinching antennas (PAs). This enables mode-domain multiplexing for efficient multi-user communications using a single waveguide. In particular, two operating protocols are proposed, namely mode selection and mode combining. Mode selection enforces each PA to predominantly radiate signal power of one single mode, while mode combining allows each PA to flexibly radiate power of multiple modes. Based on the two protocols, a sum rate maximization problem is formulated for multi-mode PASS-enabled multi-user downlink communications, where the transmit beamforming, PA positions, and PA propagation constants are jointly optimized. To address this rapidly oscillating and highly nonconvex problem, a particle swarm optimization (PSO) based Karush-Kuhn-Tucker (KKT)-parameterized beamforming (PSO- KPBF) algorithm is proposed. KKT-conditioned solutions are exploited to guide the swarm search, thus reducing the search space and achieving fast convergence. Numerical results demonstrate that: 1) Even using a simple uniform mode-combining design, the multi-mode PASS significantly outperform conventional single-mode PASS and hybrid beamforming systems; and 2) Mode combining achieves high spectral efficiency, while mode selection approximates its performance with a lower hardware complexity. Code is released at https://github.com/xiaoxiaxusummer/multi_mode_pinching_antenna

Connected autonomous vehicles (CAVs) require reliable and efficient communication frameworks to support safety critical and task-oriented applications such as collision avoidance, cooperative perception, and traffic risk assessment. Traditional communication paradigms, which focus on transmitting raw bits, often incur excessive bandwidth consumption and fail to preserve the semantic relevance of transmitted information. To bridge this gap, we propose a Graph-Based Semantic Encoder-Decoder (GBSED) architecture tailored for task-oriented communications in CAV networks. The encoder leverages scene graphs to capture spatial and semantic relationships among road entities, combined with a semantic compression algorithm that reduces the size of the extracted graph based representations by up to 99% compared to raw images, while the decoder reconstructs task relevant representations rather than raw data. This design enables a significant reduction in communication overhead while maintaining high semantic fidelity, exceeding 0.9 at SNR levels above 10dB, for downstream vehicular tasks. We evaluate the proposed framework through simulations in autonomous driving scenarios, where the semantic encoder and decoder are integrated into a MIMO OFDM physical layer system. The results demonstrate high prediction success rates for risk assessment, improved robustness under the 3GPP CDL channel, and significant compression gains, confirming that the proposed semantic communication framework is a promising solution for future 6G systems.

This paper proposes a subspace fusion sensing algorithm for cooperative integrated sensing and communication. First, we stack the received signals from access points (APs) into a third-order tensor and construct the equivalent virtual antenna (EVA) array via tensor unfolding. Then, a data association-free subspace-based fusion sensing algorithm is developed utilizing the EVA arrays from distributed APs. A derivation of Cramer-Rao lower bound (CRLB) is also presented. Finally, simulation results validate the effectiveness of the proposed algorithm compared to traditional techniques.

Radio frequency fingerprint identification (RFFI) exploits device-specific hardware impairments for transmitter recognition, but its performance is highly vulnerable to receiver variations and changing wireless channels in cross-receiver deployment. To address both challenges, this paper proposes a novel cross-receiver RFFI framework with channel robustness. In the enrollment stage, a channel-robust preprocessing method is developed to construct denoised spectral quotient (DSQ) sequences, and a DSQ-based convolutional neural network (DSQCNN) is trained using data collected from the source receiver. In the cross-receiver deployment stage, a calibration dataset is built from signals captured by both the source and target receivers, and a trainable calibration neural network (TCNN) is designed to learn the nonlinear mapping between them. The cascaded TCNN-DSQCNN framework then enables robust transmitter classification on the target receiver under varying channel conditions. To the best of our knowledge, this is the first work to jointly address channel and receiver portability through combined channel suppression and nonlinear receiver calibration. Simulations with twelve WiFi transmitters and three receivers show that the proposed method achieves reliable cross-receiver classification, reaching over 90\% accuracy at an SNR of 24 dB.

In mixed near-field and far-field systems, the nonorthogonality between near-field and far-field channels may cause severe inter-user interference and hence degrade rate performance, when the analog beamforming is designed based on the low-complexity full-array maximum ratio transmission (MRT). To tackle this issue, we propose in this paper an antenna selection-based transmission framework to effectively suppress mixed-field interference without mechanically altering antenna structures. To this end, an optimization problem is formulated to maximize the sum-rate of mixed-field systems, by jointly designing antenna selection and power allocation under the MRT-based analog beamforming. As the problem is non-convex and generally difficult to solve optimally, we first consider a typical two-user scenario to obtain useful insights. Interestingly, we analytically show that the strong mixed-field interference can be substantially mitigated by deactivating only a small portion of antennas, yet without compromising array gains too much. Moreover, an inherent tradeoff is revealed in antenna selection between interference suppression and array-gain enhancement, based on which a suboptimal number of deactivated antennas for achieving the maximum sum-rate is obtained. Next, for the general multi-user case, we develop an efficient penalty dual decomposition (PDD)-based two-layer framework to obtain its high quality solution by using the block coordinate descent (BCD) and successive convex approximation (SCA) techniques. To further reduce the computational complexity, a low-complexity antenna deactivation strategy is proposed capitalizing on an interference suppression criterion. Last, numerical results demonstrate that the proposed scheme achieves a favorable trade-off between interference suppression and array gain loss, hence achieving significant performance gains over various baseline schemes.

Electrocardiogram (ECG) analysis is vital for detecting cardiac abnormalities, yet robust automated classification is challenging due to the complexity and variability of physiological signals. In this work, we investigate transformer-based ECG classification using features derived from the Koopman operator and wavelet transforms. Two tasks are studied: (1) binary classification (Normal vs. Non-normal), and (2) four-class classification (Normal, Atrial Fibrillation, Ventricular Arrhythmia, Block). We use Extended Dynamic Mode Decomposition (EDMD) to approximate the Koopman operator. Our results show that wavelet features excel in binary classification, while Koopman features, when paired with transformers, achieve superior performance in the four-class setting. A simple hybrid of Koopman and wavelet features does not improve accuracy. However, selecting an appropriate EDMD dictionary -- specifically a radial basis function dictionary with tuned parameters -- yields significant gains, surpassing the wavelet-only baseline and the hybrid wavelet-Koopman system. We also present a Koopman-based reconstruction analysis for interpretable insights into the learned dynamics and compare against a recurrent neural network baseline. Overall, our findings demonstrate the effectiveness of Koopman-based feature learning with transformers and highlight promising directions for integrating dynamical systems theory into time-series classification.

The revolutionary convergence of fluid antenna systems (FAS) and reconfigurable intelligent surfaces (RIS) creates unprecedented opportunities for secure wireless communications, yet the practical implications of hardware impairments on this promising combination remain largely unexplored. This paper investigates the security performance of non-orthogonal multiple access (NOMA) systems when fluid antennas (FAs) meet intelligent surfaces under realistic hardware constraints. We develop a comprehensive analytical framework that captures the complex interplay between adaptive spatial diversity, intelligent signal reflection, and hardware-induced distortions in short-packet communications. Through novel piecewise linear approximations and block-correlation models, we derive tractable expressions for average secure block error rate (BLER) that reveal fundamental performance limits imposed by hardware impairments. Our analysis demonstrates that while the synergy between FAs and intelligent surfaces offers remarkable degrees of freedom for security enhancement, practical hardware imperfections create performance ceilings that persist regardless of spatial diversity gains. The theoretical framework exposes critical design trade-offs between system complexity and achievable security performance, showing that hardware quality becomes a decisive factor in realizing the full potential of FAS-RIS architectures. Extensive simulations validate our analytical insights and provide practical design guidelines for implementing secure NOMA systems that effectively balance the benefits of fluid-intelligent cooperation against the constraints of realistic hardware limitations.

The extension of 5G connectivity through Low-Earth Orbit satellite systems introduces significant technical challenges, particularly due to time-varying propagation delays and high Doppler shifts resulting from satellite motion. While the Third Generation Partnership Project Release 17 established the initial framework for non-terrestrial networks, the ongoing developments in Release 19 further enhance this effort by introducing support for regenerative payload architectures, where part of the communication protocol stack is processed directly on board the satellite. In this work, we present the design of a 5G user equipment adapted for Low-Earth Orbit satellite connectivity, with specific focus on strategies for managing variable delay and Doppler compensation. Additionally, we describe a custom experimental platform based on a drone-mounted software-defined radio platform capable of emulating both transparent and regenerative satellite payloads. Although full end-to-end system validation is not yet complete, initial laboratory tests confirm the feasibility of the architecture and lay the groundwork for future experimental campaigns.

In this paper, we propose a novel estimator of the instantaneous frequencies (IFs) of the modes making up multicomponent signals (MCSs). We are particularly interested in dealing with noisy MCSs containing close modes in the time-frequency plane. Though it is possible to adapt Prony approach to estimate IFs in such situations, interference between the modes generates oscillations in the obtained estimations. After having investigated the nature of these oscillations, we propose an algorithm to remove these in IFs estimation, based on spline approximation. Numerical applications in various situations illustrate the benefit of mixing Prony technique with spline approximation for IF estimation in noisy MCSs containing close modes.

This paper introduces a method for detecting, estimating, and localising a soft fault in wired communication networks. The proposed method is based on analysing the transmission coefficients (TC) in the time domain under both fault-free and faulty situations. An orthogonal frequency-division multiplexing (OFDM)-based scheme is used to estimate the TC. A fault-severity ratio is derived to estimate the fault intensity, while a residual-based function is proposed to determine its location. Experimental validation is carried out on a Y-shaped test setup to demonstrate the efficiency of the proposed approach.

This paper presents a novel symbiotic radio system for integrated sensing and backscatter communication (ISABC) technique that enables signal-domain interference-free coexistence of the primary communication signal and the backscatter communication (BC) signal within the same spectrum. The proposed system design allows simultaneous backscatter devices (BDs) sensing and data transmission without mutual interference by exploiting waveform-domain orthogonality between orthogonal frequency division multiplexing (OFDM) and affine frequency domain multiplexing (AFDM) signals. Specifically, a chirp-based AFDM waveform is adopted due to its inherent processing gain, which enhances the detectability and reliability of the weak backscatter signal while simultaneously supporting high-resolution sensing. Unlike conventional methods that attempt to suppress direct-link interference (DLI), this approach embeds the backscatter transmission within the affine domain while maintaining reliable OFDM-based primary communication. Furthermore, by assigning distinct affine-domain shifts to each backscatter device, the proposed framework inherently suppresses inter-backscatter device interference (IBDI). Comprehensive simulation results demonstrate that the proposed coexistence scheme effectively mitigates interference without affecting the error rate of the primary link and improves the miss-detection probability performance of the BC, making it a promising candidate for future low-power and interferenceresilient systems.

The proliferation of smart and autonomous systems has motivated a shift toward executing intelligence directly on edge devices. This shift becomes particularly challenging for zero-energy devices (ZEDs), where severe constraints on memory, energy availability, and inference accuracy must be addressed simultaneously. In this paper, we present a unified approach to managing these constraints for smart ZEDs. Specifically, we design, train, and deploy a tiny machine learning (TinyML) model for person detection on a ZED. The proposed architecture stores a single model in memory while enabling adaptive inference through multiple exit points, allowing computational effort to scale with input difficulty. As a result, low-energy inference is performed for easy instances, while higher-precision inference is selectively employed for harder cases. This strategy significantly reduces energy consumption without sacrificing detection accuracy. Furthermore, to enhance device autonomy and prevent power failures, we introduce auxiliary energy-aware circuits that dynamically regulate system operation based on available energy. Compared with a state-of-the-art energy-aware single-exit TinyML approach, the proposed method achieves an energy consumption reduction of approximately $29.6\%$. Overall, the proposed framework is appealing for enabling accurate and energy-efficient intelligence on ZED platforms.

Stacked intelligent metasurfaces (SIMs) have recently emerged as a key enabler for realizing electromagnetic wave-domain signal processing in next-generation wireless networks. However, practical SIM implementations often suffer from noticeable mismatches between theoretical models and measured responses due to fabrication and assembly imperfections. This article systematically investigates the problem of interlayer error calibration in SIMs. We first classify representative modeling and hardware-induced imperfections. Then, we outline the major challenges in SIM calibration and further develop a general framework that integrates a calibration protocol with the relevant solution strategies. Moreover, we investigate the effectiveness of the multi-stage calibration approach in mitigating geometric deviations and improving the alignment between the calibrated and practical propagation coefficients. Finally, we elaborate on key research opportunities and practical challenges toward realizing physically consistent and hardware-compliant SIM implementations for future research.

Non-orthogonal multiple access (NOMA) systems allowing multiple users sharing the same resource block offer significant gains in spectral efficiency which can enable the required massive access in future wireless systems. However, they face several challenges due to their sensitivity to power allocation coefficients, fading effects, and imperfect channel state information (CSI). To address these limitations, this paper proposes Hadamard-NOMA, an approach leveraging the Hadamard Transform (HT) at the source level prior to modulation. By introducing HT, the system mitigates the adverse impact of fading and CSI imperfections, reducing bit error rates (BER) and enhancing overall system reliability. Theoretical analysis and Monte Carlo simulations validate the effectiveness of this technique, demonstrating robust NOMA transmission in dynamic wireless environments. The proposed method offers a promising solution for next-generation wireless networks, ensuring more reliable performance under diverse transmission conditions. Simulation results confirm analytical predictions, demonstrating significant performance improvements over state-of-the-art T-NOMA and Usman-NOMA schemes. Specifically, for the Near user, a gain of 15 dB is achieved at a Bit Error Rate (BER) of $10^{-2}$, while the Far user benefits from a 10 dB gain at a BER of $10^{-1}$. Compared to Usman-NOMA, the proposed method provides an improvement of 15 dB for the Far user at BER $10^{-1}$. Additionally, in a two-user scenario with imperfect Successive Interference Cancelation (SIC), user 1 requires an SNR at least 14 dB lower than user 2 to achieve a BER of $10^{-3}$. These findings highlight the effectiveness of applying HT at the source stage, significantly mitigating CSI errors and making NOMA more resilient for next-generation wireless networks.

This paper presents the design and simulation of a new curved monopole antenna optimized for skywave HF radar applications, with a systematic investigation of the effects of curvature and fixed-section length on antenna performance. The proposed design achieves improved impedance matching, broader bandwidth, and enhanced realized gain compared to a conventional quarter-wavelength monopole at 15 MHz. Parametric analysis shows that fully bending the monopole degrades performance, whereas introducing a straight section and carefully optimizing the curvature enables a 18.5% gain increase and a 400 kHz bandwidth expansion. The single-element design is further extended to a 12-element linear array with 0.45λ spacing (where λ is the wavelength), demonstrating stable embedded-element behavior and improved low-to- moderate elevation gain for skywave over-the-horizon radar operation. At θ = 30°, the proposed array achieves 14.04 dBi compared to 13.11 dBi for the reference array, corresponding to 24% gain enhancement, which is significant in high-power HF radar systems. These results confirm that the proposed curved monopole antenna provides a compact, broadband, and scalable solution for next-generation HF radar arrays.

Simultaneous transmitting and reflecting reconfigurable intelligent surfaces (STAR-RIS) offer a transformative approach for integrated sensing and communication (ISAC) systems, particularly for enhancing physical layer security (PLS). This paper investigates a robust, secure downlink transmission framework for a STAR-RIS empowered multi-user (MU) multiple-input multiple-output (MU-MIMO) system, where a multi-antenna dual-function radar and communication base station (DFRC-BS) that simultaneously transmits confidential messages to multiple intended users (IUs) and performs target sensing in the presence of malicious eavesdroppers. To optimize system security, we formulate a worst-case robust beamforming problem to maximize the secrecy rate. This formulation jointly designs the active transmit beamforming at the BS and the passive reflection, transmission coefficients at the STAR-RIS, adheres to transmit power budgets, user quality-of-service (QoS) thresholds, sensing signal-to-interference-plus-noise ratio (SINR) requirements, maximum tolerable eavesdropping leakage, and practical phase shifts constraints. To efficiently tackle the formulated problem, we develop an alternating optimization (AO) algorithm. Specifically, the S-procedure is employed to solve semi-infinite channel uncertainty constraints, while semidefinite relaxation (SDR) and penalty convex-concave programming (CCP) are applied to obtain tractable suboptimal solutions. Extensive simulation results validate the efficacy of the proposed framework and demonstrate significant improvement in spectral efficiency compared to conventional reflecting-only RIS (R-RIS) systems under stringent sensing conditions.

This paper investigates a downlink multi-satellite integrated sensing and communication (ISAC) network, in which multiple satellites simultaneously transmit ISAC signals to provide communication services to ground user equipments and enable cooperative sensing of airborne targets through multiple gateways. To support this dual functionality, we introduce communication and sensing beamforming designs based on uniform planar arrays with optimized power allocation. Building on these designs, we propose two cooperative sensing frameworks, namely centralized and distributed. In the centralized framework, each gateway forwards its sensing observations to a central unit (CU), where the positions of multiple targets are jointly estimated from the aggregated data using a sparse signal recovery formulation. To mitigate the signaling overhead inherent in centralized processing, a distributed framework is further proposed, in which each gateway independently estimates target positions and transmits only the local estimates to the CU. To associate estimates from different gateways, a data association problem based on the squared Euclidean distance is formulated and efficiently solved using the Hungarian algorithm. The final target positions are then obtained by minimizing the distance estimation error. Simulation results demonstrate that the proposed centralized and distributed frameworks significantly outperform existing sensing schemes while satisfying communication performance requirements. We also evaluate the sensing-communication trade-off from the viewpoints of sensing accuracy and communication power consumption under the proposed frameworks.

