arXiv Daily Digest - 2026-03-02

GPU-accelerated server platforms that share most of their hardware architecture often require separate firmware images due to minor hardware differences--different component identifiers, thermal profiles, or interconnect topologies. I built nvidia-pcm to eliminate that overhead. nvidia-pcm is a platform configuration manager for NVBMC, NVIDIA's OpenBMC-based firmware distribution, that enables a single firmware image to serve multiple platform variants. At boot, nvidia-pcm queries hardware identity data over D-Bus and exports the correct platform-specific configuration as environment variables. Downstream services read those variables without knowing or caring which hardware variant they are running on. The result is that platform differences are captured entirely in declarative JSON files, not in separate build artifacts. This paper describes the architecture, implementation, and deployment impact of nvidia-pcm, and shares lessons learned from solving the platform-identity problem at a deliberately minimal level of abstraction--prioritizing adoption simplicity over comprehensive hardware modeling.

Safety-critical task planning in robotic systems remains challenging: classical planners suffer from poor scalability, Reinforcement Learning (RL)-based methods generalize poorly, and base Large Language Models (LLMs) cannot guarantee safety. To address this gap, we propose safety-generalizable large language models, named SafeGen-LLM. SafeGen-LLM can not only enhance the safety satisfaction of task plans but also generalize well to novel safety properties in various domains. We first construct a multi-domain Planning Domain Definition Language 3 (PDDL3) benchmark with explicit safety constraints. Then, we introduce a two-stage post-training framework: Supervised Fine-Tuning (SFT) on a constraint-compliant planning dataset to learn planning syntax and semantics, and Group Relative Policy Optimization (GRPO) guided by fine-grained reward machines derived from formal verification to enforce safety alignment and by curriculum learning to better handle complex tasks. Extensive experiments show that SafeGen-LLM achieves strong safety generalization and outperforms frontier proprietary baselines across multi-domain planning tasks and multiple input formats (e.g., PDDLs and natural language).

We present improved learning-augmented algorithms for finding an approximate minimum spanning tree (MST) for points in an arbitrary metric space. Our work follows a recent framework called metric forest completion (MFC), where the learned input is a forest that must be given additional edges to form a full spanning tree. Veldt et al. (2025) showed that optimally completing the forest takes $Ω(n^2)$ time, but designed a 2.62-approximation for MFC with subquadratic complexity. The same method is a $(2γ+ 1)$-approximation for the original MST problem, where $γ\geq 1$ is a quality parameter for the initial forest. We introduce a generalized method that interpolates between this prior algorithm and an optimal $Ω(n^2)$-time MFC algorithm. Our approach considers only edges incident to a growing number of strategically chosen ``representative'' points. One corollary of our analysis is to improve the approximation factor of the previous algorithm from 2.62 for MFC and $(2γ+1)$ for metric MST to 2 and $2γ$ respectively. We prove this is tight for worst-case instances, but we still obtain better instance-specific approximations using our generalized method. We complement our theoretical results with a thorough experimental evaluation.

In this paper, we provide the first investigation into adaptive combinatorial experimental design, focusing on the trade-off between regret minimization and statistical power in combinatorial multi-armed bandits (CMAB). While minimizing regret requires repeated exploitation of high-reward arms, accurate inference on reward gaps requires sufficient exploration of suboptimal actions. We formalize this trade-off through the concept of Pareto optimality and establish equivalent conditions for Pareto-efficient learning in CMAB. We consider two relevant cases under different information structures, i.e., full-bandit feedback and semi-bandit feedback, and propose two algorithms MixCombKL and MixCombUCB respectively for these two cases. We provide theoretical guarantees showing that both algorithms are Pareto optimal, achieving finite-time guarantees on both regret and estimation error of arm gaps. Our results further reveal that richer feedback significantly tightens the attainable Pareto frontier, with the primary gains arising from improved estimation accuracy under our proposed methods. Taken together, these findings establish a principled framework for adaptive combinatorial experimentation in multi-objective decision-making.

Calibration$\unicode{x2014}$the problem of ensuring that predicted probabilities align with observed class frequencies$\unicode{x2014}$is a basic desideratum for reliable prediction with machine learning systems. Calibration error is traditionally assessed via a divergence function, using the expected divergence between predictions and empirical frequencies. Accurately estimating this quantity is challenging, especially in the multiclass setting. Here, we show how to extend a recent variational framework for estimating calibration errors beyond divergences induced induced by proper losses, to cover a broad class of calibration errors induced by $L_p$ divergences. Our method can separate over- and under-confidence and, unlike non-variational approaches, avoids overestimation. We provide extensive experiments and integrate our code in the open-source package probmetrics (https://github.com/dholzmueller/probmetrics) for evaluating calibration errors.

Reconstructing fluid flows from sparse sensor measurements is a fundamental challenge in science and engineering. Widely separated measurements and complex, multiscale dynamics make accurate recovery of fine-scale structures difficult. In addition, existing methods face a persistent tradeoff: high-accuracy models are often computationally expensive, whereas faster approaches typically compromise fidelity. In this work, we introduce BLISSNet, a model that strikes a strong balance between reconstruction accuracy and computational efficiency for both flow reconstruction and nudging-based data assimilation. The model follows a DeepONet-like architecture, enabling zero-shot inference on domains of arbitrary size. After the first model call on a given domain, certain network components can be precomputed, leading to low inference cost for subsequent evaluations on large domains. Consequently, the model can achieve faster inference than classical interpolation methods such as radial basis function or bicubic interpolation. This combination of high accuracy, low cost, and zero-shot generalization makes BLISSNet well-suited for large-scale real-time flow reconstruction and data assimilation tasks.

Modern microscopy routinely produces gigapixel images that contain structures across multiple spatial scales, from fine cellular morphology to broader tissue organization. Many analysis tasks require combining these scales, yet most vision models operate at a single resolution or derive multi-scale features from one view, limiting their ability to exploit the inherently multi-resolution nature of microscopy data. We introduce MuViT, a transformer architecture built to fuse true multi-resolution observations from the same underlying image. MuViT embeds all patches into a shared world-coordinate system and extends rotary positional embeddings to these coordinates, enabling attention to integrate wide-field context with high-resolution detail within a single encoder. Across synthetic benchmarks, kidney histopathology, and high-resolution mouse-brain microscopy, MuViT delivers consistent improvements over strong ViT and CNN baselines. Multi-resolution MAE pretraining further produces scale-consistent representations that enhance downstream tasks. These results demonstrate that explicit world-coordinate modelling provides a simple yet powerful mechanism for leveraging multi-resolution information in large-scale microscopy analysis.

Quantum machine learning applies principles such as superposition and entanglement to data processing and optimization. Variational quantum models operate on qubits in high-dimensional Hilbert spaces and provide an alternative approach to model expressivity. We compare classical models and a variational quantum classifier on the XOR problem. Logistic regression, a one-hidden-layer multilayer perceptron, and a two-qubit variational quantum classifier with circuit depths 1 and 2 are evaluated on synthetic XOR datasets with varying Gaussian noise and sample sizes using accuracy and binary cross-entropy. Performance is determined primarily by model expressivity. Logistic regression and the depth-1 quantum circuit fail to represent XOR reliably, whereas the multilayer perceptron and the depth-2 quantum circuit achieve perfect test accuracy under representative conditions. Robustness analyses across noise levels, dataset sizes, and random seeds confirm that circuit depth is decisive for quantum performance on this task. Despite matching accuracy, the multilayer perceptron achieves lower binary cross-entropy and substantially shorter training time. Hardware execution preserves the global XOR structure but introduces structured deviations in the decision function. Overall, deeper variational quantum classifiers can match classical neural networks in accuracy on low-dimensional XOR benchmarks, but no clear empirical advantage in robustness or efficiency is observed in the examined settings.

AI agents powered by reasoning models require access to sensitive user data. However, their reasoning traces are difficult to control, which can result in the unintended leakage of private information to external parties. We propose training models to follow instructions not only in the final answer, but also in reasoning traces, potentially under different constraints. We hypothesize that improving their instruction following abilities in the reasoning traces can improve their privacy-preservation skills. To demonstrate this, we fine-tune models on a new instruction-following dataset with explicit restrictions on reasoning traces. We further introduce a generation strategy that decouples reasoning and answer generation using separate LoRA adapters. We evaluate our approach on six models from two model families, ranging from 1.7B to 14B parameters, across two instruction-following benchmarks and two privacy benchmarks. Our method yields substantial improvements, achieving gains of up to 20.9 points in instruction-following performance and up to 51.9 percentage points on privacy benchmarks. These improvements, however, can come at the cost of task utility, due to the trade-off between reasoning performance and instruction-following abilities. Overall, our results show that improving instruction-following behavior in reasoning models can significantly enhance privacy, suggesting a promising direction for the development of future privacy-aware agents. Our code and data are available at https://github.com/UKPLab/arxiv2026-controllable-reasoning-models

Federated learning (FL) is an effective paradigm for distributed environments such as the Internet of Things (IoT), where data from diverse devices with varying functionalities remains localized while contributing to a shared global model. By eliminating the need to transmit raw data, FL inherently preserves privacy. However, the heterogeneous nature of IoT data, stemming from differences in device capabilities, data formats, and communication constraints, poses significant challenges to maintaining both global model performance and privacy. In the context of IoT-based anomaly detection, unsupervised FL offers a promising means to identify abnormal behavior without centralized data aggregation. Nevertheless, feature heterogeneity across devices complicates model training and optimization, hindering effective implementation. In this study we propose an efficient unsupervised FL framework that enhances anomaly detection by leveraging shared features from two distinct IoT datasets: one focused on anomaly detection and the other on device identification, while preserving dataset-specific features. To improve transparency and interpretability, we employ explainable AI techniques, such as SHAP, to identify key features influencing local model decisions. Experiments conducted on real-world IoT datasets demonstrate that the proposed method significantly outperforms conventional FL approaches in anomaly detection accuracy. This work underscores the potential of using shared features from complementary datasets to optimize unsupervised federated learning and achieve superior anomaly detection results in decentralized IoT environments.

Diffusion models achieve state-of-the-art video generation quality, but their inference remains expensive due to the large number of sequential denoising steps. This has motivated a growing line of research on accelerating diffusion inference. Among training-free acceleration methods, caching reduces computation by reusing previously computed model outputs across timesteps. Existing caching methods rely on heuristic criteria to choose cache/reuse timesteps and require extensive tuning. We address this limitation with a principled sensitivity-aware caching framework. Specifically, we formalize the caching error through an analysis of the model output sensitivity to perturbations in the denoising inputs, i.e., the noisy latent and the timestep, and show that this sensitivity is a key predictor of caching error. Based on this analysis, we propose Sensitivity-Aware Caching (SenCache), a dynamic caching policy that adaptively selects caching timesteps on a per-sample basis. Our framework provides a theoretical basis for adaptive caching, explains why prior empirical heuristics can be partially effective, and extends them to a dynamic, sample-specific approach. Experiments on Wan 2.1, CogVideoX, and LTX-Video show that SenCache achieves better visual quality than existing caching methods under similar computational budgets.

The use of algorithmic predictions in decision-making leads to a feedback loop where the models we deploy actively influence the data distributions we see, and later use to retrain on. This dynamic was formalized by Perdomo et al. 2020 in their work on performative prediction. Our main result is an unconditional reduction showing that any no-regret algorithm deployed in performative settings converges to a (mixed) performatively stable equilibrium: a solution in which models actively shape data distributions in ways that their own predictions look optimal in hindsight. Prior to our work, all positive results in this area made strong restrictions on how models influenced distributions. By using a martingale argument and allowing randomization, we avoid any such assumption and sidestep recent hardness results for finding stable models. Lastly, on a more conceptual note, our connection sheds light on why common algorithms, like gradient descent, are naturally stabilizing and prevent runaway feedback loops. We hope our work enables future technical transfer of ideas between online optimization and performativity.

Estimating density ratios between pairs of intractable data distributions is a core problem in probabilistic modeling, enabling principled comparisons of sample likelihoods under different data-generating processes across conditions and covariates. While exact-likelihood models such as normalizing flows offer a promising approach to density ratio estimation, naive flow-based evaluations are computationally expensive, as they require simulating costly likelihood integrals for each distribution separately. In this work, we leverage condition-aware flow matching to derive a single dynamical formulation for tracking density ratios along generative trajectories. We demonstrate competitive performance on simulated benchmarks for closed-form ratio estimation, and show that our method supports versatile tasks in single-cell genomics data analysis, where likelihood-based comparisons of cellular states across experimental conditions enable treatment effect estimation and batch correction evaluation.

Despite their capabilities, Multimodal Large Language Models (MLLMs) may produce plausible but erroneous outputs, hindering reliable deployment. Accurate uncertainty metrics could enable escalation of unreliable queries to human experts or larger models for improved performance. However, existing uncertainty metrics have practical constraints, such as being designed only for specific modalities, reliant on external tools, or computationally expensive. We introduce UMPIRE, a training-free uncertainty quantification framework for MLLMs that works efficiently across various input and output modalities without external tools, relying only on the models' own internal modality features. UMPIRE computes the incoherence-adjusted semantic volume of sampled MLLM responses for a given task instance, effectively capturing both the global semantic diversity of samples and the local incoherence of responses based on internal model confidence. We propose uncertainty desiderata for MLLMs and provide theoretical analysis motivating UMPIRE's design. Extensive experiments show that UMPIRE consistently outperforms baseline metrics in error detection and uncertainty calibration across image, audio, and video-text benchmarks, including adversarial and out-of-distribution settings. We also demonstrate UMPIRE's generalization to non-text output tasks, including image and audio generation.

We study the problem of resilient strategies in the presence of uncertainty. Resilient strategies enable an agent to make decisions that are robust against disturbances. In particular, we are interested in those disturbances that are able to flip a decision made by the agent. Such a disturbance may, for instance, occur when the intended action of the agent cannot be executed due to a malfunction of an actuator in the environment. In this work, we introduce the concept of resilience in the stochastic setting and present a comprehensive set of fundamental problems. Specifically, we discuss such problems for Markov decision processes with reachability and safety objectives, which also smoothly extend to stochastic games. To account for the stochastic setting, we provide various ways of aggregating the amounts of disturbances that may have occurred, for instance, in expectation or in the worst case. Moreover, to reason about infinite disturbances, we use quantitative measures, like their frequency of occurrence.

We present a scalable methodology for evaluating language models in multi-turn interactions, using a suite of collaborative games that require effective communication about private information. This enables an interactive scaling analysis, in which a fixed token budget is divided over a variable number of turns. We find that in many cases, language models are unable to use interactive collaboration to improve over the non-interactive baseline scenario in which one agent attempts to summarize its information and the other agent immediately acts -- despite substantial headroom. This suggests that state-of-the-art models still suffer from significant weaknesses in planning and executing multi-turn collaborative conversations. We analyze the linguistic features of these dialogues, assessing the roles of sycophancy, information density, and discourse coherence. While there is no single linguistic explanation for the collaborative weaknesses of contemporary language models, we note that humans achieve comparable task success at superior token efficiency by producing dialogues that are more coherent than those produced by most language models. The proactive management of private information is a defining feature of real-world communication, and we hope that MT-PingEval will drive further work towards improving this capability.

Despite advances in machine learning-based medical image classifiers, the safety and reliability of these systems remain major concerns in practical settings. Existing auditing approaches mainly rely on unimodal features or metadata-based subgroup analyses, which are limited in interpretability and often fail to capture hidden systematic failures. To address these limitations, we introduce the first automated auditing framework that extends slice discovery methods to multimodal representations specifically for medical applications. Comprehensive experiments were conducted under common failure scenarios using the MIMIC-CXR-JPG dataset, demonstrating the framework's strong capability in both failure discovery and explanation generation. Our results also show that multimodal information generally allows more comprehensive and effective auditing of classifiers, while unimodal variants beyond image-only inputs exhibit strong potential in scenarios where resources are constrained.

Optimizing the consolidation process in container-based fulfillment centers requires trading off competing objectives such as processing speed, resource usage, and space utilization while adhering to a range of real-world operational constraints. This process involves moving items between containers via a combination of human and robotic workstations to free up space for inbound inventory and increase container utilization. We formulate this problem as a large-scale Multi-Objective Reinforcement Learning (MORL) task with high-dimensional state spaces and dynamic system behavior. Our method builds on recent theoretical advances in solving constrained RL problems via best-response and no-regret dynamics in zero-sum games, enabling principled minimax policy learning. Policy evaluation on realistic warehouse simulations shows that our approach effectively trades off objectives, and we empirically observe that it learns a single policy that simultaneously satisfies all constraints, even if this is not theoretically guaranteed. We further introduce a theoretical framework to handle the problem of error cancellation, where time-averaged solutions display oscillatory behavior. This method returns a single iterate whose Lagrangian value is close to the minimax value of the game. These results demonstrate the promise of MORL in solving complex, high-impact decision-making problems in large-scale industrial systems.

Pre-trained vision encoders like DINOv2 have demonstrated exceptional performance on unimodal tasks. However, we observe that their feature representations are poorly aligned across different modalities. For instance, the feature embedding for an RGB image and its corresponding depth map of the same scene exhibit a cosine similarity that is nearly identical to that of two random, unrelated images. To address this, we propose the Omnivorous Vision Encoder, a novel framework that learns a modality-agnostic feature space. We train the encoder with a dual objective: first, to maximize the feature alignment between different modalities of the same scene; and second, a distillation objective that anchors the learned representations to the output of a fully frozen teacher such as DINOv2. The resulting student encoder becomes "omnivorous" by producing a consistent, powerful embedding for a given scene, regardless of the input modality (RGB, Depth, Segmentation, etc.). This approach enables robust cross-modal understanding while retaining the discriminative semantics of the original foundation model.

The Flexible Job Shop Scheduling Problem (FJSP) originates from real production lines, while some practical constraints are often ignored or idealized in current FJSP studies, among which the limited buffer problem has a particular impact on production efficiency. To this end, we study an extended problem that is closer to practical scenarios--the Flexible Job Shop Scheduling Problem with Limited Buffers and Material Kitting. In recent years, deep reinforcement learning (DRL) has demonstrated considerable potential in scheduling tasks. However, its capacity for state modeling remains limited when handling complex dependencies and long-term constraints. To address this, we leverage a heterogeneous graph network within the DRL framework to model the global state. By constructing efficient message passing among machines, operations, and buffers, the network focuses on avoiding decisions that may cause frequent pallet changes during long-sequence scheduling, thereby helping improve buffer utilization and overall decision quality. Experimental results on both synthetic and real production line datasets show that the proposed method outperforms traditional heuristics and advanced DRL methods in terms of makespan and pallet changes, and also achieves a good balance between solution quality and computational cost. Furthermore, a supplementary video is provided to showcase a simulation system that effectively visualizes the progression of the production line.

Recent work has shown the surprising power of low-degree sandwiching polynomial approximators in the context of challenging learning settings such as learning with distribution shift, testable learning, and learning with contamination. A pair of sandwiching polynomials approximate a target function in expectation while also providing pointwise upper and lower bounds on the function's values. In this paper, we give a new method for constructing low-degree sandwiching polynomials that yield greatly improved degree bounds for several fundamental function classes and marginal distributions. In particular, we obtain degree $\mathrm{poly}(k)$ sandwiching polynomials for functions of $k$ halfspaces under the Gaussian distribution, improving exponentially over the prior $2^{O(k)}$ bound. More broadly, our approach applies to function classes that are low-dimensional and have smooth boundary. In contrast to prior work, our proof is relatively simple and directly uses the smoothness of the target function's boundary to construct sandwiching Lipschitz functions, which are amenable to results from high-dimensional approximation theory. For low-dimensional polynomial threshold functions (PTFs) with respect to Gaussians, we obtain doubly exponential improvements without applying the FT-mollification method of Kane used in the best previous result.

Small language models (SLMs) have emerged as efficient alternatives to large language models for task-specific applications. However, they are often employed in high-volume, low-latency settings, where efficiency is crucial. We propose TASC, Task-Adaptive Sequence Compression, a framework for SLM acceleration comprising two use-cases: When performing SLM fine-tuning, we propose TASC-ft, which iteratively enriches the tokenizer vocabulary with high-frequency output n-grams and then fine-tunes the model to utilize the expanded vocabulary. Next, we propose an inference-time method, termed TASC-spec. TASC-spec is a lightweight, training-free speculative decoding method that constructs an n-gram draft model from the task's output corpus, mixing task and context n-gram information.TASC-spec avoids any additional training, while bypassing draft-target vocabulary alignment constraints. We demonstrate the effectiveness of both methods across multiple low output-variability generation tasks. Our methods show consistent improvements in inference efficiency while maintaining task performance.

We present a new approach for benchmarking Large Language Model (LLM) capabilities on research-level mathematics. Existing benchmarks largely rely on static, hand-curated sets of contest or textbook-style problems as proxies for mathematical research. Instead, we establish an updatable benchmark evaluating models directly on the latest research results in mathematics. This consists of an automatic pipeline that extracts lemmas from arXiv and rewrites them into self-contained statements by making all assumptions and required definitions explicit. It results in a benchmark that can be updated regularly with new problems taken directly from human mathematical research, while previous instances can be used for training without compromising future evaluations. We benchmark current state-of-the-art LLMs, which obtain around 10-15$\%$ accuracy in theorem proving (pass@1) depending on the model, showing that there is currently a large margin of progression for LLMs to reach human-level proving capabilities in a research context.

Argumentative LLMs (ArgLLMs) are an existing approach leveraging Large Language Models (LLMs) and computational argumentation for decision-making, with the aim of making the resulting decisions faithfully explainable to and contestable by humans. Here we propose a web-based system implementing ArgLLM-empowered agents for binary tasks. ArgLLM-App supports visualisation of the produced explanations and interaction with human users, allowing them to identify and contest any mistakes in the system's reasoning. It is highly modular and enables drawing information from trusted external sources. ArgLLM-App is publicly available at https://argllm.app, with a video demonstration at https://youtu.be/vzwlGOr0sPM.

Recently, RISC-V has contributed to the development of IoT devices, requiring architectures that balance energy efficiency, compact area, and integrated security. However, most recent RISC-V cores for IoT prioritize either area footprint or energy efficiency, while adding cryptographic support further compromises compactness. As a result, truly integrated architectures that simultaneously optimize efficiency and security remain largely unexplored, leaving constrained IoT environments vulnerable to performance and security trade-offs. In this paper, we introduce SAILOR, an energy-efficient and scalable ultra-lightweight RISC-V core family for cryptographic applications in IoT. Our design is modular and spans 1-, 2-, 4-, 8-, 16-, and 32-bit serialized execution data-paths, prioritizing minimal area. This modular design and adaptable data-path minimizes the overhead of integrating RISC-V cryptography extensions, achieving low hardware cost while significantly improving energy efficiency. We validate our design approach through a comprehensive analysis of area, energy, and efficiency trade-offs. The results surpass state-of-the-art solutions in both performance and energy efficiency by up to 13x and reduce area by up to 59 %, demonstrating that lightweight cryptographic features can be added without prohibitive overhead, and that energy- or area-efficient designs need not compromise performance.

Switching uniformity, as a major challenge, hinders the practical implementation of \ac{RRAM} in memory application. Operating \ac{RRAM} in current mode, is proposed as an efficient method to improve programming schemes accuracy within the finite readout window. In this article, we demonstrate a current generator circuit to perform current programming on \ac{RRAM}. Current mirror topology is used in our circuit to convert an external pulse voltage into a pulse current fed to \ac{RRAM} directly with an amplitude equivalent with the DC reference current. The targeting ranges of \ac{RRAM}'s programming current are up to 400\,\textmu A and, in that case, our proposed circuit achieved minimum current mismatch of 1\%.

Vision transformers have recently made a breakthrough in computer vision showing excellent performance in terms of precision for numerous applications. However, their computational cost is very high compared to alternative approaches such as Convolutional Neural Networks. To address this problem, we propose a novel framework for image classification called RAViT based on a multi-branch network that operates on several copies of the same image with different resolutions to reduce the computational cost while preserving the overall accuracy. Furthermore, our framework includes an early exit mechanism that makes our model adaptive and allows to choose the appropriate trade-off between accuracy and computational cost at run-time. For example in a two-branch architecture, the original image is first resized to reduce its resolution, then a prediction is performed on it using a first transformer and the resulting prediction is reused together with the original-size image to perform a final prediction on a second transformer with less computation than a classical Vision transformer architecture. The early-exit process allows the model to make a final prediction at intermediate branches, saving even more computation. We evaluated our approach on CIFAR-10, Tiny ImageNet, and ImageNet. We obtained an equivalent accuracy to the classical Vision transformer model with only around 70% of FLOPs.

Scaling the number of qubits available across multiple quantum devices is an active area of research within distributed quantum computing (DQC). This includes quantum circuit compilation and execution management on multiple quantum devices in the network. The latter aspect is very challenging because, while reducing the makespan of job batches remains a relevant objective, novel quantum-specific constraints must be considered, including QPU utilization, non-local gate density, and the latency associated with queued DQC jobs. In this work, a range of scheduling strategies is proposed, simulated, and evaluated, including heuristics that prioritize resource maximization for QPU utilization, node selection based on heterogeneous network connectivity, asynchronous node release upon job completion, and a scheduling strategy based on reinforcement learning with proximal policy optimization. These approaches are benchmarked against traditional FIFO and LIST schedulers under varying DQC job types and network conditions for the allocation of DQC jobs to devices within a network.

At present, there are no easily understood explainable artificial intelligence (AI) methods for discrete token inputs, like text. Most explainable AI techniques do not extend well to token sequences, where both local and global features matter, because state-of-the-art models, like transformers, tend to focus on global connections. Therefore, existing explainable AI algorithms fail by (i) identifying disparate tokens of importance, or (ii) assigning a large number of tokens a low value of importance. This method for explainable AI for tokens-based classifiers generalizes a mask-based explainable AI algorithm for images. It starts with an Explainer neural network that is trained to create masks to hide information not relevant for classification. Then, the Hadamard product of the mask and the continuous values of the classifier's embedding layer is taken and passed through the classifier, changing the magnitude of the embedding vector but keeping the orientation unchanged. The Explainer is trained for a taxonomic classifier for nucleotide sequences and it is shown that the masked segments are less relevant to classification than the unmasked ones. This method focused on the importance the token as a whole (i.e., a segment of the input sequence), producing a human-readable explanation.

In many applications, evaluating the effectiveness of different alternatives comes with varying costs or resource usage. Motivated by such heterogeneity, we study the Best Arm Identification with Resource Constraints (BAIwRC) problem, where an agent seeks to identify the best alternative (aka arm) in the presence of resource constraints. Each arm pull consumes one or more types of limited resources. We make two key contributions. First, we propose the Successive Halving with Resource Rationing (SH-RR) algorithm, which integrates resource-aware allocation into the classical successive halving framework on best arm identification. The SH-RR algorithm unifies the theoretical analysis for both the stochastic and deterministic consumption settings, with a new \textit{effective consumption measure

Mobile Agents can autonomously execute user instructions, which requires hybrid-capabilities reasoning, including screen summary, subtask planning, action decision and action function. However, existing agents struggle to achieve both decoupled enhancement and balanced integration of these capabilities. To address these challenges, we propose Channel-of-Mobile-Experts (CoME), a novel agent architecture consisting of four distinct experts, each aligned with a specific reasoning stage, CoME activates the corresponding expert to generate output tokens in each reasoning stage via output-oriented activation. To empower CoME with hybrid-capabilities reasoning, we introduce a progressive training strategy: Expert-FT enables decoupling and enhancement of different experts' capability; Router-FT aligns expert activation with the different reasoning stage; CoT-FT facilitates seamless collaboration and balanced optimization across multiple capabilities. To mitigate error propagation in hybrid-capabilities reasoning, we propose InfoGain-Driven DPO (Info-DPO), which uses information gain to evaluate the contribution of each intermediate step, thereby guiding CoME toward more informative reasoning. Comprehensive experiments show that CoME outperforms dense mobile agents and MoE methods on both AITZ and AMEX datasets.

Recognizing surgical phases and steps from video is a fundamental problem in computer-assisted interventions. Recent approaches increasingly rely on large-scale pre-training on thousands of labeled surgical videos, followed by zero-shot transfer to specific procedures. While effective, this strategy incurs substantial computational and data collection costs. In this work, we question whether such heavy pre-training is truly necessary. We propose Text-Augmented Action Segmentation Optimal Transport (TASOT), an unsupervised method for surgical phase and step recognition that extends Action Segmentation Optimal Transport (ASOT) by incorporating textual information generated directly from the videos. TASOT formulates temporal action segmentation as a multimodal optimal transport problem, where the matching cost is defined as a weighted combination of visual and text-based costs. The visual term captures frame-level appearance similarity, while the text term provides complementary semantic cues, and both are jointly regularized through a temporally consistent unbalanced Gromov-Wasserstein formulation. This design enables effective alignment between video frames and surgical actions without surgical-specific pretraining or external web-scale supervision. We evaluate TASOT on multiple benchmark surgical datasets and observe consistent and substantial improvements over existing zero-shot methods, including StrasBypass70 (+23.7), BernBypass70 (+4.5), Cholec80 (+16.5), and AutoLaparo (+19.6). These results demonstrate that fine-grained surgical understanding can be achieved by exploiting information already present in standard visual and textual representations, without resorting to increasingly complex pre-training pipelines. The code will be available at https://github.com/omar8ahmed9/TASOT.

The expansion of retrieval-augmented generation (RAG) into multimodal domains has intensified the challenge for processing complex visual documents, such as financial reports. While page-level chunking and retrieval is a natural starting point, it creates a critical bottleneck: delivering entire pages to the generator introduces excessive extraneous context. This not only overloads the generator's attention mechanism but also dilutes the most salient evidence. Moreover, compressing these information-rich pages into a limited visual token budget further increases the risk of hallucinations. To address this, we introduce AgenticOCR, a dynamic parsing paradigm that transforms optical character recognition (OCR) from a static, full-text process into a query-driven, on-demand extraction system. By autonomously analyzing document layout in a "thinking with images" manner, AgenticOCR identifies and selectively recognizes regions of interest. This approach performs on-demand decompression of visual tokens precisely where needed, effectively decoupling retrieval granularity from rigid page-level chunking. AgenticOCR has the potential to serve as the "third building block" of the visual document RAG stack, operating alongside and enhancing standard Embedding and Reranking modules. Experimental results demonstrate that AgenticOCR improves both the efficiency and accuracy of visual RAG systems, achieving expert-level performance in long document understanding. Code and models are available at https://github.com/OpenDataLab/AgenticOCR.

Large-scale homogeneous detectors with optical readouts are widely used in particle detection, with Cherenkov and scintillator neutrino detectors as prominent examples. Analyses in experimental physics rely on high-fidelity simulators to translate sensor-level information into physical quantities of interest. This task critically depends on accurate calibration, which aligns simulation behavior with real detector data, and on tracking, which infers particle properties from optical signals. We present the first end-to-end differentiable optical particle detector simulator, enabling simultaneous calibration and reconstruction through gradient-based optimization. Our approach unifies simulation, calibration, and tracking, which are traditionally treated as separate problems, within a single differentiable framework. We demonstrate that it achieves smooth and physically meaningful gradients across all key stages of light generation, propagation, and detection while maintaining computational efficiency. We show that gradient-based calibration and reconstruction greatly simplify existing analysis pipelines while matching or surpassing the performance of conventional non-differentiable methods in both accuracy and speed. Moreover, the framework's modularity allows straightforward adaptation to diverse detector geometries and target materials, providing a flexible foundation for experiment design and optimization. The results demonstrate the readiness of this technique for adoption in current and future optical detector experiments, establishing a new paradigm for simulation and reconstruction in particle physics.

This study presents the first systematic, reference-free human evaluation of large language model (LLM) machine translation (MT) for Ancient Greek (AG) technical prose. We evaluate translations by three commercial LLMs (Claude, Gemini, ChatGPT) of twenty paragraph-length passages from two works by the Greek physician Galen of Pergamum (ca. 129-216 CE): On Mixtures, which has two published English translations, and On the Composition of Drugs according to Kinds, which has never been fully translated into English. We assess translation quality using both standard automated evaluation metrics (BLEU, chrF++, METEOR, ROUGE-L, BERTScore, COMET, BLEURT) and expert human evaluation via a modified Multidimensional Quality Metrics (MQM) framework applied to all 60 translations by a team of domain specialists. On the previously translated expository text, LLMs achieved high translation quality (mean MQM score 95.2/100), with performance approaching expert level. On the untranslated pharmacological text, aggregate quality was lower (79.9/100) but with high variance driven by two passages presenting extreme terminological density; excluding these, scores converged to within 4 points of the translated text. Terminology rarity, operationalized via corpus frequency in the literary Diorisis Ancient Greek Corpus, emerged as a strong predictor of translation failure (r = -.97 for passage-level quality on the untranslated text). Automated metrics showed moderate correlation with human judgment overall on the text with a wide quality spread (Composition), but no metric discriminated among high-quality translations. We discuss implications for the use of LLMs in Classical scholarship and for the design of automated evaluation pipelines for low-resource ancient languages.

Open RAN (O-RAN) exposes rich control and telemetry interfaces across the Non-RT RIC, Near-RT RIC, and distributed units, but also makes it harder to operate multi-tenant, multi-objective RANs in a safe and auditable manner. In parallel, agentic AI systems with explicit planning, tool use, memory, and self-management offer a natural way to structure long-lived control loops. This article surveys how such agentic controllers can be brought into O-RAN: we review the O-RAN architecture, contrast agentic controllers with conventional ML/RL xApps, and organise the task landscape around three clusters: network slice life-cycle, radio resource management (RRM) closed loops, and cross-cutting security, privacy, and compliance. We then introduce a small set of agentic primitives (Plan-Act-Observe-Reflect, skills as tool use, memory and evidence, and self-management gates) and show, in a multi-cell O-RAN simulation, how they improve slice life-cycle and RRM performance compared to conventional baselines and ablations that remove individual primitives. Security, privacy, and compliance are discussed as architectural constraints and open challenges for standards-aligned deployments. This framework achieves an average 8.83\% reduction in resource usage across three classic network slices.

Vision-language models (VLMs) show promise in drafting radiology reports, yet they frequently suffer from logical inconsistencies, generating diagnostic impressions unsupported by their own perceptual findings or missing logically entailed conclusions. Standard lexical metrics heavily penalize clinical paraphrasing and fail to capture these deductive failures in reference-free settings. Toward guarantees for clinical reasoning, we introduce a neurosymbolic verification framework that deterministically audits the internal consistency of VLM-generated reports. Our pipeline autoformalizes free-text radiographic findings into structured propositional evidence, utilizing an SMT solver (Z3) and a clinical knowledge base to verify whether each diagnostic claim is mathematically entailed, hallucinated, or omitted. Evaluating seven VLMs across five chest X-ray benchmarks, our verifier exposes distinct reasoning failure modes, such as conservative observation and stochastic hallucination, that remain invisible to traditional metrics. On labeled datasets, enforcing solver-backed entailment acts as a rigorous post-hoc guarantee, systematically eliminating unsupported hallucinations to significantly increase diagnostic soundness and precision in generative clinical assistants.

Reinforcement Learning from Verifiable Rewards (RLVR) has emerged as a powerful paradigm for enhancing the complex reasoning capabilities of Large Reasoning Models. However, standard outcome-based supervision suffers from a critical limitation that penalizes trajectories that are largely correct but fail due to several missteps as heavily as completely erroneous ones. This coarse feedback signal causes the model to discard valuable largely correct rollouts, leading to a degradation in rollout diversity that prematurely narrows the exploration space. Process Reward Models have demonstrated efficacy in providing reliable step-wise verification for test-time scaling, naively integrating these signals into RLVR as dense rewards proves ineffective.Prior methods attempt to introduce off-policy guided whole-trajectory replacement that often outside the policy model's distribution, but still fail to utilize the largely correct rollouts generated by the model itself and thus do not effectively mitigate the narrowing of the exploration space. To address these issues, we propose SCOPE (Step-wise Correction for On-Policy Exploration), a novel framework that utilizes Process Reward Models to pinpoint the first erroneous step in suboptimal rollouts and applies fine-grained, step-wise off-policy rectification. By applying precise refinement on partially correct rollout, our method effectively salvages partially correct trajectories and increases diversity score by 13.5%, thereby sustaining a broad exploration space. Extensive experiments demonstrate that our approach establishes new state-of-the-art results, achieving an average accuracy of 46.6% on math reasoning and exhibiting robust generalization with 53.4% accuracy on out-of-distribution reasoning tasks.

Can narratives make arguments more persuasive? And to this end, which narrative features matter most? Although stories are often seen as powerful tools for persuasion, their specific role in online, unstructured argumentation remains underexplored. To address this gap, we present ARGUS, a framework for studying the impact of narration on persuasion in argumentative discourse. ARGUS introduces a new ChangeMyView corpus annotated for story presence and six key narrative features, integrating insights from two established theoretical frameworks that capture both textual narrative features and their effects on recipients. Leveraging both encoder-based classifiers and zero-shot large language models (LLMs), ARGUS identifies stories and narrative features and applies them at scale to examine how different narrative dimensions influence persuasion success in online argumentation.

Functional testing is essential for verifying that the business logic of mobile applications aligns with user requirements, serving as the primary methodology for quality assurance in software development. Despite its importance, functional testing remains heavily dependent on manual effort due to two core challenges. First, acquiring and reusing complex business logic from unstructured requirements remains difficult, which hinders the understanding of specific functionalities. Second, a significant semantic gap exists when adapting business logic to the diverse GUI environments, which hinders the generation of test cases for specific mobile applications. To address the preceding challenges, we propose LogiDroid, a two-stage approach that generates individual functional test cases by extracting business logic and adapting it to target applications. First, in the Knowledge Retrieval and Fusion stage, we construct a dataset to retrieve relevant cases and extract business logic for the target functionality. Second, in the Context-Aware Test Generation stage, LogiDroid jointly analyzes the extracted business logic and the real-time GUI environment to generate functional test cases. This design allows LogiDroid to accurately understand application semantics and use domain expertise to generate complete test cases with verification assertions. We assess the effectiveness of LogiDroid using two widely-used datasets that cover 28 real-world applications and 190 functional requirements. Experimental results show that LogiDroid successfully tested 40% of functional requirements on the FrUITeR dataset (an improvement of over 48% compared to the state-of-the-art approaches) and 65% on the Lin dataset (an improvement of over 55% compared to the state-of-the-art approaches). These results demonstrate the significant effectiveness of LogiDroid in functional test generation.

This paper presents the Artificial Agency Program (AAP), a position and research agenda for building AI systems as reality embedded, resource-bounded agents whose development is driven by curiosity-as-learning-progress under physical and computational constraints. The central thesis is that AI is most useful when treated as part of an extended human--tool system that increases sensing, understanding, and actuation capability while reducing friction at the interface between people, tools, and environments. The agenda unifies predictive compression, intrinsic motivation, empowerment and control, interface quality (unification), and language/self-communication as selective information bottlenecks. We formulate these ideas as a falsifiable program with explicit costs, staged experiments, and a concrete multimodal tokenized testbed in which an agent allocates limited budget among observation, action, and deliberation. The aim is to provide a conceptual and experimental framework that connects intrinsic motivation, information theory, thermodynamics, bounded rationality, and modern reasoning systems

Winter road maintenance is critical for ensuring public safety and reducing environmental impacts, yet existing methods struggle to manage large-scale routing problems effectively and mostly reply on human decision. This study presents a novel, scalable bi-level optimization framework, validated on real operational data on UK strategic road networks (M25, M6, A1), including interconnected local road networks in surrounding areas for vehicle traversing, as part of the highway operator's efforts to solve existing planning challenges. At the upper level, a reinforcement learning (RL) agent strategically partitions the road network into manageable clusters and optimally allocates resources from multiple depots. At the lower level, a multi-objective vehicle routing problem (VRP) is solved within each cluster, minimizing the maximum vehicle travel time and total carbon emissions. Unlike existing approaches, our method handles large-scale, real-world networks efficiently, explicitly incorporating vehicle-specific constraints, depot capacities, and road segment requirements. Results demonstrate significant improvements, including balanced workloads, reduced maximum travel times below the targeted two-hour threshold, lower emissions, and substantial cost savings. This study illustrates how advanced AI-driven bi-level optimization can directly enhance operational decision-making in real-world transportation and logistics.

Simulation is essential to the development and evaluation of autonomous robots such as self-driving vehicles. Neural reconstruction is emerging as a promising solution as it enables simulating a wide variety of scenarios from real-world data alone in an automated and scalable way. However, while methods such as NeRF and 3D Gaussian Splatting can produce visually compelling results, they often exhibit artifacts particularly when rendering novel views, and fail to realistically integrate inserted dynamic objects, especially when they were captured from different scenes. To overcome these limitations, we introduce DiffusionHarmonizer, an online generative enhancement framework that transforms renderings from such imperfect scenes into temporally consistent outputs while improving their realism. At its core is a single-step temporally-conditioned enhancer that is converted from a pretrained multi-step image diffusion model, capable of running in online simulators on a single GPU. The key to training it effectively is a custom data curation pipeline that constructs synthetic-real pairs emphasizing appearance harmonization, artifact correction, and lighting realism. The result is a scalable system that significantly elevates simulation fidelity in both research and production environments.

Machine learning models -- including large language models (LLMs) -- are often said to exhibit monoculture, where outputs agree strikingly often. But what does it actually mean for models to agree too much? We argue that this question is inherently subjective, relying on two key decisions. First, the analyst must specify a baseline null model for what "independence" should look like. This choice is inherently subjective, and as we show, different null models result in dramatically different inferences about excess agreement. Second, we show that inferences depend on the population of models and items under consideration. Models that seem highly correlated in one context may appear independent when evaluated on a different set of questions, or against a different set of peers. Experiments on two large-scale benchmarks validate our theoretical findings. For example, we find drastically different inferences when using a null model with item difficulty compared to previous works that do not. Together, our results reframe monoculture evaluation not as an absolute property of model behavior, but as a context-dependent inference problem.

Cox processes model overdispersed point process data via a latent stochastic intensity, but both nonparametric estimation of the intensity model and posterior inference over intensity paths are typically intractable, relying on expensive MCMC methods. We introduce Neural Diffusion Intensity Models, a variational framework for Cox processes driven by neural SDEs. Our key theoretical result, based on enlargement of filtrations, shows that conditioning on point process observations preserves the diffusion structure of the latent intensity with an explicit drift correction. This guarantees the variational family contains the true posterior, so that ELBO maximization coincides with maximum likelihood estimation under sufficient model capacity. We design an amortized encoder architecture that maps variable-length event sequences to posterior intensity paths by simulating the drift-corrected SDE, replacing repeated MCMC runs with a single forward pass. Experiments on synthetic and real-world data demonstrate accurate recovery of latent intensity dynamics and posterior paths, with orders-of-magnitude speedups over MCMC-based methods.

Resource-efficient training optimization techniques are becoming increasingly important as the size of large language models (LLMs) continues to grow. In particular, batch packing is commonly used in pre-training and supervised fine-tuning to achieve resource-efficient training. We propose preference packing, a method to enhance resource efficiency in training techniques that use data with different responses for the same input prompt, such as reward models or Direct Preference Optimization (DPO). Preference packing improves resource efficiency by reducing the attention operations for duplicate input prompts and decreasing KV cache memory usage. We conducted experiments on text-only datasets and image-included datasets and achieved at least 37% reduction in training time. Notably, this method can be applied alongside existing optimization techniques such as batch sorting, resulting in a 3.22x speedup.

We propose ACWI (Adaptive Correlation Weighted Intrinsic), an adaptive intrinsic reward scaling framework designed to dynamically balance intrinsic and extrinsic rewards for improved exploration in sparse reward reinforcement learning. Unlike conventional approaches that rely on manually tuned scalar coefficients, which often result in unstable or suboptimal performance across tasks, ACWI learns a state dependent scaling coefficient online. Specifically, ACWI introduces a lightweight Beta Network that predicts the intrinsic reward weight directly from the agent state through an encoder based architecture. The scaling mechanism is optimized using a correlation based objective that encourages alignment between the weighted intrinsic rewards and discounted future extrinsic returns. This formulation enables task adaptive exploration incentives while preserving computational efficiency and training stability. We evaluate ACWI on a suite of sparse reward environments in MiniGrid. Experimental results demonstrate that ACWI consistently improves sample efficiency and learning stability compared to fixed intrinsic reward baselines, achieving superior performance with minimal computational overhead.

The pursuit of human-like conversational agents has long been guided by the Turing test. For modern speech-to-speech (S2S) systems, a critical yet unanswered question is whether they can converse like humans. To tackle this, we conduct the first Turing test for S2S systems, collecting 2,968 human judgments on dialogues between 9 state-of-the-art S2S systems and 28 human participants. Our results deliver a clear finding: no existing evaluated S2S system passes the test, revealing a significant gap in human-likeness. To diagnose this failure, we develop a fine-grained taxonomy of 18 human-likeness dimensions and crowd-annotate our collected dialogues accordingly. Our analysis shows that the bottleneck is not semantic understanding but stems from paralinguistic features, emotional expressivity, and conversational persona. Furthermore, we find that off-the-shelf AI models perform unreliably as Turing test judges. In response, we propose an interpretable model that leverages the fine-grained human-likeness ratings and delivers accurate and transparent human-vs-machine discrimination, offering a powerful tool for automatic human-likeness evaluation. Our work establishes the first human-likeness evaluation for S2S systems and moves beyond binary outcomes to enable detailed diagnostic insights, paving the way for human-like improvements in conversational AI systems.

Modern supply networks are complex interconnected systems. Multi-agent models are increasingly explored to optimise their performance. Most research assumes agents will have full observability of the system by having a single policy represent the agents, which seems unrealistic as this requires companies to share their data. The alternative is to develop a Hidden-Markov Process with separate policies, making the problem challenging to solve. In this paper, we propose a multi-agent system where the factory agent can share information downstream, increasing the observability of the environment. It can choose to share no information, lie, tell the truth or combine these in a mixed strategy. The results show that data sharing can boost the performance, especially when combined with a cooperative reward shaping. In the high demand scenario there is limited ability to change the strategy and therefore no data sharing approach benefits both agents. However, lying benefits the factory enough for an overall system improvement, although only by a relatively small amount compared to the overall reward. In the low demand scenario, the most successful data sharing is telling the truth which benefits all actors significantly.

Recently, there have been significant advancements in music generation. However, existing models primarily focus on creating modern pop songs, making it challenging to produce ancient music with distinct rhythms and styles, such as ancient Chinese SongCi. In this paper, we introduce SongSong, the first music generation model capable of restoring Chinese SongCi to our knowledge. Our model first predicts the melody from the input SongCi, then separately generates the singing voice and accompaniment based on that melody, and finally combines all elements to create the final piece of music. Additionally, to address the lack of ancient music datasets, we create OpenSongSong, a comprehensive dataset of ancient Chinese SongCi music, featuring 29.9 hours of compositions by various renowned SongCi music masters. To assess SongSong's proficiency in performing SongCi, we randomly select 85 SongCi sentences that were not part of the training set for evaluation against SongSong and music generation platforms such as Suno and SkyMusic. The subjective and objective outcomes indicate that our proposed model achieves leading performance in generating high-quality SongCi music.

Leveraging non-linear dimension reduction techniques, we remove the low dimension constraint from node embedding and propose COVE, an explainable high dimensional embedding that, when reduced to low dimension with UMAP, slightly increases performance on clustering and link prediction tasks. The embedding is inspired by neural embedding methods that use co-occurrence on a random walk as an indication of similarity, and is closely related to a diffusion process. Extending on recent community detection benchmarks, we find that a COVE UMAP HDBSCAN pipeline performs similarly to the popular Louvain algorithm.

Large language model (LLM) agents, such as OpenAI's Operator and Claude's Computer Use, can automate workflows but unable to handle payment tasks. Existing agentic solutions have gained significant attention; however, even the latest approaches face challenges in implementing end-to-end agentic payment workflows. To address this gap, this research proposes the Hierarchical Multi-Agent System for Payments (HMASP), which provides an end-to-end agentic method for completing payment workflows. The proposed HMASP leverages either open-weight or proprietary LLMs and employs a modular architecture consisting of the Conversational Payment Agent (CPA - first agent level), Supervisor agents (second agent level), Routing agents (third agent level), and the Process summary agent (fourth agent level). The CPA serves as the central entry point, handling all external requests and coordinating subsequent tasks across hierarchical levels. HMASP incorporates architectural patterns that enable modular task execution across agents and levels for payment operations, including shared state variables, decoupled message states, and structured handoff protocols that facilitate coordination across agents and workflows. Experimental results demonstrate the feasibility of the proposed HMASP. To our knowledge, HMASP is the first LLM-based multi-agent system to implement end-to-end agentic payment workflows. This work lays a foundation for extending agentic capabilities into the payment domain.

Path signatures provide a rich representation of sequential data, with strong theoretical guarantees and good performance in a variety of machine-learning tasks. While signatures have progressed from fixed feature extractors to trainable components of machine-learning models, existing libraries often lack the required scalability for large-scale, gradient-based learning. To address this gap, this paper introduces pathsig, a PyTorch-native library that computes path signatures directly in the word basis. By using CUDA kernels to update signature coefficients in parallel over prefix-closed word sets, pathsig achieves high GPU throughput and near-minimal peak memory. Compared with other libraries, pathsig achieves 10-30x speedups for computation of truncated signatures and up to 4-10x speedups in training that require backpropagation through the signature. Beyond regular truncation, pathsig supports projections of the (infinite-dimensional) signature onto user-specified sets of words and anisotropic truncation motivated by inhomogeneous path regularity, enabling more compact representations that can reduce dimensionality, redundancy, and computational cost.

Large language models (LLMs) with reasoning capabilities have fueled a compelling narrative that reasoning universally improves performance across language tasks. We test this claim through a comprehensive evaluation of 504 configurations across seven model families--including adaptive, conditional, and reinforcement learning-based reasoning architectures--on sentiment analysis datasets of varying granularity (binary, five-class, and 27-class emotion). Our findings reveal that reasoning effectiveness is strongly task-dependent, challenging prevailing assumptions: (1) Reasoning shows task-complexity dependence--binary classification degrades up to -19.9 F1 percentage points (pp), while 27-class emotion recognition gains up to +16.0pp; (2) Distilled reasoning variants underperform base models by 3-18 pp on simpler tasks, though few-shot prompting enables partial recovery; (3) Few-shot learning improves over zero-shot in most cases regardless of model type, with gains varying by architecture and task complexity; (4) Pareto frontier analysis shows base models dominate efficiency-performance trade-offs, with reasoning justified only for complex emotion recognition despite 2.1x-54x computational overhead. We complement these quantitative findings with qualitative error analysis revealing that reasoning degrades simpler tasks through systematic over-deliberation, offering mechanistic insight beyond the high-level overthinking hypothesis.

Post-Training Quantization (PTQ) has emerged as an effective technique for alleviating the substantial computational and memory overheads of Vision-Language Models (VLMs) by compressing both weights and activations without retraining the full model. Existing PTQ methods primarily rely on static identification and global compensation of sensitive or outlier channels, yet they often overlook the distributional differences of these important channels across inputs, leading to unsatisfactory quantization. In this work, we observe that the distributions and occurrence frequencies of important channels vary significantly both across modalities and among tokens, even within the same modality. Accordingly, we propose \textbf{Quant Experts (QE)}, a token-aware adaptive error compensation with mixture-of-experts for VLMs quantization. QE divides the important channels into token-independent and token-dependent groups. For the former, a shared expert is designed for most tokens to compensate for global quantization error using a low-rank adapter. For the latter, routed experts including multiple routed low-rank adapters are elaborated to compensate for local quantization error related to specific tokens. Extensive experiments demonstrate that QE consistently enhances task accuracy across various quantization settings and model scales, ranging from 2B to 70B parameters, while maintaining performance comparable to full-precision models.

This paper proposes CIRCLE, a six-stage, lifecycle-based framework to bridge the reality gap between model-centric performance metrics and AI's materialized outcomes in deployment. While existing frameworks like MLOps focus on system stability and benchmarks measure abstract capabilities, decision-makers outside the AI stack lack systematic evidence about the behavior of AI technologies under real-world user variability and constraints. CIRCLE operationalizes the Validation phase of TEVV (Test, Evaluation, Verification, and Validation) by formalizing the translation of stakeholder concerns outside the stack into measurable signals. Unlike participatory design, which often remains localized, or algorithmic audits, which are often retrospective, CIRCLE provides a structured, prospective protocol for linking context-sensitive qualitative insights to scalable quantitative metrics. By integrating methods such as field testing, red teaming, and longitudinal studies into a coordinated pipeline, CIRCLE produces systematic knowledge: evidence that is comparable across sites yet sensitive to local context. This can enable governance based on materialized downstream effects rather than theoretical capabilities.

The growth and heterogeneity of IoT devices create security challenges where static identification models can degrade as traffic evolves. This paper presents a two-stage, flow-feature-based pipeline for unsupervised IoT device traffic profiling and incremental model updating, evaluated on selected long-duration captures from the Deakin IoT dataset. For baseline profiling, density-based clustering (DBSCAN) isolates a substantial outlier portion of the data and produces the strongest alignment with ground-truth device labels among tested classical methods (NMI 0.78), outperforming centroid-based clustering on cluster purity. For incremental adaptation, we evaluate stream-oriented clustering approaches and find that BIRCH supports efficient updates (0.13 seconds per update) and forms comparatively coherent clusters for a held-out novel device (purity 0.87), but with limited capture of novel traffic (share 0.72) and a measurable trade-off in known-device accuracy after adaptation (0.71). Overall, the results highlight a practical trade-off between high-purity static profiling and the flexibility of incremental clustering for evolving IoT environments.

Large Language Model (LLM) adapters enable low-cost model specialization, but introduce complex caching and scheduling challenges in distributed serving systems where hundreds of adapters must be hosted concurrently. While prior work has largely focused on latency minimization, resource efficiency through throughput maximization remains underexplored. This paper presents a data-driven pipeline that, for a given workload, computes an adapter placement that serves the workload with the minimum number of GPUs while avoiding request starvation and GPU memory errors. To that end, the approach identifies the maximum feasible throughput attainable on each GPU by leveraging accurate performance predictions learned from real serving behavior. The proposed pipeline integrates three components: (i) a Digital Twin (DT) tailored to LLM-adapter serving, (ii) a distilled machine learning (ML) model trained on DT-generated data, and (iii) a greedy placement algorithm that exploits ML-based performance estimates to maximize GPU efficiency. The DT emulates real system dynamics with high fidelity, achieving below 5% throughput estimation error while executing up to 90 times faster than full LLM benchmarking across both predictable and unpredictable workloads. The learned ML models further accelerate performance estimation with marginal accuracy degradation, enabling scalable optimization. Experimental results demonstrate that the pipeline substantially improves GPU efficiency by reducing the number of GPUs required to sustain target workloads. Beyond GPU efficiency, the pipeline can be adapted to alternative objectives, such as latency minimization, highlighting its versatility for future large-scale LLM serving infrastructures.

Reward models are central to aligning large language models (LLMs) with human preferences. Yet most approaches rely on pointwise reward estimates that overlook the epistemic uncertainty in reward models arising from limited human feedback. Recent work suggests that quantifying this uncertainty can reduce the costs of human annotation via uncertainty-guided active learning and mitigate reward overoptimization in LLM post-training. However, uncertainty-aware reward models have so far been adopted without thorough comparison, leaving them poorly understood. This work introduces a unified framework, RewardUQ, to systematically evaluate uncertainty quantification for reward models. We compare common methods along standard metrics measuring accuracy and calibration, and we propose a new ranking strategy incorporating both dimensions for a simplified comparison. Our experimental results suggest that model size and initialization have the most meaningful impact on performance, and most prior work could have benefited from alternative design choices. To foster the development and evaluation of new methods and aid the deployment in downstream applications, we release our open-source framework as a Python package. Our code is available at https://github.com/lasgroup/rewarduq.

Market regime shifts induce distribution shifts that can degrade the performance of portfolio rebalancing policies. We propose macro-conditioned scenario-context rollout (SCR) that generates plausible next-day multivariate return scenarios under stress events. However, doing so faces new challenges, as history will never tell what would have happened differently. As a result, incorporating scenario-based rewards from rollouts introduces a reward--transition mismatch in temporal-difference learning, destabilizing RL critic training. We analyze this inconsistency and show it leads to a mixed evaluation target. Guided by this analysis, we construct a counterfactual next state using the rollout-implied continuations and augment the critic agent's bootstrap target. Doing so stabilizes the learning and provides a viable bias-variance tradeoff. In out-of-sample evaluations across 31 distinct universes of U.S. equity and ETF portfolios, our method improves Sharpe ratio by up to 76% and reduces maximum drawdown by up to 53% compared with classic and RL-based portfolio rebalancing baselines.

The rapid advancement of Vision-Language models (VLMs) has raised growing concerns that their black-box reasoning processes could lead to unintended forms of social bias. Current debiasing approaches focus on mitigating surface-level bias signals through post-hoc learning or test-time algorithms, while leaving the internal dynamics of the model largely unexplored. In this work, we introduce an interpretable, model-agnostic bias mitigation framework, DeBiasLens, that localizes social attribute neurons in VLMs through sparse autoencoders (SAEs) applied to multimodal encoders. Building upon the disentanglement ability of SAEs, we train them on facial image or caption datasets without corresponding social attribute labels to uncover neurons highly responsive to specific demographics, including those that are underrepresented. By selectively deactivating the social neurons most strongly tied to bias for each group, we effectively mitigate socially biased behaviors of VLMs without degrading their semantic knowledge. Our research lays the groundwork for future auditing tools, prioritizing social fairness in emerging real-world AI systems.

Contrastive learning has become a cornerstone of modern representation learning, allowing training with massive unlabeled data for both task-specific and general (foundation) models. A prototypical loss in contrastive training is InfoNCE and its variants. In this work, we show that the InfoNCE objective induces Gaussian structure in representations that emerge from contrastive training. We establish this result in two complementary regimes. First, we show that under certain alignment and concentration assumptions, projections of the high-dimensional representation asymptotically approach a multivariate Gaussian distribution. Next, under less strict assumptions, we show that adding a small asymptotically vanishing regularization term that promotes low feature norm and high feature entropy leads to similar asymptotic results. We support our analysis with experiments on synthetic and CIFAR-10 datasets across multiple encoder architectures and sizes, demonstrating consistent Gaussian behavior. This perspective provides a principled explanation for commonly observed Gaussianity in contrastive representations. The resulting Gaussian model enables principled analytical treatment of learned representations and is expected to support a wide range of applications in contrastive learning.

Property checking of RTL designs is a central task in formal verification. Among available engines, IC3/PDR is a widely used backbone whose performance critically depends on inductive generalization, the step that generalizes a concrete counterexample-to-induction (CTI) cube into a lemma. Prior work has explored machine learning to guide this step and achieved encouraging results, yet most methods adopt a per-clause graph analysis paradigm: for each clause they repeatedly build and analyze graphs, incurring heavy overhead and creating a scalability bottleneck. We introduce LeGend, which replaces this paradigm with one-time global representation learning. LeGend pre-trains a domain-adapted self-supervised model to produce latch embeddings that capture global circuit properties. These precomputed embeddings allow a lightweight model to predict high-quality lemmas with negligible overhead, effectively decoupling expensive learning from fast inference. Experiments show LeGend accelerates two state-of-the-art IC3/PDR engines across a diverse set of benchmarks, presenting a promising path to scale up formal verification.

Jailbreak techniques for large language models (LLMs) evolve faster than benchmarks, making robustness estimates stale and difficult to compare across papers due to drift in datasets, harnesses, and judging protocols. We introduce JAILBREAK FOUNDRY (JBF), a system that addresses this gap via a multi-agent workflow to translate jailbreak papers into executable modules for immediate evaluation within a unified harness. JBF features three core components: (i) JBF-LIB for shared contracts and reusable utilities; (ii) JBF-FORGE for the multi-agent paper-to-module translation; and (iii) JBF-EVAL for standardizing evaluations. Across 30 reproduced attacks, JBF achieves high fidelity with a mean (reproduced-reported) attack success rate (ASR) deviation of +0.26 percentage points. By leveraging shared infrastructure, JBF reduces attack-specific implementation code by nearly half relative to original repositories and achieves an 82.5% mean reused-code ratio. This system enables a standardized AdvBench evaluation of all 30 attacks across 10 victim models using a consistent GPT-4o judge. By automating both attack integration and standardized evaluation, JBF offers a scalable solution for creating living benchmarks that keep pace with the rapidly shifting security landscape.

Protein function relies on dynamic conformational ensembles, yet current generative models like AlphaFold3 often fail to produce ensembles that match experimental data. Recent experiment-guided generators attempt to address this by steering the reverse diffusion process. However, these methods are limited by fixed sampling horizons and sensitivity to initialization, often yielding thermodynamically implausible results. We introduce a general inference-time optimization framework to solve these challenges. First, we optimize over latent representations to maximize ensemble log-likelihood, rather than perturbing structures post hoc. This approach eliminates dependence on diffusion length, removes initialization bias, and easily incorporates external constraints. Second, we present novel sampling schemes for drawing Boltzmann-weighted ensembles. By combining structural priors from AlphaFold3 with force-field-based priors, we sample from their product distribution while balancing experimental likelihoods. Our results show that this framework consistently outperforms state-of-the-art guidance, improving diversity, physical energy, and agreement with data in X-ray crystallography and NMR, often fitting the experimental data better than deposited PDB structures. Finally, inference-time optimization experiments maximizing ipTM scores reveal that perturbing AlphaFold3 embeddings can artificially inflate model confidence. This exposes a vulnerability in current design metrics, whose mitigation could offer a pathway to reduce false discovery rates in binder engineering.

Spanish is the official language of twenty-one countries and is spoken by over 441 million people. Naturally, there are many variations in how Spanish is spoken across these countries. Media platforms such as YouTube rely on automatic speech recognition systems to make their content accessible to different groups of users. However, YouTube offers only one option for automatically generating captions in Spanish. This raises the question: could this captioning system be biased against certain Spanish dialects? This study examines the potential biases in YouTube's automatic captioning system by analyzing its performance across various Spanish dialects. By comparing the quality of captions for female and male speakers from different regions, we identify systematic disparities which can be attributed to specific dialects. Our study provides further evidence that algorithmic technologies deployed on digital platforms need to be calibrated to the diverse needs and experiences of their user populations.

The next generation of autonomous agents must not only learn efficiently but also act reliably and adapt their behavior in open worlds. Standard approaches typically assume fixed tasks and environments with little or no novelty, which limits world models' ability to support agents that must evolve their policies as conditions change. This paper outlines a vision for foundation world models: persistent, compositional representations that unify reinforcement learning, reactive/program synthesis, and abstraction mechanisms. We propose an agenda built around four components: (i) learnable reward models from specifications to support optimization with clear objectives; (ii) adaptive formal verification integrated throughout learning; (iii) online abstraction calibration to quantify the reliability of the model's predictions; and (iv) test-time synthesis and world-model generation guided by verifiers. Together, these components enable agents to synthesize verifiable programs, derive new policies from a small number of interactions, and maintain correctness while adapting to novelty. The resulting framework positions foundation world models as a substrate for learning, reasoning, and adaptation, laying the groundwork for agents that not only act well but can explain and justify the behavior they adopt.

Alzheimer's disease is a progressive neurodegenerative disorder in which mild cognitive impairment (MCI) marks a critical transition between aging and dementia. Neuroimaging modalities, such as structural MRI, provide biomarkers of this transition; however, their high costs and infrastructure needs limit their deployment at a population scale. Speech analysis offers a non-invasive alternative, but speech-only classifiers are developed independently of neuroimaging, leaving decision boundaries biologically ungrounded and limiting reliability on the subtle CN-versus-MCI distinction. We propose MINT (Multimodal Imaging-to-Speech Knowledge Transfer), a three-stage cross-modal framework that transfers biomarker structure from MRI into a speech encoder at training time. An MRI teacher, trained on 1,228 subjects, defines a compact neuroimaging embedding space for CN-versus-MCI classification. A residual projection head aligns speech representations to this frozen imaging manifold via a combined geometric loss, adapting speech to the learned biomarker space while preserving imaging encoder fidelity. The frozen MRI classifier, which is never exposed to speech, is applied to aligned embeddings at inference and requires no scanner. Evaluation on ADNI-4 shows aligned speech achieves performance comparable to speech-only baselines (AUC 0.720 vs 0.711) while requiring no imaging at inference, demonstrating that MRI-derived decision boundaries can ground speech representations. Multimodal fusion improves over MRI alone (0.973 vs 0.958). Ablation studies identify dropout regularization and self-supervised pretraining as critical design decisions. To our knowledge, this is the first demonstration of MRI-to-speech knowledge transfer for early Alzheimer's screening, establishing a biologically grounded pathway for population-level cognitive triage without neuroimaging at inference.

We present gradiend, an open-source Python package that operationalizes the GRADIEND method for learning feature directions from factual-counterfactual MLM and CLM gradients in language models. The package provides a unified workflow for feature-related data creation, training, evaluation, visualization, persistent model rewriting via controlled weight updates, and multi-feature comparison. We demonstrate GRADIEND on an English pronoun paradigm and on a large-scale feature comparison that reproduces prior use cases.

Real-world data frequently exhibit latent hierarchical structures, which can be naturally represented by hyperbolic geometry. Although recent hyperbolic neural networks have demonstrated promising results, many existing architectures remain partially intrinsic, mixing Euclidean operations with hyperbolic ones or relying on extrinsic parameterizations. To address it, we propose the \emph{Intrinsic Lorentz Neural Network} (ILNN), a fully intrinsic hyperbolic architecture that conducts all computations within the Lorentz model. At its core, the network introduces a novel \emph{point-to-hyperplane} fully connected layer (FC), replacing traditional Euclidean affine logits with closed-form hyperbolic distances from features to learned Lorentz hyperplanes, thereby ensuring that the resulting geometric decision functions respect the inherent curvature. Around this fundamental layer, we design intrinsic modules: GyroLBN, a Lorentz batch normalization that couples gyro-centering with gyro-scaling, consistently outperforming both LBN and GyroBN while reducing training time. We additionally proposed a gyro-additive bias for the FC output, a Lorentz patch-concatenation operator that aligns the expected log-radius across feature blocks via a digamma-based scale, and a Lorentz dropout layer. Extensive experiments conducted on CIFAR-10/100 and two genomic benchmarks (TEB and GUE) illustrate that ILNN achieves state-of-the-art performance and computational cost among hyperbolic models and consistently surpasses strong Euclidean baselines. The code is available at \href{https://github.com/Longchentong/ILNN}{\textcolor{magenta}{this url}}.

Offline reinforcement learning aims to learn an agent from pre-collected datasets, avoiding unsafe and inefficient real-time interaction. However, inevitable access to out-ofdistribution actions during the learning process introduces approximation errors, causing the error accumulation and considerable overestimation. In this paper, we construct a new pessimistic auxiliary policy for sampling reliable actions. Specifically, we develop a pessimistic auxiliary strategy by maximizing the lower confidence bound of the Q-function. The pessimistic auxiliary strategy exhibits a relatively high value and low uncertainty in the vicinity of the learned policy, avoiding the learned policy sampling high-value actions with potentially high errors during the learning process. Less approximation error introduced by sampled action from pessimistic auxiliary strategy leads to the alleviation of error accumulation. Extensive experiments on offline reinforcement learning benchmarks reveal that utilizing the pessimistic auxiliary strategy can effectively improve the efficacy of other offline RL approaches.

Sycophancy, the tendency of large language models to favour user-affirming responses over critical engagement, has been identified as an alignment failure, particularly in high-stakes advisory and social contexts. While prior work has documented conversational features correlated with sycophancy, we lack a systematic understanding of what provokes or prevents AI sycophancy. Here, we present a set of controlled experimental studies where we first isolate how input framing influences sycophancy, and second, leverage these findings to develop mitigation strategies. In a nested factorial design, we compare questions to various non-questions where we vary three orthogonal factors: epistemic certainty (statement, belief, conviction), perspective (I- vs user-perspective), and affirmation vs negation. We show that (1) sycophancy is substantially higher in response to non-questions compared to questions. Additionally, we find that (2) sycophancy increases monotonically with epistemic certainty conveyed by the user, and (3) is amplified by I-perspective framing. Building on this, we show that asking a model to convert non-questions into questions before answering significantly reduces sycophancy. Importantly, this effect is stronger than a simple baseline prompt asking models "not to be sycophantic". Our work offers a practical and effective input-level mitigation that both developers and users can easily adopt.

Masked discrete diffusion models (MDMs) are a promising new approach to generative modelling, offering the ability for parallel token generation and therefore greater efficiency than autoregressive counterparts. However, achieving an optimal balance between parallel generation and sample quality remains an open problem. Current approaches primarily address this issue through fixed, heuristic parallel sampling methods. There exist some recent learning based approaches to this problem, but its formulation from the perspective of variational inference remains underexplored. In this work, we propose a variational inference framework for learning parallel generation orders for MDMs. As part of our method, we propose a parameterisation for the approximate posterior of generation orders which facilitates parallelism and efficient sampling during training. Using this method, we conduct preliminary experiments on the GSM8K dataset, where our method performs competitively against heuristic sampling strategies in the regime of highly parallel generation. For example, our method achieves 33.1\% accuracy with an average of only only 4 generation steps, compared to 23.7-29.0\% accuracy achieved by standard competitor methods in the same number of steps. We believe further experiments and analysis of the method will yield valuable insights into the problem of parallel generation with MDMs.

How natural speech is represented in the brain constitutes a major challenge for cognitive neuroscience, with cortical envelope-following responses playing a central role in speech decoding. This paper presents our approach to the Speech Detection task in the LibriBrain Competition 2025, utilizing over 50 hours of magnetoencephalography (MEG) signals from a single participant listening to LibriVox audiobooks. We introduce the proposed Sequential Hierarchical Integration Network for EEG and MEG (SHINE) to reconstruct the binary speech-silence sequences from MEG signals. In the Extended Track, we further incorporated auxiliary reconstructions of speech envelopes and Mel spectrograms to enhance training. Ensemble methods combining SHINE with baselines (BrainMagic, AWavNet, ConvConcatNet) achieved F1-macro scores of 0.9155 (Standard Track) and 0.9184 (Extended Track) on the leaderboard test set.

Background. Automated test execution is an important activity to gather information about the quality of a software project. So-called flaky tests, however, negatively affect this process. Such tests fail seemingly at random without changes to the code and thus do not provide a clear signal. Previous work proposed to identify flaky tests based on the source code identifiers in the test code. So far, these approaches have not been evaluated in a large-scale industrial setting. Aims. We evaluate approaches to identify flaky tests and their root causes based on source code identifiers in the test code in a large-scale industrial project. Method. First, we replicate previous work by Pinto et al. in the context of SAP HANA. Second, we assess different feature extraction techniques, namely TF-IDF and TF-IDFC-RF. Third, we evaluate CodeBERT and XGBoost as classification models. For a sound comparison, we utilize both the data set from previous work and two data sets from SAP HANA. Results. Our replication shows similar results on the original data set and on one of the SAP HANA data sets. While the original approach yielded an F1-Score of 0.94 on the original data set and 0.92 on the SAP HANA data set, our extensions achieve F1-Scores of 0.96 and 0.99, respectively. The reliance on external data sources is a common root cause for test flakiness in the context of SAP HANA. Conclusions. The vocabulary of a large industrial project seems to be slightly different with respect to the exact terms, but the categories for the terms, such as remote dependencies, are similar to previous empirical findings. However, even with rather large F1-Scores, both finding source code identifiers for flakiness and a black box prediction have limited use in practice as the results are not actionable for developers.

Micro-expressions, characterized by transience and subtlety, pose challenges to existing optical flow-based recognition methods. To address this, this paper proposes a dual-branch micro-expression feature extraction network integrated with parallel attention. Key contributions include: 1) a residual network designed to alleviate gradient anishing and network degradation; 2) an Inception network constructed to enhance model representation and suppress interference from irrelevant regions; 3) an adaptive feature fusion module developed to integrate dual-branch features. Experiments on the CASME II dataset demonstrate that the proposed method achieves 74.67% accuracy, outperforming LBP-TOP (by 11.26%), MSMMT (by 3.36%), and other comparative methods.

Agentic search has emerged as a promising paradigm for adaptive retrieval systems powered by large language models (LLMs). However, existing benchmarks primarily focus on quality, overlooking efficiency factors that are critical for real-world deployment. Moreover, real-world user queries often contain underspecified preferences, a challenge that remains largely underexplored in current agentic search evaluation. As a result, many agentic search systems remain impractical despite their impressive performance. In this work, we introduce HotelQuEST, a benchmark comprising 214 hotel search queries that range from simple factual requests to complex queries, enabling evaluation across the full spectrum of query difficulty. We further address the challenge of evaluating underspecified user preferences by collecting clarifications that make annotators' implicit preferences explicit for evaluation. We find that LLM-based agents achieve higher accuracy than traditional retrievers, but at substantially higher costs due to redundant tool calls and suboptimal routing that fails to match query complexity to model capability. Our analysis exposes inefficiencies in current agentic search systems and demonstrates substantial potential for cost-aware optimization.

Modern deep neural networks remain challenging to interpret due to the opacity of their latent representations, impeding model understanding, debugging, and debiasing. Concept Embedding Models (CEMs) address this by mapping inputs to human-interpretable concept representations from which tasks can be predicted. Yet, CEMs fail to represent inter-concept relationships and require concept annotations at different granularities during training, limiting their applicability. In this paper, we introduce Hierarchical Concept Embedding Models (HiCEMs), a new family of CEMs that explicitly model concept relationships through hierarchical structures. To enable HiCEMs in real-world settings, we propose Concept Splitting, a method for automatically discovering finer-grained sub-concepts from a pretrained CEM's embedding space without requiring additional annotations. This allows HiCEMs to generate fine-grained explanations from limited concept labels, reducing annotation burdens. Our evaluation across multiple datasets, including a user study and experiments on PseudoKitchens, a newly proposed concept-based dataset of 3D kitchen renders, demonstrates that (1) Concept Splitting discovers human-interpretable sub-concepts absent during training that can be used to train highly accurate HiCEMs, and (2) HiCEMs enable powerful test-time concept interventions at different granularities, leading to improved task accuracy.

While Multimodal Large Language Models (MLLMs) demonstrate proficiency in 2D scenes, extending their perceptual intelligence to 3D point cloud understanding remains a significant challenge. Current approaches focus primarily on aligning 3D features with pre-trained models. However, they typically treat geometric reasoning as an implicit mapping process. These methods bypass intermediate logical steps and consequently suffer from geometric hallucinations. They confidently generate plausible responses that fail to ground in precise structural details. To bridge this gap, we present PointCoT, a novel framework that empowers MLLMs with explicit Chain-of-Thought (CoT) reasoning for 3D data. We advocate for a \textit{Look, Think, then Answer} paradigm. In this approach, the model is supervised to generate geometry-grounded rationales before predicting final answers. To facilitate this, we construct Point-Reason-Instruct, a large-scale benchmark comprising $\sim$86k instruction-tuning samples with hierarchical CoT annotations. By leveraging a dual-stream multi-modal architecture, our method synergizes semantic appearance with geometric truth. Extensive experiments demonstrate that PointCoT achieves state-of-the-art performance on complex reasoning tasks.

Memory systems address the challenge of context loss in Large Language Model during prolonged interactions. However, compared to human cognition, the efficacy of these systems in processing emotion-related information remains inconclusive. To address this gap, we propose an emotion-enhanced memory evaluation benchmark to assess the performance of mainstream and state-of-the-art memory systems in handling affective information. We developed the \textbf{H}uman-\textbf{L}ike \textbf{M}emory \textbf{E}motion (\textbf{HLME}) dataset, which evaluates memory systems across three dimensions: emotional information extraction, emotional memory updating, and emotional memory question answering. Experimental results indicate that none of the evaluated systems achieve robust performance across all three tasks. Our findings provide an objective perspective on the current deficiencies of memory systems in processing emotional memories and suggest a new trajectory for future research and system optimization.

This paper introduces a dataset of enriched geographic coordinates retrieved from Diderot and d'Alembert's eighteenth-century Encyclopedie. Automatically recovering geographic coordinates from historical texts is a complex task, as they are expressed in a variety of ways and with varying levels of precision. To improve retrieval of coordinates from similar digitized early modern texts, we have created a gold standard dataset, trained models, published the resulting inferred and normalized coordinate data, and experimented applying these models to new texts. From 74,000 total articles in each of the digitized versions of the Encyclopedie from ARTFL and ENCCRE, we examined 15,278 geographical entries, manually identifying 4,798 containing coordinates, and 10,480 with descriptive but non-numerical references. Leveraging our gold standard annotations, we trained transformer-based models to retrieve and normalize coordinates. The pipeline presented here combines a classifier to identify coordinate-bearing entries and a second model for retrieval, tested across encoder-decoder and decoder architectures. Cross-validation yielded an 86% EM score. On an out-of-domain eighteenth-century Trevoux dictionary (also in French), our fine-tuned model had a 61% EM score, while for the nineteenth-century, 7th edition of the Encyclopaedia Britannica in English, the EM was 77%. These findings highlight the gold standard dataset's usefulness as training data, and our two-step method's cross-lingual, cross-domain generalizability.