The transition to Extremely Large Antenna Arrays (ELAA) in 6G introduces significant near-field effects, necessitating robust near-field beam training strategies in multi-path environments. Because signal phases are frequently compromised by hardware impairments such as phase noise and frequency offsets, amplitude-only channel recovery is a critical alternative to coherent beam training. However, existing near-field amplitude-based training methods often assume simplistic line-of-sight conditions. Conversely, far-field phase retrieval (PR) methods lack the sensing flexibility required to optimize training efficiency and are fundamentally limited by plane-wave models, making them ill-suited for near-field propagation. We propose a two-stage sparse PR framework for amplitude-only near-field beam training in multipath channels. Stage I performs adaptive support discovery on the standard 2D DFT beamspace by exploiting a physics-guided prior induced by near-field beam patterns. Stage II then refines the channel estimate by restricting sensing and sparse PR to the learned subspace. Numerical results show that the proposed adaptive pipeline consistently outperforms non-adaptive baselines, improving beamforming gain by over 70% at low SNR.

When the concept of fluid antenna system (FAS) is applied to multiple-input multiple-output (MIMO) systems, this gives rise to MIMO-FAS, a.k.a.~fluid MIMO. Under rich scattering, the spatial correlation matrices are governed by the zeroth-order Bessel function $J_0(\cdot)$ through the continuously adjustable antenna positions, creating a highly non-convex landscape for optimization with fluctuating local optima -- the \emph{Bessel landscape}. In this paper, we tackle the joint transmitter (TX) and receiver (RX) antenna position optimization problem in fluid MIMO to maximize the ergodic capacity by shaping this landscape. Using Kronecker channel decomposition, we firstly develop a suite of analytical results that expose the problem's intrinsic structure: (i) a high signal-to-noise ratio (SNR) capacity approximation that decomposes the objective into separable log-determinant terms of the TX and RX correlation matrices, $\mathbf{R}_T$ and $\mathbf{R}_R$, respectively, (ii) a closed-form capacity loss bound linking $\det(\mathbf{R}_T)\det(\mathbf{R}_R)$ to the performance gap relative to the independent and identically distributed (i.i.d.) ideal MIMO channel, and (iii) the globally optimal inter-element spacing when the number of fluid elements at the TX is $N=2$ at the first zero of $J_0$. Guided by these insights, we propose two algorithms within an alternating optimization (AO) framework. The first algorithm is AO with particle swarm optimization (PSO) which deploys a particle swarm to explore the Bessel landscape globally without gradient information. Then in the second method, we use successive convex approximation (SCA) to obtain the gradient in closed form via $J_1(\cdot)$ to construct convex surrogates for orders-of-magnitude faster convergence.

Networked low Earth orbit (LEO) satellite constellations enabled by inter-satellite links offer a promising path toward ubiquitous broadband non-terrestrial services. However, fast orbital motion induces frequent scheduling updates and handovers, while stringent on-board constraints (e.g., limited radio-frequency chains) tightly couple user scheduling with cooperative beamforming. This paper investigates handover-aware power-efficient downlink transmission in networked LEO systems under statistical channel state information. We introduce a two-segment frame structure that separates handover-related operations from user-plane transmission, and propose a power consumption model that captures both the switching cost of newly established satellite-user links and the reduced effective transmission window during handover. Using a hardening-bound ergodic-rate metric, we formulate a per-frame network-wide power minimization problem with joint cooperative beamforming and implicit scheduling under segmented quality-of-service constraints, per-satellite power budgets, and serving-cardinality limits. To address scheduling-induced combinatorial sparsity and nonconvex fractional rate constraints, we develop an iterative algorithm that combines a reweighted $\ell_2$ surrogate with a penalty-based relaxation and a fractional-programming inner loop, yielding a sequence of convex second-order cone programs. Simulations based on time-varying orbital dynamics with frame-wise serving-set evolution and maritime user data quantify the power-handover tradeoff and demonstrate consistent power savings and improved feasibility over non-cooperative and pre-scheduled cooperative baselines.

We are concerned with the challenge of reliably classifying and assessing intracranial aneurysms using deep learning without compromising clinical transparency. While traditional black-box models achieve high predictive accuracy, their lack of inherent interpretability remains a significant barrier to clinical adoption and regulatory approval. Explainability is paramount in medical modeling to ensure that AI-driven diagnoses align with established neurosurgical principles. Unlike traditional eXplainable AI (XAI) methods -- such as saliency maps, which often provide post-hoc, non-causal visual correlations -- Concept Bottleneck Models (CBMs) offer a robust alternative by constraining the model's internal logic to human-understandable clinical indices. In this article, we propose an end-to-end 3D Concept Bottleneck framework that maps high-dimensional neuroimaging features to a discrete set of morphological and hemodynamic concepts for aneurysm identification. We implemented this pipeline using a pre-trained 3D ResNet-34 backbone and a 3D DenseNet-121 to extract features from CTA volumes, which were subsequently processed through a soft bottleneck layer representing human-interpretable clinical concepts. The model was optimized using a joint-loss function to balance diagnostic focal loss and concept mean squared error (MSE), validated via stratified five-fold cross-validation. Our results demonstrate a peak task classification accuracy of 93.33% +/- 4.5% for the ResNet-34 architecture and 91.43% +/- 5.8% for the DenseNet-121 model. Furthermore, the implementation of 8-pass Test-Time Augmentation (TTA) yielded a robust mean accuracy of 88.31%, ensuring diagnostic stability during inference. By maintaining an accuracy-generalization gap of less than 0.04, this framework proves that high predictive performance can be achieved without sacrificing interpretability.

Vehicle tracking, motion estimation, and collision prediction are fundamental components of traffic safety and management in Intelligent Transportation Systems (ITS). Many recent approaches rely on computationally intensive prediction models, which limits their practical deployment on resource-constrained edge devices. This paper presents a lightweight digital-twin-based framework for vehicle tracking and spatiotemporal collision prediction that relies solely on object detection, without requiring complex trajectory prediction networks. The framework is implemented and evaluated in Quanser Interactive Labs (QLabs), a high-fidelity digital twin of an urban traffic environment that enables controlled and repeatable scenario generation. A YOLO-based detector is deployed on simulated edge cameras to localize vehicles and extract frame-level centroid trajectories. Offline path maps are constructed from multiple traversals and indexed using K-D trees to support efficient online association between detected vehicles and road segments. During runtime, consistent vehicle identifiers are maintained, vehicle speed and direction are estimated from the temporal evolution of path indices, and future positions are predicted accordingly. Potential collisions are identified by analyzing both spatial proximity and temporal overlap of predicted future trajectories. Our experimental results across diverse simulated urban scenarios show that the proposed framework predicts approximately 88% of collision events prior to occurrence while maintaining low computational overhead suitable for edge deployment. Rather than introducing a computationally intensive prediction model, this work introduces a lightweight digital-twin-based solution for vehicle tracking and collision prediction, tailored for real-time edge deployment in ITS.

Radars provide robust perception of vehicle surroundings by effectively functioning in poor light and adverse weather conditions. Synthetic aperture radar (SAR) algorithms are employed to address the limited angular resolution of radars by enlarging antenna aperture size synthetically as the radar moves. An autofocus algorithm is essential to improve the SAR image quality by compensating for errors mainly caused by inaccurate radar localization. Existing autofocus algorithms are mostly tailored for the frequency domain SAR techniques which are prevalent in aviation and spaceborne applications thanks to their lower complexity in large data processing. However, in the automotive context, the backprojection algorithm (BPA) is often preferred since it provides less distorted images at the cost of more complexity. Addressing the gap in efficient autofocus solutions for time-domain algorithms, this paper introduces a dual-layered autofocus strategy that integrates the Polar Format Algorithm (PFA) with BPA. The first layer employs a novel Localization Error Compensation Autofocus (LECA) processing pipeline to estimate and correct the localization errors within the PFA domain, leveraging its computational efficiency. The second layer seamlessly transfers these corrections to BPA, enabling high-quality SAR imaging while maintaining low complexity. Additionally, the strategy extends Phase Gradient Autofocus (PGA) techniques to enhance the efficiency of localization error compensation for BPA. Validated through real-world automotive experiments, the proposed pipeline delivers state-of-the-art image focus and resolution, setting a new benchmark for computationally efficient SAR imaging.

IEEE 802.11ad standard uses analog beamforming for high-speed directional communication with mobile user (MU) in the millimeter wave (mmWave) spectrum. However, the lengthy beam alignment procedures involving large data packets between the base station (BS) and the MU introduce considerable overhead, deteriorating the overall throughput. Prior works have proposed 802.11ad-based integrated sensing and communication (ISAC) BS transceivers to eliminate time-consuming beam alignment. Instead, the radar and communication functionalities use the same waveform, spectrum, and millimeter wave front end (MFE) with a common spatial field of view. The radar detects and localizes the MU, enabling the subsequent directional communication with the MU. This work proposes an end-to-end IEEE 802.11ad-based ISAC BS transceiver prototype, wherein the digital baseband hardware frontend on edge is integrated with a Simulink-based MFE. The proposed prototype facilitates a systematic link budget and detailed performance analysis for different wireless channels, target motions, signal-to-noise ratios, hardware configurations, and impairments. We also investigate how these impairments affect radar performance and, in turn, the communication metrics since the performances of both systems are uniquely interrelated in an ISAC system. Our results show that even with hardware impairments, the 802.11ad-based ISAC offers 34% higher throughput than the standard with an ideal MFE.

An integrated sensing and communication (ISAC) framework comprises radar sensing to enable reliable direction beam-based communication between a base station (BS) and mobile user (MU). The ISAC will be an integral part of 6G with potential applications for high-speed vehicular communications. Existing works have explored azimuth and Doppler velocity estimated via radar sensing for beam identification and identification in dynamic environments. In this work, we propose radar-enabled modulation scheme selection for ISAC, thereby eliminating conventional time-consuming downlink-uplink feedback-based modulation scheme selection. We have analyzed the performance of the proposed approach for four different trajectories and shown an improvement in throughput between 54-209% over state-of-the-art ISAC.

The millimeter wave (mmW) frequency spectrum has been explored recently for large bandwidth communication. At these frequencies, narrow directional beams are required for communication since the signal attenuation is high due to atmospheric absorption. This work presents an AMD RFSoC and Sivers Semiconductors analog front-end based hardware testbed capable of directional communication via analog beamforming at mmW. The proposed testbed comprises orthogonal frequency division multiplexing (OFDM) based baseband physical layer and digital front-end on an ARM processor and field programmable gate array (FPGA), respectively, integrated with high-speed data converters of the RFSoC. The RFSoC output at sub-6GHz is integrated with a mmW multi-antenna analog-front end for over- the-air communication at 29.8 GHz. We demonstrate end-to-end communication over the air and present bit error rate (BER) analysis in the presence of radio frequency impairments and beam misalignments in real radio channels.

Cognitive Radars (CRs) employ perception-action cycle to adapt their sensing and transmission strategies based on its' perception of the target kinematic states and mission objectives. This paper considers an inverse learning Electronic Counter Measure (ECM) that infers both the perception and perception-driven action policy of the adversarial CR's from the actions of the CR, i.e. the sensing and transmission actions taken by the CR. Existing frameworks, in the literature, assume the knowledge of either the perception or the perception-action policy and infer the other. However, this assumption is unrealistic in an adversarial setting. We address this gap by proposing an online, nonparametric Bayesian machine learning framework and developing the Inverse Particle Filter with Dependent Dirichlet Process (IPFDDP) algorithm, which characterizes the perception-dependent action policy using a Dependent Dirichlet Process (DDP) and embeds kernel-based DDP inference within a Bayesian inverse particle filtering framework to jointly estimate the CR's perception and perception-action policy. Extensive numerical simulations demonstrate that IPFDDP outperforms existing inverse learning methods in terms of mean squared error, Kullback-Leibler divergence between the estimated and true policy, and accuracy in identifying relative action preferences. Unlike the existing techniques, the proposed Bayesian formulation naturally quantifies uncertainty in inferred perception and perception-action policy, enabling active probing strategies for sample efficient inverse learning. Simulation results show that active probing integrated with IPFDDP achieves, on average, a 40% faster reduction in KL divergence compared to randomized probing.

Radio Frequency Interference (RFI) is a growing concern for Global Navigation Satellite System (GNSS) reliability. The Cyclone GNSS (CYGNSS) constellation, designed for ocean wind retrieval via GNSS reflectometry (GNSS-R), provides Delay-Doppler Maps (DDMs) with noise floor metrics exploitable for spaceborne RFI detection. This study proposes a maximum-based DDM noise floor strategy that selects the highest noise floor value among four simultaneous GNSS reflections at each 0.5-second epoch, rather than their mean, preventing dilution of anomalous signals by unaffected channels. To suppress false alarms, a two-tier verification framework is introduced: (1) multi-satellite concurrent detection, confirming RFI when two or more CYGNSS satellites independently flag the same geographic region, and (2) temporal persistence verification, confirming a single-satellite detection only if threshold exceedance persists over a 10-second window. The physical basis for this criterion is established through slant-range geometry analysis between a ground-based jammer and the orbiting satellite. Performance is evaluated using CYGNSS Level 1 data from May 2025 in two regions: White Sands Missile Range, where NOTAM-announced GPS jamming tests were conducted, and the Middle East, where persistent RFI has been documented. The proposed method is compared against NASA's kurtosis-based RFI flags and a mean-based noise floor method. Results show that it detected RFI on three dates where the other methods produced negligible detections, and flagged 62% of total epochs in the Middle East compared to 46% (mean-based) and 33% (kurtosis-based). It also demonstrated capability to detect the early onset of gradually intensifying interference and atypical abnormal patterns not previously reported, highlighting the potential of maximum-based DDM noise floor analysis for sensitive and reliable spaceborne RFI detection.

Product codes are a class of quantum error correcting codes built from two or more constituent codes. They have recently gained prominence for a breakthrough yielding quantum low-density parity-check (qLDPC) codes with favorable scaling of both code distance and encoding rate. However, despite its powerful algebraic formulation, the physical mechanism for assembling a general product code from its constituents remains unclear. In this letter, we show that the tensor and balanced product codes admit an intuitive coupled-layer construction by taking a stack of one code and condensing a set of excitations in the pattern given by the checks of the other code. Our framework accommodates both classical or quantum CSS input codes, unifies known physical mechanisms for constructing higher dimensional topological phases via anyon condensation, and naturally extends to non-topological codes.

We propose a definition of asymptotically flat spacetimes that is consistent with both null infinities and compatible with known properties of gravitational scattering, incoming and outgoing radiation, and interactions with matter. For this class of spacetimes, we prove three antipodal matching conditions at spatial infinity: one for the so-called dual mass aspect, one for the leading tail of the shear, and one that non-trivially relates the peeling properties of the spacetime at past and null infinities to the leading tail and mass aspect at spatial infinity. Furthermore, we reformulate these identities as asymptotic conservation laws defined on the boundary hyperboloid at spatial infinity.

Quantum cellular automata (QCAs) are automorphisms of tensor product algebras that preserve locality, with local quantum circuits as a simple example. We study approximate QCAs, where the locality condition is only satisfied up to a small error, as occurs for local quantum dynamics on the lattice. A priori, approximate QCAs could exhibit genuinely new behavior, failing to be well-approximated by any exact QCA. We show this does not occur in one dimension: every approximate QCA on a finite circle can be rounded to a strict QCA with approximately the same action on local operators, so these systems are classified by the same index as in the exact case. Previous work considered the case of the infinite line, by using global methods not amenable to finite systems. Our new approach proceeds locally and now applies to finite systems, including circles or homomorphisms from sub-intervals. We extract exact local boundary algebras from the approximate QCA restricted to local patches, then glue these to form a strict QCA. The key technical ingredient is a robust notion of the intersection of two subalgebras: when the projections onto two subalgebras approximately commute, we construct an exact subalgebra that serves as a stable proxy for their intersection. This construction uses a recent theorem of Kitaev on the rigidity of approximate $C^*$-algebras.