Transformer-based models such as BERT have significantly advanced Natural Language Processing (NLP) across many languages. However, Nepali, a low-resource language written in Devanagari script, remains relatively underexplored. This study benchmarks multilingual, Indic, Hindi, and Nepali BERT variants to evaluate their effectiveness in Nepali topic classification. Ten pre-trained models, including mBERT, XLM-R, MuRIL, DevBERT, HindiBERT, IndicBERT, and NepBERTa, were fine-tuned and tested on the balanced Nepali dataset containing 25,006 sentences across five conceptual domains and the performance was evaluated using accuracy, weighted precision, recall, F1-score, and AUROC metrics. The results reveal that Indic models, particularly MuRIL-large, achieved the highest F1-score of 90.60%, outperforming multilingual and monolingual models. NepBERTa also performed competitively with an F1-score of 88.26%. Overall, these findings establish a robust baseline for future document-level classification and broader Nepali NLP applications.

Serverless computing simplifies cloud deployment but introduces new challenges in managing service latency and carbon emissions. Reducing cold-start latency requires retaining warm function instances, while minimizing carbon emissions favors reclaiming idle resources. This balance is further complicated by time-varying grid carbon intensity and varying workload patterns, under which static keep-alive policies are inefficient. We present LACE-RL, a latency-aware and carbon-efficient management framework that formulates serverless pod retention as a sequential decision problem. LACE-RL uses deep reinforcement learning to dynamically tune keep-alive durations, jointly modeling cold-start probability, function-specific latency costs, and real-time carbon intensity. Using the Huawei Public Cloud Trace, we show that LACE-RL reduces cold starts by 51.69% and idle keep-alive carbon emissions by 77.08% compared to Huawei's static policy, while achieving better latency-carbon trade-offs than state-of-the-art heuristic and single-objective baselines, approaching Oracle performance.

This paper presents a novel autonomous robotic assembly framework for constructing stable structures without relying on predefined architectural blueprints. Instead of following fixed plans, construction tasks are defined through targets and obstacles, allowing the system to adapt more flexibly to environmental uncertainty and variations during the building process. A reinforcement learning (RL) policy, trained using deep Q-learning with successor features, serves as the decision-making component. As a proof of concept, we evaluate the approach on a benchmark of 15 2D robotic assembly tasks of discrete block construction. Experiments using a real-world closed-loop robotic setup demonstrate the feasibility of the method and its ability to handle construction noise. The results suggest that our framework offers a promising direction for more adaptable and robust robotic construction in real-world environments.

We show that large language models (LLMs) have an astonishing ability to recover meaning from severely degraded English texts. Texts in which content words have been randomly substituted by nonsense strings, e.g., "At the ghybe of the swuint, we are haiveed to Wourge Phrear-gwurr, who sproles into an ghitch flount with his crurp", can be translated to conventional English that is, in many cases, close to the original text, e.g., "At the start of the story, we meet a man, Chow, who moves into an apartment building with his wife." These results show that structural cues (e.g., morphosyntax, closed-class words) constrain lexical meaning to a much larger degree than imagined. Although the abilities of LLMs to make sense of "Jabberwockified" English are clearly superhuman, they are highly relevant to understanding linguistic structure and suggest that efficient language processing either in biological or artificial systems likely benefits from very tight integration between syntax, lexical semantics, and general world knowledge.

We present a multiparty session type (MST) framework with asynchronous mixed choice (MC). We propose a core construct for MC that allows transient inconsistencies in protocol state between distributed participants, but ensures all participants can always eventually reach a mutually consistent state. We prove the correctness of our system by establishing a progress property and an operational correspondence between global types and distributed local type projections. Based on our theory, we implement a practical toolchain for specifying and validating asynchronous MST protocols featuring MC, and programming compliant gen_statem processes in Erlang/OTP. We test our framework by using our toolchain to specify and reimplement part of the amqp_client of the RabbitMQ broker for Erlang.

Microservice architectures are an emergent technology that builds business logic into a suite of small services. Each microservice runs in its process and the communication is made through lightweight mechanisms, usually HTTP resource API. These architectures are built upon independently deployable and, supposedly, reliable pieces of software that may, or may not, have been developed by the team using it. Nowadays, industries are dangerously migrating into microservice architectures without an effective and automatic process for testing the software being used. Furthermore, current API specification languages are not expressive enough to be used for testing purposes. To solve this problem it is necessary to extend currently broadly used API specification languages. APOSTL is a specification language to annotate APIs specifications based on first-order logic, with some restrictions. It has the purpose of extending the currently used API description languages with properties that can be useful for testing purposes, transforming these description documents into useful testing artefacts. Besides providing information needed for testing an application, APOSTL also provides an API with semantic. This additional information is then leveraged to automate microservice testing. The work developed in this thesis aims to fully automate the microservice testing process. It is achieved by the implementation of PETIT a tool able to test microservices when provided with an OpenAPI Specification document, written in JSON and properly annotated with the previously proposed specification language, APOSTL. The tool is able to analyze microservices independently from the source code availability.

Why do some speakers capture a room almost instantly while others fail to connect? The real-time architecture of audience engagement remains largely a black box. Here, we used motion-captured animations to present the pure nonverbal performance of public speakers to audiences - either in silence (nonverbal-only) or paired with the verbal content (nonverbal-plus-verbal). Using continuous response measurement (CRM), we find that audience judgments solidify with remarkable speed: Moment-to-moment engagement ratings become highly predictive of subsequent evaluations within the initial 10 seconds of the performance. Most notably, this predictive relationship emerged faster and slightly stronger in the nonverbal-only condition, with predictive information being present already after less than 5 seconds. These findings elucidate the social impact a speaker's nonverbal performance has on audience impressions, even when dissociated from the verbal content of the speech. Our approach provides a high-resolution temporal map of social impression formation, pointing to an early "moment of capture" that appears to set the stage for the reception of the following message. On a broader scale, this research validates a powerful new method to isolate different communicative channels, to scientifically deconstruct rhetorical skill, and to study the pervasive impact of nonverbal behavior more broadly. It also enables us to translate the ancient art of rhetoric into a modern science of social impression formation, yielding an empirical basis that can inform human-centered feedback, develop AI-based augmentation tools, and guide the design of engaging, socially present avatars in an increasingly AI-mediated and virtual world.

The advent of large-scale self-supervised learning (SSL) has produced a vast zoo of medical foundation models. However, selecting optimal medical foundation models for specific segmentation tasks remains a computational bottleneck. Existing Transferability Estimation (TE) metrics, primarily designed for classification, rely on global statistical assumptions and fail to capture the topological complexity essential for dense prediction. We propose a novel Topology-Driven Transferability Estimation framework that evaluates manifold tractability rather than statistical overlap. Our approach introduces three components: (1) Global Representation Topology Divergence (GRTD), utilizing Minimum Spanning Trees to quantify feature-label structural isomorphism; (2) Local Boundary-Aware Topological Consistency (LBTC), which assesses manifold separability specifically at critical anatomical boundaries; and (3) Task-Adaptive Fusion, which dynamically integrates global and local metrics based on the semantic cardinality of the target task. Validated on the large-scale OpenMind benchmark across diverse anatomical targets and SSL foundation models, our approach significantly outperforms state-of-the-art baselines by around \textbf{31\%} relative improvement in the weighted Kendall, providing a robust, training-free proxy for efficient model selection without the cost of fine-tuning. The code will be made publicly available upon acceptance.

AI coding agents allow software developers to generate code quickly, which raises a practical question for project managers and open source maintainers: can vibe coders with less development experience substitute for expert developers? To explore whether developer experience still matters in AI-assisted development, we study $22,953$ Pull Requests (PRs) from $1,719$ vibe coders in the GitHub repositories of the AIDev dataset. We split vibe coders into lower experience vibe coders ($\mathit{Exp}_{Low}$) and higher experience vibe coders ($\mathit{Exp}_{High}$) and compare contribution magnitude and PR acceptance rates across PR categories. We find that $\mathit{Exp}_{Low}$ submits PRs with larger volume ($2.15\times$ more commits and $1.47\times$ more files changed) than $\mathit{Exp}_{High}$. Moreover, $\mathit{Exp}_{Low}$ PRs, when compared to $\mathit{Exp}_{High}$, receive $4.52\times$ more review comments, and have $31\%$ lower acceptance rates, and remain open $5.16\times$ longer before resolution. Our results indicate that low-experienced vibe coders focus on generating more code while shifting verification burden onto reviewers. For practice, project managers may not be able to safely replace experienced developers with low-experience vibe coders without increasing review capacity. Development teams should therefore combine targeted training for novices with adaptive PR review cycles.

State-of-the-art models for medical image segmentation achieve excellent accuracy but require substantial computational resources, limiting deployment in resource-constrained clinical settings. We present SegMate, an efficient 2.5D framework that achieves state-of-the-art accuracy, while considerably reducing computational requirements. Our efficient design is the result of meticulously integrating asymmetric architectures, attention mechanisms, multi-scale feature fusion, slice-based positional conditioning, and multi-task optimization. We demonstrate the efficiency-accuracy trade-off of our framework across three modern backbones (EfficientNetV2-M, MambaOut-Tiny, FastViT-T12). We perform experiments on three datasets: TotalSegmentator, SegTHOR and AMOS22. Compared with the vanilla models, SegMate reduces computation (GFLOPs) by up to 2.5x and memory footprint (VRAM) by up to 2.1x, while generally registering performance gains of around 1%. On TotalSegmentator, we achieve a Dice score of 93.51% with only 295MB peak GPU memory. Zero-shot cross-dataset evaluations on SegTHOR and AMOS22 demonstrate strong generalization, with Dice scores of up to 86.85% and 89.35%, respectively. We release our open-source code at https://github.com/andreibunea99/SegMate.

We propose an experience-guided cascaded multi-agent framework for Breast Ultrasound Screening and Diagnosis, called BUSD-Agent, that aims to reduce diagnostic escalation and unnecessary biopsy referrals. Our framework models screening and diagnosis as a two-stage, selective decision-making process. A lightweight `screening clinic' agent, restricted to classification models as tools, selectively filters out benign and normal cases from further diagnostic escalation when malignancy risk and uncertainty are estimated as low. Cases that have higher risks are escalated to the `diagnostic clinic' agent, which integrates richer perception and radiological description tools to make a secondary decision on biopsy referral. To improve agent performance, past records of pathology-confirmed outcomes along with image embeddings, model predictions, and historical agent actions are stored in a memory bank as structured decision trajectories. For each new case, BUSD-Agent retrieves similar past cases based on image, model response and confidence similarity to condition the agent's current decision policy. This enables retrieval-conditioned in-context adaptation that dynamically adjusts model trust and escalation thresholds from prior experiences without parameter updates. Evaluation across 10 breast ultrasound datasets shows that the proposed experience-guided workflow reduces diagnostic escalation in BUSD-Agent from 84.95% to 58.72% and overall biopsy referrals from 59.50% to 37.08%, compared to the same architecture without trajectory conditioning, while improving average screening specificity by 68.48% and diagnostic specificity by 6.33%.

Referring Expression Comprehension (REC) links language to region level visual perception. Standard benchmarks (RefCOCO, RefCOCO+, RefCOCOg) have progressed rapidly with multimodal LLMs but remain weak tests of visual reasoning and grounding: (i) many expressions are very short, leaving little reasoning demand; (ii) images often contain few distractors, making the target easy to find; and (iii) redundant descriptors enable shortcut solutions that bypass genuine text understanding and visual reasoning. We introduce Ref-Adv, a modern REC benchmark that suppresses shortcuts by pairing linguistically nontrivial expressions with only the information necessary to uniquely identify the target. The dataset contains referring expressions on real images, curated with hard distractors and annotated with reasoning facets including negation. We conduct comprehensive ablations (word order perturbations and descriptor deletion sufficiency) to show that solving Ref-Adv requires reasoning beyond simple cues, and we evaluate a broad suite of contemporary multimodal LLMs on Ref-Adv. Despite strong results on RefCOCO, RefCOCO+, and RefCOCOg, models drop markedly on Ref-Adv, revealing reliance on shortcuts and gaps in visual reasoning and grounding. We provide an in depth failure analysis and aim for Ref-Adv to guide future work on visual reasoning and grounding in MLLMs.

Personal care formulations often contain synthetic and non-biodegradable ingredients, such as silicone and mineral oils, which can offer a unique performance. However, due to regulations like the EU ban of Octamethylcyclotetrasiloxane (D4), Decamethyl-cyclopentasiloxane (D5), Dodecamethylcyclohexasiloxane (D6) already in effect for rinse off and for leave on cosmetics by June 2027 coupled with growing consumer awareness and expectations on sustainability, personal care brands face significant pressure to replace these synthetic ingredients with natural alternatives without compromising performance and cost. As a result, formulators are confronted with the challenge to find natural-based solutions within a short timeframe. In this study, we propose a pioneering approach that utilizes predicting modelling and simulation-based digital services to obtain natural-based ingredient combinations as recommendations to commonly used synthetic ingredients. We will demonstrate the effectiveness of our predictions through the application of these proposals in specific formulations. By offering a platform of digital services, it is aimed to empower formulators to explore good performing novel and environmentally friendly alternatives, ultimately driving a substantial and genuine transformation in the personal care industry.

Speculative decoding accelerates autoregressive large language model (LLM) inference by using a lightweight draft model to propose candidate tokens that are then verified in parallel by the target model. The speedup is significantly determined by the acceptance rate, yet standard training minimizes Kullback-Leibler (KL) divergence as a proxy objective. While KL divergence and acceptance rate share the same global optimum, small draft models, having limited capacity, typically converge to suboptimal solutions where minimizing KL does not guarantee maximizing acceptance rate. To address this issue, we propose LK losses, special training objectives that directly target acceptance rate. Comprehensive experiments across four draft architectures and six target models, ranging from 8B to 685B parameters, demonstrate consistent improvements in acceptance metrics across all configurations compared to the standard KL-based training. We evaluate our approach on general, coding and math domains and report gains of up to 8-10% in average acceptance length. LK losses are easy to implement, introduce no computational overhead and can be directly integrated into any existing speculator training framework, making them a compelling alternative to the existing draft training objectives.

This paper develops a theory of graph classification under domain shift through a random-graph generative lens, where we consider intra-class graphs sharing the same random graph model (RGM) and the domain shift induced by changes in RGM components. While classic domain adaptation (DA) theories have well-underpinned existing techniques to handle graph distribution shift, the information of graph samples, which are itself structured objects, is less explored. The non-Euclidean nature of graphs and specialized architectures for graph learning further complicate a fine-grained analysis of graph distribution shifts. In this paper, we propose a theory that assumes RGM as the data generative process, exploiting its connection to hypothesis complexity in function space perspective for such fine-grained analysis. Building on a vector-valued reproducing kernel Hilbert space (vRKHS) formulation, we derive a generalization bound whose shift penalty admits a factorization into (i) a domain discrepancy term, (ii) a spectral-geometry term summarized by the accessible truncated spectrum, and (iii) an amplitude term that aggregates convergence and construction-stability effects. We empirically verify the insights on these terms in both real data and simulations.

Designing efficient reward functions for low-level control tasks is a challenging problem. Recent research aims to reduce reliance on expert experience by using Large Language Models (LLMs) with task information to generate dense reward functions. These methods typically rely on training results as feedback, iteratively generating new reward functions with greedy or evolutionary algorithms. However, they suffer from poor utilization of historical feedback and inefficient search, resulting in limited improvements in complex control tasks. To address this challenge, we propose RF-Agent, a framework that treats LLMs as language agents and frames reward function design as a sequential decision-making process, enhancing optimization through better contextual reasoning. RF-Agent integrates Monte Carlo Tree Search (MCTS) to manage the reward design and optimization process, leveraging the multi-stage contextual reasoning ability of LLMs. This approach better utilizes historical information and improves search efficiency to identify promising reward functions. Outstanding experimental results in 17 diverse low-level control tasks demonstrate the effectiveness of our method. The source code is available at https://github.com/deng-ai-lab/RF-Agent.

Cross-domain intrusion detection remains a critical challenge due to significant variability in network traffic characteristics and feature distributions across environments. This study evaluates the transferability of three widely used flow-based feature sets (Argus, Zeek and CICFlowMeter) across four widely used datasets representing heterogeneous IoT and Industrial IoT network conditions. Through extensive experiments, we evaluate in- and cross-domain performance across multiple classification models and analyze feature importance using SHapley Additive exPlanations (SHAP). Our results show that models trained on one domain suffer significant performance degradation when applied to a different target domain, reflecting the sensitivity of IoT intrusion detection systems to distribution shifts. Furthermore, the results evidence that the choice of classification algorithm and feature representations significantly impact transferability. Beyond reporting performance differences and thorough analysis of the transferability of features and feature spaces, we provide practical guidelines for feature engineering to improve robustness under domain variability. Our findings suggest that effective intrusion detection requires both high in-domain performance and resilience to cross-domain variability, achievable through careful feature space design, appropriate algorithm selection and adaptive strategies.

A key challenge for autonomous driving lies in maintaining real-time situational awareness regarding surrounding obstacles under strict latency constraints. The high processing requirements coupled with limited onboard computational resources can cause delay issues, particularly in complex urban settings. To address this, we propose leveraging Vehicle-to-Everything (V2X) communication to partially offload processing to the cloud, where compute resources are abundant, thus reducing overall latency. Our approach utilizes transformer-based models to fuse multi-camera sensor data into a comprehensive Bird's-Eye View (BEV) representation, enabling accurate 360-degree 3D object detection. The computation is dynamically split between the vehicle and the cloud based on the number of layers processed locally and the quantization level of the features. To further reduce network load, we apply feature vector clipping and compression prior to transmission. In a real-world experimental evaluation, our hybrid strategy achieved a 72 \% reduction in end-to-end latency compared to a traditional onboard solution. To adapt to fluctuating network conditions, we introduce a dynamic optimization algorithm that selects the split point and quantization level to maximize detection accuracy while satisfying real-time latency constraints. Trace-based evaluation under realistic bandwidth variability shows that this adaptive approach improves accuracy by up to 20 \% over static parameterization with the same latency performance.

Software engineering agents (SWE) are improving rapidly, with recent gains largely driven by reinforcement learning (RL). However, RL training is constrained by the scarcity of large-scale task collections with reproducible execution environments and reliable test suites. Although a growing number of benchmarks have emerged, datasets suitable for training remain limited in scale and diversity or often target a limited set of high-resource language ecosystems. We introduce SWE-rebench V2, a language-agnostic automated pipeline for harvesting executable real-world SWE tasks and constructing RL training environments at scale. The pipeline synthesizes repository-specific installation and test procedures via an interactive setup agent, and filters unsound instances using an ensemble of LLM judges, validated against human-verified SWE-bench annotations. Using this pipeline, we construct a dataset of 32,000+ tasks spanning 20 languages and 3,600+ repositories, with pre-built images for reproducible execution. To further scale training data, we additionally release 120,000+ tasks with installation instructions, fail-to-pass tests and rich metadata, where the problem statement is generated based on the original pull request description. We validate the collected instances through a diagnostic study that covers a subset of tasks in five programming languages across seven popular models, and provide instance-level metadata that flags common confounders such as overly restrictive tests and underspecified descriptions. We release the datasets, the collection and execution code, and associated artifacts to enable large-scale training of SWE agents across diverse languages and repositories.

Multi-agent debate (MAD) systems leverage collective intelligence to enhance reasoning capabilities, yet existing approaches struggle to simultaneously optimize accuracy, consensus formation, and computational efficiency. Static topology methods lack adaptability to task complexity variations, while external LLM-based coordination risks introducing privileged knowledge that compromises debate neutrality. This work presents RUMAD (Reinforcement-Unifying Multi-Agent Debate), a novel framework that formulates dynamic communication topology control in MAD as a reinforcement learning (RL) problem. RUMAD employs a content-agnostic observation scheme that captures high-level debate dynamics avoiding access to raw agent reasoning content. RUMAD uses a multi-objective reward to model solution quality, cohesion and efficiency. A PPO-trained controller dynamically adjusts edge weights in the communication graph, while a dual-threshold mechanism enables fine-grained control over both agent activation and information visibility. Experimental evaluation across MMLU, GSM8K, and GPQA benchmarks demonstrates that RUMAD achieves substantial efficiency gains, reducing token costs by over 80\%, while still improving reasoning accuracy compared to single LLM model and multiple MAD baselines. Notably, RUMAD trained exclusively on MMLU exhibits robust zero-shot generalization to out-of-domain (OOD) tasks, indicating that the learned communication strategies capture task-independent principles of effective multi-agent coordination. These results establish RUMAD as a efficient and robust approach for deploying multi-agent reasoning application with practical resource constraints.

With the aim of detecting AI-generated images and identifying the specific models responsible for their generation, we propose a multi-modal multi-task model. The model leverages pre-trained BERT and CLIP Vision encoders for text and image feature extraction, respectively, and employs cross-modal feature fusion with a tailored multi-task loss function. Additionally, a pseudo-labeling-based data augmentation strategy was utilized to expand the training dataset with high-confidence samples. The model achieved fifth place in both Tasks A and B of the `CT2: AI-Generated Image Detection' competition, with F1 scores of 83.16\% and 48.88\%, respectively. These findings highlight the effectiveness of the proposed architecture and its potential for advancing AI-generated content detection in real-world scenarios. The source code for our method is published on https://github.com/xxxxxxxxy/AIGeneratedImageDetection.

This paper studies a downlink multi-user multiple-input multiple-output (MU-MIMO) system, where the precoding matrix is computed at a baseband unit (BBU) and then transmitted to the remote antenna array over a limited-capacity digital fronthaul. The limited bit resolution of the fronthaul introduces quantization effects that are explicitly modeled. We propose a novel sum rate maximization framework that directly incorporates the quantizer's constraints into the precoding design. The resulting maximization problem, a non-convex mixed-integer program, is addressed using a new iterative algorithm inspired by the weighted minimum mean square error (WMMSE) methodology. The precoding optimization subproblem is reformulated as an integer least-squares problem and solved using a novel sphere decoding (SD) algorithm. Additionally, a low-complexity expectation propagation (EP)-based method is introduced to enable the practical implementation of quantized precoding in MU-massive MIMO (MU-mMIMO) systems. Furthermore, numerical evaluations demonstrate that the proposed precoding schemes outperform conventional approaches that optimize infinite-resolution precoding followed by element-wise quantization. We also propose a heuristic quantization-aware precoding method with comparable complexity to the baseline but superior performance. In particular, the EP-based approach offers near-optimal performance with substantial complexity reduction, making it well-suited for real-time MU-mMIMO applications.

This paper proposes a novel distributed semismooth Newton based augmented Lagrangian method for solving a class of optimization problems over networks, where the global objective is defined as the sum of locally held cost functions, and communication is restricted to neighboring agents. Specifically, we employ the augmented Lagrangian method to solve an equivalently reformulated constrained version of the original problem. Each resulting subproblem is solved inexactly via a distributed semismooth Newton method. By fully leveraging the structure of the generalized Hessian, a distributed accelerated proximal gradient method is proposed to compute the Newton direction efficiently, eliminating the need to communicate with full Hessian matrices. Theoretical results are also obtained to guarantee the convergence of the proposed algorithm. Numerical experiments demonstrate the efficiency and superiority of our algorithm compared to state-of-the-art distributed algorithms.

Automatic sleep stage scoring is crucial for the diagnosis and treatment of sleep disorders. Although deep learning models have advanced the field, many existing models are computationally demanding and designed for single-channel electroencephalography (EEG), limiting their practicality for multimodal polysomnography (PSG) data. To overcome this, we propose ULW-SleepNet, an ultra-lightweight multimodal sleep stage scoring framework that efficiently integrates information from multiple physiological signals. ULW-SleepNet incorporates a novel Dual-Stream Separable Convolution (DSSC) Block, depthwise separable convolutions, channel-wise parameter sharing, and global average pooling to reduce computational overhead while maintaining competitive accuracy. Evaluated on the Sleep-EDF-20 and Sleep-EDF-78 datasets, ULW-SleepNet achieves accuracies of 86.9% and 81.4%, respectively, with only 13.3K parameters and 7.89M FLOPs. Compared to state-of-the-art methods, our model reduces parameters by up to 98.6% with only marginal performance loss, demonstrating its strong potential for real-time sleep monitoring on wearable and IoT devices. The source code for this study is publicly available at https://github.com/wzw999/ULW-SLEEPNET.

The rapid expansion of Industrial IoT (IIoT) systems has amplified security challenges, as heterogeneous devices and dynamic traffic patterns increase exposure to sophisticated and previously unseen cyberattacks. Traditional intrusion detection systems often struggle in such environments due to their reliance on extensive labeled data and limited ability to detect new threats. To address these challenges, we propose MI$^2$DAS, a multi-layer intrusion detection framework that integrates anomaly-based hierarchical traffic pooling, open-set recognition to distinguish between known and unknown attacks and incremental learning for adapting to novel attack types with minimal labeling. Experiments conducted on the Edge-IIoTset dataset demonstrate strong performance across all layers. In the first layer, GMM achieves superior normal-attack discrimination (accuracy = 0.953, TPR = 1.000). In open-set recognition, GMM attains a recall of 0.813 for known attacks, while LOF achieves 0.882 recall for unknown attacks. For fine-grained classification of known attacks, Random Forest achieves a macro-F1 of 0.941. Finally, the incremental learning module maintains robust performance when incorporation novel attack classes, achieving a macro-F1 of 0.8995. These results showcase MI$^2$DAS as an effective, scalable and adaptive framework for enhancing IIoT security against evolving threats.

Chinese text correction has traditionally focused on spelling and grammar, while factual error correction is usually treated separately. However, in paragraph-level Chinese professional writing, linguistic (word/grammar/punctuation) and factual errors frequently co-occur and interact, making unified correction both necessary and challenging. This paper introduces CLFEC (Chinese Linguistic & Factual Error Correction), a new task for joint linguistic and factual correction. We construct a mixed, multi-domain Chinese professional writing dataset spanning current affairs, finance, law, and medicine. We then conduct a systematic study of LLM-based correction paradigms, from prompting to retrieval-augmented generation (RAG) and agentic workflows. The analysis reveals practical challenges, including limited generalization of specialized correction models, the need for evidence grounding for factual repair, the difficulty of mixed-error paragraphs, and over-correction on clean inputs. Results further show that handling linguistic and factual Error within the same context outperform decoupled processes, and that agentic workflows can be effective with suitable backbone models. Overall, our dataset and empirical findings provide guidance for building reliable, fully automatic proofreading systems in industrial settings.

Recent work applies Large Language Models (LLMs) to source-code vulnerability detection, but most evaluations still rely on random train-test splits that ignore time and overestimate real-world performance. In practice, detectors are deployed on evolving code bases and must recognise future vulnerabilities under temporal distribution shift. This paper investigates continual fine-tuning of a decoder-style language model (microsoft/phi-2 with LoRA) on a CVE-linked dataset spanning 2018-2024, organised into bi-monthly windows. We evaluate eight continual learning strategies, including window-only and cumulative training, replay-based baselines and regularisation-based variants. We propose Hybrid Class-Aware Selective Replay (Hybrid-CASR), a confidence-aware replay method for binary vulnerability classification that prioritises uncertain samples while maintaining a balanced ratio of VULNERABLE and FIXED functions in the replay buffer. On bi-monthly forward evaluation Hybrid-CASR achieves a Macro-F1 of 0.667, improving on the window-only baseline (0.651) by 0.016 with statistically significant gains ($p = 0.026$) and stronger backward retention (IBR@1 of 0.741). Hybrid-CASR also reduces training time per window by about 17 percent compared to the baseline, whereas cumulative training delivers only a minor F1 increase (0.661) at a 15.9-fold computational cost. Overall, the results show that selective replay with class balancing offers a practical accuracy-efficiency trade-off for LLM-based temporal vulnerability detection under continuous temporal drift.

Dynamic programming (DP) algorithms, such as All-Pairs Shortest Path (APSP) and genomic sequence alignment, are fundamental to many scientific domains but are severely bottlenecked by data movement on conventional architectures. While Processing-in-Memory (PIM) offers a promising solution, existing accelerators often address only a fraction of the work-flow, creating new system-level bottlenecks in host-accelerator communication and off-chip data streaming. In this work, we propose GenDRAM, a massively parallel PIM accelerator that overcomes these limitations. GenDRAM leverages the immense capacity and internal bandwidth of monolithic 3D DRAM(M3D DRAM) to integrate entire data-intensive pipelines, such as the full genomics workflow from seeding to alignment, onto a single heterogeneous chip. At its core is a novel architecture featuring specialized Search PUs for memory-intensive tasks and universal, multiplier-less Compute PUs for diverse DP calculations. This is enabled by a 3D-aware data mapping strategy that exploits the tiered latency of M3D DRAM for performance optimization. Through comprehensive simulation, we demonstrate that GenDRAM achieves a transformative performance leap, outperforming state-of-the-art GPU systems by over 68x on APSP and over 22x on the end-to-end genomics pipeline.

In federated learning (FL), multi-step local updates and data heterogeneity usually lead to sharper global minima, which degrades the performance of the global model. Popular FL algorithms integrate sharpness-aware minimization (SAM) into local training to address this issue. However, in the high data heterogeneity setting, the flatness in local training does not imply the flatness of the global model. Therefore, minimizing the sharpness of the local loss surfaces on the client data does not enable the effectiveness of SAM in FL to improve the generalization ability of the global model. We define the \textbf{flatness distance} to explain this phenomenon. By rethinking the SAM in FL and theoretically analyzing the \textbf{flatness distance}, we propose a novel \textbf{FedNSAM} algorithm that accelerates the SAM algorithm by introducing global Nesterov momentum into the local update to harmonize the consistency of global and local flatness. \textbf{FedNSAM} uses the global Nesterov momentum as the direction of local estimation of client global perturbations and extrapolation. Theoretically, we prove a tighter convergence bound than FedSAM by Nesterov extrapolation. Empirically, we conduct comprehensive experiments on CNN and Transformer models to verify the superior performance and efficiency of \textbf{FedNSAM}. The code is available at https://github.com/junkangLiu0/FedNSAM.

We present GLUScope, an open-source tool for analyzing neurons in Transformer-based language models, intended for interpretability researchers. We focus on more recent models than previous tools do; specifically we consider gated activation functions such as SwiGLU. This introduces a new challenge: understanding positive activations is not enough. Instead, both the gate and the in activation of a neuron can be positive or negative, leading to four different possible sign combinations that in some cases have quite different functionalities. Accordingly, for any neuron, our tool shows text examples for each of the four sign combinations, and indicates how often each combination occurs. We describe examples of how our tool can lead to novel insights. A demo is available at https: //sjgerstner.github.io/gluscope.

Longitudinal electronic health record (EHR) data are often left-censored, making diagnosis records incomplete and unreliable for determining disease onset. In contrast, outpatient prescriptions form renewal-based trajectories that provide a continuous signal of disease management. We propose a probabilistic framework to infer chronic treatment onset by modeling prescription dynamics as a renewal process and detecting transitions from sporadic to sustained therapy via change-point detection between a baseline Poisson (sporadic prescribing) regime and a regime-specific Weibull (sustained therapy) renewal model. Using a nationwide ePrescription dataset of 2.4 million individuals, we show that the approach yields more temporally plausible onset estimates than naive rule-based triggering, substantially reducing implausible early detections under strong left censoring. Detection performance varies across diseases and is strongly associated with prescription density, highlighting both the strengths and limits of treatment-based onset inference.

Given a set of trajectories demonstrating the execution of a task safely in a constrained MDP with observable rewards but with unknown constraints and non-observable costs, we aim to find a policy that maximizes the likelihood of demonstrated trajectories trading the balance between being conservative and increasing significantly the likelihood of high-rewarding trajectories but with potentially unsafe steps. Having these objectives, we aim towards learning a policy that maximizes the probability of the most $promising$ trajectories with respect to the demonstrations. In so doing, we formulate the ``promise" of individual state-action pairs in terms of $Q$ values, which depend on task-specific rewards as well as on the assessment of states' safety, mixing expectations in terms of rewards and safety. This entails a safe Q-learning perspective of the inverse learning problem under constraints: The devised Safe $Q$ Inverse Constrained Reinforcement Learning (SafeQIL) algorithm is compared to state-of-the art inverse constraint reinforcement learning algorithms to a set of challenging benchmark tasks, showing its merits.

We investigate the theoretical aspects of offline reinforcement learning (RL) under general function approximation. While prior works (e.g., Xie et al., 2021) have established the theoretical foundations of learning a good policy from offline data via pessimism, existing algorithms that are computationally tractable (often in an oracle-efficient sense), such as PSPI, only apply to finite and small action spaces. Moreover, these algorithms rely on state-wise mirror descent and require actors to be implicitly induced from the critic functions, failing to accommodate standalone policy parameterization which is ubiquitous in practice. In this work, we address these limitations and extend the theoretical guarantees to parameterized policy classes over large or continuous action spaces. When extending mirror descent to parameterized policies, we identify contextual coupling as the core difficulty, and show how connecting mirror descent to natural policy gradient leads to novel analyses, guarantees, and algorithmic insights, including a surprising unification between offline RL and imitation learning.

Explaining opaque Machine Learning (ML) models has become an increasingly important challenge. However, current eXplanation in AI (XAI) methods suffer several shortcomings, including insufficient abstraction, limited user interactivity, and inadequate integration of symbolic knowledge. We propose ReasonX, an explanation tool based on expressions (or, queries) in a closed algebra of operators over theories of linear constraints. ReasonX provides declarative and interactive explanations for decision trees, which may represent the ML models under analysis or serve as global or local surrogate models for any black-box predictor. Users can express background or common sense knowledge as linear constraints. This allows for reasoning at multiple levels of abstraction, ranging from fully specified examples to under-specified or partially constrained ones. ReasonX leverages Mixed-Integer Linear Programming (MILP) to reason over the features of factual and contrastive instances. We present here the architecture of ReasonX, which consists of a Python layer, closer to the user, and a Constraint Logic Programming (CLP) layer, which implements a meta-interpreter of the query algebra. The capabilities of ReasonX are demonstrated through qualitative examples, and compared to other XAI tools through quantitative experiments.

Pre-trained perception models excel in generic image domains but degrade significantly in novel environments like indoor scenes. The conventional remedy is fine-tuning on downstream data which incurs catastrophic forgetting of prior knowledge and demands costly, scene-specific annotations. We propose a paradigm shift through Sea$^2$ (See, Act, Adapt): rather than adapting the perception modules themselves, we adapt how they are deployed through an intelligent pose-control agent. Sea$^2$ keeps all perception modules frozen, requiring no downstream labels during training, and uses only scalar perceptual feedback to navigate the agent toward informative viewpoints. Specially, we transform a vision-language model (VLM) into a low-level pose controller through a two-stage training pipeline: first fine-tuning it on rule-based exploration trajectories that systematically probe indoor scenes, and then refining the policy via unsupervised reinforcement learning that constructs rewards from the perception module's outputs and confidence. Unlike prior active perception methods that couple exploration with specific models or collect data for retraining them, Sea$^2$ directly leverages off-the-shelf perception models for various tasks without the need for retraining. We conducted experiments on three visual perception tasks, including visual grounding, segmentation and 3D box estimation, with performance improvements of 13.54%, 15.92% and 27.68% respectively on dataset ReplicaCAD.

Reinforcement learning (RL) actor-critic algorithms enable autonomous learning but often require a large number of environment interactions, which limits their applicability in robotics. Leveraging expert data can reduce the number of required environment interactions. A common approach is actor pretraining, where the actor network is initialized via behavioral cloning on expert demonstrations and subsequently fine-tuned with RL. In contrast, the initialization of the critic network has received little attention, despite its central role in policy optimization. This paper proposes a pretraining approach for actor-critic algorithms like Proximal Policy Optimization (PPO) that uses expert demonstrations to initialize both networks. The actor is pretrained via behavioral cloning, while the critic is pretrained using returns obtained from rollouts of the pretrained policy. The approach is evaluated on 15 simulated robotic manipulation and locomotion tasks. Experimental results show that actor-critic pretraining improves sample efficiency by 86.1% on average compared to no pretraining and by 30.9% to actor-only pretraining.

Multimodal Large Language Models (MLLMs) have shown remarkable progress in visual reasoning and understanding tasks but still struggle to capture the complexity and subjectivity of human emotions. Existing approaches based on supervised fine-tuning often suffer from limited generalization and poor interpretability, while reinforcement learning methods such as Group Relative Policy Optimization fail to align with the intrinsic characteristics of emotional cognition. To address these challenges, we propose Reflective Reinforcement Learning for Emotional Reasoning (EMO-R3), a framework designed to enhance the emotional reasoning ability of MLLMs. Specifically, we introduce Structured Emotional Thinking to guide the model to perform step-by-step emotional reasoning in a structured and interpretable manner, and design a Reflective Emotional Reward that enables the model to re-evaluate its reasoning based on visual-text consistency and emotional coherence. Extensive experiments demonstrate that EMO-R3 significantly improves both the interpretability and emotional intelligence of MLLMs, achieving superior performance across multiple visual emotional understanding benchmarks.

Causal discovery has achieved substantial theoretical progress, yet its deployment in large-scale longitudinal systems remains limited. A key obstacle is that operational data are generated under institutional workflows whose induced partial orders are rarely formalized, enlarging the admissible graph space in ways inconsistent with the recording process. We characterize a workflow-induced constraint class for longitudinal causal discovery that restricts the admissible directed acyclic graph space through protocol-derived structural masks and timeline-aligned indexing. Rather than introducing a new optimization algorithm, we show that explicitly encoding workflow-consistent partial orders reduces structural ambiguity, especially in mixed discrete--continuous panels where within-time orientation is weakly identified. The framework combines workflow-derived admissible-edge constraints, measurement-aligned time indexing and block structure, bootstrap-based uncertainty quantification for lagged total effects, and a dynamic representation supporting intervention queries. In a nationwide annual health screening cohort in Japan with 107,261 individuals and 429,044 person-years, workflow-constrained longitudinal LiNGAM yields temporally consistent within-time substructures and interpretable lagged total effects with explicit uncertainty. Sensitivity analyses using alternative exposure and body-composition definitions preserve the main qualitative patterns. We argue that formalizing workflow-derived constraint classes improves structural interpretability without relying on domain-specific edge specification, providing a reproducible bridge between operational workflows and longitudinal causal discovery under standard identifiability assumptions.

Machine unlearning for large language models often faces a privacy dilemma in which strict constraints prohibit sharing either the server's parameters or the client's forget set. To address this dual non-disclosure constraint, we propose MPU, an algorithm-agnostic privacy-preserving Multiple Perturbed Copies Unlearning framework that primarily introduces two server-side modules: Pre-Process for randomized copy generation and Post-Process for update aggregation. In Pre-Process, the server distributes multiple perturbed and reparameterized model instances, allowing the client to execute unlearning locally on its private forget set without accessing the server's exact original parameters. After local unlearning, the server performs Post-Process by inverting the reparameterization and aggregating updates with a harmonic denoising procedure to alleviate the impact of perturbation. Experiments with seven unlearning algorithms show that MPU achieves comparable unlearning performance to noise-free baselines, with most algorithms' average degradation well below 1% under 10% noise, and can even outperform the noise-free baseline for some algorithms under 1% noise. Code is available at https://github.com/Tristan-SHU/MPU.

Structured deep model compression methods are hardware-friendly and substantially reduce memory and inference costs. However, under aggressive compression, the resulting accuracy degradation often necessitates post-compression finetuning, which can be impractical due to missing labeled data or high training cost. We propose post-hoc blockwise compensation, called GRAIL, a simple zero-finetuning step applied after model compression that restores each block's input-output behavior using a small calibration set. The method summarizes hidden activations via a Gram matrix and applies ridge regression to linearly reconstruct the original hidden representation from the reduced one. The resulting reconstruction map is absorbed into the downstream projection weights, while the upstream layer is compressed. The approach is selector-agnostic (Magnitude, Wanda, Gram-based selection, or folding), data-aware (requiring only a few forward passes without gradients or labels), and recovers classic pruning or folding when the Gram matrix is near identity, indicating weak inter-channel correlations. Across ResNets, ViTs, and decoder-only LLMs, GRAIL consistently improves accuracy or perplexity over data-free and data-aware pruning or folding baselines in practical compression regimes, with manageable overhead and no backpropagation. The code is available at https://github.com/TWWinde/GRAIL.

Diffusion-based large language models (dLLMs) have shown promising performance across various reasoning tasks, establishing themselves as an alternative to autoregressive large language models (LLMs). Unlike autoregressive LLMs that generate one token per step based on all previous tokens, dLLMs theoretically enable parallel generation of multiple tokens at each decoding step. However, recent dLLMs still favor one-token-per-step generation in practice, as directly decoding multiple masked tokens often leads to degraded generation quality and stability. This reveals a substantial gap between the theoretical parallelism and practical performance of dLLMs. To bridge this gap, we introduce an adaptive parallel decoding approach, namely DiCo, which features a three-phase divide-and-conquer paradigm to unleash the inherent parallelism of dLLMs. During the Divide phase, DiCo first explores the input masked sequence and identifies masked tokens as seed tokens, which are then expanded to construct a set of local clusters. During the Conquer phase, DiCo performs parallel decoding across different local clusters constructed in the Divide phase. The divide-and-conquer process repeatedly alternates between the Divide and Conquer phases until convergence. During the Finalize phase, DiCo decodes the remaining few masked tokens using an effective fine-grained compound decoding scheme to finalize the generation. Extensive experiments demonstrate that DiCo can achieve significant inference speedups while maintaining competitive generation quality.

The performance of search algorithms for grid-based pathfinding, e.g. A*, critically depends on the heuristic function that is used to focus the search. Recent studies have shown that informed heuristics that take the positions/shapes of the obstacles into account can be approximated with the deep neural networks. Unfortunately, the existing learning-based approaches mostly rely on the assumption that training and test grid maps are drawn from the same distribution (e.g., city maps, indoor maps, etc.) and perform poorly on out-of-distribution tasks. This naturally limits their application in practice when often a universal solver is needed that is capable of efficiently handling any problem instance. In this work, we close this gap by designing an universal heuristic predictor: a model trained once, but capable of generalizing across a full spectrum of unseen tasks. Our extensive empirical evaluation shows that the suggested approach halves the computational effort of A* by up to a factor of 2.2, while still providing solutions within 3% of the optimal cost on average altogether on the tasks that are completely different from the ones used for training $\unicode{x2013}$ a milestone reached for the first time by a learnable solver.

Point cloud processing is a computational bottleneck in autonomous driving systems, especially for real-time applications, while energy efficiency remains a critical system constraint. This work presents FPPS, an FPGA-accelerated point cloud processing system designed to optimize the iterative closest point (ICP) algorithm, a classic cornerstone of 3D localization and perception pipelines. Evaluated on the widely used KITTI benchmark dataset, the proposed system achieves up to 35$\times$ (and a runtime-weighted average of 15.95x) speedup over a state-of-the-art CPU baseline while maintaining equivalent registration accuracy. Notably, the design improves average power efficiency by 8.58x, offering a compelling balance between performance and energy consumption. These results position FPPS as a viable solution for resource-constrained embedded autonomous platforms where both latency and power are key design priorities.

We investigate the identifiability of nonlinear Canonical Correlation Analysis (CCA) in a multi-view setup, where each view is generated by an unknown nonlinear map applied to a linear mixture of shared latents and view-private noise. Rather than attempting exact unmixing, a problem proven to be ill-posed, we instead reframe multi-view CCA as a basis-invariant subspace identification problem. We prove that, under suitable latent priors and spectral separation conditions, multi-view CCA recovers the pairwise correlated signal subspaces up to view-wise orthogonal ambiguity. For $N \geq 3$ views, the objective provably isolates the jointly correlated subspaces shared across all views while eliminating view-private variations. We further establish finite-sample consistency guarantees by translating the concentration of empirical cross-covariances into explicit subspace error bounds via spectral perturbation theory. Experiments on synthetic and rendered image datasets validate our theoretical findings and confirm the necessity of the assumed conditions.

Foundation models have transformed domains from language to genomics by learning general-purpose representations from large-scale, heterogeneous data. We introduce TradeFM, a 524M-parameter generative Transformer that brings this paradigm to market microstructure, learning directly from billions of trade events across >9K equities. To enable cross-asset generalization, we develop scale-invariant features and a universal tokenization scheme that map the heterogeneous, multi-modal event stream of order flow into a unified discrete sequence -- eliminating asset-specific calibration. Integrated with a deterministic market simulator, TradeFM-generated rollouts reproduce key stylized facts of financial returns, including heavy tails, volatility clustering, and absence of return autocorrelation. Quantitatively, TradeFM achieves 2-3x lower distributional error than Compound Hawkes baselines and generalizes zero-shot to geographically out-of-distribution APAC markets with moderate perplexity degradation. Together, these results suggest that scale-invariant trade representations capture transferable structure in market microstructure, opening a path toward synthetic data generation, stress testing, and learning-based trading agents.

This paper addresses the domain generalization (DG) problem in deep learning. While most DG methods focus on enforcing visual feature invariance, we leverage the reasoning capability of multimodal large language models (MLLMs) and explore the potential of constructing reasoning chains that derives image categories to achieve more robust predictions under domain shift. To this end, we systematically study the role of reasoning in DG using DomainBed-Reasoning, a newly constructed extension of DomainBed dataset, in which each sample is paired with class-relevant reasoning chains. Our analysis reveals two key challenges: (i) fine-tuning MLLMs with reasoning chains for classification is more challenging than direct label supervision, since the model must optimize complex reasoning sequences before label prediction; and (ii) mismatches in reasoning patterns between supervision signals and fine-tuned MLLMs lead to a trade-off between semantic richness (informative but harder to optimize) and optimization efficiency (easier to optimize but less informative). To address these issues, we propose RD-MLDG (Reasoning-Driven Multimodal LLM for Domain Generalization), a framework with two components: (i) MTCT (Multi-Task Cross-Training), which introduces an additional direct classification pathway to guide reasoning supervision; and (ii) SARR (Self-Aligned Reasoning Regularization), which preserves the semantic richness of reasoning chains while mitigating reasoning-pattern mismatches via iterative self-labeling. Experiments on standard DomainBed datasets (PACS, VLCS, OfficeHome, TerraInc) demonstrate that RD-MLDG achieves state-of-the-art performances, highlighting reasoning as a promising complementary signal for robust out-of-domain generalization.

Generative models have gained significant traction in offline reinforcement learning (RL) due to their ability to model complex trajectory distributions. However, existing generation-based approaches still struggle with long-horizon tasks characterized by sparse rewards. Some hierarchical generation methods have been developed to mitigate this issue by decomposing the original problem into shorter-horizon subproblems using one policy and generating detailed actions with another. While effective, these methods often overlook the multi-scale temporal structure inherent in trajectories, resulting in suboptimal performance. To overcome these limitations, we propose MAGE, a Multi-scale Autoregressive GEneration-based offline RL method. MAGE incorporates a condition-guided multi-scale autoencoder to learn hierarchical trajectory representations, along with a multi-scale transformer that autoregressively generates trajectory representations from coarse to fine temporal scales. MAGE effectively captures temporal dependencies of trajectories at multiple resolutions. Additionally, a condition-guided decoder is employed to exert precise control over short-term behaviors. Extensive experiments on five offline RL benchmarks against fifteen baseline algorithms show that MAGE successfully integrates multi-scale trajectory modeling with conditional guidance, generating coherent and controllable trajectories in long-horizon sparse-reward settings.

Optical design is the process of configuring optical elements to precisely manipulate light for high-fidelity imaging. It is inherently a highly non-convex optimization problem that relies heavily on human heuristic expertise and domain-specific knowledge. While Large Language Models (LLMs) possess extensive optical knowledge, their capabilities in leveraging the knowledge in designing lens system remain significantly constrained. This work represents the first attempt to employ LLMs in the field of optical design. We bridge the expertise gap by enabling users without formal optical training to successfully develop functional lens systems. Concretely, we curate a comprehensive dataset, named OptiDesignQA, which encompasses both classical lens systems sourced from standard optical textbooks and novel configurations generated by automated design algorithms for training and evaluation. Furthermore, we inject domain-specific optical expertise into the LLM through a hybrid objective of full-system synthesis and lens completion. To align the model with optical principles, we employ Group Relative Policy Optimization Done Right (DrGRPO) guided by Optical Lexicographic Reward for physics-driven policy alignment. This reward system incorporates structural format rewards, physical feasibility rewards, light-manipulation accuracy, and LLM-based heuristics. Finally, our model integrates with specialized optical optimization routines for end-to-end fine-tuning and precision refinement. We benchmark our proposed method against both traditional optimization-based automated design algorithms and LLM counterparts, and experimental results show the superiority of our method.

Modern cloud servers routinely co-locate multiple latency-sensitive microservice instances to improve resource efficiency. However, the diversity of microservice behaviors, coupled with mutual performance interference under simultaneous multithreading (SMT), makes large-scale placement increasingly complex. Existing interference aware schedulers and isolation techniques rely on coarse core-level profiling or static resource partitioning, leaving asymmetric hyperthread-level heterogeneity and SMT contention dynamics largely unmodeled. We present Hestia, a hyperthread-level, interference-aware scheduling framework powered by self-attention. Through an extensive analysis of production traces encompassing 32,408 instances across 3,132 servers, we identify two dominant contention patterns -- sharing-core (SC) and sharing-socket (SS) -- and reveal strong asymmetry in their impact. Guided by these insights, Hestia incorporates (1) a self-attention-based CPU usage predictor that models SC/SS contention and hardware heterogeneity, and (2) an interference scoring model that estimates pairwise contention risks to guide scheduling decisions. We evaluate Hestia through large-scale simulation and a real production deployment. Hestia reduces the 95th-percentile service latency by up to 80\%, lowers overall CPU consumption by 2.3\% under the same workload, and surpasses five state-of-the-art schedulers by up to 30.65\% across diverse contention scenarios.

This study addresses the issues of semantic entanglement, unclear label structure, and insufficient feature representation in few-shot text classification, and proposes an optimization framework based on structured prompts to enhance semantic understanding and task adaptation under low-resource conditions. The framework first uses a pretrained language model to encode the input text and obtain basic semantic representations. It then introduces structured prompts composed of multi-dimensional semantic factors and integrates them with text features through a learnable combination mechanism, which forms task-related representations with clear boundaries in the latent space. To further strengthen the consistency between text representations and label semantics, the method constructs a structured label embedding matrix and employs a cross-space alignment mechanism to ensure stable matching between textual features and label attributes. In addition, the model applies prompt orthogonality constraints and a joint optimization objective to maintain independence across different semantic factors in the prompts, allowing the structured prompts to provide transparent and controllable guidance for classification decisions. Three types of sensitivity experiments, including learning rate sensitivity, prompt length sensitivity, and data scale sensitivity, are designed to evaluate the stability and robustness of the framework under different conditions. Experimental results show that the proposed structured prompt optimization framework effectively alleviates semantic conflicts and label ambiguity in few-shot text classification. It significantly improves performance on accuracy, precision, recall, and AUC, and demonstrates strong cross-task applicability.

Predictive hotspot mapping is an important problem in crime prediction and control. An accurate hotspot mapping helps in appropriately targeting the available resources to manage crime in cities. With an aim to make data-driven decisions and automate policing and patrolling operations, police departments across the world are moving towards predictive approaches relying on historical data. In this paper, we create a non-parametric model using a spatio-temporal kernel density formulation for the purpose of crime prediction based on historical data. The proposed approach is also able to incorporate expert inputs coming from humans through alternate sources. The approach has been extensively evaluated in a real-world setting by collaborating with the Delhi police department to make crime predictions that would help in effective assignment of patrol vehicles to control street crime. The results obtained in the paper are promising and can be easily applied in other settings. We release the algorithm and the dataset (masked) used in our study to support future research that will be useful in achieving further improvements.

Cross-domain reinforcement learning (RL) aims to learn transferable policies under dynamics shifts between source and target domains. A key challenge lies in the lack of target-domain environment interaction and reward supervision, which prevents direct policy learning. To address this challenge, we propose Bridging Dynamics Gaps for Cross-Domain Reinforcement Learning (BDGxRL), a novel framework that leverages Diffusion Schrödinger Bridge (DSB) to align source transitions with target-domain dynamics encoded in offline demonstrations. Moreover, we introduce a reward modulation mechanism that estimates rewards based on state transitions, applying to DSB-aligned samples to ensure consistency between rewards and target-domain dynamics. BDGxRL performs target-oriented policy learning entirely within the source domain, without access to the target environment or its rewards. Experiments on MuJoCo cross-domain benchmarks demonstrate that BDGxRL outperforms state-of-the-art baselines and shows strong adaptability under transition dynamics shifts.

PoCo is a technique that aims to enhance modern coverage-based seed selection (CSS) techniques (such as afl-cmin) by gradually removing obstacle conditional statements and conducting deeper seed selection.

One-stream Transformer-based trackers achieve advanced performance in visual object tracking but suffer from significant computational overhead that hinders real-time deployment. While token pruning offers a path to efficiency, existing methods are fragmented. They typically prune the search region, dynamic template, and static template in isolation, overlooking critical inter-component dependencies, which yields suboptimal pruning and degraded accuracy. To address this, we introduce UTPTrack, a simple and Unified Token Pruning framework that, for the first time, jointly compresses all three components. UTPTrack employs an attention-guided, token type-aware strategy to holistically model redundancy, a design that seamlessly supports unified tracking across multimodal and language-guided tasks within a single model. Extensive evaluations on 10 benchmarks demonstrate that UTPTrack achieves a new state-of-the-art in the accuracy-efficiency trade-off for pruning-based trackers, pruning 65.4% of vision tokens in RGB-based tracking and 67.5% in unified tracking while preserving 99.7% and 100.5% of baseline performance, respectively. This strong performance across both RGB and multimodal scenarios underlines its potential as a robust foundation for future research in efficient visual tracking. Code will be released at https://github.com/EIT-NLP/UTPTrack.

Recent advancements in Multimodal Large Language Models (MLLMs) pursue omni-perception capabilities, yet integrating robust sensory grounding with complex reasoning remains a challenge, particularly for underrepresented regions. In this report, we introduce the research preview of MERaLiON2-Omni (Alpha), a 10B-parameter multilingual omni-perception tailored for Southeast Asia (SEA). We present a progressive training pipeline that explicitly decouples and then integrates "System 1" (Perception) and "System 2" (Reasoning) capabilities. First, we establish a robust Perception Backbone by aligning region-specific audio-visual cues (e.g., Singlish code-switching, local cultural landmarks) with a multilingual LLM through orthogonal modality adaptation. Second, to inject cognitive capabilities without large-scale supervision, we propose a cost-effective Generate-Judge-Refine pipeline. By utilizing a Super-LLM to filter hallucinations and resolve conflicts via a consensus mechanism, we synthesize high-quality silver data that transfers textual Chain-of-Thought reasoning to multimodal scenarios. Comprehensive evaluation on our newly introduced SEA-Omni Benchmark Suite reveals an Efficiency-Stability Paradox: while reasoning acts as a non-linear amplifier for abstract tasks (boosting mathematical and instruction-following performance significantly), it introduces instability in low-level sensory processing. Specifically, we identify Temporal Drift in long-context audio, where extended reasoning desynchronizes the model from acoustic timestamps, and Visual Over-interpretation, where logic overrides pixel-level reality. This report details the architecture, the data-efficient training recipe, and a diagnostic analysis of the trade-offs between robust perception and structured reasoning.

The evaluation of large language models (LLMs) has predominantly relied on static datasets, which offer limited scalability and fail to capture the evolving reasoning capabilities of recent models. To overcome these limitations, we propose an agent-centric benchmarking paradigm that moves beyond static datasets by introducing a dynamic protocol in which autonomous agents iteratively generate, validate, and solve problems. Within this protocol, a teacher agent generates candidate problems, an orchestrator agent rigorously verifies their validity and guards against adversarial attacks, and a student agent attempts to solve the validated problems. An invalid problem is revised by the teacher agent until it passes validation. If the student correctly solves the problem, the orchestrator prompts the teacher to generate more challenging variants. Consequently, the benchmark scales in difficulty automatically as more capable agents are substituted into any role, enabling progressive evaluation of large language models without manually curated datasets. Adopting text anomaly detection as our primary evaluation format, which demands cross-sentence logical inference and resists pattern-matching shortcuts, we demonstrate that this protocol systematically exposes corner-case reasoning errors that conventional benchmarks fail to reveal. We further advocate evaluating systems along several complementary axes including cross-model pairwise performance and progress between the initial and orchestrator-finalized problems. By shifting the focus from fixed datasets to dynamic protocols, our approach offers a sustainable direction for evaluating ever-evolving language models and introduces a research agenda centered on the co-evolution of agent-centric benchmarks.

Embodied AI requires sub-second inference near the Radio Access Network (RAN), but deployments span heterogeneous tiers (on-device, RAN-edge, cloud) and must not disrupt real-time baseband processing. We report measurements from a 5G Standalone (SA) AI-RAN testbed using a fixed baseline policy for repeatability. The setup includes an on-device tier, a three-node RAN-edge cluster co-hosting a containerized 5G RAN, and a cloud tier. We find that on-device execution remains multi-second and fails to meet sub-second budgets. At the RAN edge, SLA feasibility is primarily determined by model variant choice: quantized models concentrate below 0.5\,s, while unquantized and some larger quantized models incur deadline misses due to stalls and queuing. In the cloud tier, meeting a 0.5\,s deadline is challenging on the measured WAN path (up to 32.9\% of requests complete within 0.5\,s), but all evaluated variants meet a 1.0\,s deadline (100\% within 1.0\,s). Under saturated downlink traffic and up to $N{=}20$ concurrent inference clients, Multi-Instance GPU (MIG) isolation preserves baseband timing-health proxies, supporting safe co-location under fixed partitioning.

The field of Artificial Intelligence is undergoing a transition from Generative AI -- probabilistic generation of text and images -- to Agentic AI, in which autonomous systems execute actions within external environments on behalf of users. This transition exposes a fundamental architectural mismatch: Large Language Models (LLMs) produce stochastic, unstructured outputs, whereas the backend infrastructure they must control -- databases, APIs, cloud services -- requires deterministic, schema-conformant inputs. The present paper describes the Auton Agentic AI Framework, a principled architecture for standardizing the creation, execution, and governance of autonomous agent systems. The framework is organized around a strict separation between the Cognitive Blueprint, a declarative, language-agnostic specification of agent identity and capabilities, and the Runtime Engine, the platform-specific execution substrate that instantiates and runs the agent. This separation enables cross-language portability, formal auditability, and modular tool integration via the Model Context Protocol (MCP). The paper formalizes the agent execution model as an augmented Partially Observable Markov Decision Process (POMDP) with a latent reasoning space, introduces a hierarchical memory consolidation architecture inspired by biological episodic memory systems, defines a constraint manifold formalism for safety enforcement via policy projection rather than post-hoc filtering, presents a three-level self-evolution framework spanning in-context adaptation through reinforcement learning, and describes runtime optimizations -- including parallel graph execution, speculative inference, and dynamic context pruning -- that reduce end-to-end latency for multi-step agent workflows.

In UAV dynamic decision, complex and variable hazardous factors pose severe challenges to the generalization capability of algorithms. Despite offering semantic understanding and scene generalization, Large Language Models (LLM) lack domain-specific UAV control knowledge and formal safety assurances, restricting their direct applicability. To bridge this gap, this paper proposes a train-free two-layer decision architecture based on LLMs, integrating high-level safety planning with low-level precise control. The framework introduces three key contributions: 1) A fuzzy Control Barrier Function verification mechanism for semantically-augmented actions, providing provable safety certification for LLM outputs. 2) A star-hierarchical graph-based retrieval-augmented generation system, enabling efficient, elastic, and interpretable scene adaptation. 3) Systematic experimental validation in pursuit-evasion scenarios with unknown obstacles and emergent threats, demonstrating that our SAGE-LLM maintains performance while significantly enhancing safety and generalization without online training. The proposed framework demonstrates strong extensibility, suggesting its potential for generalization to broader embodied intelligence systems and safety-critical control domains.

Large Language Model (LLM)-based agents show promise for e-commerce conversational shopping, yet existing implementations lack the interaction depth and contextual breadth required for complex product research. Meanwhile, the Deep Research paradigm, despite advancing information synthesis in web search, suffers from domain gaps when transferred to e-commerce. We propose ProductResearch, a multi-agent framework that synthesizes high-fidelity, long-horizon tool-use trajectories for training robust e-commerce shopping agents. The framework employs a User Agent to infer nuanced shopping intents from behavioral histories, and a Supervisor Agent that orchestrates iterative collaboration with a Research Agent to generate synthetic trajectories culminating in comprehensive, insightful product research reports. These trajectories are rigorously filtered and distilled through a reflective internalization process that consolidates multi-agent supervisory interactions into coherent single-role training examples, enabling effective fine-tuning of LLM agents for complex shopping inquiries. Extensive experiments show that a compact MoE model fine-tuned on our synthetic data achieves substantial improvements over its base model in response comprehensiveness, research depth, and user-perceived utility, approaching the performance of frontier proprietary deep research systems and establishing multi-agent synthetic trajectory training as an effective and scalable paradigm for enhancing LLM-based shopping assistance.

This paper presents an efficient mesh deformation method based on boundary integration and neural operators, formulating the problem as a linear elasticity boundary value problem (BVP). To overcome the high computational cost of traditional finite element methods and the limitations of existing neural operators in handling Dirichlet boundary conditions for vector fields, we introduce a direct boundary integral representation using a Dirichlet-type Green's tensor. This formulation expresses the internal displacement field solely as a function of boundary displacements, eliminating the need to solve for unknown tractions. Building on this, we design a Boundary-Integral-based Neural Operator (BINO) that learns the geometry- and material-aware Green's traction kernel. A key technical advantage of our framework is the mathematical decoupling of the physical integration process from the geometric representation via geometric descriptors. While this study primarily demonstrates robust generalization across diverse boundary conditions, the architecture inherently possesses potential for cross-geometry adaptation. Numerical experiments, including large deformations of flexible beams and rigid-body motions of NACA airfoils, confirm the model's high accuracy and strict adherence to the principles of linearity and superposition. The results demonstrate that the proposed framework ensures mesh quality and computational efficiency, providing a reliable new paradigm for parametric mesh generation and shape optimization in engineering.

LLM-powered Multi-Agent Systems (MAS) have demonstrated remarkable capabilities in complex domains but suffer from inherent fragility and opaque failure mechanisms. Existing failure attribution methods, whether relying on direct prompting, costly replays, or supervised fine-tuning, typically treat execution logs as flat sequences. This linear perspective fails to disentangle the intricate causal links inherent to MAS, leading to weak observability and ambiguous responsibility boundaries. To address these challenges, we propose CHIEF, a novel framework that transforms chaotic trajectories into a structured hierarchical causal graph. It then employs hierarchical oracle-guided backtracking to efficiently prune the search space via sybthesized virtual oracles. Finally, it implements counterfactual attribution via a progressive causal screening strategy to rigorously distinguish true root causes from propagated symptoms. Experiments on Who&When benchmark show that CHIEF outperforms eight strong and state-of-the-art baselines on both agent- and step-level accuracy. Ablation studies further confirm the critical role of each proposed module.

The quadratic computational cost of processing vision tokens in Multimodal Large Language Models (MLLMs) hinders their widespread adoption. While progressive vision token pruning offers a promising solution, current methods misinterpret shallow layer functions and use rigid schedules, which fail to unlock the full efficiency potential. To address these issues, we propose HiDrop, a framework that aligns token pruning with the true hierarchical function of MLLM layers. HiDrop features two key innovations: (1) Late Injection, which bypasses passive shallow layers to introduce visual tokens exactly where active fusion begins; and (2) Concave Pyramid Pruning with an Early Exit mechanism to dynamically adjust pruning rates across middle and deep layers. This process is optimized via an inter-layer similarity measure and a differentiable top-k operator. To ensure practical efficiency, HiDrop further incorporates persistent positional encoding, FlashAttention-compatible token selection, and parallel decoupling of vision computation to eliminate hidden overhead associated with dynamic token reduction. Extensive experiments show that HiDrop compresses about 90% visual tokens while matching the original performance and accelerating training by 1.72 times. Our work not only sets a new state-of-the-art for efficient MLLM training and inference but also provides valuable insights into the hierarchical nature of multimodal fusion. The code is released at https://github.com/EIT-NLP/HiDrop.

We study the geometry of training trajectories in small transformer models and find that parameter updates organize into a dominant drift direction with transverse residual dynamics. Using uncentered, row-normalized trajectory PCA, we show that a single direction captures a large fraction of cumulative parameter movement early in training, while remaining components encode oscillatory behavior in auxiliary probe performance. Instantaneous gradients exhibit little alignment with this dominant direction, indicating that it arises from accumulated optimizer updates rather than per-batch gradient structure. Comparing AdamW with SGD variants at matched loss levels reveals substantial differences in trajectory geometry: AdamW develops multi-dimensional drift structure, whereas SGD-family optimizers produce nearly colinear parameter evolution and weaker probe dynamics. Reheating selectively perturbs transverse components with minimal effect on the dominant drift coordinate. These findings suggest that optimizer choice shapes the effective dimensionality and structure of learning trajectories beyond what is apparent from loss values alone.

Human operators are still frequently exposed to hazardous environments such as disaster zones and industrial facilities, where intuitive and reliable teleoperation of mobile robots and Unmanned Aerial Vehicles (UAVs) is essential. In this context, hands-free teleoperation enhances operator mobility and situational awareness, thereby improving safety in hazardous environments. While vision-based gesture recognition has been explored as one method for hands-free teleoperation, its performance often deteriorates under occlusions, lighting variations, and cluttered backgrounds, limiting its applicability in real-world operations. To overcome these limitations, we propose a multimodal gesture recognition framework that integrates inertial data (accelerometer, gyroscope, and orientation) from Apple Watches on both wrists with capacitive sensing signals from custom gloves. We design a late fusion strategy based on the log-likelihood ratio (LLR), which not only enhances recognition performance but also provides interpretability by quantifying modality-specific contributions. To support this research, we introduce a new dataset of 20 distinct gestures inspired by aircraft marshalling signals, comprising synchronized RGB video, IMU, and capacitive sensor data. Experimental results demonstrate that our framework achieves performance comparable to a state-of-the-art vision-based baseline while significantly reducing computational cost, model size, and training time, making it well suited for real-time robot control. We therefore underscore the potential of sensor-based multimodal fusion as a robust and interpretable solution for gesture-driven mobile robot and drone teleoperation.

The paradigm of large language model (LLM) reasoning is shifting from parameter scaling to test-time compute scaling, yet many existing approaches still rely on uniform brute-force sampling (for example, fixed best-of-N or self-consistency) that is costly, hard to attribute, and can trigger overthinking with diminishing returns. We propose ODAR-Expert, an adaptive routing framework that optimizes the accuracy-efficiency trade-off via principled resource allocation. ODAR uses a difficulty estimator grounded in amortized active inference to dynamically route queries between a heuristic Fast Agent and a deliberative Slow Agent. We further introduce a free-energy-principled, risk-sensitive fusion mechanism that selects answers by minimizing a variational free energy objective, balancing log-likelihood with epistemic uncertainty (varentropy) as a principled alternative to ad hoc voting over heterogeneous candidates. Extensive evaluation across 23 benchmarks shows strong and consistent gains, including 98.2% accuracy on MATH and 54.8% on Humanity's Last Exam (HLE), while improving the compute-accuracy frontier under compute-matched settings. We also validate reproducibility on a fully open-source stack (Llama 4 + DeepSeek), where ODAR surpasses homogeneous sampling strategies while reducing computational costs by 82%. Overall, our results suggest that thinking-optimal scaling requires adaptive resource allocation with free-energy-based decision-making rather than simply increasing test-time compute.

This paper argues that the two leading AGI firms -- OpenAI and Anthropic -- construct sociotechnical imaginaries through a structurally consistent rhetorical strategy, despite meaningful differences in execution. Drawing on Jasanoff (2015)'s framework of sociotechnical imaginaries, the paper analyzes two essays published in late 2024: Sam Altman's "The Intelligence Age" and Dario Amodei's "Machines of Loving Grace." Close comparative reading identifies four shared rhetorical operations: the self-exemption move, which disavows prophetic authority while exercising it; teleological naturalization, which embeds AGI's arrival in narratives of historical inevitability; qualified acknowledgment, which absorbs concessions to risk into an optimistic frame; and implicit indispensability, which positions each firm as central to the imagined future without naming it as a commercial actor. That two competing institutions with different cultures, risk philosophies, and leaders with notably different public personae converge on the same rhetorical architecture suggests the imaginary reflects not only firm-level strategy but the institutional position these firms occupy. The paper extends the sociotechnical imaginaries framework from nation-states to private firms at the frontier of transformative technology development, identifies the discursive mechanism through which corporate authority over technological futures is projected and stabilized, and demonstrates that this mechanism is at minimum structural rather than idiosyncratic. The findings raise the question of what institutional arrangements would make that authority contestable from outside the firms that produce it.

The remote sensing community is witnessing a rapid growth of foundation models, which provide powerful embeddings for a wide range of downstream tasks. However, practical adoption and fair comparison remain challenging due to substantial heterogeneity in model release formats, platforms and interfaces, and input data specifications. These inconsistencies significantly increase the cost of obtaining, using, and benchmarking embeddings across models. To address this issue, we propose rs-embed, a Python library that offers a unified, region of interst (ROI) centric interface: with a single line of code, users can retrieve embeddings from any supported model for any location and any time range. The library also provides efficient batch processing to enable large-scale embedding generation and evaluation. The code is available at: https://github.com/cybergis/rs-embed

This study proposes the General Bayes framework for policy learning. We consider decision problems in which a decision-maker chooses an action from an action set to maximize its expected welfare. Typical examples include treatment choice and portfolio selection. In such problems, the statistical target is a decision rule, and the prediction of each outcome $Y(a)$ is not necessarily of primary interest. We formulate this policy learning problem by loss-based Bayesian updating. Our main technical device is a squared-loss surrogate for welfare maximization. We show that maximizing empirical welfare over a policy class is equivalent to minimizing a scaled squared error in the outcome difference, up to a quadratic regularization controlled by a tuning parameter $ζ>0$. This rewriting yields a General Bayes posterior over decision rules that admits a Gaussian pseudo-likelihood interpretation. We clarify two Bayesian interpretations of the resulting generalized posterior, a working Gaussian view and a decision-theoretic loss-based view. As one implementation example, we introduce neural networks with tanh-squashed outputs. Finally, we provide theoretical guarantees in a PAC-Bayes style.

Large language model (LLM) agents typically rely on reactive decision-making paradigms such as ReAct, selecting actions conditioned on growing execution histories. While effective for short tasks, these approaches often lead to redundant tool usage, unstable reasoning, and high token consumption in complex long-horizon tasks involving branching, iteration, or multi-tool coordination. To address these limitations, this paper introduces PseudoAct, a novel framework for flexible planning and action control in LLM agents through pseudocode synthesis. Leveraging the ability of LLMs to express task-solving strategies as code, PseudoAct synthesizes a structured pseudocode plan that decomposes a task into subtasks and explicitly encodes control flow, including sequencing, conditionals, loops, parallel composition, and combinations of these logic primitives. Actions are then executed by following this global plan, making the decision logic explicit and temporally coherent. This design reduces redundant actions, prevents infinite loops, and avoids uninformative alternative exploration, enabling consistent and efficient long-horizon decision-making. Experiments on benchmark datasets show that our method significantly outperforms existing reactive agent approaches, achieving a 20.93% absolute gain in success rate on FEVER and setting a new state-of-the-art on HotpotQA.

Due to the scalability and portability, low-altitude intelligent networks (LAINs) are essential in various fields such as surveillance and disaster rescue. However, in LAINs, unmanned aerial vehicles (UAVs) are characterized by the distributed topology and high mobility, thus vulnerable to security threats, which may degrade routing performances for data transmissions. Hence, how to ensure the routing stability and security of LAINs is challenging. In this paper, we focus on the routing with multiple UAV clusters in LAINs. To minimize the damage caused by potential threats, we present the zero-trust architecture with the software-defined perimeter and blockchain techniques to manage the identify and mobility of UAVs. Besides, we formulate the routing problem to optimize the end-to-end (E2E) delay and transmission success ratio (TSR) simultaneously, which is an integer nonlinear programming problem and intractable to solve. Therefore, we reformulate the problem into a decentralized partially observable Markov decision process. We design the multi-agent double deep Q-network-based routing algorithms to solve the problem, empowered by the soft-hierarchical experience replay buffer and prioritized experience replay mechanisms. Finally, extensive simulations are conducted and the numerical results demonstrate that the proposed framework reduces the average E2E delay by 59\% and improves the TSR by 29\% on average compared to benchmarks, while simultaneously enabling faster and more robust identification of low-trust UAVs.

The increasing size and heterogeneity of exoplanet catalogs have made systematic habitability assessment challenging, particularly given the extreme scarcity of potentially habitable planets and the evolving nature of their labels. In this study, we explore the use of pool-based active learning to improve the efficiency of habitability classification under realistic observational constraints. We construct a unified dataset from the Habitable World Catalog and the NASA Exoplanet Archive and formulate habitability assessment as a binary classification problem. A supervised baseline based on gradient-boosted decision trees is established and optimized for recall in order to prioritize the identification of rare potentially habitable planets. This model is then embedded within an active learning framework, where uncertainty-based margin sampling is compared against random querying across multiple runs and labeling budgets. We find that active learning substantially reduces the number of labeled instances required to approach supervised performance, demonstrating clear gains in label efficiency. To connect these results to a practical astronomical use case, we aggregate predictions from independently trained active-learning models into an ensemble and use the resulting mean probabilities and uncertainties to rank planets originally labeled as non-habitable. This procedure identifies a single robust candidate for further study, illustrating how active learning can support conservative, uncertainty-aware prioritization of follow-up targets rather than speculative reclassification. Our results indicate that active learning provides a principled framework for guiding habitability studies in data regimes characterized by label imbalance, incomplete information, and limited observational resources.

We present Geodesic Semantic Search (GSS), a retrieval system that learns node-specific Riemannian metrics on citation graphs to enable geometry-aware semantic search. Unlike standard embedding-based retrieval that relies on fixed Euclidean distances, \gss{} learns a low-rank metric tensor $\mL_i \in \R^{d \times r}$ at each node, inducing a local positive semi-definite metric $\mG_i = \mL_i \mL_i^\top + \eps \mI$. This parameterization guarantees valid metrics while keeping the model tractable. Retrieval proceeds via multi-source Dijkstra on the learned geodesic distances, followed by Maximal Marginal Relevance reranking and path coherence filtering. On citation prediction benchmarks with 169K papers, \gss{} achieves 23\% relative improvement in Recall@20 over SPECTER+FAISS baselines while providing interpretable citation paths. Our hierarchical coarse-to-fine search with k-means pooling reduces computational cost by 4$\times$ compared to flat geodesic search while maintaining 97\% retrieval quality. We provide theoretical analysis of when geodesic distances outperform direct similarity, characterize the approximation quality of low-rank metrics, and validate predictions empirically. Code and trained models are available at https://github.com/YCRG-Labs/geodesic-search.

Multi-mode tensor time series (TTS) can be found in many domains, such as search engines and environmental monitoring systems. Learning representations of a TTS benefits various applications, but it is also challenging since the complexities inherent in the tensor hinder the realization of rich representations. In this paper, we propose a novel representation learning method designed specifically for TTS, namely MoST. Specifically, MoST uses a tensor slicing approach to reduce the complexity of the TTS structure and learns representations that can be disentangled into individual non-temporal modes. Each representation captures mode-specific features, which are the relationship between variables within the same mode, and mode-invariant features, which are in common in representations of different modes. We employ a contrastive learning framework to learn parameters; the loss function comprises two parts intended to learn representation in a mode-specific way and mode-invariant way, effectively exploiting disentangled representations as augmentations. Extensive experiments on real-world datasets show that MoST consistently outperforms the state-of-the-art methods in terms of classification and forecasting accuracy. Code is available at https://github.com/KoheiObata/MoST.