Using Vakarelov's theory of lattice logics with negation, we introduce the (co)quasiintuitionistic logic, and prove its soundness and completeness with respect to the class of (co)quasiintuitionistic algebras. Combining these algebras together, we obtain biquasiintuitionistic algebras and the corresponding logic. Their further extension with the Skolem algebra structure defines Akchurin algebras and the respective logic. Next we generalise the framework of spectral presheaves (which is a main object in the Butterfield--Isham--Döring topos theoretic approach to quantum mechanics) to arbitrary complete orthocomplemented lattices, and show that the orthocomplementation determines two negation operators on the spectral presheaf (one paraconsistent, another paracomplete), equipping the set of all closed-and-open subpresheaves of a spectral presheaf with the structure of a biquasiintuitionistic algebra. Combined with the generic Skolem (i.e. Heyting and Brouwer) algebra structure of this set, this gives a particular instance of an Akchurin algebra, which is a sound model of a product of biquasiintuitionistic and biintuitionistic logics, featuring four distinct negations. We also show that the underlying orthocomplemented lattice can be reconstructed as an internal object of the spectral presheaf, resulting as the image of a double coquasiintuitionistic (resp., quasiintuitionistic) negation monad (resp., comonad). Finally, we prove a no-go theorem for the claim that the spectral presheaf is a model of a dialectical (or any other) relevance logic.

The large overhead imposed by quantum error correction is a critical challenge to the realization of quantum computers, and motivates searching for alternative error correcting codes and fault-tolerant circuit constructions. Postselection is a powerful tool that builds large programs out of probabilistically generated sub-circuits, and has been shown to increase the threshold of quantum error correction based on fusing fixed-size resource states or concatenated codes. In this work, we present an approach to lower the overhead of quantum computing using scalable postselection, based on directly postselecting sub-circuits with a size extensive in the code distance using decoder soft information. We introduce a metric, the partial gap, that estimates what the logical gap of a resource state will be after it is consumed, and show that postselection based on the partial gap leads to scalable improvements in the logical error rate. In the specific context of implementing logical gates via teleportation through a cluster state, we demonstrate that scalable postselection provides a $4\times$ reduction in the overhead per logical gate, at the same logical error probability.

We report quantum chemistry calculations performed on superconducting quantum processors for a molecule exhibiting the half-Möbius electronic topology originally introduced by Rončević et al. Using SqDRIFT, a randomized sample-based Krylov quantum diagonalization algorithm, we achieve reliable quantum simulations on active spaces corresponding to 36 orbitals (72 qubits) and extend previous studies up to 50 orbitals (100 qubits). We demonstrate that a systematic increase of active space sizes, which has a concrete impact on the accuracy of the electronic structure description, is achievable with state-of-the-art quantum processors, thus offering a promising path towards practically relevant quantum-assisted electronic-structure calculations.

We study the superradiant phase transition of a mesoscopic Fermi gas comprising between a few tens and a few thousand $^6$Li atoms in a high-finesse cavity across a wide range of densities. We observe a non-monotonic variation of the superradiant threshold as a function of density, with a minimum reached when the Fermi and recoil wavevectors are comparable. The minimum corresponds to a crossover between Fermi pressure-assisted ordering and Pauli blocking of photon scattering, in good agreement with theory. This interpretation is confirmed by a study of the atom-number dependence of the ordering threshold and photon number scaling. Lastly, we demonstrate the operation of our mesoscopic system in a regime where light-induced forces are opposite for the two spin components, leading to an ordered phase with a spin-density-wave character. Our system opens the perspective of studying few-fermion systems with strong and coherent light-matter coupling.

The fragmented landscape of quantum computer benchmarks, characterized by system-specific tools and inconsistent evaluation methodologies, hinders reliable cross-platform performance assessment. We introduce Metriq, an open-source collaborative platform for reproducible cross-platform quantum benchmarking that integrates benchmark definition and execution, data collection, and public presentation into a unified workflow. The Metriq benchmark suite spans both system-level metrics that characterize fundamental device properties such as entanglement quality, gate performance, and circuit speed, as well as application-inspired protocols that assess performance on quantum machine learning, optimization, and quantum simulation tasks. Benchmarks are chosen to scale with processor size, and the framework incorporates cost and resource estimation to support practical evaluation. Using Metriq, we collect and publicly release results from more than ten quantum computers across multiple hardware vendors, enabling systematic cross-platform comparison. The resulting curated dataset also reveals the practical strengths and limitations of individual benchmarks, creating a feedback loop that informs the ongoing refinement of the suite. To summarize performance across the benchmark suite, we introduce the Metriq Score, a composite index aggregating benchmark outcomes. We further present cross-benchmark analyses enabled by the shared dataset and their correlations with hardware calibration metrics. Through open development and data sharing, Metriq provides a practical foundation for reproducible benchmarking of quantum computers as hardware and benchmarking methods continue to evolve.

In the forthcoming years the LHC experiments are going to be upgraded to benefit from the substantial increase of the LHC instantaneous luminosity, which will lead to larger, denser events, and, consequently, greater complexity in reconstructing charged particle tracks, motivating frontier research in new technologies. Quantum machine learning models are being investigated as potential new approaches to high energy physics (HEP) tasks. We characterize and upgrade a quantum graph neural network (QGNN) architecture for charged particle track reconstruction on a simulated high luminosity dataset. The model operates on a set of event graphs, each built from the hits generated in tracking detector layers by particles produced in proton collisions, performing a classification of the possible hit connections between adjacent layers. In this approach the QGNN is designed as a hybrid architecture, interleaving classical feedforward networks with parametrized quantum circuits. We characterize the interplay between the classical and quantum components. We report on the principal upgrades to the original design, and present new evidence of improved training behavior, specifically in terms of convergence toward the final trained configuration.

Black holes are the most efficient scramblers in nature. By mapping the instantaneous mass and angular momentum of two spinless black holes in a quasi-circular binary onto those of an effective Kerr black hole, we demonstrate that the final state of the merger remnant corresponds with remarkable accuracy to the configuration that renders null geodesics unstable at the highest possible rate. This suggests a deep connection between the properties of black holes resulting from binary mergers and their unstable null orbits.

This article is devoted to the investigation of the observational properties of the Schwarzschild black hole (BH) surrounded by a dark matter (DM) halo. Our study commences with a brief review of spacetime, including the horizon structure and curvature invariants, which are the Ricci scalar, the square of the Ricci tensor, and the Kretschmann scalar. Subsequently, we explore the massive and massless particle dynamics around the Schwarzschild BH surrounded by a dark matter halo, including the innermost stable circular orbit (ISCO) and photon sphere radii. It was found that the radius of the ISCO increases under the influence of the spacetime parameters. Additionally, we investigate the weak gravitational lensing with the assumption that the BH is surrounded by a uniform and non-uniform plasma. Finally, we examine the impact of a plasma on the BH shadow and employ Event Horizon Telescope (EHT) observational data to constrain the BH's parameters.

We investigate the geodesic structure of realistic static and spherically symmetric spacetimes embedding neutron stars in metric $f(R)$ gravity, focusing on the quadratic Starobinsky model $f(R)=aR^2$ with $a<0$. Neutron-star solutions are obtained by numerically solving the modified Tolman-Oppenheimer-Volkoff system for several realistic equations of state. Such solutions are then matched consistently to the exterior vacuum geometry by enforcing the full set of junction conditions required in metric $f(R)$ theories. Using an effective potential approach, we show that stable circular orbits appear in discrete radial bands separated by forbidden regions, with a dominant principal band of stability that depends sensitively on the stellar central pressure, the equation of state, and the magnitude of the parameter $|a|$. Outside the stable bands, massive particles can have bound but unstable precessing trajectories as well as unbounded motions. On the other hand, for null geodesics, we find no evidence for photon spheres outside the neutron star within the parameter range studied.

One of the main targets for space-borne gravitational wave detectors is the detection of Extreme Mass Ratio Inspirals (EMRIs). The data analysis of EMRIs requires waveform models that are both accurate and fast. The major challenge for the fast generation of such waveforms is the generation of the Teukolsky amplitudes for generic (eccentric and inclined) Kerr orbits. The requirement for the modeling of $\sim10^5$ harmonic modes across a four-dimensional parameter space makes traditional approaches, including direct computation or dense interpolation, computationally prohibitive. To overcome this issue, we introduce a convolutional encoder-decoder architecture for a fast and end-to-end global fitting of the Teukolsky amplitudes. We also adopt a transfer learning strategy to reduce the size of the training dataset, and the model is trained gradually from the simplest Schwarzschild circular orbits to generic Kerr orbits step by step. Within this framework, we obtain a surrogate model based on a semi-analytical Post-Newtonian dataset, and the full harmonic amplitudes can be generated within milliseconds, while the median mode-distribution error for generic orbits is approximately $\sim10^{-3}$. This result indicates that the framework is viable for constructing efficient waveform models for EMRIs.

We develop a multi-reference perturbation theory for electronic structure calculations based on symmetries of the Hamiltonian. The reference Hamiltonian in the symmetry-based perturbation theory (SBPT) is chosen such that it possesses more symmetries than the original Hamiltonian, leading to a larger reduction of computational resources in terms of both the number of configurations in the configuration interaction expansion and the number of required qubits in quantum computing applications. We provide approximate, scalable solutions for the second-order correction, as well as an application to selected configuration interaction. We show that SBPT is an extension of other existing multi-reference perturbation theories and that it can give better results for some molecular systems in a robust way.

A proper-time method for constructing models of dynamic gravitational-wave fields is presented. Using the proper-time method, analytical (not numerical) models of secondary gravitational waves are constructed as perturbative solutions of linearized field equations against the background of the exact wave solution of Einstein's equations for the vacuum in the Bianchi VI universe in a privileged wave coordinate system. Relations for the proper time of test particles against the background of a strong gravitational wave are used. The analytical form of the metric components for secondary gravitational waves is found from compatibility conditions for the field equations. From the field equations, an explicit form of ordinary differential equations and their solutions is obtained for functions included in small corrections to the metric for secondary gravitational waves. It is shown that there exists a continuum of gravitational wave parameters for which the perturbative solutions are stable.

We study bipartite spin-singlet correlations when rotational symmetry is described by a quantum group rather than an ordinary Lie group. We show that, even though the single-spin observables act as in the undeformed theory, the non-trivial coproduct reshapes the notion of "total" symmetry and leads to a deformed analogue of the Bell singlet state. We show that implementing local measurements with the conventional tensor-factor observables yields a striking effect: perfect anticorrelation is preserved, yet the one-site outcome statistics become deformation-dependent and biased. Using instead the symmetry-covariant, R-matrix-dressed embedding of local observables restores unbiased statistics while maintaining perfect anticorrelation. Our results demonstrate that, in a quantum group symmetry setting, strict tensor-factor locality is not stable under the symmetry and must be replaced by a braided notion of locality to formulate consistent local measurements.

Identifying the physical grounds distinguishing quantum theory from broader probabilistic frameworks remains an open challenge. Communication-based proposals -- most notably the principles of impossibility of superluminal signaling and information causality (IC) -- highlight the role of communication in ruling out unphysical theories and offer an operational rationale on why quantum predictions prevail over these alternative models. Nevertheless, most such developments rely on communicating parts optimizing over specific tasks, such as communication complexity problems and random access codes (RAC). In this work, we systematically extend this communication-based approach. We characterize the class of communication tasks relevant for this context, and employ the general information-theoretic framework to derive new operational constraints preventing such unphysical behaviors. Remarkably, our results reveal a broad family of previously undetected implausible behaviors, independent of any particular encoding or decoding strategy, reinforcing the role of communication as a fundamental lens through which physically meaningful theories can be identified.

Harada's recently announced Kerr-Schild solution of Einstein's equations in general relativity [Phys. Rev. D {\bf 112}, 024020 92025)] is identified as isometric with one of the well-known NUT solutions.

We present a quantum sensing protocol for three-dimensional estimation of the difference between the momenta of two photons based on spatially resolved interferometric sampling measurements. The protocol attains ultimate quantum precision in the simultaneous estimation of the components of the relative momentum for any values of the parameters already with $\sim 2000$ sampling measurements and a bias below $1\%$. These results identify 3D spatially resolved two-photon interference as an efficient tool for multi-parameter quantum sensing, with potential applications in high-precision 3D localization, refractometry, and near-field calibration in free-space quantum technologies.

We study a massless Dirac field subjected to two alternative boundary conditions on two parallel thin walls, in d + 1 dimensions. The two configurations correspond to the system being even or odd under reflection about the midplane between the two walls, and lead to qualitatively different behaviors. The even (symmetric) configuration produces an attractive Casimir force, whereas the odd (antisymmetric) one yields repulsion, in agreement with a general theorem linking parity to the sign of the fermionic Casimir effect. We complement this result by studying two phenomena associated with the vacuum fluctuations responsible for the Casimir interaction, both of which are also sensitive to parity: the correlation between currents concentrated on the walls, and the induced bulk current under the influence of an external electric field. For the latter we show that, in 2 + 1 dimensions, an induced transverse (Hall-like) current arises, whose spatial profile inherits the symmetry of the confining potential.

High-fidelity gate implementation requires sophisticated control pulses that steer the quantum system to undergo the desired transformation. Quantum Optimal Control allows to derive these control pulses in an open-loop fashion based on numerical simulations. However, their precision can be limited by incomplete knowledge of the system. Closed-loop optimization overcomes this limitation by incorporating feedback from measurements, provided a suitable and efficient measure of the gate performance can be defined. In this article, we present an efficient method to evaluate the performance of a two-qubit gate by preparation and measurement of only two quantum states, enabling experimental closed-loop optimization with a metric previously believed to be limited to open-loop control. We tailor the approach to nitrogen-vacancy centers in diamond and, through numerical simulations, demonstrate how the method can optimize a two-qubit gate while reducing the number of required measurements by two orders of magnitude compared to standard process tomography under realistic experimental settings.

This work places the invariant $ds^2$ at the center of the gravitational interaction, interpreting it not as a purely geometric object but as the differential of proper time, endowed with direct physical meaning. Starting from the extension of Fermat's principle to massive particles--namely, the requirement that freely falling bodies follow trajectories that extremize proper time, which for timelike motion corresponds to a local maximum--and invoking the universality of Galilean free fall, we derive the form of $ds^2$ in a static gravitational field. Lorentz invariance then provides the natural framework to extend this result to systems involving moving matter. The invariant derived through this procedure matches the weak-field limit of General Relativity formulated in the harmonic gauge. Within this linearized regime, we show that the structure of the theory already contains the seeds of its non-linear completion: any dynamically consistent extension to strong gravitational fields necessarily involves the Ricci tensor. From this viewpoint, Einstein's field equations appear not as a postulated geometric law, but as the unique covariant closure required to ensure energy momentum conservation and the self consistency of the gravitational interaction.

Recent work has noted that a space-sharing argument proves the tightness of the entropic quantum Singleton bounds, which was left open in the literature for various settings involving only-quantum messages, only-classical messages, or both classical and quantum messages. Focusing on the setting of entanglement-assisted classical coding (EACC), in this letter we first elaborate upon the space-sharing argument and the tight Singleton bound for this setting, and then establish a new tight entropic Singleton bound for EACC codes with entanglement assistance distributed across a subset of encoders when only local quantum operations are allowed at each encoder.

Continuous-variable (CV) quantum systems provide a versatile platform for quantum information processing, in which quantum states can be represented in the quadrature phase space. In realistic implementations, environmental noise, primarily photon loss and dephasing, progressively degrades these states. Machine-learning-based quantum error mitigation (QEM) has recently emerged as a promising approach to suppress such noise; however, existing methods are typically limited to the training horizon and require training data that cover the entire evolution, which is experimentally demanding. Here we introduce a framework for extrapolative quantum error mitigation based on a time-conditioned Swin Transformer. By explicitly embedding the evolution time via adaptive layer normalization, the model learns a correction map that accounts for the continuous accumulation of noise while capturing nonlocal phase-space correlations. Numerical simulations under both Markovian and non-Markovian noise demonstrate accurate state recovery in the long-time regime, where existing approaches deteriorate. Our results establish extrapolative QEM as a practical route to mitigating noise in CV quantum systems without exhaustive training data.

In nuclear magnetic resonance (NMR), Composite Pulses (CPs) are widely used to correct for pulse imperfections, e.g., RF field inhomogeneity and resonance offset. Although robust pulse sequences have been developed throughout the years, the imperfection of the initial state has not been widely discussed in the literature as an additional systematic error. In previous work, we developed a classical canonical framework to perform stability analysis and used this as a measure of CP robustness. In that work, a single initial condition was allowed to evolve under various pulse imperfections. The current work extends this approach to $2D$ distributions of initial conditions on the Bloch Sphere; the objective is to minimize the area in order to preserve coherence, while maximizing population inversion of the entire distribution. As a case study, we investigate Levitt's $90(x)180(y)90(x)$ pulse sequence, when there is a spread in initial conditions. The canonical framework enables us to assess the robustness of Levitt's pulse sequence, and we find that it is maintained to a great extent even when considering a spread of initial conditions. Nevertheless, by conducting a numerical optimization, we have identified several variants of Levitt's pulse sequence that produce a larger coherent population inversion when there is a spread in initial conditions.

Stress-tensor deformations suggest a geometric origin of emergent gravity but are typically non-local for $d>2$. We couple a seed QFT to Einstein gravity with deformation parameter $λ$ and evaluate the gravitational path integral at the metric saddle. Around a fixed reference background, the leading deformation is universal: a bilocal term quadratic in the stress tensor with kernel set by the graviton Green's function, plus a systematic higher-order expansion. Expressed on the saddle-point (deformed) metric, the flow becomes local. We then provide two constructive completions on deformed backgrounds--Palatini $f(R)$ gravity and an eigenvalue method for general Ricci-based theories--and apply them to scalar generalized Nambu-Goto and $\det T$ deformations (arbitrary $d$), two-dimensional multi-scalar ModMax and Born-Infeld models, and four-dimensional root-$T\bar T$ and $T\bar T$ flows of Maxwell theory yielding ModMax and Born-Infeld electrodynamics. In free field theory, an off-shell analysis further shows that the leading quantum correction generates an Einstein-Hilbert term with controlled higher-derivative terms.

The transfer tensor method is a versatile tool for analyzing and propagating general open quantum systems. It captures in a compact manner all memory effects in a non-Markovian system through a straightforward transformation of a set of dynamical maps. Transfer tensors provide the exact convolutional propagator associated with a given time discretization over the past evolution of an open quantum system. Here we show that, for any finite time discretization, the memory kernel of the Nakajima Zwanzig equation deviates from the exact transfer tensors, although both converge in the continuous-time limit, as expected. We examine this behaviour in the context of an analytically solvable model: a two level atom resonant with a lossy cavity in the Jaynes Cummings limit. The atomic dynamics separate into two decoupled degrees of freedom -- the coherence and the population inversion. We derive exact expressions for the dynamical map, the transfer tensors and the memory kernel governing the coherence, and we relate them to their counterparts for the population inversion. As a function of the ratio between the cavity loss rate and the atom-cavity coupling strength, we identify regions of enhanced non-Markovianity in which the system can be described as fully Markovian for certain time-step choices.

We study black hole shadows in nonminimally coupled Weyl connection gravity, a metric-affine extension of general relativity in which spacetime is described by a metric and a Weyl vector field encoding non-metricity. Despite going beyond the Riemannian framework, the presence of a non-dynamical Weyl vector ensures second-order field equations. The theory admits Schwarzschild- and Reissner--Nordström-like solutions modified by a Weyl integration constant that parametrizes deviations from General Relativity. By computing the corresponding shadow radii and confronting them with the Event Horizon Telescope constraints on Sgr A*, we place observational bounds on the Weyl parameter. Assuming an observer distance $r_O = 4.1\times 10^{10}M$ and requiring consistency at the $2σ$ level, we obtain $ω\gtrsim 10^{11.7}M$ (model I), $ω\gtrsim 10^{10.5}M$ (model II), and $ω\sim 10^{12}M$ (model III). Our results show that present horizon-scale imaging already sets meaningful limits on spacetime non-metricity. This work highlights the power of black hole shadow observations as probes of extended gravitational dynamics and establishes a direct link between Weyl-based theories and current astrophysical data.

Tensor network contraction is a fundamental computational challenge underlying quantum many-body physics, statistical mechanics, and machine learning. Belief propagation (BP) provides an efficient approximate solution, but introduces systematic errors on graphs with loops. Here, we introduce a hybrid method that achieves exact results by stochastically sampling loop corrections to BP and showcase our method by applying it to the two-dimensional ferromagnetic Ising model. For any pairwise Markov random field with symmetric edge potentials, our approach exploits an exact factorization of the partition function into the BP contribution and a loop correction factor summing over all valid loop configurations, weighted by edge weights derived directly from the potentials. We sample this sum using Markov chain Monte Carlo with moves that preserve the loop constraint, combined with umbrella sampling to ensure efficient exploration across all correlation strengths. Our stochastic approach provides unbiased estimates with controllable statistical error in any parameter regime.

We investigate a simple $f(R, T_φ)$ gravity model coupled to a scalar field and demonstrate that the theory admits classical degenerate metric solutions, analogous to those known in general relativity. In particular, we identify a class of solutions that exhibits a smooth transition from a Euclidean to a Lorentzian domain, thus yielding a classical dynamical realization of signature change.

We show that the hidden symmetry and separability characteristic of Kerr geometry arise as an inevitable \emph{local} consequence of the Einstein equations for rotating spacetimes. Without assuming separability, algebraic speciality, Killing-Yano symmetry, or global boundary conditions, we analyze general stationary and axisymmetric geometries in a locally non-rotating orthonormal frame. Imposing a minimal physical requirement, the local equilibrium condition, we find that the mixed Einstein equations enforce a rigid projective alignment between the radial and angular sectors. This result does not rely on vacuum: the mixed equations are insensitive to the detailed form of a comoving stress-energy tensor. Consistency then requires equality of Schwarzian derivatives, leading to a universal classification of local solutions into Möbius, exponential, and trigonometric branches. Global regularity generically excludes the trigonometric branch, leaving precisely the Kerr-type sector and the emergence of Petrov type D structure. This provides a structural precursor to Kerr uniqueness, demonstrating that the kinematical core of Kerr geometry is encoded locally within the Einstein equations.

In this paper, we extend Horndeski's theory into the Palatini approach, assuming that the metric tensor and the (symmetric) connection are a priori independent objects. We introduce an additional transformation of the connection and write down the action functional being form-invariant under both the disformal transformation of the metric and the new transformation of the connection. We show that such a theory reduces on-shell to a metric subclass of Horndeski's gravity called kinetic gravity braiding. We also introduce an invariant metric and connection, and demonstrate that quantities defined in such a way lead to a metric theory. In the second part of the paper, we consider a simple cosmological model within the theory and explore its potential links with $ k$-essence-type theories, with a non-trivial coupling between the scalar field and the matter part of the action in the Einstein frame. We show that there exists a model that reproduces late-time cosmic acceleration, approaching asymptotically the de Sitter phase, motivating further study of the theories.

Quantum algorithms present a quadratically improved complexity over classical ones for certain sampling tasks. For instance, the Quantum Amplitude Estimation (QAE) algorithm promises to speedup the estimation of the mean of certain functions, given access to the quantum state corresponding to the probability distribution to be sampled from. Classically, samples are often obtained by running steps a Markov Chain. In this work, we experimentally use encodings of Markov chains to prepare quantum states and run a quantum Markov Chain Monte Carlo algorithm (qMCMC) on Quantinuum's H2 and Helios quantum computers. We demonstrate that it is possible to obtain accurate results on current Noisy Intermediate Scale Quantum (NISQ) hardware, operating directly on the physical qubits.

The finite square potential well is a staple problem in introductory quantum mechanics. There is an extensive literature on the determination of the allowed energies, which requires the solution of a transcendental equation by numerical, graphical or approximate analytic methods. Here we investigate the less explored problem of a particle in a semi-infinite potential well. The energy eigenvalues, which are also determined by a transcendental equation, are found by a standard graphical method, and a simple rule that yields the number of stationary states is provided. Next a simplification of the aforementioned transcendental equation is attempted. During the process pitfalls are encountered and a purportedly simpler graphical treatment of the problem given in the solutions manual to a fine textbook is shown to be flawed. A more careful analysis brings forth the correct simplification, which is shown to be particularly suitable for finding highly accurate approximations to the energy levels. Finally, a class of exact solutions is produced, the associated normalized eigenfunctions are constructed and the probability of finding the particle inside the well is computed.

Dynamical Lie algebras, i.e. Lie subalgebras of $\mathfrak{su}(2^n)$, generated by Pauli strings have recently been studied intensively. They are also called Pauli Lie algebras or Hamiltonian Lie algebras. In this paper we provide a uniform mathematical approach to various recent results on Pauli Lie algebras. Moreover, we present an algorithm that on input of a set of Pauli strings determines the isomorphism type of the dynamical Lie algebra generated by these Pauli's in time $\mathcal{O}(\max(n,m)^3)$ where $m$ is the size of the generating set.

Trions -- Coulomb-bound three-particle excitations composed of two like-charge carriers and one oppositely charged carrier -- are central quasiparticles in two-dimensional semiconductors. Reduced dielectric screening and quantum confinement strongly enhance their binding energies, making them robust and experimentally accessible. This review surveys theoretical and experimental advances in trion physics, emphasizing rigorous few-body approaches and the role of dielectric environment, anisotropy, and external electric and magnetic fields. We analyze computational methods for describing trions in two-dimensional configuration spaces and discuss how reduced dimensionality modifies their structure and stability. Connections to many-body phenomena, including screening, Landau-level mixing, and exciton--polaron crossover, are also highlighted.

We study the propagation and emission of gravitational waves in the metric-affine formulation of the bumblebee model, where spontaneous Lorentz symmetry breaking arises from a vector field acquiring a nonvanishing vacuum expectation value. Working in the geometric-optics limit of the linearized theory, we derive the modified dispersion relation governing the graviton modes and show that it depends on the orientation of the wave vector relative to the background vector. The polarization sector is examined for timelike and spacelike configurations of the Lorentz-violating vacuum. In both cases only two independent tensor modes propagate, although their propagation properties and tensor structure depend on the orientation of the background field. We then construct the retarded Green function associated with the modified wave operator and determine the radiation-zone produced by localized sources. In the timelike configuration the Lorentz-violating effects appear through a modified propagation speed and an overall amplitude renormalization, leading to a shifted retarded time while preserving the quadrupole structure of the waveform. In contrast, the spacelike sector introduces anisotropic corrections to the quadrupole amplitude together with an additional contribution proportional to the third time derivative of the quadrupole moment. As an astrophysical application, the gravitational radiation emitted by a circular binary black hole system is evaluated, allowing observational constraints on the Lorentz-violating combination $ξb^{2}$ to be estimated using multimessenger bounds from GW170817/GRB~170817A and waveform consistency requirements from gravitational wave observations.

Quantum-enhanced sensing is commonly benchmarked using the quantum Fisher information (QFI), often interpreted as a direct indicator of achievable precision. However, this quantity acquires operational meaning only within a fully specified inference framework that consistently incorporates state preparation, measurement design, resource accounting, estimator construction, prior information, and finite data effects. Here we establish a realistic end-to-end framework for quantum sensing under finite resources and identify general principles required for operationally meaningful performance assessment. A central conceptual point is that the relevant unit of estimation is not a single detection event but the inference data set required to construct a consistent estimator. We apply this approach to several paradigmatic sensing strategies frequently cited in the literature. Revisiting phase estimation with NOON states within a Bayesian framework under equal total photon resources, we explicitly construct optimal estimators and show that such schemes offer no performance advantage over repeated classical interferometry for global phase estimation with finite prior width. The apparent Heisenberg-like scaling arises predominantly from prior constraints rather than from information gained in the measurement, which is operationally negligible in the resource-normalized sense considered here. We further analyse Holland-Burnett interferometry and homodyne detection with squeezed states, demonstrating how estimator construction and repetition number determine the attainable precision and when QFI provides a reliable diagnostic. Our results clarify the conditions under which nonclassical resources lead to genuine metrological advantages and provide a practical methodology for designing and evaluating quantum sensing protocols under realistic experimental constraints.

In this work, we test the geometrical structure and thermodynamic properties of the Einstein-Maxwell-Power-Yang-Mills black hole (BH) models, which constitute a nonlinear generalization of the standard Einstein-Yang-Mills theory through the inclusion of a power-law Yang-Mills invariant. Also, we begin by analyzing the spacetime geometry via the metric function $f(r)$ and examine the modifications induced by the electromagnetic charge and nonlinear Yang-Mills parameter on the horizon structure, causal structure, and gravitational potential. Subsequently, the dynamics of photons and massive particles are explored through the study of null and timelike geodesics, allowing the determination of the effective potential, photon sphere radius, and associated BH shadow. Also, the stability of circular photon orbits is quantified using the Lyapunov exponent, which characterizes the timescale of orbital instability and provides a direct link to observable photon ring features. For massive particles, the innermost stable circular orbit (ISCO) is calculated, illustrating the influence of BH parameters on the dynamics of accretion disks. From the thermodynamic viewpoint, we compute the principal thermodynamic quantities, including the BH mass, Hawking temperature, Bekenstein-Hawking entropy, heat capacity, and Gibbs free energy, to assess both local and global stability of the system. The divergence of the heat capacity signals the occurrence of second-order phase transitions, whereas the Gibbs free energy analysis identifies possible first-order phase transitions between distinct thermodynamic configurations. In this context, our results demonstrate that the nonlinear Yang-Mills parameter strongly affects the spacetime geometry, particle dynamics, and thermodynamic phase structure, shifting the location of stability regions and critical points associated with phase transitions.

In this paper, we introduce a new family of stabilizer quantum LDPC codes derived from the classical linear codes $L_k$ and $L_k^{+}$, defined via sub-exceding functions. In previous work, these codes demonstrated strong performance in minimum distance, decoding efficiency, and structural simplicity. By combining the hypergraph product framework with a generalized Shor construction, we obtain a scalable class of quantum codes with parameters $[[6k^2,\, k^2,\, d]]$. The resulting quantum codes exhibit a rich combinatorial structure and promising properties, particularly in terms of locality, low-density parity-check (LDPC) structure, and asymptotic behavior. The minimum distance satisfies $d=3$ for $k=3$ and $d=4$ for $k\ge4$, establishing a new framework for structured quantum LDPC code design and optimization.

We propose a bipartite quantum key distribution (QKD) protocol based on causal nonseparability: the presence of a resource -- a process matrix -- that does not correspond to any definite causal order between two parties. In our protocol, Alice and Bob perform local operations arranged in a ``causal-order guessing game,'' whereby each round yields an 85.35\% probability of matching bits when the communication is undisturbed. This raw matching probability (or equivalently, a $\sim14.65\%$ error rate) is amenable to standard forward error-correction strategies. We further discuss the practical construction of the QKD protocol using indefinite causal order, where several different scenarios are deeply analyzed.

This work integrates the physics-informed neural network (PINN) approach into the neural quantum state framework to simulate open quantum system dynamics, to circumvent the computationally expensive time-dependent variational principle required in conventional variational methods. The proposed PINN-DQME method employs time-encoded neural networks within a time-domain decomposition strategy to represent the evolution governed by the dissipaton-embedded quantum master equation (DQME). We implement and validate this approach in the single-impurity Anderson model, benchmarking the PINN-DQME results against the numerically exact hierarchical equations of motion. The PINN-DQME method demonstrates high accuracy in simulating quantum dissipative dynamics at high temperatures, where non-Markovian effects are weak. However, for strongly non-Markovian dynamics at low temperatures, it encounters challenges with error accumulation during time propagation, highlighting an area for future refinement in applying PINNs to complex quantum dynamical settings.

Quantum gate teleportation enables the implementation of nonlocal quantum operations without direct interactions between distant nodes. We propose an efficient protocol for implementing arbitrary controlled-unitary (CU) gates acting on two spatially separated parties via indefinite causal order (ICO). By establishing a maximally entanglement between two remote nodes and coherently superposing orders of single-qubit gates, our protocol circumvents the drawback of complex local two-qubit operations. This ICO-based approach enables full programmability of CU gates by adjusting the inherent single-qubit operations, offering advantages over conventional fixed causal-order methods in terms of reduced circuit complexity and improved experimental flexibility. Furthermore, we develop an optical construction to implement the polarization CU gate using a stable and reciprocal Sagnac interferometer. Our work establishes a practical framework for scalable distributed quantum computation with flexible operations.

We investigate the interior structure of asymptotically flat hairy black holes (HBHs) arising in the Einstein-Klein-Gordon theory with nonpositive-definite scalar potentials, where nontrivial scalar hair exists at the event horizon. While exterior properties, including shadow imaging for HBHs supported by an inverted Higgs-like potential have been extensively investigated, their interior structure remains largely unexplored. In many gravitational theories, backreaction of classical fields can significantly eliminate the Cauchy horizon, which is known to be highly unstable due to the mass inflation effect, raising important questions regarding the validity of the Strong Cosmic Censorship conjecture. These considerations motivate us to examine the interior structure of HBHs by numerically integrating the field equations inward from the outer horizon. We find that the scalar field and the metric functions increase monotonically inside the horizon and diverge as $r \rightarrow 0$. The Ricci and Kretschmann scalars also diverge at $r=0$, confirming the presence of a genuine curvature singularity. No additional root of the metric function is observed, indicating the absence of a Cauchy horizon in the electrically neutral HBHs considered here. Furthermore, the weak energy condition is violated throughout the interior region, and the degree of violation becomes more pronounced as the scalar field at the horizon increases. These results provide new insight into the global structure of HBHs and their implications for cosmic censorship.

Quantum algorithms for electronic-structure simulations are actively being developed, yet many hybrid quantum-classical approaches are bottlenecked by the measurement overhead associated with large molecular Hamiltonians. Here we introduce a diagonalization-driven framework that progressively drives the electronic Hamiltonian toward a (block-)diagonal form in the Slater-determinant basis using sequential Givens rotations. In contrast to Schrödinger-picture methods that variationally optimize a wave function, our approach adopts a Heisenberg-picture viewpoint: the Hamiltonian is iteratively transformed, and rotation angles are determined classically from low-dimensional effective blocks, reducing the quantum workload to a small, fixed set of matrix-element measurements per iteration. Candidate generators are estimated via approximate Baker-Campbell-Hausdorff updates with truncation and cumulant-based approximations that control Hamiltonian growth, complemented by stochastic selection to avoid stagnation. We further introduce an angle-merging procedure that reduces circuit depth by consolidating repeated small-angle rotations. We benchmark the framework on N$_2$ and strongly correlated hydrogen systems, assessing convergence behavior, residual-structure diagnostics, measurement-accuracy trade-offs, circuit costs, and robustness under finite sampling.