Time series anomaly detection (TSAD) has been an important area of research for decades, with reconstruction-based methods, mostly based on generative models, gaining popularity and demonstrating success. Diffusion models have recently attracted attention due to their advanced generative capabilities. Existing diffusion-based methods for TSAD rely on a conditional strategy, which reconstructs input instances from white noise with the aid of the conditioner. However, this poses challenges in accurately reconstructing the normal parts, resulting in suboptimal detection performance. In response, we propose a novel diffusion-based method, named AnomalyFilter, which acts as a selective filter that only denoises anomaly parts in the instance while retaining normal parts. To build such a filter, we mask Gaussian noise during the training phase and conduct the denoising process without adding noise to the instances. The synergy of the two simple components greatly enhances the performance of naive diffusion models. Extensive experiments on five datasets demonstrate that AnomalyFilter achieves notably low reconstruction error on normal parts, providing empirical support for its effectiveness in anomaly detection. AnomalyFilter represents a pioneering approach that focuses on the noise design of diffusion models specifically tailored for TSAD.

Cloud computing enables cost-effective on-demand network access to a shared pool of configurable computing resources. The purpose of this paper is to examine and identifying the use of Cloud computing in the critical infrastructure domain among small and medium sized enterprises (SMEs). The data for this study were gathered from a survey of different academic, industry, governmental and online literature related to the use of Cloud computing in SMEs. The result revealed that there are risks involved in the use of Cloud computing, SMEs are deploying Cloud computing using different deployment models and reaching a high level of deployment within the critical infrastructure. The research findings are useful for SMEs that are planning or are in the use of Cloud computing, as well as for SMEs policymakers and business support community that engaged with Cloud computing initiatives.

TRIZ-based contradiction mining is a fundamental task in patent analysis and systematic innovation, as it enables the identification of improving and worsening technical parameters that drive inventive problem solving. However, existing approaches largely rely on rule-based systems or traditional machine learning models, which struggle with semantic ambiguity, domain dependency, and limited generalization when processing complex patent language. Recently, large language models (LLMs) have shown strong semantic understanding capabilities, yet their direct application to TRIZ parameter extraction remains challenging due to hallucination and insufficient grounding in structured TRIZ knowledge. To address these limitations, this paper proposes TRIZ-RAGNER, a retrieval-augmented large language model framework for TRIZ-aware named entity recognition in patent-based contradiction mining. TRIZ-RAGNER reformulates contradiction mining as a semantic-level NER task and integrates dense retrieval over a TRIZ knowledge base, cross-encoder reranking for context refinement, and structured LLM prompting to extract improving and worsening parameters from patent sentences. By injecting domain-specific TRIZ knowledge into the LLM reasoning process, the proposed framework effectively reduces semantic noise and improves extraction consistency. Experiments on the PaTRIZ dataset demonstrate that TRIZ-RAGNER consistently outperforms traditional sequence labeling models and LLM-based baselines. The proposed framework achieves a precision of 85.6%, a recall of 82.9%, and an F1-score of 84.2% in TRIZ contradiction pair identification. Compared with the strongest baseline using prompt-enhanced GPT, TRIZ-RAGNER yields an absolute F1-score improvement of 7.3 percentage points, confirming the effectiveness of retrieval-augmented TRIZ knowledge grounding for robust and accurate patent-based contradiction mining.

Large-scale Vision-Language Models (VLMs) exhibit strong zero-shot recognition, yet their real-world deployment is challenged by distribution shifts. While Test-Time Adaptation (TTA) can mitigate this, existing VLM-based TTA methods operate under a closed-set assumption, failing in open-set scenarios where test streams contain both covariate-shifted in-distribution (csID) and out-of-distribution (csOOD) data. This leads to a critical difficulty: the model must discriminate unknown csOOD samples to avoid interference while simultaneously adapting to known csID classes for accuracy. Current open-set TTA (OSTTA) methods rely on hard thresholds for separation and entropy minimization for adaptation. These strategies are brittle, often misclassifying ambiguous csOOD samples and inducing overconfident predictions, and their parameter-update mechanism is computationally prohibitive for VLMs. To address these limitations, we propose Prototype-based Double-Check Separation (ProtoDCS), a robust framework for OSTTA that effectively separates csID and csOOD samples, enabling safe and efficient adaptation of VLMs to csID data. Our main contributions are: (1) a novel double-check separation mechanism employing probabilistic Gaussian Mixture Model (GMM) verification to replace brittle thresholding; and (2) an evidence-driven adaptation strategy utilizing uncertainty-aware loss and efficient prototype-level updates, mitigating overconfidence and reducing computational overhead. Extensive experiments on CIFAR-10/100-C and Tiny-ImageNet-C demonstrate that ProtoDCS achieves state-of-the-art performance, significantly boosting both known-class accuracy and OOD detection metrics. Code will be available at https://github.com/O-YangF/ProtoDCS.

Vision-language models (VLMs) show strong potential for complex diagnostic tasks in medical imaging. However, applying VLMs to multi-organ medical imaging introduces two principal challenges: (1) modality-specific vision-language alignment and (2) cross-modal feature fusion. In this work, we propose MedMAP, a Medical Modality-Aware Pretraining framework that enhances vision-language representation learning in 3D MRI. MedMAP comprises a modality-aware vision-language alignment stage and a fine-tuning stage for multi-organ abnormality detection. During the pre-training stage, the modality-aware encoders implicitly capture the joint modality distribution and improve alignment between visual and textual representations. We then fine-tune the pre-trained vision encoders (while keeping the text encoder frozen) for downstream tasks. To this end, we curated MedMoM-MRI3D, comprising 7,392 3D MRI volume-report pairs spanning twelve MRI modalities and nine abnormalities tailored for various 3D medical analysis tasks. Extensive experiments on MedMoM-MRI3D demonstrate that MedMAP significantly outperforms existing VLMs in 3D MRI-based multi-organ abnormality detection. Our code is available at https://github.com/RomantiDr/MedMAP.

We introduce AudioCapBench, a benchmark for evaluating audio captioning capabilities of large multimodal models. \method covers three distinct audio domains, including environmental sound, music, and speech, with 1,000 curated evaluation samples drawn from established datasets. We evaluate 13 models across two providers (OpenAI, Google Gemini) using both reference-based metrics (METEOR, BLEU, ROUGE-L) and an LLM-as-Judge framework that scores predictions on three orthogonal dimensions: \textit{accuracy} (semantic correctness), \textit{completeness} (coverage of reference content), and \textit{hallucination} (absence of fabricated content). Our results reveal that Gemini models generally outperform OpenAI models on overall captioning quality, with Gemini~3~Pro achieving the highest overall score (6.00/10), while OpenAI models exhibit lower hallucination rates. All models perform best on speech captioning and worst on music captioning. We release the benchmark as well as evaluation code to facilitate reproducible audio understanding research.

The rapid advancement of Large Language Models (LLMs) has led to the emergence of intelligent agents capable of autonomously interacting with environments and invoking external tools. Recently, agent-based software repair approaches have received widespread attention, as repair agents can automatically analyze and localize bugs, generate patches, and achieve state-of-the-art performance on repository-level benchmarks. However, existing approaches usually adopt a localize-then-fix paradigm, jumping directly from "where the bug is" to "how to fix it", leaving a fundamental reasoning gap. To this end, we propose SGAgent, a Suggestion-Guided multi-Agent framework for repository-level software repair, which follows a localize-suggest-fix paradigm. SGAgent introduces a suggestion phase to strengthen the transition from localization to repair. The suggester starts from the buggy locations and incrementally retrieves relevant context until it fully understands the bug, and then provides actionable repair suggestions. Moreover, we construct a Knowledge Graph from the target repository and develop a KG-based toolkit to enhance SGAgent's global contextual awareness and repository-level reasoning. Three specialized sub-agents (i.e., localizer, suggester, and fixer) collaborate to achieve automated end-to-end software repair. Experimental results on SWE-Bench show that SGAgent with Claude-3.5 achieves 51.3% repair accuracy, 81.2% file-level and 52.4% function-level localization accuracy with an average cost of $1.48 per instance, outperforming all baselines using the same base model. Furthermore, SGAgent attains 48% accuracy on VUL4J and VJBench for vulnerability repair, demonstrating strong generalization across tasks and programming languages.

Everyone from AI executives and researchers to doomsayers, politicians, and activists is talking about Artificial General Intelligence (AGI). Yet, they often don't seem to agree on its exact definition. One common definition of AGI is an AI that can do everything a human can do, but are humans truly general? In this paper, we address what's wrong with our conception of AGI, and why, even in its most coherent formulation, it is a flawed concept to describe the future of AI. We explore whether the most widely accepted definitions are plausible, useful, and truly general. We argue that AI must embrace specialization, rather than strive for generality, and in its specialization strive for superhuman performance, and introduce Superhuman Adaptable Intelligence (SAI). SAI is defined as intelligence that can learn to exceed humans at anything important that we can do, and that can fill in the skill gaps where humans are incapable. We then lay out how SAI can help hone a discussion around AI that was blurred by an overloaded definition of AGI, and extrapolate the implications of using it as a guide for the future.

Federated LoRA provides a communication-efficient mechanism for fine-tuning large language models on decentralized data. In practice, however, a discrepancy between the factor-wise averaging used to preserve low rank and the mathematically correct aggregation of local updates can cause significant aggregation error and unstable training. We argue that a major source of this problem is rotational misalignment, arising from the rotational invariance of low-rank factorizations -- semantically equivalent updates can be represented in different latent subspaces across clients since $(B_i R_i)(R_i^\top A_i) = B_i A_i$. When such misaligned factors are averaged directly, they interfere destructively and degrade the global update. To address this issue, we propose FedRot-LoRA, a federated LoRA framework that aligns client updates via orthogonal transformations prior to aggregation. This alignment preserves the semantic update while reducing cross-client subspace mismatch, without increasing communication cost or restricting model expressivity. We provide a convergence analysis that examines the aggregation error induced by factor-wise averaging and shows how rotational alignment yields a tighter upper bound on this error. Extensive experiments on natural language understanding and generative tasks demonstrate that FedRot-LoRA consistently outperforms existing federated LoRA baselines across a range of heterogeneity levels and LoRA ranks.

Ensuring the safety of LLM-generated content is essential for real-world deployment. Most existing guardrail models formulate moderation as a fixed binary classification task, implicitly assuming a fixed definition of harmfulness. In practice, enforcement strictness - how conservatively harmfulness is defined and enforced - varies across platforms and evolves over time, making binary moderators brittle under shifting requirements. We first introduce FlexBench, a strictness-adaptive LLM moderation benchmark that enables controlled evaluation under multiple strictness regimes. Experiments on FlexBench reveal substantial cross-strictness inconsistency in existing moderators: models that perform well under one regime can degrade substantially under others, limiting their practical usability. To address this, we propose FlexGuard, an LLM-based moderator that outputs a calibrated continuous risk score reflecting risk severity and supports strictness-specific decisions via thresholding. We train FlexGuard via risk-alignment optimization to improve score-severity consistency and provide practical threshold selection strategies to adapt to target strictness at deployment. Experiments on FlexBench and public benchmarks demonstrate that FlexGuard achieves higher moderation accuracy and substantially improved robustness under varying strictness. We release the source code and data to support reproducibility.

Stochastic Bilevel Optimization has emerged as a fundamental framework for meta-learning and hyperparameter optimization. Despite the practical prevalence of single-loop algorithms--which update lower and upper variables concurrently--their theoretical understanding, particularly in the stochastic regime, remains significantly underdeveloped compared to their multi-loop counterparts. Existing analyses often yield suboptimal convergence rates or obscure the critical dependence on the lower-level condition number $κ$, frequently burying it within generic Lipschitz constants. In this paper, we bridge this gap by providing a refined convergence analysis of the Single-loop Stochastic Approximate Implicit Differentiation (SSAID) algorithm. We prove that SSAID achieves an $ε$-stationary point with an oracle complexity of $\mathcal{O}(κ^7 ε^{-2})$. Our result is noteworthy in two aspects: (i) it matches the optimal $\mathcal{O}(ε^{-2})$ rate of state-of-the-art multi-loop methods (e.g., stocBiO) while maintaining the computational efficiency of a single-loop update; and (ii) it provides the first explicit, fine-grained characterization of the $κ$-dependence for stochastic AID-based single-loop methods. This work demonstrates that SSAID is not merely a heuristic approach, but admits a rigorous theoretical foundation with convergence guarantees competitive with mainstream multi-loop frameworks.

Synthesizing high-quality training data is crucial for enhancing domain models' reasoning abilities. Existing methods face limitations in long-tail knowledge coverage, effectiveness verification, and interpretability. Knowledge-graph-based approaches still fall short in functionality, granularity, customizability, and evaluation. To address these issues, we propose MMKG-RDS, a flexible framework for reasoning data synthesis that leverages multimodal knowledge graphs. It supports fine-grained knowledge extraction, customizable path sampling, and multidimensional data quality scoring. We validate MMKG-RDS with the MMKG-RDS-Bench dataset, covering five domains, 17 task types, and 14,950 samples. Experimental results show fine-tuning Qwen3 models (0.6B/8B/32B) on a small number of synthesized samples improves reasoning accuracy by 9.2%. The framework also generates distinct data, challenging existing models on tasks involving tables and formulas, useful for complex benchmark construction. The dataset and code are available at https://github.com/360AILAB-NLP/MMKG-RDS

Hyperparameter optimization (HPO) is known to be costly in deep learning, especially when leveraging automated approaches. Most of the existing automated HPO methods are accuracy-based, i.e., accuracy metrics are used to guide the trials of different hyperparameter configurations amongst a specific search space. However, many trials may encounter severe training problems, such as vanishing gradients and insufficient convergence, which can hardly be reflected by accuracy metrics in the early stages of the training and often result in poor performance. This leads to an inefficient optimization trajectory because the bad trials occupy considerable computation resources and reduce the probability of finding excellent hyperparameter configurations within a time limitation. In this paper, we propose \textbf{Bad Trial Tackler (BTTackler)}, a novel HPO framework that introduces training diagnosis to identify training problems automatically and hence tackles bad trials. BTTackler diagnoses each trial by calculating a set of carefully designed quantified indicators and triggers early termination if any training problems are detected. Evaluations are performed on representative HPO tasks consisting of three classical deep neural networks (DNN) and four widely used HPO methods. To better quantify the effectiveness of an automated HPO method, we propose two new measurements based on accuracy and time consumption. Results show the advantage of BTTackler on two-fold: (1) it reduces 40.33\% of time consumption to achieve the same accuracy comparable to baseline methods on average and (2) it conducts 44.5\% more top-10 trials than baseline methods on average within a given time budget. We also released an open-source Python library that allows users to easily apply BTTackler to automated HPO processes with minimal code changes.

We propose a Multivariate Spatio-Temporal Neural Hawkes Process for modeling complex multivariate event data with spatio-temporal dynamics. The proposed model extends continuous-time neural Hawkes processes by integrating spatial information into latent state evolution through learned temporal and spatial decay dynamics, enabling flexible modeling of excitation and inhibition without predefined triggering kernels. By analyzing fitted intensity functions of deep learning-based temporal Hawkes process models, we identify a modeling gap in how fitted intensity behavior is captured beyond likelihood-based performance, which motivates the proposed spatio-temporal approach. Simulation studies show that the proposed method successfully recovers sensible temporal and spatial intensity structure in multivariate spatio-temporal point patterns, while existing temporal neural Hawkes process approach fails to do so. An application to terrorism data from Pakistan further demonstrates the proposed model's ability to capture complex spatio-temporal interaction across multiple event types.

Significant progress has been made in the field of Instruction-based Image Editing Models (IIEMs). However, while these models demonstrate plausible adherence to instructions and strong reasoning ability on current benchmarks, their ability to edit small objects remains underexplored, despite its importance for precise local editing and refining details in both real and generated images. In this paper, we introduce DeepLookEditBench (DLEBench), the first benchmark dedicated to assessing the abilities of IIEMs in editing small-scale objects. Specifically, we construct a challenging testbed comprising 1889 samples across seven instruction types. In these samples, target objects occupy only 1%-10% of the image area, covering complex scenarios such as partial occlusion and multi-object editing. To ensure robust evaluation on this benchmark, we propose an evaluation protocol with refined score rubrics to minimize subjectivity and ambiguity in two criteria: Instruction Following and Visual Consistency. This protocol also introduces a dual-mode evaluation framework (Tool-driven and Oracle-guided Modes) addressing the misalignment between LMM-as-a-Judge and human judgements on DLEBench. Empirical results on 10 IIEMs reveal significant performance gaps in small-scale object editing, highlighting the need for specialized benchmarks to advance this ability.

Resonant tunneling diodes (RTDs) provide room-temperature terahertz oscillation and strong nonlinear mixing, enabling compact monostatic sensors in which a single device acts as both a bias-tunable oscillator and a self-oscillating mixer. This paper presents a 280 GHz-band radar concept enabled by a single RTD, which exploits self-mixing to generate a low-frequency radar interferometric signal while sweeping the RTD oscillation frequency. We show that the RTD self-mixing waveform can be interpreted from a radar perspective and processed to extract micrometer-scale displacement and thin-film thickness changes from repeated sweeps. Using the proposed technique, we experimentally demonstrate a minimum detectable displacement of ~5 um and quantitatively resolve polymer film thicknesses of 12.5, 25, and 50 um.

Diffractive optical neural networks (DONNs) have demonstrated unparalleled energy efficiency and parallelism by processing information directly in the optical domain. However, their computational expressivity is constrained by static, passive diffractive phase masks that lack efficient nonlinear responses and reprogrammability. To address these limitations, we introduce the Recurrent Diffractive Optical Neural Processor (ReDON), a novel architecture featuring reconfigurable, recurrent self-modulated nonlinearity. This mechanism enables dynamic, input-dependent optical transmission through in-situ electro-optic self-modulation, providing a highly efficient and reprogrammable approach to optical computation. Inspired by the gated linear unit (GLU) used in large language models, ReDON senses a fraction of the propagating optical field and modulates its phase or intensity via a lightweight parametric function, enabling effective nonlinearity with minimal inference overhead. As a non-von Neumann architecture in which the primary weighting elements (metasurfaces) remain fixed, ReDON substantially extends the nonlinear representational capacity and task adaptability of conventional DONNs through recurrent optical hardware reuse and dynamically tunable nonlinearity. We systematically investigate various self-modulation configurations to characterize the trade-offs between hardware efficiency and computational expressivity. On image recognition and segmentation benchmarks, ReDON improves test accuracy and mean intersection-over-union (mIoU) by up to 20% compared with prior DONNs employing either optical or digital nonlinearities at comparable model complexity and negligible additional power consumption. This work establishes a new paradigm for reconfigurable nonlinear optical computing, uniting recurrence and self-modulation within non-von Neumann analog processors.

Machine learning holds promise for advancing clinical decision support, yet it remains unclear when multimodal learning truly helps in practice, particularly under modality missingness and fairness constraints. In this work, we conduct a systematic benchmark of multimodal fusion between Electronic Health Records (EHR) and chest X-rays (CXR) on standardized cohorts from MIMIC-IV and MIMIC-CXR, aiming to answer four fundamental questions: when multimodal fusion improves clinical prediction, how different fusion strategies compare, how robust existing methods are to missing modalities, and whether multimodal models achieve algorithmic fairness. Our study reveals several key insights. Multimodal fusion improves performance when modalities are complete, with gains concentrating in diseases that require complementary information from both EHR and CXR. While cross-modal learning mechanisms capture clinically meaningful dependencies beyond simple concatenation, the rich temporal structure of EHR introduces strong modality imbalance that architectural complexity alone cannot overcome. Under realistic missingness, multimodal benefits rapidly degrade unless models are explicitly designed to handle incomplete inputs. Moreover, multimodal fusion does not inherently improve fairness, with subgroup disparities mainly arising from unequal sensitivity across demographic groups. To support reproducible and extensible evaluation, we further release a flexible benchmarking toolkit that enables plug-and-play integration of new models and datasets. Together, this work provides actionable guidance on when multimodal learning helps, when it fails, and why, laying the foundation for developing clinically deployable multimodal systems that are both effective and reliable. The open-source toolkit can be found at https://github.com/jakeykj/CareBench.

Algorithmic decisions about individuals require predictions that are not only accurate but also fair with respect to sensitive attributes such as gender and race. Causal notions of fairness align with legal requirements, yet many methods assume access to detailed knowledge of the underlying causal graph, which is a demanding assumption in practice. We propose a learning framework that achieves interventional fairness by leveraging a causal graph over \textit{clusters of variables}, which is substantially easier to estimate than a variable-level graph. With possible \textit{adjustment cluster sets} identified from such a cluster causal graph, our framework trains a prediction model by reducing the worst-case discrepancy between interventional distributions across these sets. To this end, we develop a computationally efficient barycenter kernel maximum mean discrepancy (MMD) that scales favorably with the number of sensitive attribute values. Extensive experiments show that our framework strikes a better balance between fairness and accuracy than existing approaches, highlighting its effectiveness under limited causal graph knowledge.

The reasoning capability of large language models (LLMs), defined as their ability to analyze, infer, and make decisions based on input information, is essential for building intelligent task-oriented dialogue systems. However, existing benchmarks do not sufficiently reflect the complexity of real-world scenarios, which limits their effectiveness in evaluating and enhancing LLM reasoning in practical contexts. Many current reasoning datasets are overly simplistic and abstract, often disconnected from realistic task flows, domain constraints, and operational rules, making it difficult to effectively evaluate LLMs' logical reasoning ability. In addition, data contamination from pretraining corpora undermines the reliability of evaluation results, and traditional crowdsourcing methods for dataset construction are labor-intensive and difficult to scale. To address these challenges, we propose a LLM-driven framework for synthesizing multi-turn, task-oriented dialogues grounded in realistic reasoning scenarios, leveraging trilevel optimization to enhance dialogue quality. Our method generates dialogues grounded in authentic task scenarios, enriched with real-world information, and exhibiting strong contextual coherence. Corresponding reasoning tasks are carefully designed around these dialogues and iteratively refined to continuously improve the tasks' quality and challenge. The resulting dataset serves as a valuable benchmark for assessing and advancing the realistic logical reasoning capabilities of LLMs. Experimental results show that our synthetic data-based reasoning tasks introduce non-trivial reasoning challenges and provide meaningful support for improving the reasoning capabilities of LLMs.

We present SleepLM, a family of sleep-language foundation models that enable human sleep alignment, interpretation, and interaction with natural language. Despite the critical role of sleep, learning-based sleep analysis systems operate in closed label spaces (e.g., predefined stages or events) and fail to describe, query, or generalize to novel sleep phenomena. SleepLM bridges natural language and multimodal polysomnography, enabling language-grounded representations of sleep physiology. To support this alignment, we introduce a multilevel sleep caption generation pipeline that enables the curation of the first large-scale sleep-text dataset, comprising over 100K hours of data from more than 10,000 individuals. Furthermore, we present a unified pretraining objective that combines contrastive alignment, caption generation, and signal reconstruction to better capture physiological fidelity and cross-modal interactions. Extensive experiments on real-world sleep understanding tasks verify that SleepLM outperforms state-of-the-art in zero-shot and few-shot learning, cross-modal retrieval, and sleep captioning. Importantly, SleepLM also exhibits intriguing capabilities including language-guided event localization, targeted insight generation, and zero-shot generalization to unseen tasks. All code and data will be open-sourced.

Long-form question answering (LFQA) demands nuanced evaluation of multi-sentence explanatory responses, yet existing metrics often fail to reflect human judgment. We present LFQA-HP-1M, a large-scale dataset comprising 1.3M human pairwise preference annotations for LFQA. We propose nine rubrics for answer quality evaluation, and show that simple linear models based on these features perform comparably to state-of-the-art LLM evaluators. We further examine transitivity consistency, positional bias, and verbosity biases in LLM evaluators and demonstrate their vulnerability to adversarial perturbations. Overall, this work provides one of the largest public LFQA preference datasets and a rubric-driven framework for transparent and reliable evaluation.

Heteroscedasticity -- where the variance of a variable changes with other variables -- is pervasive in real data, and elucidating why it arises from the perspective of statistical moments is crucial in scientific knowledge discovery and decision-making. However, standard causal discovery does not reveal which causes act on the mean versus the variance, as it returns a single moment-agnostic graph, limiting interpretability and downstream intervention design. We propose a Bayesian, moment-driven causal discovery framework that infers separate \textit{mean} and \textit{variance} causal graphs from observational heteroscedastic data. We first derive the identification results by establishing sufficient conditions under which these two graphs are separately identifiable. Building on this theory, we develop a variational inference method that learns a posterior distribution over both graphs, enabling principled uncertainty quantification of structural features (e.g., edges, paths, and subgraphs). To address the challenges of parameter optimization in heteroscedastic models with two graph structures, we take a curvature-aware optimization approach and develop a prior incorporation technique that leverages domain knowledge on node orderings, improving sample efficiency. Experiments on synthetic, semi-synthetic, and real data show that our approach accurately recovers mean and variance structures and outperforms state-of-the-art baselines.

Graph neural networks (GNNs) offer a principled approach to financial fraud detection by jointly learning from node features and transaction graph topology. However, their effectiveness on real-world anti-money laundering (AML) benchmarks depends critically on training practices such as specifically weight initialisation and normalisation that remain underexplored. We present a systematic ablation of initialisation and normalisation strategies across three GNN architectures (GCN, GAT, and GraphSAGE) on the Elliptic Bitcoin dataset. Our experiments reveal that initialisation and normalisation are architecture-dependent: GraphSAGE achieves the strongest performance with Xavier initialisation alone, GAT benefits most from combining GraphNorm with Xavier initialisation, while GCN shows limited sensitivity to these modifications. These findings offer practical, architecture-specific guidance for deploying GNNs in AML pipelines for datasets with severe class imbalance. We release a reproducible experimental framework with temporal data splits, seeded runs, and full ablation results.

With the increasing importance of distributed scientific workflows, there is a critical need to ensure Quality of Service (QoS) constraints, such as minimizing time or limiting execution to resource subsets. However, the unpredictable nature of workflow behavior, even with similar configurations, makes it difficult to provide QoS guarantees. For effective reasoning about QoS scheduling, we introduce QoSFlow, a performance modeling method that partitions a workflow's execution configuration space into regions with similar behavior. Each region groups configurations with comparable execution times according to a given statistical sensitivity, enabling efficient QoS-driven scheduling through analytical reasoning rather than exhaustive testing. Evaluation on three diverse workflows shows that QoSFlow's execution recommendations outperform the best-performing standard heuristic by 27.38%. Empirical validation confirms that QoSFlow's recommended configurations consistently match measured execution outcomes across different QoS constraints.

Memory-augmented Large Language Models (LLMs) have demonstrated remarkable capability for complex and long-horizon embodied planning. By keeping track of past experiences and environmental states, memory enables LLMs to maintain a global view, thereby avoiding repetitive exploration. However, existing approaches often store the memory as raw text, leading to excessively long prompts and high prefill latency. While it is possible to store and reuse the KV caches, the efficiency benefits are greatly undermined due to frequent KV cache updates. In this paper, we propose KEEP, a KV-cache-centric memory management system for efficient embodied planning. KEEP features 3 key innovations: (1) a Static-Dynamic Memory Construction algorithm that reduces KV cache recomputation by mixed-granularity memory group; (2) a Multi-hop Memory Re-computation algorithm that dynamically identifies important cross-attention among different memory groups and reconstructs memory interactions iteratively; (3) a Layer-balanced Memory Loading that eliminates unbalanced KV cache loading and cross-attention computation across different layers. Extensive experimental results have demonstrated that KEEP achieves 2.68x speedup with negligible accuracy loss compared with text-based memory methods on ALFRED dataset. Compared with the KV re-computation method CacheBlend (EuroSys'25), KEEP shows 4.13% success rate improvement and 1.90x time-to-first-token (TTFT) reduction. Our code is available on https://github.com/PKU-SEC-Lab/KEEP_Embodied_Memory.

Recent multimodal models such as Contrastive Language-Image Pre-training (CLIP) have shown remarkable ability to align visual and linguistic representations. However, domains where small visual differences carry large semantic significance, such as diagram understanding, remain challenging due to the models' limited sensitivity to fine-grained structural variations. We propose a new training paradigm designed to enhance diagram comprehension in vision-language models. Our approach introduces pseudo contrastive samples generated by a diagram renderer that creates synthetic diagrams using randomly picked text elements. These samples highlight structural differences in diagrammatic imagery without requiring any modification or editing of the original data. By incorporating these pseudo contrastive samples into the training objective, the model learns to capture more precise and semantically consistent diagram structures. Empirical evaluations on a benchmark dataset of flowcharts demonstrate substantial improvements over standard CLIP and hard-negative CLIP training in both image-text matching and visual question answering tasks. The results underscore the value of domain-specific training strategies and contribute to advancing diagrammatic understanding within the broader context of vision-language learning.

Large unimodal foundation models for vision and language encode rich semantic structures, yet aligning them typically requires computationally intensive multimodal fine-tuning. Such approaches depend on large-scale parameter updates, are resource intensive, and can perturb pretrained representations. Emerging evidence suggests, however, that independently trained foundation models may already exhibit latent semantic compatibility, reflecting shared structures in the data they model. This raises a fundamental question: can cross-modal alignment be achieved without modifying the models themselves? Here we introduce HDFLIM (HyperDimensional computing with Frozen Language and Image Models), a framework that establishes cross-modal mappings while keeping pretrained vision and language models fully frozen. HDFLIM projects unimodal embeddings into a shared hyperdimensional space and leverages lightweight symbolic operations -- binding, bundling, and similarity-based retrieval to construct associative cross-modal representations in a single pass over the data. Caption generation emerges from high-dimensional memory retrieval rather than iterative gradient-based optimization. We show that HDFLIM achieves performance comparable to end-to-end vision-language training methods and produces captions that are more semantically grounded than zero-shot baselines. By decoupling alignment from parameter tuning, our results suggest that semantic mapping across foundation models can be realized through symbolic operations on hyperdimensional encodings of the respective embeddings. More broadly, this work points toward an alternative paradigm for foundation model alignment in which frozen models are integrated through structured representational mappings rather than through large-scale retraining. The codebase for our implementation can be found at https://github.com/Abhishek-Dalvi410/HDFLIM.

Knowledge distillation transfers large teacher models to compact student models, enabling deployment on resource-limited platforms while suffering minimal performance degradation. However, this paradigm could lead to various security risks, especially model theft. Existing defenses against model theft, such as watermarking and secure enclaves, focus primarily on identity authentication and incur significant resource costs. Aiming to provide post-theft accountability and traceability, we propose a novel fingerprinting framework that superimposes device-specific Physical Unclonable Function (PUF) signatures onto teacher logits during distillation. Compared with watermarking or secure enclaves, our approach is lightweight, requires no architectural changes, and enables traceability of any leaked or cloned model. Since the signatures are based on PUFs, this framework is robust against reverse engineering and tampering attacks. In this framework, the signature recovery process consists of two stages: first a neural network-based decoder and then a Hamming distance decoder. Furthermore, we also propose a bit compression scheme to support a large number of devices. Experiment results demonstrate that our framework achieves high key recovery rate and negligible accuracy loss while allowing a tunable trade-off between these two key metrics. These results show that the proposed framework is a practical and robust solution for protecting distilled models.

Multivariate time series forecasting is widely applied in fields such as transportation, energy, and finance. However, the data commonly suffers from issues of multi-scale characteristics, weak correlations, and noise interference, which limit the predictive performance of existing models. This paper proposes a dual-stream sparse Mixer prediction framework that extracts global trends and local dynamic features from sequences in both the frequency and time domains, respectively. It employs a sparsity mechanism to filter out invalid information, thereby enhancing the accuracy of cross-variable dependency modeling. Experimental results demonstrate that this method achieves leading performance on multiple real-world scenario datasets, validating its effectiveness and generality. The code is available at https://github.com/SDMixer/SDMixer

In the field of educational assessment, automated scoring systems increasingly rely on deep learning and large language models (LLMs). However, these systems face significant risks of bias amplification, where model prediction gaps between student groups become larger than those observed in training data. This issue is especially severe for underrepresented groups such as English Language Learners (ELLs), as models may inherit and further magnify existing disparities in the data. We identify that this issue is closely tied to representation bias: the scarcity of minority (high-scoring ELL) samples makes models trained with empirical risk minimization favor majority (non-ELL) linguistic patterns. Consequently, models tend to under-predict ELL students who even demonstrate comparable domain knowledge but use different linguistic patterns, thereby undermining the fairness of automated scoring outcomes. To mitigate this, we propose BRIDGE, a Bias-Reducing Inter-group Data GEneration framework designed for low-resource assessment settings. Instead of relying on the limited minority samples, BRIDGE synthesizes high-scoring ELL samples by "pasting" construct-relevant (i.e., rubric-aligned knowledge and evidence) content from abundant high-scoring non-ELL samples into authentic ELL linguistic patterns. We further introduce a discriminator model to ensure the quality of synthetic samples. Experiments on California Science Test (CAST) datasets demonstrate that BRIDGE effectively reduces prediction bias for high-scoring ELL students while maintaining overall scoring performance. Notably, our method achieves fairness gains comparable to using additional real human data, offering a cost-effective solution for ensuring equitable scoring in large-scale assessments.

The Multiple Traveling Salesman Problem (mTSP) extends the Traveling Salesman Problem to m tours that start and end at a common depot and jointly visit all customers exactly once. In the min-max variant, the objective is to minimize the longest tour, reflecting workload balance. We propose a hybrid approach, Construct, Merge, Solve & Adapt with Reinforcement Learning (RL-CMSA), for the symmetric single-depot min-max mTSP. The method iteratively constructs diverse solutions using probabilistic clustering guided by learned pairwise q-values, merges routes into a compact pool, solves a restricted set-covering MILP, and refines solutions via inter-route remove, shift, and swap moves. The q-values are updated by reinforcing city-pair co-occurrences in high-quality solutions, while the pool is adapted through ageing and pruning. This combination of exact optimization and reinforcement-guided construction balances exploration and exploitation. Computational results on random and TSPLIB instances show that RL-CMSA consistently finds (near-)best solutions and outperforms a state-of-the-art hybrid genetic algorithm under comparable time limits, especially as instance size and the number of salesmen increase.

Quantum machine learning models for sequential data face scalability challenges with complex multivariate signals. We introduce the Hybrid Quantum Temporal Convolutional Network (HQTCN), which combines classical temporal windowing with a quantum convolutional neural network core. By applying a shared quantum circuit across temporal windows, HQTCN captures long-range dependencies while achieving significant parameter reduction. Evaluated on synthetic NARMA sequences and high-dimensional EEG time-series, HQTCN performs competitively with classical baselines on univariate data and outperforms all baselines on multivariate tasks. The model demonstrates particular strength under data-limited conditions, maintaining high performance with substantially fewer parameters than conventional approaches. These results establish HQTCN as a parameter-efficient approach for multivariate time-series analysis.

Suicide remains a pressing global public health concern. While social media platforms offer opportunities for early risk detection through online conversation trees, existing approaches face two major limitations: (1) They rely on predefined rules (e.g., quotes or relies) to log conversations that capture only a narrow spectrum of user interactions, and (2) They overlook hidden influences such as user conformity and suicide copycat behavior, which can significantly affect suicidal expression and propagation in online communities. To address these limitations, we propose a Multi-Agent Causal Reasoning (MACR) framework that collaboratively employs a Reasoning Agent to scale user interactions and a Bias-aware Decision-Making Agent to mitigate harmful biases arising from hidden influences. The Reasoning Agent integrates cognitive appraisal theory to generate counterfactual user reactions to posts, thereby scaling user interactions. It analyses these reactions through structured dimensions, i.e., cognitive, emotional, and behavioral patterns, with a dedicated sub-agent responsible for each dimension. The Bias-aware Decision-Making Agent mitigates hidden biases through a front-door adjustment strategy, leveraging the counterfactual user reactions produced by the Reasoning Agent. Through the collaboration of reasoning and bias-aware decision making, the proposed MACR framework not only alleviates hidden biases, but also enriches contextual information of user interactions with counterfactual knowledge. Extensive experiments on real-world conversational datasets demonstrate the effectiveness and robustness of MACR in identifying suicide risk.

Transforming image features from perspective view (PV) space to bird's-eye-view (BEV) space remains challenging in autonomous driving due to depth ambiguity and occlusion. Although several view transformation (VT) paradigms have been proposed, the challenge still remains. In this paper, we propose a new regularization framework, dubbed CycleBEV, that enhances existing VT models for BEV semantic segmentation. Inspired by cycle consistency, widely used in image distribution modeling, we devise an inverse view transformation (IVT) network that maps BEV segmentation maps back to PV segmentation maps and use it to regularize VT networks during training through cycle consistency losses, enabling them to capture richer semantic and geometric information from input PV images. To further exploit the capacity of the IVT network, we introduce two novel ideas that extend cycle consistency into geometric and representation spaces. We evaluate CycleBEV on four representative VT models covering three major paradigms using the large-scale nuScenes dataset. Experimental results show consistent improvements -- with gains of up to 0.74, 4.86, and 3.74 mIoU for drivable area, vehicle, and pedestrian classes, respectively -- without increasing inference complexity, since the IVT network is used only during training. The implementation code is available at https://github.com/JeongbinHong/CycleBEV.

Understanding sources of uncertainty is fundamental to trustworthy three-dimensional scene modeling. While recent advances in neural radiance fields (NeRFs) achieve impressive accuracy in scene reconstruction and novel view synthesis, the lack of uncertainty estimation significantly limits their deployment in safety-critical settings. Existing uncertainty quantification methods for NeRFs fail to capture both aleatoric and epistemic uncertainty. Among those that do quantify one or the other, many of them either compromise rendering quality or incur significant computational overhead to obtain uncertainty estimates. To address these issues, we introduce Evidential Neural Radiance Fields, a probabilistic approach that seamlessly integrates with the NeRF rendering process and enables direct quantification of both aleatoric and epistemic uncertainty from a single forward pass. We compare multiple uncertainty quantification methods on three standardized benchmarks, where our approach demonstrates state-of-the-art scene reconstruction fidelity and uncertainty estimation quality.

For every real number $c \geq 1$ and for all $\varepsilon > 0$, there is a fitness function $f : \{0,1\}^n \to \mathbb{R}$ for which the optimal mutation rate for the $(1+1)$ evolutionary algorithm on $f$, denoted $p_n$, satisfies $p_n \approx c/n$ in that $|np_n - c| < \varepsilon$. In other words, the set of all $c \geq 1$ for which the mutation rate $c/n$ is optimal for the $(1+1)$ EA is dense in the interval $[1, \infty)$. To show this, a fitness function is introduced which is called HillPathJump.