We construct a new family of exact vacuum black brane solutions to five-dimensional Einstein gravity with a negative cosmological constant, characterized by a homogeneous horizon with Bianchi VI$_h$ symmetry. This construction generalizes the known Solv (Bianchi VI$_{-1}$) geometry via a continuous anisotropy parameter, $h$. By reducing the field equations to a cohomogeneity-one system, we derive the metric analytically. These homogeneous solutions are not asymptotically (locally) AdS, but nevertheless we analyze the thermodynamics, establishing scaling relations between entropy and temperature with anisotropic holographic system in mind. Additionally, we identify a new branch of Ricci-flat, hyperscaling-violating vacuum solutions in the case where the cosmological constant vanishes.

We analyze relativistic corrections to the wave-packet dynamics of the quantum harmonic oscillator within a perturbative framework. General expressions are derived for the leading-order relativistic contributions to the wave-packet parameters such as the average position, width and the uncertainty relations. For Gaussian wave packets, these corrections admit closed-form analytic expressions at order $1/c^{2}$. When applied to electron wave packets, the results indicate that relativistic effects become non-negligible in the keV scale harmonic confinement energies -- the uncertainty relationship deviation reaches 0.1% to 1% for an electron wave-packet moving at 15% speed of light and confined within 1-10 keV energy range which might be experimentally verifiable.

Motivated by the entropy-functional formulation of emergent gravity, in which spacetime is endowed with a thermodynamic entropy that, upon extremization, yields the Einstein field equations, we reformulate the onset of a neutron star as a bulk gravitational-thermodynamic threshold problem. By combining the Jacobson-Padmanabhan perspective of gravity as an equation of state with Tolman redshift, Tolman-Oppenheimer-Volkoff hydrostatics, and a bulk binding versus thermal energy balance, we derive four characteristic temperature scales for a static compact object: the Newtonian virial threshold, the redshifted threshold as observed at infinity, the degenerate-matter threshold, and the surface screen temperature.The paper supplies a mathematically expanded derivation, explicit TOV and Fermi-gas formulae, and thermodynamic consistency relations.

We present Lattice (L, ticker: LAT), a peer-to-peer electronic cash system designed as a post-quantum settlement layer for the era of quantum computing. Lattice combines three independent defense vectors: hardware resilience through RandomX CPU-only proof-of-work, network resilience through LWMA-1 per-block difficulty adjustment (mitigating the Flash Hash Rate vulnerability that affects fixed-interval retarget protocols), and cryptographic resilience through ML-DSA-44 post-quantum digital signatures (NIST FIPS 204, lattice-based), enforced exclusively from the genesis block with no classical signature fallback. The protocol uses a brief warm-up period of 5,670 fast blocks (53-second target, 25 LAT reduced reward) for network bootstrap, then transitions permanently to 240-second blocks, following a 295,000-block halving schedule with a perpetual tail emission floor of 0.15 LAT per block. Block weight capacity grows in stages (11M to 28M to 56M) as the network matures. The smallest unit of LAT is the shor, named after Peter Shor, where 1 LAT = 10^8 shors.

Understanding the structure of multi-qubit quantum states is essential for both quantum information research and education, yet intuitive visualization beyond the single-qubit Bloch sphere remains challenging. In this work, we propose a unified geometric framework for visualizing two- and three-qubit pure states in which local degrees of freedom and entanglement degrees of freedom are explicitly separated. For two qubits, we combine Bloch-sphere representations of reduced density operators with a complex concurrence plotted on the complex plane, enabling simultaneous visualization of entanglement strength and phase structure. For three qubits, building on the generalized Schmidt decomposition, we introduce bipartite and GHZ-type tripartite complex concurrences, which, together with local Bloch vectors, provide a complete coordinate representation of the state. Unlike classification-based approaches, our method focuses on representing a given concrete state, revealing how local properties and nonlocal correlations coexist. The framework distinguishes states with identical entanglement magnitudes but different interference structures and provides intuitive insight into the balance between pairwise and genuinely tripartite entanglement. This approach offers both conceptual clarity and potential applications in quantum education and state analysis.

Recently, Kretschmer et al. [KGD+25] presented an experimental demonstration of a proposed quantum information advantage protocol. We present an alternate proposal based on a relation derived from parallel-repeated CHSH games. Our memory measure is based on an information measure and is different from [KGD+25], where they count the number of qubits. Our proposal has an efficient verifier and a noise-robust quantum prover which is arguably much more efficient compared to [KGD+25].

A higher-order Suzuki-Trotter decomposition or Trotterization can be exploited to mitigate the Trotter error in digital quantum simulation. This work revisits the second-order symmetric Trotterization in terms of the Trotter error, where quantum many-body spin dynamics of the transverse-field Ising model is simulated. While the work presents a pedagogical way to exploit a real quantum computer, the effectiveness of the symmetric Trotterization is evaluated in a prototype superconducting quantum device on IBM Quantum Experience. It turns out that the symmetric Trotterization does not provide higher accuracy than the first-order Trotterization in the testbed using the transverse-field Ising model. The result indicates that apart from the quantum errors, such as logical gate error and readout error, the use of a higher-order Trotterization should be circumspect, and the Trotter error would play an insignificant role in particular applications in an early stage of realized noisy intermediate-scale quantum (NISQ) devices.

The quantum mechanical description of phase remains a fundamental challenge, with theoretical efforts tracing from the early works of London and Dirac to discrete formalisms. In this work, we extend the action-angle formalism to the Helmholtz-Schrödinger equation by introducing a phase-dependent wavefunction $φ(θ, t)$ residing in the Hardy space $H^2(\mathbb{D})$. This mathematical structure, defined by functions analytic on the unit disk with square-integrable boundary values, naturally ensures the positivity of the energy spectrum while providing a rigorous framework for wave dynamics in photonic systems. We demonstrate that establishing a self-adjoint phase operator requires extending the Hilbert space to $L^2$, a procedure that necessitates the admission of negative energy states. We interpret these states through an analogy with the Dirac sea, where the existence of antiphase or antiphoton modes provides a conceptual framework for understanding the fundamental limits of phase localization and quantum uncertainty. This formalism is applied to light propagation in multimode waveguides characterized by anharmonic refractive index profiles. By mapping modal dispersion to our phase representation, we show that the deviation of propagation constants from linear spacing governs the spatial evolution of the optical field. This approach offers a clear mechanism for the emergence of periodic self-imaging known as the Talbot effect, the generation of fractional revivals, and the formation of complex fractal interference patterns, providing a robust toolkit for the characterization and design of multimode interference devices.

In this work, density perturbations are investigated within the framework of a fractional Newtonian cosmology. Focusing on the matter-dominated era and employing the fluid-flow approach, the growth equation for density perturbations is derived and solved analytically. No dynamical instability arises in the physically relevant parameter space. It is shown that both the growth equation and its solutions depend explicitly on the fractional parameter $α$, and reduce to their standard Newtonian and relativistic counterparts in the special limit $α= 1$. The existence of both growing and decaying perturbative modes is confirmed, and, in accordance with current cosmological observations, the analysis is restricted to the growing mode. Using observational relations, in particular the Sachs--Wolfe equation, an observational upper bound on the parameter $α$ is obtained, which is more restrictive than the bounds inferred from background dynamics and theoretical perturbative considerations. When combined with the independent constraints arising from the background analysis, these results confine the fractional parameter $α$ to a narrow and physically viable region of parameter space. Overall, the present study indicates that, although the background evolution of the fractional model may closely mimic that of $Λ$CDM, the associated density perturbations generically carry a distinct fractional signature that can, in principle, be tested observationally.

Entanglement fidelity quantifies how well a quantum channel preserves the correlations between a transmitted system and an inaccessible reference system. We derive closed-form expressions for the entanglement fidelity associated with several standard quantum noise models, including the random Pauli-X, dephasing, depolarizing, Werner-Holevo, generalized Pauli (Weyl), and amplitude-damping channels. For each model, we express the entanglement fidelity in terms of a general input density operator $ρ$, using Schumacher's Kraus-operator approach, which provides a channel-agnostic recipe applicable to any completely positive trace-preserving (CPTP) map with a finite Kraus representation. We then specialize to a communication scenario in which the source emits a two-letter parametric alphabet, thereby making explicit the dependence of entanglement preservation on both channel and source parameters. The resulting expressions enable direct comparisons of channel performance and rankings for representative families of input states, including common qubit states.

I study a promise problem for an unknown unitary operator acting on an $n$-qubit system. The operator is promised to take one of two forms: either it implements a fixed permutation of computational basis states, or it implements the same permutation together with a conditional sign change determined by a designated input qubit. I show that these two cases can be distinguished with certainty using a single query to the unknown operator and a measurement of only one qubit. The procedure requires no ancilla qubits and uses only $n+1$ Hadamard gates in addition to the oracle call. The promise is intrinsically quantum, since the two cases differ only in their relative-phase structure and therefore have no direct classical counterpart in the usual black-box model.

We present a calculation of the full retarded Green functions of the Regge-Wheeler and Teukolsky equations obeyed by gravitational field perturbations of Schwarzschild spacetime. We perform the calculations for spacetime points along: (i) a timelike circular geodesic (where null-separated points are not at caustics); and (ii) a static worldline (where null-separated points are at caustics). These Green functions show a 4-fold singularity structure away from caustics, and 2-fold at caustics (similarly to the case of scalar field perturbations, which we also reproduce). Physical oscillations near the singularities appear in the gravitational case, which were not present in the scalar case. We obtain our results by developing various numerical and analytical methods.

Gleason's theorem is often cited as establishing the Born rule from the structure of Hilbert space, yet its original proof is mathematically sophisticated and rarely accessible to physicists. In this article we present a simple route to the essence of Gleason's result, using the generalized Bloch representation of quantum states. We show explicitly why non-Born probability rules exist for two-dimensional systems, and why they become impossible in dimension three and higher. Our argument does not reproduce Gleason's full mathematical theorem, but it clarifies why the Born rule is unavoidable in higher dimension and why qubits are truly exceptional.

We investigate elastic scattering by a compact, horizonless body in curved spacetime, considering a massless scalar wave incident on a static, spherically symmetric, uniform-density star of radius $R$ and mass $M$ with a Schwarzschild exterior. We introduce an exact Debye-series decomposition of the scattering matrix, in the spirit of Debye expansions in Mie scattering. This decomposition separates direct surface reflection from contributions involving transmission into the interior and subsequent propagation, and admits a natural trajectory interpretation. We then determine the associated Regge-Debye pole spectrum in the complex angular-momentum plane. For neutron-star-like tenuities ($R>3M$), the spectrum exhibits two pole families: a surface-wave branch associated with the surface matching condition and a broad-resonance branch associated with the interior regularity condition. For ultracompact objects ($R<3M$), the surface-wave branch persists, while the interior-resonance sector splits into broad- and narrow-resonance branches. We next reconstruct the scattering amplitude from the Debye partial-wave contributions and find excellent agreement with direct partial-wave calculations. Finally, we develop complex angular-momentum representations order by order in the Debye series, making explicit how the pole families and non-pole sectors contribute to each Debye term. In the neutron-star-like regime, we find a genuine competition between Regge-Debye pole sums and branch-cut contributions, and show that, at high frequency, the rainbow-like enhancement already arises from the first interior-transmission contribution and is dominated by the interior-resonance Regge-Debye poles. By contrast, in the ultracompact regime, the Debye amplitudes are overwhelmingly pole dominated.

We give a general construction of genuinely multipartite entanglement signals from families of lower-partite symmetric local-unitary invariants satisfying a natural compatibility condition. Möbius inversion on the partition lattice plays a key role in this construction. We show that many examples of multipartite entanglement signals considered in the literature fit naturally into this framework. We also explain how the genuinely multipartite signal can be extracted from a general, not necessarily symmetric, multi-invariant.

This reply is also a friendly introduction to the impossibility of emergence of preferred structures from the Hamiltonian $\mathsf{H}$ and the unit vector $|ψ\rangle$ only. The obstructions to emergence are illustrated on the concrete construction of a tensor product structure (TPS) from Soulas et al., 2025 (arXiv:2512.07468v2). Soulas et al. offer their TPS as a counterexample to the proof from Stoica, 2022a (arXiv:2102.08620) that structures constructed only from $\mathsf{H}$ and $|ψ\rangle$ either contradict physical observations or can't describe them unambiguously. Soulas et al.'s construction of a unique TPS can't be both invariant and compatible with physical observations, so it can't be a counterexample. Its incompatibility becomes visible by examining how the relation between $|ψ(t)\rangle$ and the TPS, encoding the entanglement, changes in time. Therefore their TPS doesn't refute, but confirms (Stoica, 2022a). Besides this, since Soulas et al.'s method to construct preferred structures consists of choosing their invariants, by the same logic one could claim as well that the masses of elementary particles emerge uniquely just by fixing their values by hand. Soulas et al.'s construction is concrete and can illustrate the major obstructions for emergent structures, confirming them despite doing the best possible to avoid them. This makes it an excellent pedagogical tool to illustrate the trilemma, but also the relational and structural aspects of quantum theory and its symmetries.

Distributed quantum computing (DQC) connects many small quantum processors into a single logical machine, offering a practical route to scalable quantum computation. However, most existing DQC paradigms are structure-agnostic. Circuit cutting proposed by Peng et al. in [Phys. Rev. Lett., Oct. 2020] reduces per-device qubits at the cost of exponential classical post-processing, while search-space partitioning proposed by Avron et al. in [Phys. Rev. A., Nov. 2021] distributes the workload but weakens Grover's ideal quadratic speedup. In this paper, we introduce a structure-aware framework for distributed quantum optimization that resolves this complexity-resource trade-off. We model the objective function as a factor graph and expose its sparse interaction structure. We cut the graph along its natural ``seams'', i.e., a separator of boundary variables, to obtain loosely coupled subproblems that fit on resource-constrained processors. We coordinate these subproblems with shared entanglement, so the network executes a single globally coherent search rather than independent local searches. We prove that this design preserves Grover-like scaling: for a search space of size $N$, our framework achieves $O(\sqrt{N})$ query complexity up to processors and separator dependent factors, while relaxing the qubit requirement of each processor. We extend the framework with a hierarchical divide-and-conquer strategy that scales to large-scale optimization problems and supports two operating modes: a fully coherent mode for fault-tolerant networks and a hybrid mode that inserts measurements to cap circuit depth on near-term devices. We validate the predicted query-entanglement trade-offs through simulations over diverse network topologies, and we show that structure-aware decomposition delivers a practical path to scalable distributed quantum optimization on quantum networks.

Previous work on the gravitational-wave background generated in a two-scalar-field cosmological model, in which warm inflation and the dark sector are unified within a single framework, has shown that the resulting spectrum could be potentially detectable by planned next-generation gravitational-wave observatories. In this work, we extend this analysis to the weak dissipation regime of warm inflation, highlighting how the features of the inflationary scenario play a crucial role in the production of gravitational waves. The full gravitational-wave energy spectrum is calculated using the formalism of continuous Bogoliubov coefficients. By comparing our results with those obtained in the strong dissipation regime and with the sensitivity curves of future detectors, we find that the weak dissipation regime improves the prospects for observational detection.

Extreme mass ratio inspirals (EMRIs), as a core target for future space-based gravitational wave detection, offer crucial observational grounds for testing strong-field gravitational theories and classifying black holes through their orbital dynamics and gravitational wave radiation characteristics. This study systematically investigates the periodic orbit characteristics and gravitational wave radiation properties of EMRIs in the spacetime of a non-minimally coupled Einstein-Yang-Mills (EYM) black hole. The results show that as the magnetic charge parameter \(Q\) and the non-minimal coupling constant \(ξ\) increase, the radii of the marginally bound orbit (\(r_{\text{MBO}}\)) and the innermost stable circular orbit (\(r_{\text{ISCO}}\)) decrease significantly, with corresponding reductions in the orbital energy \(E\) and angular momentum \(L\). Furthermore, the allowable parameter space of energy and angular momentum (\(E\)-\(L\)) for bound orbits shifts towards the left. We then plot typical periodic orbits through orbit classification, finding that the non-minimal coupling effect suppresses the orbital contraction induced by the magnetic charge, leading to degeneracy of orbits with different \(Q\) values towards the Schwarzschild case, as well as phase shifts, amplitude enhancement, and period shortening in the gravitational waveforms with increasing \(Q\) and \(ξ\). These results provide theoretical predictions for distinguishing different black hole models through future gravitational wave observations.

Starting from generalized position operators, we derive complex and quaternionic angular momentum operators along with their commutation algebra as well. These algebras differ from the standard Hermitian ones, especially in terms of commutation relations involving partial and total angular momentum operators. Despite these differences, the effective quantum expectation values obtained from slightly deformed algebras align with those from the conventional Hermitian algebra. This suggests that even though the wave functions and resulting dynamics differ from standard quantum Hermitian behavior, these deformed algebras can still be effectively understood as valid angular momentum algebras.

Quadratic Unconstrained Binary Optimization (QUBO) is a standard NP-hard optimization problem. Recently, it has gained renewed interest through quantum computing, as QUBOs directly reduce to the Ising model, on which quantum annealing devices are based. We introduce a QUBO formulation for permutations using compare-exchange networks, with only $O(n \log^2 n)$ binary variables. This is a substantial improvement over the standard permutation matrix encoding, which requires $n^2$ variables and has a much denser interaction graph. A central feature of our approach is uniformity: each permutation corresponds to a unique variable assignment, enabling unbiased sampling. Our construction also allows additional constraints, including fixed points and parity. Moreover, it provides a representation of permutations that supports the operations multiplication and inversion, and also makes it possible to check the order of a permutation. This can be used to uniformly generate permutations of a given order or, for example, permutations that commute with a specified permutation. To our knowledge, this is the first result linking oblivious compare-exchange networks with QUBO encodings. While similar functionality can be achieved using permutation matrices, our method yields QUBOs that are both smaller and sparser. We expect this method to be practically useful in areas where unbiased sampling of constrained permutations is important, including cryptography and combinatorial design.

Black hole solutions are studied here within the symmetric teleparallel formulation of gravity, employing the $f(Q)$ model in which the gravitational dynamics are governed by the non-metricity scalar $Q$. We focus on static, circularly symmetric spacetimes in $(2+1)$-dimensions, analyzing both charged and uncharged cases. By adopting a power-law form for $f(Q)$, we derive exact black hole solutions and explore their thermodynamic and geometric properties. Curvature and non-metricity scalars reveal central singularities stronger than those in general relativity. we find that the horizon radii increase with the charge parameter while higher values of the non-metricity coefficient, $c_{4}$, or the cosmological constant $Λ$ tend to merge or eliminate horizons, reducing their total number and altering the near-origin structure of the spacetime. We perform a detailed topological analysis based on the Euler characteristic and examine the geodesic completeness of the spacetime. Our findings show that, depending on the presence of electric charge, the singularity may or may not be reachable by geodesics. The thermodynamic stability is confirmed via temperature, entropy, and heat capacity calculations. This study highlights the rich structure of $f(Q)$ gravity in lower-dimensional settings and offers new insights into the nature of singularities and black hole topologies in modified gravity theories.

We consider an optical interferometer with coherent light in one input and a squeezed vacuum in another. Such an interferometer is known to beat the standard quantum limit of sensitivity to the difference of phase shifts in its arms. We find the ultimate limit of precision for such an interferometer by calculating quantum Fisher information of the mixed quantum state in one of the interferometer's outputs. We show that this information is asymptotically close to quantum Fisher information for the two-mode readout. We conclude that the single-mode readout is optimal for phase estimation in squeezing-enhanced interferometry.

We investigate Josephson phenomena in a slowly rotating stationary spacetime, emphasizing the distinct roles of gravitational redshift and rotational frame dragging motivated by [10.1007/JHEP02(2026)006]. Using a covariant formulation based on gauge-invariant phase dynamics and conserved currents within a $3+1$ decomposition, we analyze both AC and DC Josephson effects and interferometric configurations. Restricting attention to linear order in the rotation parameter $a$, and working in the Eulerian/ZAMO frame, we show that in the slow-rotation slicing adopted here the lapse function agrees with its static (Schwarzschild-type) form up to $\mathcal{O}(a^2)$, while rotational effects enter through the shift vector. Consequently, redshift effects on Josephson frequencies and DC critical currents remain unchanged relative to the non-rotating case at $\mathcal{O}(a)$. The AC Josephson relation retains its redshifted structure when expressed in terms of proper voltages and reduces to the standard flat-spacetime form when formulated in terms of asymptotic (Killing-time) observables. Likewise, the DC critical current measured at infinity scales with a single power of the lapse function and is unaffected by rotation at linear order in the absence of azimuthal condensate momentum. Rotational effects become relevant only in configurations sensitive to spatial phase transport or to synchronization with respect to the global time coordinate. In particular, RF-driven interferometric setups can acquire Sagnac-type phase offsets associated with frame dragging, whereas the DC fluxoid constraint remains unshifted at linear order in the present approximation. Our results provide a clean separation between lapse-driven redshift effects and shift-driven rotational contributions in Josephson physics and furnish a consistent framework for superconducting circuits in stationary spacetimes.

Spectral crowding of collective motional modes limits the fidelity of entangling interactions in trapped-ion quantum processors by inducing off-resonant coupling to spectator modes. We introduce a geometric-phase entangling interaction driven by a transverse, time-dependent structured-light force. By applying the force in a plane orthogonal to the optical propagation direction, we suppress the effects of spectral crowding while preserving single-ion addressing. The scheme is compatible with arbitrary qubit encodings, provided that the qubit states experience a differential AC Stark shift. We experimentally realise high-fidelity two-qubit gates with error rates below $5\times10^{-3}$ in ion crystals containing up to 12 ions confined within a single potential well. These results establish gradient-field light-shift gates as a scalable approach to high-fidelity entangling generation in spectrally crowded trapped-ion systems.

Quantum key distribution is unbreakable in theory but may be hacked via imperfections in its hardware implementations. While many imperfections have been mitigated by countermeasures and advanced security proofs, several remain unsolved. One of these is a superlinear behaviour in single-photon detectors, when the click probability rises faster with the photon number of an incoming light pulse than expected from individual independent photon detections. Here we test an avalanche single-photon detector sinusoidally-gated at 312.5 MHz for superlinearity. Its click probability is moderately superlinear. However, we notice that the click timing depends strongly on the incoming pulse energy. The click occurs progressively earlier, shifting more than 2 ns as the energy rises over a wide 50-dB range. An attacker might use this energy-time effect to conditionally toggle the click between adjacent key bit slots, violating an implicit assumption in the security proofs and rendering them inapplicable. We propose two attacks that exploit this flaw.

An all-orders worldline effective action for Kerr-Newman black hole is achieved in twistor particle theory. Exact hidden symmetries are identified in self-dual backgrounds.

We introduce the concept of quantum minimal learning machine (QMLM), a supervised similarity-based learning algorithm. The algorithm is conceptually based on a classical machine learning model and adopted to work with quantum data. We will motivate the theory and run the model as an error mitigation method for various parameters.

Gate fidelity -- an average fidelity over all possible input states -- is the workhorse metric for benchmarking quantum gates or circuits, yet fault-tolerant quantum computing ultimately depends on the worst-case behavior, typically quantifiable by so-called the diamond distance. In the low-error regime, the coherent errors can inflate the worst-case error even when the reported gate fidelity is high, making the gate fidelity alone an unreliable proxy for fault-tolerance readiness. To capture the missing information, we introduce a companion observable -- what we dub the fidelity deviation -- that quantifies how strongly the state-dependent fidelities fluctuate across input states. Adopting such fluctuations in assessing the fault-tolerance is physically natural because some input directions are nearly unaffected while others form narrow "valleys" that dominate adversarial circuit behavior. For coherent (unitary) gate errors on two or more qubits, we show that the gate fidelity together with the fidelity deviation constrains the relevant spectral moments of the error unitary, enabling an explicit and tight certificate of the worst-case error. Both quantities are estimated directly from the same randomized input-measurement experiment, without full process tomography. We show that the fidelity and its deviation can provide an economical, operationally meaningful, and accurate standard for assessing the fault tolerance of the engineered quantum gates and circuits.

Network coordination - synchronising traffic signals, scheduling trains, assigning communication slots requires minimising pairwise costs across coupled systems. These problems are NP-hard yet share a common Fourier-sparse structure exploitable by quantum algorithms. We introduce the Fourier Network Coordination problem (Fourier-NC),unifying eight application domains. For abelian and dihedral groups, classical sparse Fourier transforms match quantum in the same oracle model, limiting the advantage to at most polynomial. The genuine separation emerges for the symmetric group Sk: a conditional super-exponential speedup of k! -> poly(k) for class-function costs with non-trivial minimisers. When the minimising conjugacy class is structurally determined, the problem lies in NP (int) BQP and is conditionally outside P (Corollary 6.5), placing it in the intermediate complexity regime alongside integer factorisation and graph isomorphism. We formalise the abelian index α(G) = [G : Amax] as the structural invariant governing the quantum-classical gap and identify a three-regime complexity trichotomy: abelian ({α= 1, classical sFFT suffices), nearly abelian (α= dmax, polynomial advantage), and strongly non-abelian (α>>dmax, super-exponential advantage).

Quadratic quasinormal modes (QQNMs) and Christodoulou memory effect are key nonlinear phenomena in gravitational wave physics. QQNMs characterize the near zone nonlinear response of a perturbed black hole, whereas the memory effect is a nonlinear remnant imprinted at null infinity by outgoing radiation. This naturally raises the question of whether and in what sense the two can be bridged. We show that they are related through bridge coefficients which depend primarily on remnant black hole parameters during ringdown. Future space-based gravitational-wave detectors can probe this relation. These results provide a new avenue for testing gravity and a fresh perspective on the nonlinear regime of general relativity.

We develop a unified analytic treatment of the Horowitz--Polchinski string/black hole correspondence that systematically incorporates higher-derivative corrections to gravity. Working in Euclidean signature -- where the Euclidean black hole and the thermal scalar arise as competing saddles of the same finite-temperature ensemble -- we include the Gauss--Bonnet term. The analysis is rendered tractable in this UV--sensitive regime by the large-\(D\) expansion, which sharply separates the geometry into a universal near-zone and an asymptotic far-zone. In the near-zone, the coupled large-\(D\) equations reduce the thermal-scalar sector to an exactly solvable Schrödinger problem, from which we extract the \(α'\)-corrected decay exponent and the corresponding shift of the Hagedorn temperature. In the far-zone, we construct closed-form Euclidean solutions of Einstein--Gauss--Bonnet theory at leading order in both \(1/D\) and \(α'\). Matching the two regions yields the complete corrected saddle -- fixing its temperature, horizon data, and on--shell action -- and permits a fully analytic comparison of free energies between the thermal-scalar and black hole phases. This provides a controlled derivation of the HP correspondence point with explicit higher-curvature corrections.

The quantum Fisher information matrix (QFIM) is central to multiparameter quantum metrology, dictating the attainable sensitivity via the quantum Cramér-Rao bound. In this work, we investigate the ultimate precision limits for relativistic quantum thermometry in a bipartite system of uniformly accelerated Unruh-DeWitt detectors. Utilizing the symmetric logarithmic derivative (SLD) formalism within the QFIM framework, we analyze the individual and simultaneous estimation of the Unruh temperature $T$ and the initial-state parameter $Δ_0$. In the noiseless case, we demonstrate that these two parameters are quantum compatible, allowing the multiparameter quantum Cramér-Rao bound to be saturated without a loss of precision. We then examine the impact of environmental effects by comparing Markovian and non-Markovian dynamics. In the Markovian regime, dissipation leads to a monotonic degradation of estimation precision; conversely, non-Markovian memory effects induce temporal oscillations and transient precision enhancements due to information backflow. The robustness of the estimation protocols is further analyzed under correlated noisy channels, including amplitude damping, phase flip, and phase damping. We show that dissipative noise results in more significant precision loss than purely dephasing mechanisms, whereas classical correlations in the noise mitigate this degradation. Finally, we discuss the estimation of the detector energy-level spacing $ω$ as a natural extension, highlighting its sensitivity to the environmental structure. Our results provide a unified framework for relativistic multiparameter quantum metrology within the context of open quantum systems.

We present an analysis that investigates the $H_0$ and $S_8$ tensions by considering a dark energy model. The latter is a late-time model characterized by a future abrupt event known as the Little Rip (LR) model and characterised by one extra parameter, $β$, compared to the standard model, $Λ$CDM. To test this approach, we perform a statistical analysis by the MCMC method using the most recent observational data. We obtain a positive correlation in ($H_0$, $β$) plane. We also note that the Hubble tension is less than $3σ$ when using early measurements, i.e., Cosmic Microwave Background (CMB) data, and when combining it with Baryon Acoustic Oscillation (BAO) data, but it is no longer so when we combine early and late measurements (i.e. PantheonPlus (PP)). In addition, we test the model with DESI-DR2 combined with CMB and recent SNIa measurements. We notice that our model shifts toward the quintessence field. For a complete statistical analysis, we use the Akaike Information Criteria and Bayesian analysis of the evidence. According to Bayes factors, we find that the LR model provides an improved fit only to CMB data.

In this work, we introduce a method for finding exact solutions to the vacuum Einstein field equations in higher dimensions from a given solution to the chiral equation. When considering a $n + 2$-dimensional spacetime with $n$ commutative Killing vectors, the metric tensor can take the form $\hat g = f ( ρ, ζ) ( d ρ^2 + d ζ^2 ) + g_{μν} ( ρ, ζ) d x^μd x^ν$. Then, the Einstein field equations in vacuum reduce to a chiral equation, $( ρg_{, z} g ^{-1} )_{, \bar z} + ( ρg_{, \bar z} g ^{-1} )_{, z} = 0$, and two differential equations, $( \ln f ρ^{1-1/n} )_{, Z} = \fracρ{2} \operatorname{tr} ( g_{, _Z} g^{-1} )^2$, where $g \in SL( n, \mathbb{R} )$ is the normalized matrix representation of $g_{μν}$, $z = ρ+ i ζ$ and $Z = z, \bar z$. We use the ansatz $g = g ( ξ^a )$, where the parameters $ξ^a$ depend on $z$ and $\bar z$ and satisfy a generalized Laplace equation, $( ρξ^a _{, z} )_{, \bar z} + ( ρξ^a _{, \bar z} )_{, z} = 0$. The chiral equation to the Killing equation, $A_{a , ξ^b} + A_{b , ξ^a} = 0$, where $A_a = g_{, ξ^a} g^{-1}$. Furthermore, we assume that the matrices $A_a$ commute with each other; in this way, they fulfill the Killing equation.

Quantum Key Distribution (QKD) offers theoretically unbreakable security by leveraging quantum mechanics. However, practical implementation is challenged by environmental vulnerabilities, noise, and hardware imperfections. Recently, Machine Learning (ML) has emerged as a powerful tool to address these limitations and enhance the real-world viability of QKD systems. In this survey, we review ML techniques applied to improve QKD security and performance across five applications. First, parameter optimization, covering signal calibration, polarization alignment, phase stabilization, modulation state tuning, and post-processing enhancements to maximize secure key generation and minimize error rates. Second, attack detection, where ML models identify and classify quantum threats such as photon-number-splitting and Trojan-horse attacks. Third, protocol selection, leveraging ML to dynamically choose QKD protocols based on operational conditions. Fourth, key performance prediction of core metrics such as Secret Key Rate (SKR) and Quantum Bit Error Rate (QBER). Finally, quantum network management, optimizing large-scale QKD deployments through intelligent routing, node management, and resource allocation. Performance improvements are evaluated using accuracy, reduced QBER, and increased SKR. While ML shows significant potential for finance, government, and defense applications, challenges remain in scalability, computational demands, and real-world testing. Ongoing work should focus on lightweight, generalizable models and standardized benchmarks for practical ML-enhanced QKD deployment.

Black hole-neutron star mergers, together with binary neutron star mergers, are key laboratories for neutron star physics. They enable us to probe merger dynamics imprinted in gravitational waves and potential electromagnetic counterparts. These systems link microphysics and macrophysics by placing constraints on the dense matter equations of state, potentially revealing the imprint of hadron-quark phase transitions, clarifying the role of neutrino irradiation in shaping the ejecta, its r-process nucleosynthesis, and kilonova emission, as well as assessing how magnetically driven instabilities affect mass ejection and possible electromagnetic signatures. Despite their importance, black hole-neutron star mergers remain relatively less studied and therefore not yet well understood, largely due to the lack of publicly available numerical relativity setups suitable for such investigations. In this work, we present a fully reproducible black hole-neutron star merger simulation performed exclusively using Einstein Toolkit thorns, targeting the detected event \texttt{GW230529}. The simulations are carried out at three resolutions with finest grid spacings of $162$, $222$ and $310$ meters to assess numerical robustness. The entire setup, from initial data to a parameter file with some of the analysis scripts, is publicly released as a new Einstein Toolkit gallery example and will be distributed as part of the Hypatia release, establishing a reference black hole-neutron star merger configuration within the Einstein Toolkit.

We study the Null Energy Condition (NEC) arising from the Virasoro constraint on the string worldsheet. We then analyze how the NEC in the external spacetime directions emerges under general time-dependent string compactifications. Finally, we exhibit compactifications in which the averaged Einstein-frame condition allows the lowerdimensional description of the external spacetime to violate the NEC, thereby realizing a bouncing cosmology, while the higher-dimensional NEC remains satisfied, as dictated by worldsheet symmetry.

A concise review of various mathematical formulations of the uncertainty relations in quantum mechanics discovered since 1927 is given. Besides the traditional Heisenberg inequality, the modifications made by Schrödinger and Robertson, as well as generalizations to sets of several noncommuting operators, are considered. The "entropic" inequalities and "local" uncertainty relations, together with inequalities which connect the so-called total width and the mean peak width of a wave function, are discussed. Inequalities for the products of higher order moments of the coordinate and momentum are presented. Inequalities making the uncertainty relations more accurate when the "purity" of a quantum state is fixed are demonstrated. Diverse formulations of the energy-time uncertainty relations are considered.