Wavefunction-based quantum methods are some of the most accurate tools for predicting and analyzing the electronic structure of molecules, in particular for accounting for dynamical electron correlation. However, most methods of including dynamical correlation beyond the simple second-order Møller-Plesset perturbation theory (MP2) level are too computationally expensive to apply to large molecules. Approximations which reduce scaling with system size are a potential remedy, such as the tensor hyper-contraction (THC) technique of Hohenstein et al., but also result in additional sources of error. In this work, we correct errors in THC-approximated methods using machine learning. Specifically, we apply THC to third-order Møller-Plesset theory (MP3) as a simplified model for coupled cluster with single and double excitations (CCSD), and train several regression models on observed THC errors from the Main Group Chemistry Database (MGCDB84). We compare performance of multiple linear regression models and non-linear Kernel Ridge regression models. We also investigate correlation procedures using absolute and relative corrections and evaluate the corrections for both molecule and reaction energies. We discuss the potential for using regression techniques to correct THC-MP3 errors by comparing it to the "canonical" MP3 reference values and find the optimum technique based on accuracy. We find that non-linear regression models reduced root mean squared errors between THC- and canonical MP3 by a factor of 6-9$\times$ for total molecular energies and 2-3$\times$ for reaction energies.

We study generative modeling of graphs with recurring subgraph motifs. We propose Flowette, a continuous flow matching framework, that employs a graph neural network based transformer to learn a velocity field defined over graph representations with node and edge attributes. Our model preserves topology through optimal transport based coupling, and long-range structural dependencies through regularisation. To incorporate domain driven structural priors, we introduce graphettes, a new probabilistic family of graph structure models that generalize graphons via controlled structural edits for motifs like rings, stars and trees. We theoretically analyze the coupling, invariance, and structural properties of the proposed framework, and empirically evaluate it on synthetic and small-molecule graph generation tasks. Flowette demonstrates consistent improvements, highlighting the effectiveness of combining structural priors with flow-based training for modeling complex graph distributions.

In many economically relevant contexts where machine learning is deployed, multiple platforms obtain data from the same pool of users, each of whom selects the platform that best serves them. Prior work in this setting focuses exclusively on the "local" losses of learners on the distribution of data that they observe. We find that there exist instances where learners who use existing algorithms almost surely converge to models with arbitrarily poor global performance, even when models with low full-population loss exist. This happens through a feedback-induced mechanism, which we call the overspecialization trap: as learners optimize for users who already prefer them, they become less attractive to users outside this base, which further restricts the data they observe. Inspired by the recent use of knowledge distillation in modern ML, we propose an algorithm that allows learners to "probe" the predictions of peer models, enabling them to learn about users who do not select them. Our analysis characterizes when probing succeeds: this procedure converges almost surely to a stationary point with bounded full-population risk when probing sources are sufficiently informative, e.g., a known market leader or a majority of peers with good global performance. We verify our findings with semi-synthetic experiments on the MovieLens, Census, and Amazon Sentiment datasets.

Symbolic regression has recently gained traction in AI-driven scientific discovery, aiming to recover explicit closed-form expressions from data that reveal underlying physical laws. Despite recent advances, existing methods remain dominated by heuristic search algorithms or data-intensive approaches that assume low-noise regimes and lack principled uncertainty quantification. Fully probabilistic formulations are scarce, and existing Markov chain Monte Carlo-based Bayesian methods often struggle to efficiently explore the highly multimodal combinatorial space of symbolic expressions. We introduce VaSST, a scalable probabilistic framework for symbolic regression based on variational inference. VaSST employs a continuous relaxation of symbolic expression trees, termed soft symbolic trees, where discrete operator and feature assignments are replaced by soft distributions over allowable components. This relaxation transforms the combinatorial search over an astronomically large symbolic space into an efficient gradient-based optimization problem while preserving a coherent probabilistic interpretation. The learned soft representations induce posterior distributions over symbolic structures, enabling principled uncertainty quantification. Across simulated experiments and Feynman Symbolic Regression Database within SRBench, VaSST achieves superior performance in both structural recovery and predictive accuracy compared to state-of-the-art symbolic regression methods.

We propose a Hierarchical Multi-scale Knowledge-aware Graph Network (HMKGN) that models multi-scale interactions and spatially hierarchical relationships within whole-slide images (WSIs) for cancer prognostication. Unlike conventional attention-based MIL, which ignores spatial organization, or graph-based MIL, which relies on static handcrafted graphs, HMKGN enforces a hierarchical structure with spatial locality constraints, wherein local cellular-level dynamic graphs aggregate spatially proximate patches within each region of interest (ROI) and a global slide-level dynamic graph integrates ROI-level features into WSI-level representations. Moreover, multi-scale integration at the ROI level combines coarse contextual features from broader views with fine-grained structural representations from local patch-graph aggregation. We evaluate HMKGN on four TCGA cohorts (KIRC, LGG, PAAD, and STAD; N=513, 487, 138, and 370) for survival prediction. It consistently outperforms existing MIL-based models, yielding improved concordance indices (10.85% better) and statistically significant stratification of patient survival risk (log-rank p < 0.05).

Large-scale Graph Neural Networks (GNNs) are typically trained by sampling a vertex's neighbors to a fixed distance. Because large input graphs are distributed, training requires frequent irregular communication that stalls forward progress. Moreover, fetched data changes with graph, graph distribution, sample and batch parameters, and caching polices. Consequently, any static prefetching method will miss crucial opportunities to adapt to different dynamic conditions. In this paper, we introduce Rudder, a software module embedded in the state-of-the-art AWS DistDGL framework, to autonomously prefetch remote nodes and minimize communication. Rudder's adaptation contrasts with both standard heuristics and traditional ML classifiers. We observe that the generative AI found in contemporary Large Language Models (LLMs) exhibits emergent properties like In-Context Learning (ICL) for zero-shot tasks, with logical multi-step reasoning. We find this behavior well-suited for adaptive control even with substantial undertraining. Evaluations using standard datasets and unseen configurations on the NERSC Perlmutter supercomputer show up to 91% improvement in end-to-end training performance over baseline DistDGL (no prefetching), and an 82% improvement over static prefetching, reducing communication by over 50%. Our code is available at https://github.com/aishwaryyasarkar/rudder-llm-agent.

Sentences like "She will go to France or Spain, or perhaps to Germany or France." appear formally redundant, yet become acceptable in contexts such as "Mary will go to a philosophy program in France or Spain, or a mathematics program in Germany or France." While this phenomenon has typically been analyzed using symbolic formal representations, we aim to provide a complementary account grounded in artificial neural mechanisms. We first present new behavioral evidence from humans and large language models demonstrating the robustness of this apparent non-redundancy across contexts. We then show that, in language models, redundancy avoidance arises from two interacting mechanisms: models learn to bind contextually relevant information to repeated lexical items, and Transformer induction heads selectively attend to these context-licensed representations. We argue that this neural explanation sheds light on the mechanisms underlying context-sensitive semantic interpretation, and that it complements existing symbolic analyses.

Human reasoning often involves working over limited information to arrive at probabilistic conclusions. In its simplest form, this involves making an inference that is not strictly entailed by a premise, but rather only likely given the premise. While reasoning LLMs have demonstrated strong performance on logical and mathematical tasks, their behavior on such open-ended, non-deterministic inferences remains largely unexplored. We introduce ProbCOPA, a dataset of 210 handcrafted probabilistic inferences in English, each annotated for inference likelihood by 25--30 human participants. We find that human responses are graded and varied, revealing probabilistic judgments of the inferences in our dataset. Comparing these judgments with responses from eight state-of-the-art reasoning LLMs, we show that models consistently fail to produce human-like distributions. Finally, analyzing LLM reasoning chains, we find evidence of a common reasoning pattern used to evaluate such inferences. Our findings reveal persistent differences between humans and LLMs, and underscore the need to evaluate reasoning beyond deterministic settings.

In the real world, planning is often challenged by distribution shifts. As such, a model of the environment obtained under one set of conditions may no longer remain valid as the distribution of states or the environment dynamics change, which in turn causes previously learned strategies to fail. In this work, we propose a theoretical framework for planning under partial observability using Partially Observable Markov Decision Processes (POMDPs) formulated using causal knowledge. By representing shifts in the environment as interventions on this causal POMDP, the framework enables evaluating plans under hypothesized changes and actively identifying which components of the environment have been altered. We show how to maintain and update a belief over both the latent state and the underlying domain, and we prove that the value function remains piecewise linear and convex (PWLC) in this augmented belief space. Preservation of PWLC under distribution shifts has the advantage of maintaining the tractability of planning via $α$-vector-based POMDP methods.

Previous work establishing completeness results for $\textit{counterfactual identification}$ has been circumscribed to the setting where the input data belongs to observational or interventional distributions (Layers 1 and 2 of Pearl's Causal Hierarchy), since it was generally presumed impossible to obtain data from counterfactual distributions, which belong to Layer 3. However, recent work (Raghavan & Bareinboim, 2025) has formally characterized a family of counterfactual distributions which can be directly estimated via experimental methods - a notion they call $\textit{counterfactual realizabilty}$. This leaves open the question of what $\textit{additional}$ counterfactual quantities now become identifiable, given this new access to (some) Layer 3 data. To answer this question, we develop the CTFIDU+ algorithm for identifying counterfactual queries from an arbitrary set of Layer 3 distributions, and prove that it is complete for this task. Building on this, we establish the theoretical limit of which counterfactuals can be identified from physically realizable distributions, thus implying the $\textit{fundamental limit to exact causal inference in the non-parametric setting}$. Finally, given the impossibility of identifying certain critical types of counterfactuals, we derive novel analytic bounds for such quantities using realizable counterfactual data, and corroborate using simulations that counterfactual data helps tighten the bounds for non-identifiable quantities in practice.

Automated placement of components on printed circuit boards (PCBs) is a critical stage in placement layout design. While reinforcement learning (RL) has been successfully applied to system-on-chip IP block placement and chiplet arrangement in complex packages, PCB component placement presents unique challenges due to several factors: variation in component sizes, single- and double-sided boards, wirelength constraints, board constraints, and non-overlapping placement requirements. In this work, we adopt a component-centric layout for automating PCB component placement using RL: first, the main component is fixed at the center, while passive components are placed in proximity to the pins of the main component. Free space around the main component is discretized, drastically reducing the search space while still covering all feasible placement; second, we leverage prior knowledge that each passive's position has to be near to its corresponding voltage source. This allows us to design the reward function which avoids wasted exploration of infeasible or irrelevant search space. Using the component centric layout, we implemented different methods including Deep Q-Network, Actor-Critic algorithm and Simulated Annealing. Evaluation on over nine real-world PCBs of varying complexity shows that our best proposed method approaches near human-like placements in terms of wirelength and feasibility.

We study the statistical complexity of estimating partition functions given sample access to a proposal distribution and an unnormalized density ratio for a target distribution. While partition function estimation is a classical problem, existing guarantees typically rely on structural assumptions about the domain or model geometry. We instead provide a general, information-theoretic characterization that depends only on the relationship between the proposal and target distributions. Our analysis introduces the integrated coverage profile, a functional that quantifies how much target mass lies in regions where the density ratio is large. We show that integrated coverage tightly characterizes the sample complexity of multiplicative partition function estimation and provide matching lower bounds. We further express these bounds in terms of $f$-divergences, yielding sharp phase transitions depending on the growth rate of f and recovering classical results as a special case while extending to heavy-tailed regimes. Matching lower bounds establish tightness in all regimes. As applications, we derive improved finite-sample guarantees for importance sampling and self-normalized importance sampling, and we show a strict separation between the complexity of approximate sampling and counting under the same divergence constraints. Our results unify and generalize prior analyses of importance sampling, rejection sampling, and heavy-tailed mean estimation, providing a minimal-assumption theory of partition function estimation. Along the way we introduce new technical tools including new connections between coverage and $f$-divergences as well as a generalization of the classical Paley-Zygmund inequality.

Cascaded or stacked intelligent metasurfaces (SIMs) have emerged as a promising technology to overcome the physical limitations of single-layer reconfigurable intelligent surfaces (RISs) in wideband wireless communication. By intelligently manipulating electromagnetic waves, SIMs enhance signal propagation in complex environments and offer additional degrees of freedom for beamforming. This paper proposes a coupling-aware, wideband, circuit-based framework that captures frequency-dependent mutual coupling and wideband channel responses over multiple subbands. Based on this model, we formulate a joint active and passive beamforming design that optimizes the base-station precoder to enable carrier aggregation across frequency-selective subbands, together with metasurface phase shifts, to maximize spectral efficiency. Simulation results reveal the importance of accounting for coupling and wideband effects, and show that performance depends strongly on operating conditions. Single-layer RIS configurations can be favorable in narrowband and/or low-SNR regimes, whereas SIMs can significantly outperform under wideband multi-user conditions by mitigating coupling-induced distortion and maintaining a more consistent phase response across frequencies. The results provide physical insights into design trade-offs between structural simplicity and wideband adaptability, highlighting SIMs as a scalable solution for future-generation wideband multi-user MIMO systems. We further show that partially reconfigurable SIM architectures achieve near-optimal performance with reduced complexity.

Foundation models pretrained on large-scale 3D medical imaging data face challenges when adapted to multiple downstream tasks under continual learning with limited labeled data. We address few-shot continual learning for 3D brain MRI by combining a frozen pretrained backbone with task-specific Low-Rank Adaptation (LoRA) modules. Tasks arrive sequentially -- tumor segmentation (BraTS) and brain age estimation (IXI) -- with no replay of previous task data. Each task receives a dedicated LoRA adapter; only the adapter and task-specific head are trained while the backbone remains frozen, thereby eliminating catastrophic forgetting by design (BWT=0). In continual learning, sequential full fine-tuning suffers severe forgetting (T1 Dice drops from 0.80 to 0.16 after T2), while sequential linear probing achieves strong T1 (Dice 0.79) but fails on T2 (MAE 1.45). Our LoRA approach achieves the best balanced performance across both tasks: T1 Dice 0.62$\pm$0.07, T2 MAE 0.16$\pm$0.05, with zero forgetting and $<$0.1\% trainable parameters per task, though with noted systematic age underestimation in T2 (Wilcoxon $p<0.001$). Frozen foundation models with task-specific LoRA adapters thus offer a practical solution when both tasks must be maintained under few-shot continual learning.

Subadditive set functions play a pivotal role in computational economics (especially in combinatorial auctions), combinatorial optimization or artificial intelligence applications such as interpretable machine learning. However, specifying a set function requires assigning values to an exponentially large number of subsets in general, a task that is often resource-intensive in practice, particularly when the values derive from external sources such as retraining of machine learning models. A~simple omission of certain values introduces ambiguity that becomes even more significant when the incomplete set function has to be further optimized over. Motivated by the well-known result about inapproximability of subadditive functions using deterministic value queries with respect to a multiplicative error, we study a problem of approximating an unknown subadditive (or a subclass of thereof) set function with respect to an additive error -- i. e., we aim to efficiently close the distance between minimal and maximal completions. Our contributions are threefold: (i) a thorough exploration of minimal and maximal completions of different classes of set functions with missing values and an analysis of their resulting distance; (ii) the development of methods to minimize this distance over classes of set functions with a known prior, achieved by disclosing values of additional subsets in both offline and online manner; and (iii) empirical demonstrations of the algorithms' performance in practical scenarios.

Operator learning is reshaping scientific computing by amortizing inference across infinite families of problems. While neural operators (NOs) are increasingly well understood for regression, far less is known for classification and its unsupervised analogue: clustering. We prove that sample-based neural operators can learn any finite collection of classes in an infinite-dimensional reproducing kernel Hilbert space, even when the classes are neither convex nor connected, under mild kernel sampling assumptions. Our universal clustering theorem shows that any $K$ closed classes can be approximated to arbitrary precision by NO-parameterized classes in the upper Kuratowski topology on closed sets, a notion that can be interpreted as disallowing false-positive misclassifications. Building on this, we develop an NO-powered clustering pipeline for functional data and apply it to unlabeled families of ordinary differential equation (ODE) trajectories. Discretized trajectories are lifted by a fixed pre-trained encoder into a continuous feature map and mapped to soft assignments by a lightweight trainable head. Experiments on diverse synthetic ODE benchmarks show that the resulting practical SNO recovers latent dynamical structure in regimes where classical methods fail, providing evidence consistent with our universal clustering theory.

Reachability analysis has become increasingly important in robotics to distinguish safe from unsafe states. Unfortunately, existing reachability and safety analysis methods often fall short, as they typically require known system dynamics or large datasets to estimate accurate system models, are computationally expensive, and assume full state information. A recent method, called MORALS, aims to address these shortcomings by using topological tools to estimate3DR-eEgnciodnesr of Attraction (ROA) in a low-dimensional latent space. However, MORALS still relies on full state knowledge and has not been studied when only sensor measurements are available. This paper presents Visual Morse Graph-Aided Estimation of Regions of Attraction in a Learned Latent Space (V- MORALS). V-MORALS takes in a dataset of image-based trajectories of a system under a given controller, and learns a latent space for reachability analysis. Using this learned latent space, our method is able to generate well-defined Morse Graphs, from which we can compute ROAs for various systems and controllers. V-MORALS provides capabilities similar to the original MORALS architecture without relying on state knowledge, and using only high-level sensor data. Our project website is at: https://v-morals.onrender.com.

Deep generative models (DGMs) compress high-dimensional data but often entangle distinct physical factors in their latent spaces. We present an auxiliary-variable-guided framework for disentangling representations of thermal Sunyaev-Zel'dovich (tSZ) maps of dark matter halos. We introduce halo mass and concentration as auxiliary variables and apply a lightweight alignment penalty to encourage latent dimensions to reflect these physical quantities. To generate sharp and realistic samples, we extend latent conditional flow matching (LCFM), a state-of-the-art generative model, to enforce disentanglement in the latent space. Our Disentangled Latent-CFM (DL-CFM) model recovers the established mass-concentration scaling relation and identifies latent space outliers that may correspond to unusual halo formation histories. By linking latent coordinates to interpretable astrophysical properties, our method transforms the latent space into a diagnostic tool for cosmological structure. This work demonstrates that auxiliary guidance preserves generative flexibility while yielding physically meaningful, disentangled embeddings, providing a generalizable pathway for uncovering independent factors in complex astronomical datasets.

Differentially Private Stochastic Gradient Descent (DP-SGD) is a cornerstone technique for ensuring privacy in deep learning, widely used in both training from scratch and fine-tuning large-scale language models. While DP-SGD predominantly relies on the Gaussian mechanism, the Laplace mechanism remains underutilized due to its reliance on L1 norm clipping. This constraint severely limits its practicality in high-dimensional models because the L1 norm of an n-dimensional gradient can be up to sqrt(n) times larger than its L2 norm. As a result, the required noise scale grows significantly with model size, leading to poor utility or untrainable models. In this work, we introduce Lap2, a new solution that enables L2 clipping for Laplace DP-SGD while preserving strong privacy guarantees. We overcome the dimensionality-driven clipping barrier by computing coordinate-wise moment bounds and applying majorization theory to construct a tight, data-independent upper bound over the full model. By exploiting the Schur-convexity of the moment accountant function, we aggregate these bounds using a carefully designed majorization set that respects the L2 clipping constraint. This yields a multivariate privacy accountant that scales gracefully with model dimension and enables the use of thousands of moments. Empirical evaluations demonstrate that our approach significantly improves the performance of Laplace DP-SGD, achieving results comparable to or better than Gaussian DP-SGD under strong privacy constraints. For instance, fine-tuning RoBERTa-base (125M parameters) on SST-2 achieves 87.88% accuracy at epsilon=0.54, outperforming Gaussian (87.16%) and standard Laplace (48.97%) under the same budget.

Limitations on the availability of Dynamic Vision Sensors (DVS) present a fundamental challenge to researchers of neuromorphic computer vision applications. In response, datasets have been created by the research community, but often contain a limited number of samples or scenarios. To address the lack of a comprehensive simulator of neuromorphic vision datasets, we introduce the Anomalous Neuromorphic Tool for Shapes (ANTShapes), a novel dataset simulation framework. Built in the Unity engine, ANTShapes simulates abstract, configurable 3D scenes populated by objects displaying randomly-generated behaviours describing attributes such as motion and rotation. The sampling of object behaviours, and the labelling of anomalously-acting objects, is a statistical process following central limit theorem principles. Datasets containing an arbitrary number of samples can be created and exported from ANTShapes, along with accompanying label and frame data, through the adjustment of a limited number of parameters within the software. ANTShapes addresses the limitations of data availability to researchers of event-based computer vision by allowing for the simulation of bespoke datasets to suit purposes including object recognition and localisation alongside anomaly detection.

Medical image segmentation models are typically optimised with voxel-wise losses that constrain predictions only in the output space. This leaves latent feature representations largely unconstrained, potentially limiting generalisation. We propose {SegReg}, a latent-space regularisation framework that operates on feature maps of U-Net models to encourage structured embeddings while remaining fully compatible with standard segmentation losses. Integrated with the nnU-Net framework, we evaluate SegReg on prostate, cardiac, and hippocampus segmentation and demonstrate consistent improvements in domain generalisation. Furthermore, we show that explicit latent regularisation improves continual learning by reducing task drift and enhancing forward transfer across sequential tasks without adding memory or any extra parameters. These results highlight latent-space regularisation as a practical approach for building more generalisable and continual-learning-ready models.

Background: Clinical prediction models are increasingly used to inform healthcare decisions, but determining the minimum sample size for their development remains a critical and unresolved challenge. Inadequate sample sizes can lead to overfitting, poor generalisability, and biased predictions. Existing approaches, such as heuristic rules, closed-form formulas, and simulation-based methods, vary in flexibility and accuracy, particularly for complex data structures and machine learning models. Methods: We review current methodologies for sample size estimation in prediction modelling and introduce a conceptual framework that distinguishes between mean-based and assurance-based criteria. Building on this, we propose a novel simulation-based approach that integrates learning curves, Gaussian Process optimisation, and assurance principles to identify sample sizes that achieve target performance with high probability. This approach is implemented in pmsims, an open-source, model-agnostic R package. Results: Through case studies, we demonstrate that sample size estimates vary substantially across methods, performance metrics, and modelling strategies. Compared to existing tools, pmsims provides flexible, efficient, and interpretable solutions that accommodate diverse models and user-defined metrics while explicitly accounting for variability in model performance. Conclusions: Our framework and software advance sample size methodology for clinical prediction modelling by combining flexibility with computational efficiency. Future work should extend these methods to hierarchical and multimodal data, incorporate fairness and stability metrics, and address challenges such as missing data and complex dependency structures.

Federated Learning (FL) enables a group of clients to collaboratively train a model without sharing individual data, but its performance drops when client data are heterogeneous. Clustered FL tackles this by grouping similar clients. However, existing clustered FL approaches rely solely on either data similarity or gradient similarity; however, this results in an incomplete assessment of client similarities. Prior clustered FL approaches also restrict knowledge and representation sharing to clients within the same cluster. This prevents cluster models from benefiting from the diverse client population across clusters. To address these limitations, FedDAG introduces a clustered FL framework, FedDAG, that employs a weighted, class-wise similarity metric that integrates both data and gradient information, providing a more holistic measure of similarity during clustering. In addition, FedDAG adopts a dual-encoder architecture for cluster models, comprising a primary encoder trained on its own clients' data and a secondary encoder refined using gradients from complementary clusters. This enables cross-cluster feature transfer while preserving cluster-specific specialization. Experiments on diverse benchmarks and data heterogeneity settings show that FedDAG consistently outperforms state-of-the-art clustered FL baselines in accuracy.

We introduce spiky rank, a new matrix parameter that enhances blocky rank by combining the combinatorial structure of the latter with linear-algebraic flexibility. A spiky matrix is block-structured with diagonal blocks that are arbitrary rank-one matrices, and the spiky rank of a matrix is the minimum number of such matrices required to express it as a sum. This measure extends blocky rank to real matrices and is more robust for problems with both combinatorial and algebraic character. Our conceptual contribution is as follows: we propose spiky rank as a well-behaved candidate matrix complexity measure and demonstrate its potential through applications. We show that large spiky rank implies high matrix rigidity, and that spiky rank lower bounds yield lower bounds for depth-2 ReLU circuits, the basic building blocks of neural networks. On the technical side, we establish tight bounds for random matrices and develop a framework for explicit lower bounds, applying it to Hamming distance matrices and spectral expanders. Finally, we relate spiky rank to other matrix parameters, including blocky rank, sparsity, and the $γ_2$-norm.

Collecting a high-quality dataset is a critical task that demands meticulous attention to detail, as overlooking certain aspects can render the entire dataset unusable. Autonomous driving challenges remain a prominent area of research, requiring further exploration to enhance the perception and planning performance of vehicles. However, existing datasets are often incomplete. For instance, datasets that include perception information generally lack planning data, while planning datasets typically consist of extensive driving sequences where the ego vehicle predominantly drives forward, offering limited behavioral diversity. In addition, many real datasets struggle to evaluate their models, especially for planning tasks, since they lack a proper closed-loop evaluation setup. The CARLA Leaderboard 2.0 challenge, which provides a diverse set of scenarios to address the long-tail problem in autonomous driving, has emerged as a valuable alternative platform for developing perception and planning models in both open-loop and closed-loop evaluation setups. Nevertheless, existing datasets collected on this platform present certain limitations. Some datasets appear to be tailored primarily for limited sensor configuration, with particular sensor configurations. To support end-to-end autonomous driving research, we have collected a new dataset comprising over 2.85 million frames using the CARLA simulation environment for the diverse Leaderboard 2.0 challenge scenarios. Our dataset is designed not only for planning tasks but also supports dynamic object detection, lane divider detection, centerline detection, traffic light recognition, prediction tasks and visual language action models . Furthermore, we demonstrate its versatility by training various models using our dataset. Moreover, we also provide numerical rarity scores to understand how rarely the current state occurs in the dataset.

Concept Bottleneck Models (CBMs) provide inherent interpretability by first mapping input samples to high-level semantic concepts, followed by a combination of these concepts for the final classification. However, the annotation of human-understandable concepts requires extensive expert knowledge and labor, constraining the broad adoption of CBMs. On the other hand, there are a few works that leverage the knowledge of large language models (LLMs) to construct concept bottlenecks. Nevertheless, they face two essential limitations: First, they overlook the uncertainty associated with the concepts annotated by LLMs and lack a valid mechanism to quantify uncertainty about the annotated concepts, increasing the risk of errors due to hallucinations from LLMs. Additionally, they fail to incorporate the uncertainty associated with these annotations into the learning process for concept bottleneck models. To address these limitations, we propose a novel uncertainty-aware CBM method, which not only rigorously quantifies the uncertainty of LLM-annotated concept labels with valid and distribution-free guarantees, but also incorporates quantified concept uncertainty into the CBM training procedure to account for varying levels of reliability across LLM-annotated concepts. We also provide the theoretical analysis for our proposed method. Extensive experiments on the real-world datasets validate the desired properties of our proposed methods.

Understanding and extracting structured insights from unstructured documents remains a foundational challenge in industrial NLP. While Large Language Models (LLMs) enable zero-shot extraction, traditional pipelines often fail to handle multi-document packets, complex reasoning, and strict compliance requirements. We present IDP (Intelligent Document Processing) Accelerator, a framework enabling agentic AI for end-to-end document intelligence with four key components: (1) DocSplit, a novel benchmark dataset and multimodal classifier using BIO tagging to segment complex document packets; (2) configurable Extraction Module leveraging multimodal LLMs to transform unstructured content into structured data; (3) Agentic Analytics Module, compliant with the Model Context Protocol (MCP) providing data access through secure, sandboxed code execution; and (4) Rule Validation Module replacing deterministic engines with LLM-driven logic for complex compliance checks. The interactive demonstration enables users to upload document packets, visualize classification results, and explore extracted data through an intuitive web interface. We demonstrate effectiveness across industries, highlighting a production deployment at a leading healthcare provider achieving 98% classification accuracy, 80% reduced processing latency, and 77% lower operational costs over legacy baselines. IDP Accelerator is open-sourced with a live demonstration available to the community.

Though patients are increasingly granted digital access to their electronic health records (EHRs), existing interfaces may not support precise, trustworthy answers to patient-specific questions. Large language models (LLM) show promise in clinical question answering (QA), but retrieval-based approaches are computationally inefficient, prone to hallucination, and difficult to deploy over real-life EHRs. In this work, we introduce FHIRPath-QA, the first open dataset and benchmark for patient-specific QA that includes open-standard FHIRPath queries over real-world clinical data. We propose a text-to-FHIRPath QA paradigm that shifts reasoning from free-text generation to FHIRPath query synthesis, significantly reducing LLM usage. Built on MIMIC-IV on FHIR Demo, the dataset pairs over 14k natural language questions in patient and clinician phrasing with validated FHIRPath queries and answers. Further, we demonstrate that state-of-the-art LLMs struggle to deal with ambiguity in patient language and perform poorly in FHIRPath query synthesis. However, they benefit strongly from supervised fine-tuning. Our results highlight that text-to-FHIRPath synthesis has the potential to serve as a practical foundation for safe, efficient, and interoperable consumer health applications, and our dataset and benchmark serve as a starting point for future research on the topic. The full dataset and generation code is available at: https://github.com/mooshifrew/fhirpath-qa.

Transformer architectures, capable of capturing sequential dependencies in the history of user interactions, have become the dominant approach in sequential recommender systems. Despite their success, such models consider sequence elements in isolation, implicitly accounting for the complex relationships between them. Graph neural networks, in contrast, explicitly model these relationships through higher order interactions but are often unable to adequately capture their evolution over time, limiting their use for predicting the next interaction. To fill this gap, we present a new framework that combines transformers and graph neural networks and aligns different representations for solving next-item prediction task. Our solution simultaneously encodes structural dependencies in the interaction graph and tracks their dynamic change. Experimental results on a number of open datasets demonstrate that the proposed framework consistently outperforms both pure sequential and graph approaches in terms of recommendation quality, as well as recent methods that combine both types of signals.

Multi-Agent Path Finding (MAPF) aims to move agents from their start to goal vertices on a graph. Lifelong MAPF (LMAPF) continuously assigns new goals to agents as they complete current ones. To guide agents' movement in LMAPF, prior works have proposed Guidance Graph Optimization (GGO) methods to optimize a guidance graph, which is a bidirected weighted graph whose directed edges represent moving and waiting actions with edge weights being action costs. Higher edge weights represent higher action costs. However, edge weights only provide soft guidance. An edge with a high weight only discourages agents from using it, instead of prohibiting agents from traversing it. In this paper, we explore the need to incorporate edge directions optimization into GGO, providing strict guidance. We generalize GGO to Mixed Guidance Graph Optimization (MGGO), presenting two MGGO methods capable of optimizing both edge weights and directions. The first optimizes edge directions and edge weights in two phases separately. The second applies Quality Diversity algorithms to optimize a neural network capable of generating edge directions and weights. We also incorporate traffic patterns relevant to edge directions into a GGO method, making it capable of generating edge-direction-aware guidance graphs.

Can Machine Learning (ML) autonomously recover interpretable mathematical structure from raw numerical data? We aim to answer this question using the classification of real-root configurations of polynomials up to degree five as a structured benchmark. We tested an extensive set of ML models, including decision trees, logistic regression, support vector machines, random forest, gradient boosting, XGBoost, symbolic regression, and neural networks. Neural networks achieved strong in-distribution performance on quintic classification using raw coefficients alone (84.3% + or - 0.9% balanced accuracy), whereas decision trees perform substantially worse (59.9% + or - 0.9\%). However, when provided with an explicit feature capturing sign changes at critical points, decision trees match neural performance (84.2% + or - 1.2%) and yield explicit classification rules. Knowledge distillation reveals that this single invariant accounts for 97.5% of the extracted decision structure. Out-of-distribution, data-efficiency, and noise robustness analyses indicate that neural networks learn continuous, data-dependent geometric approximations of the decision boundary rather than recovering scale-invariant symbolic rules. This distinction between geometric approximation and symbolic invariance explains the gap between predictive performance and interpretability observed across models. Although high predictive accuracy is attainable, we find no evidence that the evaluated ML models autonomously recover discrete, human-interpretable mathematical rules from raw coefficients. These results suggest that, in structured mathematical domains, interpretability may require explicit structural inductive bias rather than purely data-driven approximation.

Multi-scalar multiplication (MSM), defined as MSM(P, x) = sum_{i=1}^n x_i P_i, is a dominant computational kernel in discrete-logarithm-based cryptography and often becomes a bottleneck for verifiers and other resource-constrained clients. We present 2G2T, a simple protocol for verifiably outsourcing MSM to an untrusted server. After a one-time keyed setup for fixed bases P = (P1, ..., Pn) that produces a public merged-bases vector T and client secret state, the server answers each query x = (x1, ..., xn) with only two group elements: A claimed to equal MSM(P, x) and an auxiliary value B claimed to equal MSM(T, x). Verification requires a single length-n field inner product and a constant number of group operations (two scalar multiplications and one addition), while the server performs two MSMs. In our Ristretto255 implementation, verification is up to ~300x faster than computing the MSM locally using a highly optimized MSM routine for n up to 2^18, and the server-to-client response is constant-size (two compressed group elements, 64 bytes on Ristretto255). Despite its simplicity and efficiency, 2G2T achieves statistical soundness: for any (even computationally unbounded) adversarial server, the probability of accepting an incorrect result is at most 1/q per query, and at most e/q over e adaptive executions, in a prime-order group of size q.

Data assimilation (DA) for compressible flows with shocks is challenging because many classical DA methods generate spurious oscillations and nonphysical features near uncertain shocks. We focus here on the ensemble Kalman filter (EnKF). We show that the poor performance of the standard EnKF may be attributed to the bimodal forecast distribution that can arise in the vicinity of an uncertain shock location; this violates the assumptions underpinning the EnKF, which assume a forecast which is close to Gaussian. To address this issue we introduce the new neural EnKF. The basic idea is to systematically embed neural function approximations within ensemble DA by mapping the forecast ensemble of shocked flows to the parameter space (weights and biases) of a deep neural network (NN) and to subsequently perform DA in that space. The nonlinear mapping encodes sharp and smooth flow features in an ensemble of NN parameters. Neural EnKF updates are therefore well-behaved only if the NN parameters vary smoothly within the neural representation of the forecast ensemble. We show that such a smooth variation of network parameters can be enforced via physics-informed transfer learning, and demonstrate that in so-doing the neural EnKF avoids the spurious oscillations and nonphysical features that plague the standard EnKF. The applicability of the neural EnKF is demonstrated through a series of systematic numerical experiments with an inviscid Burgers' equation, Sod's shock tube, and a two-dimensional blast wave.

Psychiatric questionnaires are highly context sensitive and often only weakly predict subsequent symptom severity, which makes the prognostic relationship difficult to learn. Although flexible nonlinear models can improve predictive accuracy, their limited interpretability can erode clinical trust. In fields such as imaging and omics, investigators commonly address visit- and instrument-specific artifacts by extracting stable signal through preprocessing and then fitting an interpretable linear model. We adopt the same strategy for questionnaire data by decoupling preprocessing from prediction: we restrict nonlinear capacity to a baseline preprocessing module that estimates stable item values, and then learn a linear mapping from these stabilized baseline items to future severity. We refer to this two-stage method as REFINE (Redundancy-Exploiting Follow-up-Informed Nonlinear Enhancement), which concentrates nonlinearity in preprocessing while keeping the prognostic relationship transparently linear and therefore globally interpretable through a coefficient matrix, rather than through post hoc local attributions. In experiments, REFINE outperforms other interpretable approaches while preserving clear global attribution of prognostic factors across psychiatric and non-psychiatric longitudinal prediction tasks.

Lightweight neural network accelerators are essential for edge devices with limited resources and power constraints. While quantization and binarization can efficiently reduce hardware cost, they still rely on the conventional Artificial Neural Network (ANN) computation pattern. The recently proposed Kolmogorov-Arnold Network (KAN) presents a novel network paradigm built on learnable nonlinear functions. However, it is computationally expensive for hardware deployment. Inspired by KAN, we propose BiKA, a multiply-free architecture that replaces nonlinear functions with binary, learnable thresholds, introducing an extremely lightweight computational pattern that requires only comparators and accumulators. Our FPGA prototype on Ultra96-V2 shows that BiKA reduces hardware resource usage by 27.73% and 51.54% compared with binarized and quantized neural network systolic array accelerators, while maintaining competitive accuracy. BiKA provides a promising direction for hardware-friendly neural network design on edge devices.

Scientific research relies on accurate citation for attribution and integrity, yet large language models (LLMs) introduce a new risk: fabricated references that appear plausible but correspond to no real publications. Such hallucinated citations have already been observed in submissions and accepted papers at major machine learning venues, exposing vulnerabilities in peer review. Meanwhile, rapidly growing reference lists make manual verification impractical, and existing automated tools remain fragile to noisy and heterogeneous citation formats and lack standardized evaluation. We present the first comprehensive benchmark and detection framework for hallucinated citations in scientific writing. Our multi-agent verification pipeline decomposes citation checking into claim extraction, evidence retrieval, passage matching, reasoning, and calibrated judgment to assess whether a cited source truly supports its claim. We construct a large-scale human-validated dataset across domains and define unified metrics for citation faithfulness and evidence alignment. Experiments with state-of-the-art LLMs reveal substantial citation errors and show that our framework significantly outperforms prior methods in both accuracy and interpretability. This work provides the first scalable infrastructure for auditing citations in the LLM era and practical tools to improve the trustworthiness of scientific references.

Detection of rare lesions in whole-body CT is fundamentally limited by extreme class imbalance and low target-to-volume ratios, producing precision collapse despite high AUROC. Synthetic augmentation with diffusion models offers promise, yet pixel-space diffusion is computationally expensive, and existing mask-conditioned approaches lack controllable attribute-level regulation and paired supervision for accountable training. We introduce SALIENT, a mask-conditioned wavelet-domain diffusion framework that synthesizes paired lesion-masking volumes for controllable CT augmentation under long-tail regimes. Instead of denoising in pixel space, SALIENT performs structured diffusion over discrete wavelet coefficients, explicitly separating low-frequency brightness from high-frequency structural detail. Learnable frequency-aware objectives disentangle target and background attributes (structure, contrast, edge fidelity), enabling interpretable and stable optimization. A 3D VAE generates diverse volumetric lesion masks, and a semi-supervised teacher produces paired slice-level pseudo-labels for downstream mask-guided detection. SALIENT improves generative realism, as reflected by higher MS-SSIM (0.63 to 0.83) and lower FID (118.4 to 46.5). In a separate downstream evaluation, SALIENT-augmented training improves long-tail detection performance, yielding disproportionate AUPRC gains across low prevalences and target-to-volume ratios. Optimal synthetic ratios shift from 2x to 4x as labeled seed size decreases, indicating a seed-dependent augmentation regime under low-label conditions. SALIENT demonstrates that frequency-aware diffusion enables controllable, computationally efficient precision rescue in long-tail CT detection.