Spin defects in diamond are promising platforms for quantum sensing. The longest electron spin relaxation times ($T_1$) at room temperature for solid-state defects are observed in nitrogen vacancy centers in diamond, which can reach 6.67 ms, and substitutional nitrogen ("P1 centers") in diamond, which exhibit a $T_1$ of 2 ms. No other solid-state defect has exhibited millisecond-scale spin relaxation times at room temperature thus far. Here, we characterize the spin properties of the WAR5 defect in diamond with pulsed electron spin resonance. The observed $T_1$ is one of the longest for solid-state spin defects: 0.97(27) ms at room temperature and 14.38(19) min at 4 K. The observed coherence time ($T_2$) is 246(7) $μ$s, which can be extended to 6.49(34) ms at 4 K with dynamical decoupling. Furthermore, we demonstrate optical spin polarization with a range of wavelengths from 405 nm to 500 nm and propose potential zero-phonon line candidates.

For any unitary representation $ρ$ on a finite-dimensional Hilbert space \(V\) with differential \(dρ: \mathfrak{g} \to \mathfrak{u}(V)\) for the Lie algebra $\mathfrak g$, we consider the Hamiltonian evolution \[ U_X(t) \coloneqq ρ(\exp(tX)) = e^{t\,dρ(X)}, \qquad t\in\mathbb{R}. \] For any complexification $ X_\mathbb{C} = X_0 + \sum\limits_{α\inΔ} x_αE_α$ associated with the root system $Δ$, we introduce the numerical invariants %\emph{root activity} and \emph{root curvature} functionals \begin{align*} \mathcal{A}_p(X) &\coloneqq \Bigl(\sum_{α\inΔ} |x_α|^p \,\|dρ(E_α)\|_{\mathrm{op}}^p\Bigr)^{1/p}, \quad 1\le p<\infty\\ \mathcal{C}(X) &\coloneqq \Bigl(\sum_{α\inΔ} |α(X_0)|^2\,|x_α|^2 \,\|dρ(E_α)\|_{\mathrm{op}}^2\Bigr)^{1/2}, \end{align*} where \(\|\cdot\|_{\mathrm{op}}\) is the operator norm on \(\mathrm{End}(V)\). We first describe how the Hamiltonian \(dρ(X)\) is distributed along the directions of root spaces $\mathfrak{g}_α$. Our main result shows that for each fixed \(X\in\mathfrak{g}\) there exists a constant \(C_X>0\) such that \[ \bigl\| e^{t(dρ(X_0)+dρ(X_{\mathrm{root}}))} - e^{\frac{t}{2}dρ(X_0)} e^{t dρ(X_{\mathrm{root}})} e^{\frac{t}{2}dρ(X_0)} \bigr\|_{\mathrm{op}} \le C_X\,t^{3}\,\bigl(\mathcal{C}(X)+\mathcal{A}_1(X_{\mathrm{root}})\bigr) \] for all sufficiently small \(|t|\). We also introduce a root-gate circuit model and test this on spin$-$chain Hamiltonians on \((\mathbb{C}^2)^{\otimes n}\subset\mathfrak{su}(2^n)\), where root spaces are spanned by matrix units, \(\mathcal{A}_p\), and \(\mathcal{C}\), which gives sharper complexity bounds and dimension$-$free representation$-$theoretic invariants.

We study graphene in an external magnetic field within a noncommutative (NC) framework. A gauge-invariant NC Hamiltonian is derived, and the system is analyzed using the ladder-operator formalism, yielding deformed Landau levels and eigenstates. The thermal properties of gauge-invariant NC graphene are then investigated via the partition function, constructed using Euler and Hurwitz zeta functions. Analytical expressions for the partition function, free energy, internal energy, entropy, and specific heat are obtained and numerically evaluated.

The maintenance of quantum entanglement lays the elementary building block of quantum information processing, requiring an integration of long coherence time, sufficient storage capacity, and high-fidelity entangling gates. Here we encode two-qubit entangled states into the decoherence-free subspaces (DFS) of four ions in a cryogenic trap. By crosstalk-free sympathetic cooling under dual-type encoding and multi-state detection which discards the collision-induced leakage error, we achieve a storage lifetime of about one hour for the entangled logical states. We further study the second-order DFS and show its advantage in suppressing the spatially nonuniform noise over the first-order DFS. Our work paves the way for applications of DFS quantum memories in quantum computing, quantum network and precision measurement.

We adopt a general relativistic ray-tracing approach to study the shadows and polarization images of spherically symmetric Gauss-Bonnet (GB) black holes enveloped by geometrically thick accretion flows. Specifically, we adopt a phenomenological RIAF-like model and an analytical Hou disk model. In the RIAF-like model, increasing the GB coupling parameter $λ$ reduces both the size and brightness of the higher-order image, while increasing $θ$ alters the shape of the higher-order image and obscures the horizon's outline. The main difference between isotropic and anisotropic emission is that the latter produce distortion of the high-order image in the vertical direction, leading to an elliptical morphology. For the Hou disk model, due to specific regions being geometrically thinner with the conical approximation, the high-order images are narrower with the increase in $λ$ than the RIAF model. While increasing $θ$ enhances the brightness of the direct images outside the higher-order images, but hardly changes the size of the higher-order images, which is in sharp contrast to the RIAF model. Meanwhile, the Hou disk produces polarization patterns that trace the brightness configuration and are affected by $λ$ and $θ$, reflecting the intrinsic structure of spacetime. These results illustrate that intensity and polarization in thick-disk models provide probes of GB black holes and near-horizon accretion dynamics.

Krylov complexity provides a powerful framework for characterizing the dynamical evolution of quantum systems through the spreading of states in Krylov space. The motivation for this is rooted in the optimality of the Krylov basis for the analyzed cost function. In this work, we reinterpret the motivation for the Krylov basis from a dynamical perspective and show that it corresponds to a first-order approximation of the time-evolution operator. We extend this framework to higher-order generators and analytically disprove the optimality assumption by showing that an infinite-order generator can be constructed to exhibit smaller spread for arbitrary times. We propose a natural time scale for the construction of these higher-order generators and discuss results for matrices sampled from Gaussian Unitary Ensembles, demonstrating smaller Krylov complexity at all higher orders. These results extend the framework of Krylov complexity beyond the conventional Krylov basis by disproving the widely held assumption of optimality, extending the construction to higher-order generators, and introducing a physically motivated method for their construction. Our findings therefore suggest that previous statements and results concerning Krylov complexity may need to be reconsidered.

The classical no-hair theorem states that stationary black holes in general relativity can be completely described by only a small set of global parameters. Within this framework, no additional geometric structures are expected to persist outside the event horizon. However, quantum vacuum polarization may introduce small modifications to the near-horizon geometry, effectively giving rise to what is known as quantum hair. Such corrections may provide a possible window into the microscopic structure and thermodynamic properties of black holes. In this work, we examine how the quantum hair parameter γ influences the periodic orbital dynamics of test bodies in extreme mass-ratio inspirals (EMRIs) and their associated gravitational-wave emission. We find that γ significantly modifies the characteristic radii and angular momenta of two important circular orbits, namely the marginally bound orbit (MBO) and the innermost stable circular orbit (ISCO), leading to a shift in the allowed region of the energy-angular momentum (E-L) phase space. Based on the rational number q classification, we further show that quantum corrections tend to enhance the zoom-whirl orbital behavior.Gravitational-wave calculations using the Numerical Kludge approach indicate that quantum hair alters the effective spacetime potential, produces small drifts in the fundamental orbital frequencies, and consequently leads to observable phase dephasing in longduration signals. These results provide a dynamical signature for distinguishing quantum-corrected black holes from classical Schwarzschild ones and offer theoretical motivation for testing quantum gravity effects with future space-based gravitational-wave observatories.

Quantum measurement is a dynamical process of an apparatus coupled to a test system. Ideal measurement of the $z$-component of a spin-$\frac{1}{2}$ ($s_z=\pm\frac{1}{2}$) has been modeled by the Curie-Weiss model for quantum measurement. Recently, the model was generalized to higher spin and the thermodynamics was solved. Here the dynamics is considered. To this end, the dynamics for spin-$\frac{1}{2}$ case are cast in general notation. The dynamics of the measurement of the $z$-component of a spin-1 ($s_z=0,\pm 1$) are solved in detail and evaluated numerically. Energy costs of the measurement, which are macroscopic, are evaluated. Generalization to higher spin is straightforward.

We study the general black hole solutions of dimensionally reduced five-dimensional Einstein-Gauss-Bonnet gravity. The reduced theory contains gravity, electromagnetism and a scalar field, with nonlinear corrections to the action and nontrivial couplings. The solutions can be classified through mass and electric and magnetic charge. They present peculiar features with respect to the solutions of the standard Kaluza-Klein theory without Gauss-Bonnet corrections, like the existence of an extremal mass even in the neutral case. Also the thermodynamics is affected, for example, extremal black holes display nonvanishing temperature and entropy.

In the theories with extra dimensions, gravitational waves can leak into extra dimensions, resulting in a reduction in the amplitude of the observed gravitational waves. Such an effect modifies the standard luminosity distance of gravitational wave sources, which enables constraining extra dimensions by measuring both the luminosity distance and redshift from gravitational wave events. The main purpose of this paper is to assess the capacity of the third-generation gravitational wave detector network, Einstein Telescope and Cosmic Explorer, to constrain the theory of extra dimensions through observations of precessing binary black hole mergers. To this end, we generate a dataset of signals of precessing binary black hole mergers detectable by the detector network with one year of observations. We then employ the Fisher information matrix method for parameter estimation, together with a dark siren approach that obtains the redshift information from the galaxy catalog within a hierarchical Bayesian framework. Our results show that the precession in the binary system can significantly improve the precision of the luminosity distances and narrow their redshift ranges, thus leading to a tighter constraint on the extra dimensions. Based on this approach, we constrain the number of spacetime dimensions $D$ and the screening scale $R_{\rm c}$ of the extra dimensions, obtaining $D = 3.99^{+0.07}_{-0.06}$ and $\log_{10}(R_c/\mathrm{Mpc}) > 3.76$ at $68\%$ credible level. This result improves previous ones from an analysis with GWTC-3 by about one order of magnitude. This work provides theoretical foundations and projected sensitivities for future third-generation gravitational wave observations in exploring the effects of extra dimensions.

We compute highly accurate first principle based \textit{ab initio} adiabatic potential energy surfaces (PESs) using State-Averaged Multi-Configurational Self-Consistent Field (SA-MCSCF) followed by internally contracted Multi-Reference Configuration Interaction method incorporating fixed-reference Davidson corrections [ic-MRCI(Q)], where a full valence active space of 18 electrons in 12 orbitals and aug-cc-pVQZ basis set are employed for the low-lying four singlet electronic states of ozone ($\tilde{X}^1A'$, $1~^1A''$, $1~^1A'$ and $2^1A''$). It accurately reproduces the dissociation energies of ozone (1.101 eV) as well as the molecular oxygen (5.106 eV) along with vibrational frequencies of O$_3$ in comparison with experimental data. To ensure appropriate accuracy and proper convergence in the interaction as well as asymptotic regions, we (a) extend the number of electronic states in SA-MCSCF calculation (singlet as well as triplet and quintet); (b) systematically expand the active space [(12e,9o) $\rightarrow$ (18e,12o) $\rightarrow$ (24e,15o)] and basis set size (AVDZ $\rightarrow$ AV6Z $\rightarrow$ Complete Basis Set limit); (c) incorporate multi-reference character along with Davidson correction. Conical intersections between the adjacent electronic states (1-2, 2-3 and 3-4) are located at \textit{C}$_{2v}$, \textit{D}$_{3h}$ as well as \textit{C}$_{s}$ geometries through the four-state adiabatic-to-diabatic transformation of non-adiabatic coupling terms (NACTs) computed at Coupled-Perturbed Multi-Configurational Self-Consistent Field (CP-MCSCF) method along the circular contours. Finally, we present: (a) ic-MRCI(Q) calculated minimum energy path of incoming oxygen to the diatom (O$_2$) is devoid of any ``reef'' feature; (b) NACTs and diabatic PES matrix elements as function of hyperangles ($θ$,$φ$) at a fixed hyperradius $ρ= 4$ Bohr for a four state sub-Hilbert space.

We investigate a class of first-order scalar field theories minimally coupled to a Carrollian connection that are defined intrinsically on the Carrollian plane, i.e., the theories are not defined via limits of Lorentzian theories. The theories built are invariant under the extended Carrollian transformations which include supertranslations. The symmetry allows for a large class of Lagrangians, independence of spacetime coordinates is all that is required. However, invariance under supertranslations (which include boosts as linear supertranslations) forces the energy density to be static and the momentum density to vanish -- this precludes on-shell propagation of fields. Thus, to have propagating theories, one must move beyond single field theories that are minimally coupled to the geometry.

We investigate static, spherically symmetric black hole spacetimes induced by the spontaneous Lorentz symmetry breaking of a Kalb--Ramond (KR) two-form field non--minimally coupled to gravity, coexisting with an anisotropic fluid. By adopting a general equation of state where the radial pressure relates to the energy density via $w_1 = -1$ and the tangential pressure via an arbitrary parameter $w_2$, we derive exact analytical solutions representing black holes surrounded by diverse matter fields, including dust ($w_2=0$), radiation ($w_2=1/3$), and dark energy-like distributions ($w_2=-1/2$). A rigorous analysis of curvature invariants confirms a genuine core singularity, while the global geometry and adherence to standard energy conditions are shown to be highly sensitive to the interplay between the KR coupling ($\ell$), the fluid density parameter ($K$), and $w_2$. Furthermore, we analyse null geodesics in detail to determine the photon sphere and shadow radii. Using the Gibbons--Werner geometrical approach and the Gauss-Bonnet theorem applied to the optical metric, we compute the weak deflection angle of light, demonstrating that both the KR field and the anisotropic fluid significantly enhance light bending, particularly in dark-energy-like backgrounds. Finally, we evaluate strong deflection limit (SDL) observables for the supermassive black holes Sgr A$^*$ and M87$^*$, revealing quantifiable deviations from standard Schwarzschild geometries. These results offer novel astrophysical signatures for constraining string-inspired KR gravity and anisotropic dark matter halos using current and future observations.

Scalable characterization of quantum processors is crucial for mitigating noise and imperfections. While randomized measurement protocols enable efficient access to local observables, inferring a globally consistent description of multi-qubit processes remains challenging. Here we introduce a local-to-global reconstruction framework for one-dimensional multi-qubit states and channels. The method is efficient provided that correlations, as quantified by the conditional mutual information, decay exponentially. In particular, we prove that under this assumption, the required number of samples scales polynomially with the system size and the desired global reconstruction error. Our approach is based on combining local shadow tomography with locally optimal recovery maps obtained by convex optimization. We supplement these rigorous guarantees by studying the performance of the protocol numerically for a system evolving under a local Lindbladian and a noisy, shallow circuit. By employing a tensor networ representation, we reconstruct channels acting on up to 50 qubits and accurately recover global diagnostics such as the process fidelity, the Choi state purity, and Pauli-weight-resolved process matrix elements. Our work thus extends the powerful toolbox local shadow tomography to scalable channel characterization with access to global properties.

We extend a recently developed projective circuit quantisation approach to incorporate superconducting Higgs modes associated to gap dynamics. This approach starts from a microscopic fermionic Hamiltonian for mesoscopic superconductors, and projects the system onto its low-energy "BCS" Hilbert space. We derive analytical results for the superconducting Higgs mass, "spring" constant, and oscillation frequency of the gap dynamics, which we validate numerically. We compute anharmonic corrections to the Higgs frequency for higher excitations of small superconducting islands, and compare our results to previous long-wavelength calculations.

Generalised Probabilistic Theories (GPTs) provide a unifying framework encompassing classical theories, quantum theories, as well as hypothetical alternatives. We investigate the problem of extending a system with a finite set of transformations. We also investigate the problem of adding to a translation invariant set of systems a finite set of entangled states and effects, plus all their images by the translation symmetry. We show that determining whether such extensions are consistent with the axioms of GPTs is undecidable: they are computationally equivalent to the halting problem for Turing machines. The source of the undecidability is that these finite extensions generate infinitely many conditions which must be checked, because iterating transformations produces infinitely many new transformations, and similarly, entangled states and effects generate infinitely many new states via the analog of teleportation. Our results show that extending GPTs to include dynamics or entanglement encounters fundamental computability obstructions, which can only be circumvented by introducing additional physical or mathematical assumptions.