Large language models are trained primarily on human-generated data and feedback, yet they exhibit persistent errors arising from annotation noise, subjective preferences, and the limited expressive bandwidth of natural language. We argue that these limitations reflect structural properties of the supervision channel rather than model scale or optimization. We develop a unified theory showing that whenever the human supervision channel is not sufficient for a latent evaluation target, it acts as an information-reducing channel that induces a strictly positive excess-risk floor for any learner dominated by it. We formalize this Human-Bounded Intelligence limit and show that across six complementary frameworks (operator theory, PAC-Bayes, information theory, causal inference, category theory, and game-theoretic analyses of reinforcement learning from human feedback), non-sufficiency yields strictly positive lower bounds arising from the same structural decomposition into annotation noise, preference distortion, and semantic compression. The theory explains why scaling alone cannot eliminate persistent human-aligned errors and characterizes conditions under which auxiliary non-human signals (e.g., retrieval, program execution, tools) increase effective supervision capacity and collapse the floor by restoring information about the latent target. Experiments on real preference data, synthetic known-target tasks, and externally verifiable benchmarks confirm the predicted structural signatures: human-only supervision exhibits a persistent floor, while sufficiently informative auxiliary channels strictly reduce or eliminate excess error.

Training large language models to reason with search engines via reinforcement learning is hindered by a fundamental credit assignment problem: existing methods such as Search-R1 provide only a sparse outcome reward after an entire multi-step trajectory, making it infeasible to attribute success or failure to individual reasoning and retrieval decisions. Process-reward methods like StepSearch alleviate this by introducing step-level supervision, but rely on heuristic rewards such as TF-IDF overlap with gold documents, and still sample k complete trajectories per example, retaining high gradient variance. We propose SLATE, a framework built on two complementary ideas: (1) truncated step-level sampling, which generates k trajectories that share a common prefix and differ only at the next step, and (2) dense LLM-as-judge rewards, which replace heuristic scoring with a capable LLM evaluator that assesses the quality of each reasoning step, search query, and answer, providing richer and more reliable supervision. We theoretically prove that under the same dense reward structure, truncated sampling reduces the variance of advantage estimates by up to a factor of T compared to full-trajectory sampling for T-step trajectories, yielding lower-variance, better-targeted policy gradients. Experiments on seven QA benchmarks confirm that SLATE consistently outperforms both sparse-reward and process-reward baselines, with the largest gains on harder multi-hop tasks and smaller models.

Graphic layouts serve as an important and engaging medium for visual communication across different channels. While recent layout generation models have demonstrated impressive capabilities, they frequently fail to align with nuanced human aesthetic judgment. Existing preference datasets and reward models trained on text-to-image generation do not generalize to layout evaluation, where the spatial arrangement of identical elements determines quality. To address this critical gap, we introduce DesignSense-10k, a large-scale dataset of 10,235 human-annotated preference pairs for graphic layout evaluation. We propose a five-stage curation pipeline that generates visually coherent layout transformations across diverse aspect ratios, using semantic grouping, layout prediction, filtering, clustering, and VLM-based refinement to produce high-quality comparison pairs. Human preferences are annotated using a 4-class scheme (left, right, both good, both bad) to capture subjective ambiguity. Leveraging this dataset, we train DesignSense, a vision-language model-based classifier that substantially outperforms existing open-source and proprietary models across comprehensive evaluation metrics (54.6% improvement in Macro F1 over the strongest proprietary baseline). Our analysis shows that frontier VLMs remain unreliable overall and fail catastrophically on the full four-class task, underscoring the need for specialized, preference-aware models. Beyond the dataset, our reward model DesignSense yields tangible downstream gains in layout generation. Using our judge during RL based training improves generator win rate by about 3%, while inference-time scaling, which involves generating multiple candidates and selecting the best one, provides a 3.6% improvement. These results highlight the practical impact of specialized, layout-aware preference modeling on real-world layout generation quality.

The Lagrangian Particles (LP) module of the PLUTO code offers a powerful simulation tool to predict the non-thermal emission produced by shock accelerated particles in large-scale relativistic magnetized astrophysics flows. The LPs represent ensembles of relativistic particles with a given energy distribution which is updated by solving the relativistic cosmic ray transport equation. The approach consistently includes the effects of adiabatic expansion, synchrotron and inverse Compton emission. The large scale nature of such systems creates boundless computational demand which can only be satisfied by targeting modern computing hardware such as Graphic Processing Units (GPUs). In this work we presents the GPU-compatible C++ re-design of the LP module, that, by means of the programming model OpenACC and the Message Passing Interface library, is capable of targeting both single commercial GPUs as well as multi-node (pre-)exascale computing facilities. The code has been benchmarked up to 28672 parallel CPUs cores and 1024 parallel GPUs demonstrating $\sim(80-90)\%$ weak scaling parallel efficiency and good strong scaling capabilities. Our results demonstrated a speedup of $6$ times when solving that same benchmark test with 128 full GPU nodes (4GPUs per node) against the same amount of full high-end CPU nodes (112 cores per node). Furthermore, we conducted a code verification by comparing its prediction to corresponding analytical solutions for two test cases. We note that this work is part of broader project that aims at developing gPLUTO, the novel and revised GPU-ready implementation of its legacy.

Numerous lines of aim to control $\textit{model disagreement}$ -- the extent to which two machine learning models disagree in their predictions. We adopt a simple and standard notion of model disagreement in real-valued prediction problems, namely the expected squared difference in predictions between two models trained on independent samples, without any coordination of the training processes. We would like to be able to drive disagreement to zero with some natural parameter(s) of the training procedure using analyses that can be applied to existing training methodologies. We develop a simple general technique for proving bounds on independent model disagreement based on $\textit{anchoring}$ to the average of two models within the analysis. We then apply this technique to prove disagreement bounds for four commonly used machine learning algorithms: (1) stacked aggregation over an arbitrary model class (where disagreement is driven to 0 with the number of models $k$ being stacked) (2) gradient boosting (where disagreement is driven to 0 with the number of iterations $k$) (3) neural network training with architecture search (where disagreement is driven to 0 with the size $n$ of the architecture being optimized over) and (4) regression tree training over all regression trees of fixed depth (where disagreement is driven to 0 with the depth $d$ of the tree architecture). For clarity, we work out our initial bounds in the setting of one-dimensional regression with squared error loss -- but then show that all of our results generalize to multi-dimensional regression with any strongly convex loss.

We identify occlusion reasoning as a fundamental yet overlooked aspect for 3D layout-conditioned generation. It is essential for synthesizing partially occluded objects with depth-consistent geometry and scale. While existing methods can generate realistic scenes that follow input layouts, they often fail to model precise inter-object occlusions. We propose SeeThrough3D, a model for 3D layout conditioned generation that explicitly models occlusions. We introduce an occlusion-aware 3D scene representation (OSCR), where objects are depicted as translucent 3D boxes placed within a virtual environment and rendered from desired camera viewpoint. The transparency encodes hidden object regions, enabling the model to reason about occlusions, while the rendered viewpoint provides explicit camera control during generation. We condition a pretrained flow based text-to-image image generation model by introducing a set of visual tokens derived from our rendered 3D representation. Furthermore, we apply masked self-attention to accurately bind each object bounding box to its corresponding textual description, enabling accurate generation of multiple objects without object attribute mixing. To train the model, we construct a synthetic dataset with diverse multi-object scenes with strong inter-object occlusions. SeeThrough3D generalizes effectively to unseen object categories and enables precise 3D layout control with realistic occlusions and consistent camera control.

A dataset server must often distribute the same large payload to many clients, incurring massive communication costs. Since clients frequently operate on diverse hardware and software frameworks, transmitting a pre-trained model is often infeasible; instead, agents require raw data to train their own task-specific models locally. While dataset distillation attempts to compress training signals, current methods struggle to scale to high-resolution data and rarely achieve sufficiently small files. In this paper, we propose Pseudo-Labels as Data (PLADA), a method that completely eliminates pixel transmission. We assume agents are preloaded with a large, generic, unlabeled reference dataset (e.g., ImageNet-1K, ImageNet-21K) and communicate a new task by transmitting only the class labels for specific images. To address the distribution mismatch between the reference and target datasets, we introduce a pruning mechanism that filters the reference dataset to retain only the labels of the most semantically relevant images for the target task. This selection process simultaneously maximizes training efficiency and minimizes transmission payload. Experiments on 10 diverse datasets demonstrate that our approach can transfer task knowledge with a payload of less than 1 MB while retaining high classification accuracy, offering a promising solution for efficient dataset serving.

The Platonic Representation Hypothesis posits that neural networks trained on different modalities converge toward a shared statistical model of the world. Recent work exploits this convergence by aligning frozen pretrained vision and language models with lightweight alignment layers, but typically relies on contrastive losses and millions of paired samples. In this work, we ask whether meaningful alignment can be achieved with substantially less supervision. We introduce a semi-supervised setting in which pretrained unimodal encoders are aligned using a small number of image-text pairs together with large amounts of unpaired data. To address this challenge, we propose SOTAlign, a two-stage framework that first recovers a coarse shared geometry from limited paired data using a linear teacher, then refines the alignment on unpaired samples via an optimal-transport-based divergence that transfers relational structure without overconstraining the target space. Unlike existing semi-supervised methods, SOTAlign effectively leverages unpaired images and text, learning robust joint embeddings across datasets and encoder pairs, and significantly outperforming supervised and semi-supervised baselines.

Recent work such as AlphaEvolve has shown that combining LLM-driven optimization with evolutionary search can effectively improve programs, prompts, and algorithms across domains. In this paradigm, previously evaluated solutions are reused to guide the model toward new candidate solutions. Crucially, the effectiveness of this evolution process depends on the search strategy: how prior solutions are selected and varied to generate new candidates. However, most existing methods rely on fixed search strategies with predefined knobs (e.g., explore-exploit ratios) that remain static throughout execution. While effective in some settings, these approaches often fail to adapt across tasks, or even within the same task as the search space changes over time. We introduce EvoX, an adaptive evolution method that optimizes its own evolution process. EvoX jointly evolves candidate solutions and the search strategies used to generate them, continuously updating how prior solutions are selected and varied based on progress. This enables the system to dynamically shift between different search strategies during the optimization process. Across nearly 200 real-world optimization tasks, EvoX outperforms existing AI-driven evolutionary methods including AlphaEvolve, OpenEvolve, GEPA, and ShinkaEvolve on the majority of tasks.

The lack of reasoning capabilities in Vision-Language Models (VLMs) has remained at the forefront of research discourse. We posit that this behavior stems from a reporting bias in their training data. That is, how people communicate about visual content by default omits tacit information needed to supervise some types of reasoning; e.g., "at the game today!" is a more likely caption than "a photo of 37 people standing behind a field". We investigate the data underlying the popular VLMs OpenCLIP, LLaVA-1.5 and Molmo through the lens of theories from pragmatics, and find that reporting bias results in insufficient representation of four reasoning skills (spatial, temporal, negation, and counting), despite the corpora being of web-scale, and/or synthetically generated. With a set of curated benchmarks, we demonstrate that: (i) VLMs perform poorly on the aforementioned types of reasoning suppressed in the training data by reporting bias; (ii) contrary to popular belief, scaling data size, model size, and to multiple languages does not result in emergence of these skills by default; but, promisingly, (iii) incorporating annotations specifically collected to obtain tacit information is effective. Our findings highlight the need for more intentional training data curation methods, rather than counting on scale for emergence of reasoning capabilities.

Standard mixed-precision training of neural networks requires many bytes of accelerator memory for each model parameter. These bytes reflect not just the parameter itself, but also its gradient and one or more optimizer state variables. With each of these values typically requiring 4 bytes, training even a 7 billion parameter model can be impractical for researchers with less than 100GB of accelerator memory. We introduce FlashOptim, a suite of optimizations that reduces per-parameter memory by over 50% while preserving model quality and API compatibility. Our approach introduces two key techniques. First, we improve master weight splitting by finding and exploiting a tight bound on its quantization error. Second, we design companding functions that greatly reduce the error in 8-bit optimizer state quantization. Together with 16-bit gradients, these techniques reduce AdamW memory from 16 bytes to 7 bytes per parameter, or 5 bytes with gradient release. They also cut model checkpoint sizes by more than half. Experiments with FlashOptim applied to SGD, AdamW, and Lion show no measurable quality degradation on any task from a collection of standard vision and language benchmarks, including Llama-3.1-8B finetuning.

Coarse data arise when learners observe only partial information about samples; namely, a set containing the sample rather than its exact value. This occurs naturally through measurement rounding, sensor limitations, and lag in economic systems. We study Gaussian mean estimation from coarse data, where each true sample $x$ is drawn from a $d$-dimensional Gaussian distribution with identity covariance, but is revealed only through the set of a partition containing $x$. When the coarse samples, roughly speaking, have ``low'' information, the mean cannot be uniquely recovered from observed samples (i.e., the problem is not identifiable). Recent work by Fotakis, Kalavasis, Kontonis, and Tzamos [FKKT21] established that sample-efficient mean estimation is possible when the unknown mean is identifiable and the partition consists of only convex sets. Moreover, they showed that without convexity, mean estimation becomes NP-hard. However, two fundamental questions remained open: (1) When is the mean identifiable under convex partitions? (2) Is computationally efficient estimation possible under identifiability and convex partitions? This work resolves both questions. [...]

Recent advances in machine learning have emphasized the integration of structured optimization components into end-to-end differentiable models, enabling richer inductive biases and tighter alignment with task-specific objectives. In this work, we introduce a novel differentiable approximation to the zero-one loss-long considered the gold standard for classification performance, yet incompatible with gradient-based optimization due to its non-differentiability. Our method constructs a smooth, order-preserving projection onto the n,k-dimensional hypersimplex through a constrained optimization framework, leading to a new operator we term Soft-Binary-Argmax. After deriving its mathematical properties, we show how its Jacobian can be efficiently computed and integrated into binary and multiclass learning systems. Empirically, our approach achieves significant improvements in generalization under large-batch training by imposing geometric consistency constraints on the output logits, thereby narrowing the performance gap traditionally observed in large-batch training.

AI-powered scientific research tools are rapidly being integrated into research workflows, yet the field lacks a clear lens into how researchers use these systems in real-world settings. We present and analyze the Asta Interaction Dataset, a large-scale resource comprising over 200,000 user queries and interaction logs from two deployed tools (a literature discovery interface and a scientific question-answering interface) within an LLM-powered retrieval-augmented generation platform. Using this dataset, we characterize query patterns, engagement behaviors, and how usage evolves with experience. We find that users submit longer and more complex queries than in traditional search, and treat the system as a collaborative research partner, delegating tasks such as drafting content and identifying research gaps. Users treat generated responses as persistent artifacts, revisiting and navigating among outputs and cited evidence in non-linear ways. With experience, users issue more targeted queries and engage more deeply with supporting citations, although keyword-style queries persist even among experienced users. We release the anonymized dataset and analysis with a new query intent taxonomy to inform future designs of real-world AI research assistants and to support realistic evaluation.

Neural network accelerators have been widely applied to edge devices for complex tasks like object tracking, image recognition, etc. Previous works have explored the quantization technologies in related lightweight accelerator designs to reduce hardware resource consumption. However, low precision leads to high accuracy loss in inference. Therefore, mixed-precision quantization becomes an alternative solution by applying different precision in different layers to trade off resource consumption and accuracy. Because regular designs for multiplication on hardware cannot support the precision reconfiguration for a multi-precision Quantized Neural Network (QNN) model in runtime, we propose a runtime reconfigurable multi-precision multi-channel bitwise systolic array design for QNN accelerators. We have implemented and evaluated our work on the Ultra96 FPGA platform. Results show that our work can achieve 1.3185 to 3.5671 times speedup in inferring mixed-precision models and has less critical path delay, supporting a higher clock frequency (250MHz).

A growing number of publications address the best practices to use Large Language Models (LLMs) for software engineering in recent years. However, most of this work focuses on widely-used general purpose programming languages like Python due to their widespread usage training data. The utility of LLMs for software within the industrial process automation domain, with highly-specialized languages that are typically only used in proprietary contexts, remains underexplored. This research aims to utilize and integrate LLMs in the industrial development process, solving real-life programming tasks (e.g., generating a movement routine for a robotic arm) and accelerating the development cycles of manufacturing systems.

The advancement of large language models (LLMs) has accelerated the development of autonomous financial trading systems. While mainstream approaches deploy multi-agent systems mimicking analyst and manager roles, they often rely on abstract instructions that overlook the intricacies of real-world workflows, which can lead to degraded inference performance and less transparent decision-making. Therefore, we propose a multi-agent LLM trading framework that explicitly decomposes investment analysis into fine-grained tasks, rather than providing coarse-grained instructions. We evaluate the proposed framework using Japanese stock data, including prices, financial statements, news, and macro information, under a leakage-controlled backtesting setting. Experimental results show that fine-grained task decomposition significantly improves risk-adjusted returns compared to conventional coarse-grained designs. Crucially, further analysis of intermediate agent outputs suggests that alignment between analytical outputs and downstream decision preferences is a critical driver of system performance. Moreover, we conduct standard portfolio optimization, exploiting low correlation with the stock index and the variance of each system's output. This approach achieves superior performance. These findings contribute to the design of agent structure and task configuration when applying LLM agents to trading systems in practical settings.

Large language models (LLMs) perform increasingly well on biology benchmarks, but it remains unclear whether they uplift novice users -- i.e., enable humans to perform better than with internet-only resources. This uncertainty is central to understanding both scientific acceleration and dual-use risk. We conducted a multi-model, multi-benchmark human uplift study comparing novices with LLM access versus internet-only access across eight biosecurity-relevant task sets. Participants worked on complex problems with ample time (up to 13 hours for the most involved tasks). We found that LLM access provided substantial uplift: novices with LLMs were 4.16 times more accurate than controls (95% CI [2.63, 6.87]). On four benchmarks with available expert baselines (internet-only), novices with LLMs outperformed experts on three of them. Perhaps surprisingly, standalone LLMs often exceeded LLM-assisted novices, indicating that users were not eliciting the strongest available contributions from the LLMs. Most participants (89.6%) reported little difficulty obtaining dual-use-relevant information despite safeguards. Overall, LLMs substantially uplift novices on biological tasks previously reserved for trained practitioners, underscoring the need for sustained, interactive uplift evaluations alongside traditional benchmarks.

Using advanced machine learning techniques, we developed a method for reconstructing precisely the arrival direction and energy of ultra-high-energy cosmic rays from the voltage traces they induced on ground-based radio detector arrays. In our approach, triggered antennas are represented as a graph structure, which serves as input for a graph neural network (GNN). By incorporating physical knowledge into both the GNN architecture and the input data, we improve the precision and reduce the required size of the training set with respect to a fully data-driven approach. This method achieves an angular resolution of 0.092° and an electromagnetic energy reconstruction resolution of 16.4% on simulated data with realistic noise conditions. We also employ uncertainty estimation methods to enhance the reliability of our predictions, quantifying the confidence of the GNN's outputs and providing confidence intervals for both direction and energy reconstruction. Finally, we investigate strategies to verify the model's consistency and robustness under real life variations, with the goal of identifying scenarios in which predictions remain reliable despite domain shifts between simulation and reality.

Self-reflection enables language agents to iteratively refine solutions, yet often produces repetitive outputs that limit reasoning performance. Recent studies have attempted to address this limitation through various approaches, among which increasing reflective diversity has shown promise. Our empirical analysis reveals a strong positive correlation between reflective diversity and task success, further motivating the need for diverse reflection signals. We introduce ParamMem, a parametric memory module that encodes cross-sample reflection patterns into model parameters, enabling diverse reflection generation through temperature-controlled sampling. Building on this module, we propose ParamAgent, a reflection-based agent framework that integrates parametric memory with episodic and cross-sample memory. Extensive experiments on code generation, mathematical reasoning, and multi-hop question answering demonstrate consistent improvements over state-of-the-art baselines. Further analysis reveals that ParamMem is sample-efficient, enables weak-to-strong transfer across model scales, and supports self-improvement without reliance on stronger external model, highlighting the potential of ParamMem as an effective component for enhancing language agents.

Generalized Rapid Action Value Estimation (GRAVE) has been shown to be a strong variant within the Monte-Carlo Tree Search (MCTS) family of algorithms for General Game Playing (GGP). However, its reliance on storing additional win/visit statistics at each node makes its use impractical in memory-constrained environments, thereby limiting its applicability in practice. In this paper, we introduce the GRAVE2, GRAVER and GRAVER2 algorithms, which extend GRAVE through two-level search, node recycling, and a combination of both techniques, respectively. We show that these enhancements enable a drastic reduction in the number of stored nodes while matching the playing strength of GRAVE.

An artificial intelligence (AI) model can be viewed as a function that maps inputs to outputs in high-dimensional spaces. Once designed and well trained, the AI model is applied for inference. However, even optimized AI models can produce inference errors due to aleatoric and epistemic uncertainties. Interestingly, we observed that when inferring multiple samples based on invariant transformations of an input, inference errors can show partial independences due to epistemic uncertainty. Leveraging this insight, we propose a "resampling" based inferencing that applies to a trained AI model with multiple transformed versions of an input, and aggregates inference outputs to a more accurate result. This approach has the potential to improve inference accuracy and offers a strategy for balancing model size and performance.

Leader-follower interaction is an important paradigm in human-robot interaction (HRI). Yet, assigning roles in real time remains challenging for resource-constrained mobile and assistive robots. While large language models (LLMs) have shown promise for natural communication, their size and latency limit on-device deployment. Small language models (SLMs) offer a potential alternative, but their effectiveness for role classification in HRI has not been systematically evaluated. In this paper, we present a benchmark of SLMs for leader-follower communication, introducing a novel dataset derived from a published database and augmented with synthetic samples to capture interaction-specific dynamics. We investigate two adaptation strategies: prompt engineering and fine-tuning, studied under zero-shot and one-shot interaction modes, compared with an untrained baseline. Experiments with Qwen2.5-0.5B reveal that zero-shot fine-tuning achieves robust classification performance (86.66% accuracy) while maintaining low latency (22.2 ms per sample), significantly outperforming baseline and prompt-engineered approaches. However, results also indicate a performance degradation in one-shot modes, where increased context length challenges the model's architectural capacity. These findings demonstrate that fine-tuned SLMs provide an effective solution for direct role assignment, while highlighting critical trade-offs between dialogue complexity and classification reliability on the edge.

Generative virtual staining (VS) models for high-throughput screening (HTS) can provide an estimated posterior distribution of possible biological feature values for each input and cell. However, when evaluating a VS model, the true posterior is unavailable. Existing evaluation protocols only check the accuracy of the marginal distribution over the dataset rather than the predicted posteriors. We introduce information gain (IG) as a cell-wise evaluation framework that enables direct assessment of predicted posteriors. IG is a strictly proper scoring rule and comes with a sound theoretical motivation allowing for interpretability, and for comparing results across models and features. We evaluate diffusion- and GAN-based models on an extensive HTS dataset using IG and other metrics and show that IG can reveal substantial performance differences other metrics cannot.

Machine learning techniques are now routinely encountered in research laboratories across the globe. Impressive progress has been made through ML and AI techniques with regards to large data set processing. This progress has increased the ability of the experimenter to digest data and make novel predictions regarding phenomena of interest. However, machine learning predictors generated from data sets taken from the natural sciences are often treated as black boxes which are used broadly and generally without detailed consideration of the causal structure of the data set of interest. Work has been attempted to bring causality into discussions of machine learning models of natural phenomena; however, a firm and unified theoretical treatment is lacking. This series of three papers explores the union of chemical theory, biological theory, probability theory and causality that will correct current causal flaws of machine learning in the natural sciences. This paper, Part 1 of the series, provides the formal framework of the foundational causal structure of phenomena in chemical biology and is extended to machine learning through the novel concept of focus, defined here as the ability of a machine learning algorithm to narrow down to a hidden underpinning mechanism in large data sets. Initial proof of these principles on a family of Akt inhibitors is also provided. The second paper containing Part 2 will provide a formal exploration of chemical similarity, and Part 3 will present extensive experimental evidence of how hidden causal structures weaken all machine learning in chemical biology. This series serves to establish for chemical biology a new kind of mathematical framework for modeling mechanisms in Nature without the need for the tools of reductionism: inferential mechanics.

For each axiom of KM belief update we provide a corresponding axiom in a modal logic containing three modal operators: a unimodal belief operator $B$, a bimodal conditional operator $>$ and the unimodal necessity operator $\square$. We then compare the resulting logic to the similar logic obtained from converting the AGM axioms of belief revision into modal axioms and show that the latter contains the former. Denoting the latter by $\mathcal L_{AGM}$ and the former by $\mathcal L_{KM}$ we show that every axiom of $\mathcal L_{KM}$ is a theorem of $\mathcal L_{AGM}$. Thus AGM belief revision can be seen as a special case of KM belief update. For the strong version of KM belief update we show that the difference between $\mathcal L_{KM}$ and $\mathcal L_{AGM}$ can be narrowed down to a single axiom, which deals exclusively with unsurprising information, that is, with formulas that were not initially disbelieved.

Emotion Recognition in Conversations (ERC) presents unique challenges, requiring models to capture the temporal flow of multi-turn dialogues and to effectively integrate cues from multiple modalities. We propose Mixture of Speech-Text Experts for Recognition of Emotions (MiSTER-E), a modular Mixture-of-Experts (MoE) framework designed to decouple two core challenges in ERC: modality-specific context modeling and multimodal information fusion. MiSTER-E leverages large language models (LLMs) fine-tuned for both speech and text to provide rich utterance-level embeddings, which are then enhanced through a convolutional-recurrent context modeling layer. The system integrates predictions from three experts-speech-only, text-only, and cross-modal-using a learned gating mechanism that dynamically weighs their outputs. To further encourage consistency and alignment across modalities, we introduce a supervised contrastive loss between paired speech-text representations and a KL-divergence-based regulariza-tion across expert predictions. Importantly, MiSTER-E does not rely on speaker identity at any stage. Experiments on three benchmark datasets-IEMOCAP, MELD, and MOSI-show that our proposal achieves 70.9%, 69.5%, and 87.9% weighted F1-scores respectively, outperforming several baseline speech-text ERC systems. We also provide various ablations to highlight the contributions made in the proposed approach.

Federated learning (FL) faces challenges in uncertainty quantification (UQ). Without reliable UQ, FL systems risk deploying overconfident models at under-resourced agents, leading to silent local failures despite seemingly satisfactory global performance. Existing federated UQ approaches often address data heterogeneity or model heterogeneity in isolation, overlooking their joint effect on coverage reliability across agents. Conformal prediction is a widely used distribution-free UQ framework, yet its applications in heterogeneous FL settings remains underexplored. We provide FedWQ-CP, a simple yet effective approach that balances empirical coverage performance with efficiency at both global and agent levels under the dual heterogeneity. FedWQ-CP performs agent-server calibration in a single communication round. On each agent, conformity scores are computed on calibration data and a local quantile threshold is derived. Each agent then transmits only its quantile threshold and calibration sample size to the server. The server simply aggregates these thresholds through a weighted average to produce a global threshold. Experimental results on seven public datasets for both classification and regression demonstrate that FedWQ-CP empirically maintains agent-wise and global coverage while producing the smallest prediction sets or intervals.

In recent times, large datasets hinder efficient model training while also containing redundant concepts. Dataset distillation aims to synthesize compact datasets that preserve the knowledge of large-scale training sets while drastically reducing storage and computation. Recent advances in diffusion models have enabled training-free distillation by leveraging pre-trained generative priors; however, existing guidance strategies remain limited. Current score-based methods either perform unguided denoising or rely on simple mode-based guidance toward instance prototype centroids (IPC centroids), which often are rudimentary and suboptimal. We propose Manifold-Guided Distillation (ManifoldGD), a training-free diffusion-based framework that integrates manifold consistent guidance at every denoising timestep. Our method employs IPCs computed via a hierarchical, divisive clustering of VAE latent features, yielding a multi-scale coreset of IPCs that captures both coarse semantic modes and fine intra-class variability. Using a local neighborhood of the extracted IPC centroids, we create the latent manifold for each diffusion denoising timestep. At each denoising step, we project the mode-alignment vector onto the local tangent space of the estimated latent manifold, thus constraining the generation trajectory to remain manifold-faithful while preserving semantic consistency. This formulation improves representativeness, diversity, and image fidelity without requiring any model retraining. Empirical results demonstrate consistent gains over existing training-free and training-based baselines in terms of FID, l2 distance among real and synthetic dataset embeddings, and classification accuracy, establishing ManifoldGD as the first geometry-aware training-free data distillation framework.

Real-world Table-Text question answering (QA) tasks require models that can reason across long text and source tables, traversing multiple hops and executing complex operations such as aggregation. Yet existing benchmarks are small, manually curated - and therefore error-prone - and contain shallow questions that seldom demand more than two hops or invoke aggregations, grouping, or other advanced analytical operations expressible in natural-language queries. We present SPARTA, an end-to-end construction framework that automatically generates large-scale Table-Text QA benchmarks with lightweight human validation, requiring only one quarter of the annotation time of HybridQA. The framework first constructs a reference fact database by enriching each source table with grounding tables whose tuples are atomic facts automatically extracted from the accompanying unstructured passages, then synthesizes nested queries whose number of nested predicates matches the desired hop count. To ensure that every SQL statement is executable and that its verbalization yields a fluent, human-sounding question, we propose two novel techniques: provenance-based refinement, which rewrites any syntactically valid query that returns a non-empty result, and realistic-structure enforcement, which confines generation to post-order traversals of the query graph. The resulting pipeline produces thousands of high-fidelity question-answer pairs covering aggregations, grouping, and deep multi-hop reasoning across text and tables. On SPARTA, state-of-the-art models that reach over 70 F1 on HybridQA or over 50 F1 on OTT-QA drop by more than 30 F1 points, exposing fundamental weaknesses in current cross-modal reasoning. Our benchmark, construction code, and baseline models are available at https://github.com/pshlego/SPARTA/tree/main.

Modeling neural population dynamics is crucial for foundational neuroscientific research and various clinical applications. Conventional latent variable methods typically model continuous brain dynamics through discretizing time with recurrent architecture, which necessarily results in compounded cumulative prediction errors and failure of capturing instantaneous, nonlinear characteristics of EEGs. We propose ODEBRAIN, a Neural ODE latent dynamic forecasting framework to overcome these challenges by integrating spatio-temporal-frequency features into spectral graph nodes, followed by a Neural ODE modeling the continuous latent dynamics. Our design ensures that latent representations can capture stochastic variations of complex brain states at any given time point. Extensive experiments verify that ODEBRAIN can improve significantly over existing methods in forecasting EEG dynamics with enhanced robustness and generalization capabilities.

Offline goal-conditioned reinforcement learning (GCRL) learns goal-conditioned policies from static pre-collected datasets. However, accurate value estimation remains a challenge due to the limited coverage of the state-action space. Recent physics-informed approaches have sought to address this by imposing physical and geometric constraints on the value function through regularization defined over first-order partial differential equations (PDEs), such as the Eikonal equation. However, these formulations can often be ill-posed in complex, high-dimensional environments. In this work, we propose a physics-informed regularization derived from the viscosity solution of the Hamilton-Jacobi-Bellman (HJB) equation. By providing a physics-based inductive bias, our approach grounds the learning process in optimal control theory, explicitly regularizing and bounding updates during value iterations. Furthermore, we leverage the Feynman-Kac theorem to recast the PDE solution as an expectation, enabling a tractable Monte Carlo estimation of the objective that avoids numerical instability in higher-order gradients. Experiments demonstrate that our method improves geometric consistency, making it broadly applicable to navigation and high-dimensional, complex manipulation tasks. Open-source codes are available at https://github.com/HrishikeshVish/phys-fk-value-GCRL.

We study Stackelberg (leader--follower) tuning of network parameters (tolls, capacities, incentives) in combinatorial congestion games, where selfish users choose discrete routes (or other combinatorial strategies) and settle at a congestion equilibrium. The leader minimizes a system-level objective (e.g., total travel time) evaluated at equilibrium, but this objective is typically nonsmooth because the set of used strategies can change abruptly. We propose ZO-Stackelberg, which couples a projection-free Frank--Wolfe equilibrium solver with a zeroth-order outer update, avoiding differentiation through equilibria. We prove convergence to generalized Goldstein stationary points of the true equilibrium objective, with explicit dependence on the equilibrium approximation error, and analyze subsampled oracles: if an exact minimizer is sampled with probability $κ_m$, then the Frank--Wolfe error decays as $\mathcal{O}(1/(κ_m T))$. We also propose stratified sampling as a practical way to avoid a vanishing $κ_m$ when the strategies that matter most for the Wardrop equilibrium concentrate in a few dominant combinatorial classes (e.g., short paths). Experiments on real-world networks demonstrate that our method achieves orders-of-magnitude speedups over a differentiation-based baseline while converging to follower equilibria.

Chest X-ray plays a central role in thoracic diagnosis, and its interpretation inherently requires multi-step, evidence-grounded reasoning. However, large vision-language models (LVLMs) often generate plausible responses that are not faithfully grounded in diagnostic evidence and provide limited visual evidence for verification, while also requiring costly retraining to support new diagnostic tasks, limiting their reliability and adaptability in clinical settings. To address these limitations, we present CXReasonAgent, a diagnostic agent that integrates a large language model (LLM) with clinically grounded diagnostic tools to perform evidence-grounded diagnostic reasoning using image-derived diagnostic and visual evidence. To evaluate these capabilities, we introduce CXReasonDial, a multi-turn dialogue benchmark with 1,946 dialogues across 12 diagnostic tasks, and show that CXReasonAgent produces faithfully grounded responses, enabling more reliable and verifiable diagnostic reasoning than LVLMs. These findings highlight the importance of integrating clinically grounded diagnostic tools, particularly in safety-critical clinical settings.

Simulation code for conventional supercomputers serves as a reference for neuromorphic computing systems. The present bottleneck of distributed large-scale spiking neuronal network simulations is the communication between compute nodes. Communication speed seems limited by the interconnect between the nodes and the software library orchestrating the data transfer. Profiling reveals, however, that the variability of the time required by the compute nodes between communication calls is large. The bottleneck is in fact the waiting time for the slowest node. A statistical model explains total simulation time on the basis of the distribution of computation times between communication calls. A fundamental cure is to avoid communication calls because this requires fewer synchronizations and reduces the variability of computation times across compute nodes. The organization of the mammalian brain into areas lends itself to such an optimization strategy. Connections between neurons within an area have short delays, but the delays of the long-range connections across areas are an order of magnitude longer. This suggests a structure-aware mapping of areas to compute nodes allowing for a partition into more frequent communication between nodes simulating a particular area and less frequent global communication. We demonstrate a substantial performance gain on a real-world example. This work proposes a local-global hybrid communication architecture for large-scale neuronal network simulations as a first step in mapping the structure of the brain to the structure of a supercomputer. It challenges the long-standing belief that the bottleneck of simulation is synchronization inherent in the collective calls of standard communication libraries. We provide guidelines for the energy efficient simulation of neuronal networks on conventional computing systems and raise the bar for neuromorphic systems.

Deep Research Agents (DRAs) are promising agentic systems that gather and synthesize information to support research across domains such as financial decision-making, medical analysis, and scientific discovery. Despite recent improvements in research quality (e.g., outcome accuracy when ground truth is available), DRA system design often overlooks a critical barrier to real-world deployment: stochasticity. Under identical queries, repeated executions of DRAs can exhibit substantial variability in terms of research outcome, findings, and citations. In this paper, we formalize the study of stochasticity in DRAs by modeling them as information acquisition Markov Decision Processes. We introduce an evaluation framework that quantifies variance in the system and identify three sources of it: information acquisition, information compression, and inference. Through controlled experiments, we investigate how stochasticity from these modules across different decision steps influences the variance of DRA outputs. Our results show that reducing stochasticity can improve research output quality, with inference and early-stage stochasticity contributing the most to DRA output variance. Based on these findings, we propose strategies for mitigating stochasticity while maintaining output quality via structured output and ensemble-based query generation. Our experiments on DeepSearchQA show that our proposed mitigation methods reduce average stochasticity by 22% while maintaining high research quality.

Diffusion Language Models (DLMs) are often advertised as enabling parallel token generation, yet practical fast DLMs frequently converge to left-to-right, autoregressive (AR)-like decoding dynamics. In contrast, genuinely non-AR generation is promising because it removes AR's sequential bottleneck, better exploiting parallel hardware to reduce synchronization/communication overhead and improve latency scaling with output length. We argue that a primary driver of AR-like decoding is a mismatch between DLM objectives and the highly sequential structure of widely used training data, including standard pretraining corpora and long chain-of-thought (CoT) supervision. Motivated by this diagnosis, we propose NAP (Non-Autoregressive Parallel DLMs), a proof-of-concept, data-centric approach that better aligns supervision with non-AR parallel decoding. NAP curates examples as multiple independent reasoning trajectories and couples them with a parallel-forced decoding strategy that encourages multi-token parallel updates. Across math reasoning benchmarks, NAP yields stronger performance under parallel decoding than DLMs trained on standard long CoT data, with gains growing as parallelism increases. Our results suggest that revisiting data and supervision is a principled direction for mitigating AR-like behavior and moving toward genuinely non-autoregressive parallel generation in DLMs. Our code is available at https://github.com/pixeli99/NAP.

Function contract generation is a classical problem in program analysis that targets the automated analysis of functions in a program with multiple procedures. The problem is fundamental in inter-procedural analysis where properties of functions are first obtained via the generation of function contracts and then the generated contracts are used as building blocks to analyze the whole program. Typical objectives in function contract generation include pre-/post-conditions and assigns information (that specifies the modification information over program variables and memory segments during function execution). In programs with array manipulations, a crucial point in function contract generation is the treatment of array segments that imposes challenges in inferring invariants and assigns information over such segments. To address this challenge, we propose a novel symbolic execution framework that carries invariants and assigns information over contiguous segments of arrays. We implement our framework as a prototype within LLVM, and further integrate our prototype with the ACSL assertion format and the Frama-C software verification platform. Experimental evaluation over a variety of benchmarks from the literature and functions from realistic libraries shows that our framework is capable of handling array manipulating functions that indeed involve the carry of array information and are beyond existing approaches.