The two-dimensional effective Polyakov action is often realized as the anomalous contributions of string theories and fermions coupled to gravity in two-dimensions. However, as a result of the reparameterization invariance, one finds that the effective action produces vanishing Hamiltonians as constraints even in disparate gauges such as the dynamical light-cone and the ADM formalism of the metric. On the other hand, two-dimensional gravitational theories naturally arise as geometric actions on the coadjoint orbits of the Virasoro algebra. The Thomas-Whitehead gravity formalism extends the effective Polyakov action in such a way that the defining coadjoint element for the orbit becomes a dynamical field, viz the diffeomorphism field. In this work, we examine the constraint analysis and quantization of the Hamiltonian in the context of Thomas-Whitehead gravity using both the dynamical light-cone and the ADM formalisms of the metric. Constraint analysis is then repeated in a Minkowski background and with a dynamical action for the diffeomorphisms field arising from the Thomas-Whitehead action. Adding dynamics to the diffeomorphism field subsequently removes the vanishing Hamiltonians.

Processing qubit Hamiltonians derived from electronic-structure problems can become classically prohibitive because many downstream manipulations still rely on dense operator constructions whose cost grows exponentially with qubit number. We introduce Qronecker, a cut-aware low-rank Kronecker decomposition algorithm that turns Hamiltonian compression into a certifiable, resource-aware decision primitive. Operating entirely in Pauli coefficient space, Qronecker avoids forming dense 2^n x 2^n matrices, constructs low-rank Kronecker approximations under a chosen bipartition, and returns both an instance-specific compressibility curve and a state-independent worst-case energy certificate that links rank and cut choices to conservative energy-deviation bounds. Across molecular benchmarks comprising hundreds of systems up to 30 qubits, we find that traceless low-rank structure is common but heterogeneous: many screened systems reach high coefficient-space fidelity at low rank, yielding large savings in classical preprocessing and conditional reductions in downstream circuit-resource proxies, while the certificate remains valid but conservative on the auditable subset. The same analysis shows that fixed global fidelity targets are not generally sufficient for chemistry-level guarantees, motivating adaptive rank and cut selection. These results position Qronecker as a certifiable compression primitive for rank and cut selection in quantum-chemistry Hamiltonian processing.

Characterizing the noise affecting superconducting qubits is essential for improving their performance. Existing noise-sensing techniques use the qubit itself as a detector, but its short coherence time limits both sensitivity and accessible frequency range. Here, we demonstrate a method for measuring qubit frequency noise by converting it into photon loss in a coupled high-quality superconducting cavity. We prepare a single photon in the cavity and perform repeated mid-circuit qubit measurements with post-selection to isolate noise-induced loss from intrinsic cavity decay, placing an upper bound on the intrinsic dressed-dephasing rate of $(0.29 \, \mathrm{s})^{-1}$ at 508 MHz, corresponding to a qubit frequency-noise power spectral density below $5.4\times10^3\,\mathrm{Hz}^2/\,\mathrm{Hz}$. By exploiting the cavity's millisecond-scale lifetime, this technique provides access to high-frequency noise processes that are beyond the reach of conventional qubit-based spectroscopy and that may impose previously unexplored limits on qubit coherence.

Quantum coherence and other non-classical features are widely discussed in chemical dynamics, yet it remains difficult to quantify when such resources are operationally relevant for a given process and observable. While quantum resource theories provide a comprehensive framework for comparing free and resourceful settings, existing approaches typically rely on resource monotones or on performance bounds under free operations, and do not directly quantify the maximal influence a chosen resource can exert on a fixed chemical dynamics. Here, we introduce task specific, process level quantifiers that upper bound the largest change a quantum resource can induce in a target figure of merit. Central is a resource impact functional $\mathcal{C}_M(Λ)$, defined by comparing a state with its paired resource-free counterpart under the same quantum channel $Λ$, which admits an operational interpretation in binary hypothesis testing. We derive variation and time bounds that constrain how rapidly a resource can modify a target signal, providing resource-aware analogues of quantum speed limits. Moreover, we show that open system dynamics can be decomposed into free and resourceful components such that only the resourceful component contributes to $\mathcal{C}_M(Λ)$, thereby isolating the parts of a generator responsible for resource-induced changes in the observable. We illustrate the framework exemplary for energy transfer in a donor-acceptor dimer in two analytically solvable regimes. Our results provide a general toolbox for diagnosing and benchmarking quantum resource effects in molecular processes.

Rotating traversable wormholes allow the effects of frame dragging and rotation to be studied in the absence of event horizons. We develop a quantum field theoretic treatment of massless scalar perturbations in the rotating Teo spacetime. This spacetime is an exact, stationary, horizonless wormhole connecting two asymptotically flat regions. Using the Bogoliubov transformation formalism, we construct ``in'' and ``out'' mode solutions defined on the two asymptotic regions and compute the Bogoliubov coefficients that quantify vacuum mode mixing. The effective radial potential induced by rotation and frame dragging forms an asymmetric scattering barrier. This geometric asymmetry allows an exact analytic evaluation of reflection and transmission amplitudes via the barrier-penetration exponent. This results in closed-form expressions for the Bogoliubov coefficients, the mean particle number, and the two-mode entanglement entropy as functions of the rotation parameter. The resulting amplification arises at the level of quantum Bogoliubov mode mixing and vacuum squeezing, rather than classical superradiant flux enhancement. Since this spacetime is stationary, particle creation originates from geometric asymmetry and boundary conditions, and not from explicit time dependence. Co-rotating and counter-rotating modes experience inequivalent scattering. This renders the process intrinsically non-reciprocal. We identify this mechanism as a stationary, geometric analogue of the Asymmetric Dynamical Casimir Effect (ADCE). In the rotating Teo geometry, rotation and frame dragging play the role that moving boundaries play in the dynamical Casimir effect, acting as the source of asymmetric vacuum mode mixing.

We introduce heterogeneous quantum error-correcting codes composed of qubit types with distinct error channels and study their performance in the code-capacity regime using maximum-likelihood tensor network decoding. In the regime where both qubit types share the same noise bias but differ in physical error rate, placing noisier qubits in the bulk -- where each error triggers more syndrome bits -- and cleaner qubits on the boundary yields thresholds exceeding 0.4 (compared to ~0.2 for the reverse placement) and improvements exceeding three orders of magnitude in logical error rate at high bias, with the advantage growing exponentially with code distance. In the regime where both types share the same error rate but differ in bias, the optimal strategy reverses: placing high-bias (more predictable) qubits on the boundary increases the threshold from 0.292(5) to 0.360(9) at a bias ratio of 100, and from 0.29(1) to 0.398(4) at a bias ratio of 1000. We also observe a striking bias-inversion property: the logical error channel becomes strongly XX X- and YY Y-biased despite the physical noise being ZZ Z-biased. We propose a stabilizer-ratio hypothesis that provides a unified information-theoretic explanation for both placement rules and predicts even larger advantages for code families such as color codes.

We develop a flat-space holographic dictionary for a free massive spinor field in four-dimensional Minkowski spacetime, using the hyperbolic (Milne) slicing into $\mathbb H^3$ (Euclidean $\mathrm{AdS}_3$). Decomposing bulk fields into $\mathbb H^3$ harmonics labeled by a continuous parameter, we obtain the renormalized on-shell action as a functional of boundary data and extract the corresponding two-point correlation functions of dual spinning operators on the celestial sphere. The resulting correlators take the universal form dictated by two-dimensional conformal symmetry for spin-$\frac{1}{2}$ primaries. In this way, the four-dimensional Dirac problem is reduced to a family of effective $\mathrm{AdS}_3$ problems, closely following the logic of standard AdS/CFT. We show how the near-boundary behavior of the bulk spinor selects the appropriate celestial sources and determines the conformal dimension of the dual operators. As a further application, we construct the associated spinor conformal primary wavefunctions and clarify their relation to the flat-space bulk-to-boundary map.

Spectral-sirens inference enables the extraction of cosmological parameters from gravitational-wave data alone, without electromagnetic counterparts or galaxy catalogs. We introduce new parametric mass functions for the binary black hole population that capture significant structure across the mass spectrum and are moderately favoured by Bayesian evidence over simpler models. Analysing the latest gravitational-wave transient catalog, GWTC-4.0, we show that powerlaws-only population models constrain the Hubble constant to $23\%$ precision, $H_0 = 53.3^{+14.0}_{-10.8} ~\rm km \,s^{-1} \,Mpc^{-1}$ at $68\%$ confidence level. This represents a $\sim 50\%$ improvement over the corresponding binary black hole-only analysis by the LIGO-Virgo-KAGRA collaboration, achieving precision comparable to their joint analyses including neutron stars and galaxy catalogs. We further test alternative cosmological models, establishing competitive constraints on modified gravitational-wave propagation, while bounds on the dark energy equation-of-state parameters remain uninformative. Projecting to future O5 observing run, we forecast substantial improvements in $H_0$ and modified propagation parameters with larger datasets at higher redshifts. Our results highlight the strong interplay between the black hole mass distribution and inferred cosmology, underscoring the need for suitable population models to fully exploit gravitational-wave data.

We show that all two-dimensional Horndeski theories can arise from the reduction of pure gravities in $d \geq 4$ dimensions and therefore all onshell configurations for the two-dimensional metric and scalar field correspond to genuine $d$-dimensional gravitational vacuum solutions. We discuss separately the two-dimensional Horndeski theories which can arise from the reduction of $d$-dimensional generally covariant gravitational actions built only from curvature invariants without covariant derivatives and possessing second-order equations of motion on $2 + (d-2)$ warped-product backgrounds. The discussion is subsequently extended to generic $d$-dimensional gravitational actions with this latter property. We establish a Birkhoff theorem for all gravitational theories whose reduction yields an integrable two-dimensional Horndeski theory, in which case static spherically symmetric solutions satisfy $g_{tt} g_{rr} = -1$ in Schwarzschild gauge whereby the metric function $g_{tt} = -f$ is determined by an algebraic equation. We therefore propose to call all such theories quasi-topological gravities. These results can be used to show in reverse that any $d$-dimensional static spherically symmetric and asymptotically flat spacetime satisfying $g_{tt} g_{rr} = -1$ in Schwarzschild gauge with an invertible dependence of $f$ on the ADM mass can be reconstructed explicitly as a vacuum solution to a $d$-dimensional gravitational theory. We discuss examples of regular black holes such as the Bardeen spacetime, which could not be obtained from polynomial and non-polynomial quasi-topological gravities involving only curvature invariants without covariant derivatives.

Recent advances in quantum sensing -- optical clocks at $5.5\times 10^{-19}$ systematic uncertainty, frequency-dependent squeezing below the standard quantum limit, quantum magnetometers approaching fundamental sensitivity limits -- raise a natural question: can these technologies detect gravitational waves directly, or enhance existing detectors beyond current capabilities? We show that the answer is primarily determined by the \emph{coupling mechanism} between the gravitational wave and the sensor. Starting from the tidal Hamiltonian in Fermi normal coordinates, we identify three physically distinct coupling mechanisms and derive their transducer gains within linearized general relativity and non-relativistic quantum mechanics. Internal atomic coupling (tidal distortion of electronic wavefunctions) yields a transducer gain $G_A = 2.4\times 10^{-20}$, with vanishing first-order energy shifts for all $J=0$ clock states -- a $\sim\!10^{35}$ deficit relative to laser interferometry that exceeds any projected quantum enhancement. Center-of-mass coupling (Doppler shifts from geodesic motion) reaches strain sensitivities of $\sim\!10^{-18}$, still $10^4$ above LISA requirements. Light propagation coupling (phase accumulation over macroscopic baselines) provides the enormous transducer gain that makes laser interferometry -- and atom interferometry -- viable. For detectors exploiting this third mechanism, we quantify how much improvement quantum sensors can provide through the detector's noise architecture: LISA's noise budget is $\sim\!91\%$ classical, limiting combined quantum enhancement to $\mathcal{E} \approx 1.04$, while ground-based detectors in the shot-noise-dominated regime achieve $\mathcal{E} = 1.8$--$2.4$. Atom interferometers exploit the same light-propagation mechanism to uniquely access the 0.01--10~Hz band.

In this paper, we introduce a technique for contracting (i.e. numerically evaluating) ZX-diagrams whose complexity scales with their rank-width, a graph parameter that behaves nicely under ZX rewrite rules. Given a rank-decomposition of width $R$, our method simulates a graph-like ZX-diagram in $Õ(4^R)$ time. Applied to classical simulation of quantum circuits, it is no slower than either naive state vector simulation or stabiliser decompositions with $α= 0.5$, and in practice can be significantly faster for suitably chosen rank-decompositions. Since finding optimal rank-decompositions is NP-hard, we introduce heuristics that produce good decompositions in practice. We benchmark our simulation routine against Quimb, a popular tensor contraction library, and observe substantial reductions in floating-point operations (often by several orders of magnitude) for random and structured non-Clifford circuits as well as random ZX-diagrams.

Threshold estimation is central to fault-tolerant quantum computing, but the reported threshold depends not only on the code and noise model, but also on the decoder used to interpret syndrome data. We study this dependence for surface-code threshold estimation under both a standard Pauli noise model and a hybrid continuous-variable/discrete model motivated by GKP-style digitization. Using LiDMaS+ as a common experimental platform, we compare minimum-weight perfect matching (MWPM) and Union-Find under matched sweep grids, matched distances, and deterministic seeding, and we additionally evaluate trained neural-guided MWPM in the hybrid regime. In the Pauli baseline at distance $d=5$, MWPM consistently outperforms Union-Find, reducing the mean sampled logical error rate from $0.384$ to $0.260$, and producing a stable threshold summary with crossing median $p_c \approx 0.053$. In the hybrid fixed-distance run, Union-Find is substantially worse than MWPM (mean LER $0.1657$ versus $0.1195$), while trained neural-guided MWPM tracks MWPM closely (mean LER $0.1158$). Across hybrid multi-distance sweeps, the distance-dependent reversal in logical-error ordering remains visible, but the grid-based crossing estimator still returns boundary-valued $σ_c=0.05$ for all decoders. Neural-guided runs also show elevated decoder-failure diagnostics at high noise ($\max$ decoder-failure rate $0.1335$ at $d=7,σ=0.60$), indicating that learned guidance quality and decoder robustness must be reported alongside threshold curves. These results show that decoder choice and estimator design both materially affect threshold inference.

We show that deterministic multimode Gaussian channels admit a symmetric-space description. Passing from the n-mode Siegel disk to a doubled version of that space lets general Gaussian dynamics act by a linear-fractional (Mobius) transformation on a single matrix parameter. This doubled disk naturally parametrizes Gaussian kernels in the Fock-Bargmann representation, and contains an explicit physical subset corresponding to valid mixed Gaussian states. Starting from the standard X,Y parametrization of a deterministic Gaussian channel, we construct a normalized oscillator-semigroup element whose fractional action reproduces the channel update on that subset; Gaussian unitaries appear as the symplectic, isometric special case. This gives a bridge between covariance-matrix channel theory and the adjacency-matrix or symmetric-space picture, preserves a simple composition law given by matrix multiplication of the acting blocks, and suggests a direct route to graphical update rules beyond pure states.

We propose a nonlocal extension of the generalized Dirac oscillator (GDO) in $(1+1)$ dimensions by replacing the multiplicative interaction $f(x)$ with an integral operator $\hat F$ with kernel $f(x,x')$. The resulting Dirac equation preserves an operator factorization and decouples into two nonlocal Schrödinger-type (Sturm--Liouville) equations for the spinor components. We derive explicit expressions for the associated supersymmetric partner kernels in terms of $f$ and its derivatives, and we show that a complex-translation metric $η=e^{-θp_x}$ leads to a simple sufficient \emph{kernel-level} pseudo-Hermiticity constraint, $f(x+\ii\hbarθ,x'+\ii\hbarθ)=f^*(x',x)$, extending the familiar local complex-shift criteria. To provide a transparent \emph{nonlocal-to-local} interpretation, we adapt the Coz--Arnold--MacKellar current-based localization to each component equation, obtaining energy-dependent equivalent local potentials and multiplicative Perey (damping) factors. The mapping breaks down precisely at current zeros, thereby diagnosing the spurious solutions of the corresponding nonlocal Schrödinger problem. Finally, we illustrate the formalism with analytically tractable benchmarks (the local Dirac oscillator and a translation-invariant kernel) and with a finite-rank separable model (Gaussian form factor) that reduces the integro-differential problem to a small set of coupled ordinary differential equations and algebraic constraints.

Simulating the dynamics of molecular excitons in complex nanophotonic environments requires integrating rigorous electromagnetic simulations with accurate treatments of open quantum system dynamics. In this work, we develop MQED-QD (Macroscopic Quantum Electrodynamics for Quantum Dynamics), a robust computational package for simulating exciton dynamics in arbitrary dielectric and plasmonic environments. Based on the MQED framework, the package offers a unified workflow for constructing the dyadic Green's functions from classical electromagnetic solvers, parametrizing quantum master equations, and propagating the time evolution to determine the molecular subsystem's dynamical properties. To demonstrate the package's capabilities, we simulate exciton transport within a one-dimensional molecular chain near a silver nanostructure, including benchmarking against planar surfaces and exploring the influence of silver nanorods. Our results reveal that surface plasmon polaritons on nanorods dramatically enhance long-range dipole-dipole interactions, accelerating exciton delocalization and yielding higher participation ratios compared to planar geometries. By elucidating accurate molecular exciton dynamics in conjunction with nanophotonics and plasmonics, MQED-QD provides a powerful, open-source package that facilitates the rational design of nanoscale architectures.