Brain foundation models have achieved remarkable advances across a wide range of neuroscience tasks. However, most existing models are limited to a single functional modality, restricting their ability to exploit complementary spatiotemporal dynamics and the collective data scale across imaging techniques. To address this limitation, we propose Brain-OF, the first omnifunctional brain foundation model jointly pretrained on fMRI, EEG and MEG, capable of handling both unimodal and multimodal inputs within a unified framework. To reconcile heterogeneous spatiotemporal resolutions, we introduce the Any-Resolution Neural Signal Sampler, which projects diverse brain signals into a shared semantic space.To further manage semantic shifts, the Brain-OF backbone integrates DINT attention with a Sparse Mixture of Experts, where shared experts capture modality-invariant representations and routed experts specialize in modality-specific semantics. Furthermore, we propose Masked Temporal-Frequency Modeling, a dual-domain pretraining objective that jointly reconstructs brain signals in both the time and frequency domains. Brain-OF is pretrained on a large-scale corpus comprising around 40 datasets and demonstrates superior performance across diverse downstream tasks, highlighting the benefits of joint multimodal integration and dual-domain pretraining.

We consider the problem of contextual online RLHF with general preferences, where the goal is to identify the Nash Equilibrium. We adopt the Generalized Bilinear Preference Model (GBPM) to capture potentially intransitive preferences via low-rank, skew-symmetric matrices. We investigate general preference learning with any strongly convex regularizer and regularization strength $η^{-1}$, generalizing beyond prior work limited to reverse KL-regularization. Central to our analysis is proving that the dual gap of the greedy policy is bounded by the square of the estimation error, a result derived solely from strong convexity and the skew-symmetry of GBPM. Building on this insight and a feature diversity assumption, we establish two regret bounds via two simple algorithms: (1) Greedy Sampling achieves polylogarithmic, $e^{\mathcal{O}(η)}$-free regret $\tilde{\mathcal{O}}(ηd^4 (\log T)^2)$. (2) Explore-Then-Commit achieves $\mathrm{poly}(d)$-free regret $\tilde{\mathcal{O}}(\sqrt{ηr T})$ by exploiting the low-rank structure; this is the first statistically efficient guarantee for online RLHF in high-dimensions.

Quantum machine learning models using angle encoding naturally represent truncated Fourier series, providing universal function approximation capabilities with sufficient circuit depth. For unary fixed-frequency encodings, circuit depth scales as O(omega_max * (omega_max + epsilon^{-2})) with target frequency magnitude omega_max and precision epsilon. Trainable-frequency approaches theoretically reduce this to match the target spectrum size, requiring only as many encoding gates as frequencies in the target spectrum. Despite this compelling efficiency, their practical effectiveness hinges on a key assumption: that gradient-based optimization can drive prefactors to arbitrary target values. We demonstrate through systematic experiments that frequency prefactors exhibit limited trainability: movement is constrained to approximately +/-1 units with typical learning rates. When target frequencies lie outside this reachable range, optimization frequently fails. To overcome this frequency reachability limitation, we propose grid-based initialization using ternary encodings, which generate dense integer frequency spectra. While this approach requires O(log_3(omega_max)) encoding gates -- more than the theoretical optimum but exponentially fewer than fixed-frequency methods -- it ensures target frequencies lie within the locally reachable range. On synthetic targets with three shifted high frequencies, ternary grid initialization achieves a median R^2 score of 0.9969, compared to 0.1841 for the trainable-frequency baseline. For the real-world Flight Passengers dataset, ternary grid initialization achieves a median R^2 score of 0.9671, representing a 22.8% improvement over trainable-frequency initialization (median R^2 = 0.7876).

Deep learning (DL) libraries are widely used in critical applications, where even subtle silent bugs can lead to serious consequences. While existing DL fuzzing techniques have made progress in detecting crashes, they inherently struggle to detect silent bugs due to the lack of effective test programs and corresponding oracles. Building on the observation that historical bug reports contain rich, underutilized information about silent bugs, we leverage large language models (LLMs) to perform versatile yet controlled bug transfer for silent bug fuzzing. Specifically, our approach uses LLMs to extract context-aware bug patterns from historical issues, match semantically related Application Programming Interfaces (APIs) using functionality-based embeddings, and synthesize test cases with customized oracles. This enables proactive detection of silent bugs by transferring high-risk contexts and oracle designs from known buggy APIs to functionally similar target APIs. To ensure the reliability of our context-aware bug transfer, we introduce an LLM-powered self-validation module that systematically evaluates the validity of each transferred bug instance. We implement this methodology in a tool named TransFuzz and evaluate it on three mainstream DL libraries: PyTorch, TensorFlow, and MindSpore. TransFuzz successfully discovers 79 previously unknown bugs (12 confirmed as Common Vulnerabilities and Exposures (CVEs)) in 10 bug types, demonstrating its effectiveness and generalizability in migrating DL library bug discovery capabilities.

Semi-structured documents integrate diverse interleaved data elements (e.g., tables, charts, hierarchical paragraphs) arranged in various and often irregular layouts. These documents are widely observed across domains and account for a large portion of real-world data. However, existing methods struggle to support natural language question answering over these documents due to three main technical challenges: (1) The elements extracted by techniques like OCR are often fragmented and stripped of their original semantic context, making them inadequate for analysis. (2) Existing approaches lack effective representations to capture hierarchical structures within documents (e.g., associating tables with nested chapter titles) and to preserve layout-specific distinctions (e.g., differentiating sidebars from main content). (3) Answering questions often requires retrieving and aligning relevant information scattered across multiple regions or pages, such as linking a descriptive paragraph to table cells located elsewhere in the document. To address these issues, we propose MoDora, an LLM-powered system for semi-structured document analysis. First, we adopt a local-alignment aggregation strategy to convert OCR-parsed elements into layout-aware components, and conduct type-specific information extraction for components with hierarchical titles or non-text elements. Second, we design the Component-Correlation Tree (CCTree) to hierarchically organize components, explicitly modeling inter-component relations and layout distinctions through a bottom-up cascade summarization process. Finally, we propose a question-type-aware retrieval strategy that supports (1) layout-based grid partitioning for location-based retrieval and (2) LLM-guided pruning for semantic-based retrieval. Experiments show MoDora outperforms baselines by 5.97%-61.07% in accuracy. The code is at https://github.com/weAIDB/MoDora.

As Large Language Models (LLMs) are increasingly used, their security risks have drawn increasing attention. Existing research reveals that LLMs are highly susceptible to jailbreak attacks, with effectiveness varying across language contexts. This paper investigates the role of classical Chinese in jailbreak attacks. Owing to its conciseness and obscurity, classical Chinese can partially bypass existing safety constraints, exposing notable vulnerabilities in LLMs. Based on this observation, this paper proposes a framework, CC-BOS, for the automatic generation of classical Chinese adversarial prompts based on multi-dimensional fruit fly optimization, facilitating efficient and automated jailbreak attacks in black-box settings. Prompts are encoded into eight policy dimensions-covering role, behavior, mechanism, metaphor, expression, knowledge, trigger pattern and context; and iteratively refined via smell search, visual search, and cauchy mutation. This design enables efficient exploration of the search space, thereby enhancing the effectiveness of black-box jailbreak attacks. To enhance readability and evaluation accuracy, we further design a classical Chinese to English translation module. Extensive experiments demonstrate that effectiveness of the proposed CC-BOS, consistently outperforming state-of-the-art jailbreak attack methods.

Large language models (LLMs) have demonstrated strong capabilities in code generation, yet they remain prone to producing security vulnerabilities. Existing approaches commonly suffer from two key limitations: the scarcity of high-quality security data and coarse-grained reinforcement learning reward signals. To address these challenges, we propose Vul2Safe, a new secure code generation framework that leverages LLM self-reflection to construct high-confidence repair pairs from real-world vulnerabilities, and further generates diverse implicit prompts to build the PrimeVul+ dataset. Meanwhile, we introduce SRCode, a novel training framework that pioneers the use of token-level rewards in reinforcement learning for code security, which enables the model to continuously attend to and reinforce critical fine-grained security patterns during training. Compared with traditional instance-level reward schemes, our approach allows for more precise optimization of local security implementations. Extensive experiments show that PrimeVul+ and SRCode substantially reduce security vulnerabilities in generated code while improving overall code quality across multiple benchmarks.

This paper introduces a novel methodology for dynamic networks by leveraging a new symmetry-principled class of primitives, isotropic activation functions. This approach enables real-time neuronal growth and shrinkage of the architectures in response to task demand. This is made possible by network structural changes that are invariant under symmetry reparameterisations, leaving the computation identical under neurogenesis and well approximated under neurodegeneration. This is undertaken by leveraging the isotropic primitives' property of basis independence, resulting in the loss of the individuated neurons implicit in the elementwise functional form. Isotropy thereby allows a freedom in the basis to which layers are decomposed and interpreted as individual artificial neurons. This enables a layer-wise diagonalisation procedure, in which typical interconnected layers, such as dense layers, convolutional kernels, and others, can be reexpressed so that neurons have one-to-one, ordered connectivity within alternating layers. This indicates which one-to-one neuron-to-neuron communications are strongly impactful on overall functionality and which are not. Inconsequential neurons can thus be removed (neurodegeneration), and new inactive scaffold neurons added (neurogenesis) whilst remaining analytically invariant in function. A new tunable model parameter, intrinsic length, is also introduced to ensure this analytical invariance. This approach mathematically equates connectivity pruning with neurodegeneration. The diagonalisation also offers new possibilities for mechanistic interpretability into isotropic networks, and it is demonstrated that isotropic dense networks can asymptotically reach a sparsity factor of 50% whilst retaining exact network functionality. Finally, the construction is generalised, demonstrating a nested functional class for this form of isotropic primitive architectures.

Generative Recommendation (GenRec) typically leverages Large Language Models (LLMs) to redefine personalization as an instruction-driven sequence generation task. However, fine-tuning on user logs inadvertently encodes sensitive attributes into model parameters, raising critical privacy concerns. Existing Machine Unlearning (MU) techniques struggle to navigate this tension due to the Polysemy Dilemma, where neurons superimpose sensitive data with general reasoning patterns, leading to catastrophic utility loss under traditional gradient or pruning methods. To address this, we propose Utility-aware Contrastive AttenuatioN (U-CAN), a precision unlearning framework that operates on low-rank adapters. U-CAN quantifies risk by contrasting activations and focuses on neurons with asymmetric responses that are highly sensitive to the forgetting set but suppressed on the retention set. To safeguard performance, we introduce a utility-aware calibration mechanism that combines weight magnitudes with retention-set activation norms, assigning higher utility scores to dimensions that contribute strongly to retention performance. Unlike binary pruning, which often fragments network structure, U-CAN develop adaptive soft attenuation with a differentiable decay function to selectively down-scale high-risk parameters on LoRA adapters, suppressing sensitive retrieval pathways and preserving the topological connectivity of reasoning circuits. Experiments on two public datasets across seven metrics demonstrate that U-CAN achieves strong privacy forgetting, utility retention, and computational efficiency.

Recent deep research agents primarily improve performance by scaling reasoning depth, but this leads to high inference cost and latency in search-intensive scenarios. Moreover, generalization across heterogeneous research settings remains challenging. In this work, we propose \emph{Search More, Think Less} (SMTL), a framework for long-horizon agentic search that targets both efficiency and generalization. SMTL replaces sequential reasoning with parallel evidence acquisition, enabling efficient context management under constrained context budgets. To support generalization across task types, we further introduce a unified data synthesis pipeline that constructs search tasks spanning both deterministic question answering and open-ended research scenarios with task appropriate evaluation metrics. We train an end-to-end agent using supervised fine-tuning and reinforcement learning, achieving strong and often state of the art performance across benchmarks including BrowseComp (48.6\%), GAIA (75.7\%), Xbench (82.0\%), and DeepResearch Bench (45.9\%). Compared to Mirothinker-v1.0, SMTL with maximum 100 interaction steps reduces the average number of reasoning steps on BrowseComp by 70.7\%, while improving accuracy.

In this Master's thesis, the graph properties of a multi-level drug-protein network are studied, as well as how the network's shape has informed discoveries over the years, identifying primarily crawling discoveries and a smaller number of hopping discoveries. Finally, the network structure is used to inform a network diffusion recommendation system and to prioritize existing drugs for repurposing against proteins in organisms that cause Neglected Tropical Diseases.

Large language models are increasingly applied to operational decision-making where the underlying structure is constrained optimization. Existing benchmarks evaluate whether LLMs can formulate optimization problems as solver code, but leave open a complementary question. Can LLMs directly produce correct solutions to fully specified constrained optimization problems without access to a solver? We introduce ConstraintBench, a benchmark for evaluating LLMs on direct constrained optimization across 10 operations research domains, with all ground-truth solutions verified by the Gurobi solver. Each task presents a natural-language scenario with entities, constraints, and an optimization objective; the model must return a structured solution that a deterministic verifier checks against every constraint and the solver-proven optimum. We evaluate six frontier models on 200 tasks and find that feasibility, not optimality, is the primary bottleneck. The best model achieves only 65.0% feasibility, yet feasible solutions average 89 to 96% of the Gurobi-optimal objective. No model exceeds 30.5% on joint feasibility and optimality within 0.1% of the solver reference. Per-domain analysis shows large variation in difficulty, with average feasibility spanning from 85.0% in the facility location domain to 0.8% in the crew assignment domain. Further, systematic failure modes include duration constraint misunderstanding, entity hallucination, and a feasibility-optimality decoupling in facility location and vehicle routing where models achieve high feasibility but 0% optimality. ConstraintBench and all evaluation infrastructure will be publicly released.

Recent work has identified a subset of attention heads in Transformer as retrieval heads, which are responsible for retrieving information from the context. In this work, we first investigate retrieval heads in multilingual contexts. In multilingual language models, we find that retrieval heads are often shared across multiple languages. Expanding the study to cross-lingual setting, we identify Retrieval-Transition heads(RTH), which govern the transition to specific target-language output. Our experiments reveal that RTHs are distinct from retrieval heads and more vital for Chain-of-Thought reasoning in multilingual LLMs. Across four multilingual benchmarks (MMLU-ProX, MGSM, MLQA, and XQuaD) and two model families (Qwen-2.5 and Llama-3.1), we demonstrate that masking RTH induces bigger performance drop than masking Retrieval Heads (RH). Our work advances understanding of multilingual LMs by isolating the attention heads responsible for mapping to target languages.

Fully Sharded Data Parallel (FSDP), also known as ZeRO, is widely used for training large-scale models, featuring its flexibility and minimal intrusion on model code. However, current FSDP systems struggle with structure-aware training methods (e.g., block-wise quantized training) and with non-element-wise optimizers (e.g., Shampoo and Muon) used in cutting-edge models (e.g., Gemini, Kimi K2). FSDP's fixed element- or row-wise sharding formats conflict with the block-structured computations. In addition, today's implementations fall short in communication and memory efficiency, limiting scaling to tens of thousands of GPUs. We introduce veScale-FSDP, a redesigned FSDP system that couples a flexible sharding format, RaggedShard, with a structure-aware planning algorithm to deliver both flexibility and performance at scale. veScale-FSDP natively supports efficient data placement required by FSDP, empowering block-wise quantization and non-element-wise optimizers. As a result, veScale-FSDP achieves 5~66% higher throughput and 16~30% lower memory usage than existing FSDP systems, while scaling efficiently to tens of thousands of GPUs.

As the complexity of the HPC storage stack rapidly grows, domain scientists face increasing challenges in effectively utilizing HPC storage systems to achieve their desired I/O performance. To identify and address I/O issues, scientists largely rely on I/O experts to analyze their I/O traces and provide insights into potential problems. However, with a limited number of I/O experts and the growing demand for data-intensive applications, inaccessibility has become a major bottleneck, hindering scientists from maximizing their productivity. Rapid advances in LLMs make it possible to build an automated tool that brings trustworthy I/O performance diagnosis to domain scientists. However, key challenges remain, such as the inability to handle long context windows, a lack of accurate domain knowledge about HPC I/O, and the generation of hallucinations during complex interactions. In this work, we propose IOAgent as a systematic effort to address these challenges. IOAgent integrates a module-based pre-processor, a RAG-based domain knowledge integrator, and a tree-based merger to accurately diagnose I/O issues from a given Darshan trace file. Similar to an I/O expert, IOAgent provides detailed justifications and references for its diagnoses and offers an interactive interface for scientists to ask targeted follow-up questions. To evaluate IOAgent, we collected a diverse set of labeled job traces and released the first open diagnosis test suite, TraceBench. Using this test suite, we conducted extensive evaluations, demonstrating that IOAgent matches or outperforms state-of-the-art I/O diagnosis tools with accurate and useful diagnosis results. We also show that IOAgent is not tied to specific LLMs, performing similarly well with both proprietary and open-source LLMs. We believe IOAgent has the potential to become a powerful tool for scientists navigating complex HPC I/O subsystems in the future.

Models operating on dynamic physiologic signals must distinguish benign, label-preserving variability from true concept change. Existing concept-drift frameworks are largely distributional and provide no principled guidance on how much a model's internal representation may move when the underlying signal undergoes physiologically plausible fluctuations in energy. As a result, deep models often misinterpret harmless changes in amplitude, rate, or morphology as concept drift, yielding unstable predictions, particularly in multimodal fusion settings. This study introduces Physiologic Energy Conservation Theory (PECT), an energy-based framework for concept stability in dynamic signals. PECT posits that under virtual drift, normalized latent displacement should scale proportionally with normalized signal energy change, while persistent violations of this proportionality indicate real concept drift. We operationalize this principle through Energy-Constrained Representation Learning (ECRL), a lightweight regularizer that penalizes energy-inconsistent latent movement without modifying encoder architectures or adding inference-time cost. Although PECT is formulated for dynamic signals in general, we instantiate and evaluate it on multimodal ECG across seven unimodal and hybrid models. Experiments show that in the strongest trimodal hybrid (1D+2D+Transformer), clean accuracy is largely preserved (96.0% to 94.1%), while perturbed accuracy improves substantially (72.6% to 85.5%) and fused representation drift decreases by over 45%. Similar trends are observed across all architectures, providing empirical evidence that PECT functions as an energy-drift law governing concept stability in continuous physiologic signals.

While prior work has focused on projecting adversarial examples back onto the manifold of natural data to restore safety, we argue that a comprehensive understanding of AI safety requires characterizing the unsafe regions themselves. This paper introduces a framework for systematically mapping the Manifold of Failure in Large Language Models (LLMs). We reframe the search for vulnerabilities as a quality diversity problem, using MAP-Elites to illuminate the continuous topology of these failure regions, which we term behavioral attraction basins. Our quality metric, Alignment Deviation, guides the search towards areas where the model's behavior diverges most from its intended alignment. Across three LLMs: Llama-3-8B, GPT-OSS-20B, and GPT-5-Mini, we show that MAP-Elites achieves up to 63% behavioral coverage, discovers up to 370 distinct vulnerability niches, and reveals dramatically different model-specific topological signatures: Llama-3-8B exhibits a near-universal vulnerability plateau (mean Alignment Deviation 0.93), GPT-OSS-20B shows a fragmented landscape with spatially concentrated basins (mean 0.73), and GPT-5-Mini demonstrates strong robustness with a ceiling at 0.50. Our approach produces interpretable, global maps of each model's safety landscape that no existing attack method (GCG, PAIR, or TAP) can provide, shifting the paradigm from finding discrete failures to understanding their underlying structure.

Abstractions of causal models allow for the coarsening of models such that relations of cause and effect are preserved. Whereas abstractions focus on the relation between two models, in this paper we study a framework for causal embeddings which enable multiple detailed models to be mapped into sub-systems of a coarser causal model. We define causal embeddings as a generalization of abstraction, and present a generalized notion of consistency. By defining a multi-resolution marginal problem, we showcase the relevance of causal embeddings for both the statistical marginal problem and the causal marginal problem; furthermore, we illustrate its practical use in merging datasets coming from models with different representations.

Heterogeneous nodes that combine multi-core CPUs with diverse accelerators are rapidly becoming the norm in both high-performance computing (HPC) and AI infrastructures. Exploiting these platforms, however, requires orchestrating several low-level accelerator APIs such as CUDA, SYCL, and Triton. In some occasions they can be combined with optimized vendor math libraries: e.g., cuBLAS and oneAPI. Each API or library introduces its own abstractions, execution semantics, and synchronization mechanisms. Combining them within a single application is therefore error-prone and labor-intensive. We propose reusing a task-based data-flow methodology together with Task-Aware APIs (TA-libs) to overcome these limitations and facilitate the seamless integration of multiple accelerator programming models, while still leveraging the best-in-class kernels offered by each API. Applications are expressed as a directed acyclic graph (DAG) of host tasks and device kernels managed by an OpenMP/OmpSs-2 runtime. We introduce Task-Aware SYCL (TASYCL) and leverage Task-Aware CUDA (TACUDA), which elevate individual accelerator invocations to first-class tasks. When multiple native runtimes coexist on the same multi-core CPU, they contend for threads, leading to oversubscription and performance variability. To address this, we unify their thread management under the nOS-V tasking and threading library, to which we contribute a new port of the PoCL (Portable OpenCL) runtime. These results demonstrate that task-aware libraries, coupled with the nOS-V library, enable a single application to harness multiple accelerator programming models transparently and efficiently. The proposed methodology is immediately applicable to current heterogeneous nodes and is readily extensible to future systems that integrate even richer combinations of CPUs, GPUs, FPGAs, and AI accelerators.

Reconstructing dynamic visual experiences from brain activity provides a compelling avenue for exploring the neural mechanisms of human visual perception. While recent progress in fMRI-based image reconstruction has been notable, extending this success to video reconstruction remains a significant challenge. Current fMRI-to-video reconstruction approaches consistently encounter two major shortcomings: (i) inconsistent visual representations of salient objects across frames, leading to appearance mismatches; (ii) poor temporal coherence, resulting in motion misalignment or abrupt frame transitions. To address these limitations, we introduce SemVideo, a novel fMRI-to-video reconstruction framework guided by hierarchical semantic information. At the core of SemVideo is SemMiner, a hierarchical guidance module that constructs three levels of semantic cues from the original video stimulus: static anchor descriptions, motion-oriented narratives, and holistic summaries. Leveraging this semantic guidance, SemVideo comprises three key components: a Semantic Alignment Decoder that aligns fMRI signals with CLIP-style embeddings derived from SemMiner, a Motion Adaptation Decoder that reconstructs dynamic motion patterns using a novel tripartite attention fusion architecture, and a Conditional Video Render that leverages hierarchical semantic guidance for video reconstruction. Experiments conducted on the CC2017 and HCP datasets demonstrate that SemVideo achieves superior performance in both semantic alignment and temporal consistency, setting a new state-of-the-art in fMRI-to-video reconstruction.

LLM agents have been widely adopted in real-world applications, relying on agent frameworks for workflow execution and multi-agent coordination. As these systems scale, understanding bugs in the underlying agent frameworks becomes critical. However, existing work mainly focuses on agent-level failures, overlooking framework-level bugs. To address this gap, we conduct an empirical study of 998 bug reports from CrewAI and LangChain, constructing a taxonomy of 15 root causes and 7 observable symptoms across five agent lifecycle stages: 'Agent Initialization','Perception', 'Self-Action', 'Mutual Interaction' and 'Evolution'. Our findings show that agent framework bugs mainly arise from 'API misuse', 'API incompatibility', and 'Documentation Desync', largely concentrated in the 'Self-Action' stage. Symptoms typically appear as 'Functional Error', 'Crash', and 'Build Failure', reflecting disruptions to task progression and control flow.

In a previous work, we introduced the fuzzy Ethical Decision-Making framework (fEDM), a risk-based ethical reasoning architecture grounded in fuzzy logic. The original model combined a fuzzy Ethical Risk Assessment module (fERA) with ethical decision rules, enabled formal structural verification through Fuzzy Petri Nets (FPNs), and validated outputs against a single normative referent. Although this approach ensured formal soundness and decision consistency, it did not fully address two critical challenges: principled explainability of decisions and robustness under ethical pluralism. In this paper, we extend fEDM in two major directions. First, we introduce an Explainability and Traceability Module (ETM) that explicitly links each ethical decision rule to the underlying moral principles and computes a weighted principle-contribution profile for every recommended action. This enables transparent, auditable explanations that expose not only what decision was made but why, and on the basis of which principles. Second, we replace single-referent validation with a pluralistic semantic validation framework that evaluates decisions against multiple stakeholder referents, each encoding distinct principle priorities and risk tolerances. This shift allows principled disagreement to be formally represented rather than suppressed, thus increasing robustness and contextual sensitivity. The resulting extended fEDM, called fEDM+, preserves formal verifiability while achieving enhanced interpretability and stakeholder-aware validation, making it suitable as an oversight and governance layer for ethically sensitive AI systems.

Many emerging agentic paradigms require agents to collaborate with one another (or people) to achieve shared goals. Unfortunately, existing approaches to learning policies for such collaborative problems produce brittle solutions that fail when paired with new partners. We attribute these failures to a combination of free-riding during training and a lack of strategic robustness. To address these problems, we study the concept of strategic risk aversion and interpret it as a principled inductive bias for generalizable cooperation with unseen partners. While strategically risk-averse players are robust to deviations in their partner's behavior by design, we show that, in collaborative games, they also (1) can have better equilibrium outcomes than those at classical game-theoretic concepts like Nash, and (2) exhibit less or no free-riding. Inspired by these insights, we develop a multi-agent reinforcement learning (MARL) algorithm that integrates strategic risk aversion into standard policy optimization methods. Our empirical results across collaborative benchmarks (including an LLM collaboration task) validate our theory and demonstrate that our approach consistently achieves reliable collaboration with heterogeneous and previously unseen partners across collaborative tasks.

The discovery of chaotic quantum circuits with (partially) solvable dynamics has played a key role in our understanding of non-equilibrium quantum matter and, at the same time, has helped the development of concrete platforms for quantum computation. It was shown that solvability does not prevent the generation of chaotic dynamics, however, it imposes non-trivial constraints on the generated correlations. A natural question is then whether it is possible to gain insight into the generic case despite the latter being very hard to access. To address this question here we introduce a family of 'asymptotically solvable' quantum circuits where the solvability constraints only affect correlations on length scales beyond a tuneable threshold. This means that their dynamics are only solvable for long enough times: for times shorter than the threshold they are generic. We show this by computing both their dynamical correlations on the equilibrium (infinite temperature) state and their thermalisation dynamics following quantum quenches from compatible (asymptotically solvable) non-equilibrium initial states. The class of systems we introduce is generically ergodic but contains a non-interacting point, which we use to provide exact analytical results, complementing those of numerical experiments, on the non-solvable early time regime.

Active tissues exhibit tension fluctuations that are correlated in space and time. We study a minimal overdamped surface model in which such fluctuations enter as a zero-mean, multiplicative modulation of the local surface tension. Although the deterministic elastic dynamics (tension plus bending) stabilizes the flat state for all nonzero wave numbers, we find that sufficiently persistent active fluctuations generate positive ensemble growth rates for a finite band of Fourier modes, leading to stochastic buckling with wavelength selection. A non-Markovian theory based on the Novikov--Furutsu theorem captures the instability threshold and unstable band observed in simulations.

The quantum XXZ spin-1/2 chain features non-Gaussian spin current fluctuations in the regime of easy-axis anisotropy. Using ballistic macroscopic fluctuation theory, we derive the exact probability distribution of typical spin-current fluctuations in thermal equilibrium. The obtained nested Gaussian distribution is fully characterized by its variance which we analytically relate to the spin diffusion constant and static spin susceptibility, and compare our with numerical simulations. By unveiling how the same mechanism which leads to anomalous charge current fluctuations in single-file systems manifests in the XXZ chain, our approach establishes the universal hydrodynamic origin of the observed anomalous fluctuations.

Resolving growth mechanisms and thickness evolution of functional properties is one of the key tasks in materials discovery and optimization involving thin-film materials, traditionally requiring significant experimental budgets. Here we introduce the combination of thickness-gradient libraries and automated scanning probe microscopy as a systematic pathway to elucidate growth modes and disentangle ferroelectric and electrochemical contributions in ferroelectric thin films. As a model system, we explore the Hf0.5Zr0.5O2 (HZO) gradient thin films grown on LaxSr1-xMnO3 (LSMO) bottom electrode thin films. Automated piezoresponse force microscopy, spectroscopy, and lithography reveals that irreversible topographic deformation arises from electrochemical activity at the LSMO surface, whereas reversible phase inversion in HZO reflects ferroelectric switching. Automated topography height-map scans are used to further quantify nucleation density, particle-size evolution, and roughness correlations across the thickness-gradient, demonstrating that improved plume stabilization during growth suppresses interfacial reactions and promotes dense, fine-grained HZO conducive to ferroelectric phase formation. This combined materials-engineering and automated-SPM framework establishes a platform for high-throughput, mechanism-resolved characterization of ferroionic and ferroelectric responses in complex oxide films.

Understanding the electrical double layer (EDL), i.e, the distribution of electrolyte at an electrified interface, in concentrated electrolytes is important for various technologies, such as supercapacitors, batteries and electrocatalysis. Atomistic approaches offer unprecedented detail, but are too computationally expensive to exhaustively investigate the EDL of concentrated electrolytes, motivating the development of continuum theories. In these concentrated electrolytes, correlations between ions and solvents are strong, through electrostatic and specific interactions, as well as significant excluded volume effects of the complicated molecular species, making the development of theories challenging. Thus far, there are mainly two distinct \textit{simple} theoretical approaches to understand the EDL of concentrated electrolytes, with account of these correlations beyond mean-field. One is a local-density approximation (LDA) based on treating electrostatic and specific interactions beyond mean-field through the ionic aggregation and solvation; where a simple conceptual understanding can be gained and reasonable agreement with experiments in terms of integrated quantities, but poor agreement for ion profiles and Debye capacitance. The other approach is to treat electrostatic correlations and excluded volume effects more rigorously with beyond LDA approaches, but at the cost of simplifying the chemical interactions between species; where excellent agreement can be obtained for ion profiles, differential capacitance, etc., but mainly for the simplified hard-sphere systems that the theories are based on. Here, we describe the merits and downfalls of these two approaches, how they have contributed to understanding anomalous underscreening, and outline future directions for these theoretical approaches.

Superconducting diodes that support nonreciprocal supercurrent flow in principle constitute attractive, non-dissipative, circuit elements for superconducting electronics. But their realization faces fundamental challenges, as conventional Josephson tunnel junctions are inherently reciprocal. Existing approaches to break reciprocity typically involve magnetism or spin-orbit coupling, which often increase device complexity and limit reproducibility. Here, we demonstrate an alternative dynamical route to supercurrent nonreciprocity based on parametric driving. By applying a frequency-modulated supercurrent amplitude we show that effective higher-order, nonharmonic terms are generated in the current-phase relation. Leveraging mathematical equivalences with the Kapitza pendulum, we show that these terms dynamically break reciprocity. This establishes the concept of a Kapitza supercurrent diode and demonstrates that nonreciprocal superconducting transport can be engineered by nonequilibrium driving conventional Josephson tunnel junctions. We propose two implementations of the Kapitza supercurrent diode - via gate-controlled superconducting interferometers or flux-driven double-loop SQUIDs - to achieve nonreciprocal supercurrent transport within experimentally accessible frequencies $ω/2π\sim 1$-$10\,\mathrm{GHz}$.

Defect engineering is a key strategy to control resistive switching (RS) in oxide-based memristive devices, where oxygen vacancy (OV) dynamics governs filament formation and rupture. We investigate the effect of Ag nanoparticles (AgNPs) embedded in the top electrode of Pt/Ta2O5/TaO2/Pt memristors and analyze their RS behavior and statistical stability. Devices without AgNPs exhibit two hysteresis switching loops (HSLs) with opposite chiralities, originating from the participation of the Pt/Ta2O5 top interface and the Ta2O5/TaO2 bottom interface. Incorporating AgNPs reduces the overall device resistance and selectively suppresses one loop, yielding a single, well-defined switching mode. Moreover, devices incorporating Ag-NPs show markedly reduced cycle-to-cycle variability of the high-resistance state, as confirmed by Weibull analysis, indicating improved endurance and switching reproducibility. Within a filamentary RS framework, we attribute this behavior to local metallization of the top interface by AgNPs, which partially inhibit OV transport and confines the RS dynamics to the bottom interface. Numerical simulations with the Oxygen Vacancy Resistive Network (OVRN) model succesfully reproduce the experimental HSLs, statistical trends, and tunable ON/OFF ratios with AgNPs coverage. These findings demonstrate that targeted interface metallization via metallic nanoparticles provides an effective route to control multi-interface RS dynamics and improve switching stability in without modifying the oxide architecture.

The thermal Casimir effect, arising from fluctuating electromagnetic fields of thermally agitated charges, induces thermosensitive forces and presents a novel approach to detecting nanoscale hot electrons, elusive yet ubiquitous in modern miniaturized transistors. However, detecting thermal Casimir forces at the nanoscale remains extremely challenging due to background forces such as electrostatic force and quantum Casimir force. In this study, we present the first non-contact force measurement of hot electrons based on the thermal Casimir effect. Using an atomic force microscope (AFM) with a dual-resonant tip, we achieve thermosensitive force detection of nonequilibrium hot electrons while effectively suppressing background thermo-insensitive forces, thereby distinguishing them from cold electrons. In silicon nanoconstriction devices, the measured thermal Casimir pressure reaches approximately 3 bar at a separation of 5 nm at an electron temperature of about 10^3 K. Our work introduces a novel methodology for hot electron nanothermometry and provides critical insights into the thermo-mechanical properties of post-Moore nanoelectronics.

We propose a general mechanism for the spontaneous emergence of filling-enforced fully compensated ferrimagnetism (fFIM), characterized by zero net magnetization yet ferromagnetic-like spin-split band structures. Using Hartree-Fock mean-field calculations of the Hubbard model, we map out the stability regime of spontaneous fFIM over a broad parameter space of interaction strength and staggered potential. We show the unique quantum-geometry-governed optical selection rules and the abundant valley- and spin-related physics of electronics and optics arising from the emergence of fFIM order, with tunable spin-polarized and valley-contrasting charge and spin currents. Furthermore, based on our theory, we demonstrate that spontaneous fFIM can be realized in nominally nonmagnetic graphene via defect engineering. Our results establish a unified framework for the mechanism, emergent properties, and materials realization of spontaneous fFIM, opening new opportunities for spintronic, valleytronic, and optoelectronic applications.

Bridging the gap between atomistic detail and continuum mechanics is a central challenge in modeling biological membranes, particularly for mesoscopic phenomena spanning large length and time scales. In this work, we introduce a new, solvent-free, one-particle-thick, coarse-grained model for lipid bilayers, governed by an additive potential. Our approach treats orientational elasticity through distinct additive energy terms for tilt and splay, offering an unbiased potential form. The model is implemented in the LAMMPS molecular dynamics engine. Our simulations show spontaneous self-assembly of lamellar structures and stable vesicles from disordered states. We map the dynamical phase diagram of the system, identifying distinct gel-like, fluid, and gas regimes, controlled by temperature and the steepness of the isotropic attraction. The model accurately reproduces the theoretical $1/q^{4}$ fluctuation spectrum for tensionless membranes and exhibits tunable mechanical properties, including biologically relevant bending rigidities and area compressibility moduli. We show how we can include osmotic pressure and spontaneous curvature in our model. Finally, we demonstrate the model's applicability to complex membrane remodeling by simulating the adhesive wrapping of colloidal nanoparticles, recovering the predicted dependency on particle size and adhesion strength.

We theoretically study microwave absorption spectroscopy of fractional quantum Hall droplets in the presence of quasiparticle tunneling across a quantum point contact. This contact-free probe provides access to collective edge dynamics beyond conventional transport measurements. We develop a nonperturbative path-integral Monte Carlo approach that enables computation of the frequency-dependent response at finite temperature and for arbitrary droplet geometries, and benchmark the method against analytical results in the weak-tunneling regime. We find that tunneling produces measurable shifts and broadening of resonance peaks, with systematic dependence on tunneling strength and device geometry. Such shifts and broadenings are not obtained in perturbative treatments acting directly on the response function, but emerge when interaction-kernel effects are properly incorporated. Our results indicate experimentally accessible signatures of edge-mode interference and tunneling-induced renormalization of collective excitations, and support the use of microwave spectroscopy as a quantitative probe of quasiparticle dynamics in mesoscopic quantum Hall structures.

Entanglement between solid-state quantum emitters (QEs) is a key resource for photonic quantum technologies. Achieving such entanglement requires strong and controllable long-range interactions between QEs. However, engineering such coupling remains challenging, particularly for on-chip distant solid-state QEs. Here, we introduce a forward-designed platform that enables ultracompact nanophotonic architectures to mediate enhanced long-range QE-QE interactions via engineered surface plasmon polariton interference. Using this strategy, we realize two distinct configurations: a phase-conjugated elliptic design for energy funneling, and a co-radiating hyperbolic design for its suppression. We experimentally demonstrate large enhancement and suppression of energy transfer rates compared to bare substrates. Furthermore, we predict transient entanglement between spatially separated QEs with concurrence peaking at 0.493, approaching the theoretical bound in the transient regime. Extending to the multi-QE case, we observe enhanced energy funneling and predict QE-QE entanglement in three-QE configurations. These results establish a compact and scalable framework for on-chip entanglement engineering in integrated quantum nanophotonic systems.

Experimental data on solubilization kinetics found in literature were analyzed by using the model proposed earlier (1). The rates of oil molecular exchange between the micellar core and the surrounding aqueous solution were determined. It was concluded that the solubilization of hydrocarbon molecules by nonionic surfactants of ethylene oxide type is essentially barrier-free, that is, is diffusion controlled. It is quite different for ionic surfactants, where the rate is one-two orders of magnitude slower, indicating the existence of a potential barrier for hydrocarbons to get inside the micelles. A Fickean diffusion model of solubilization has been proposed to explain these trends. For ionic micelles, the hydrocarbons are predicted to be excluded from the micellar double layer region because of their low dielectric constant. The Poisson-Boltzmann model was used to model this effect; the diffusion retardation factors were compared with the experiment and a fair agreement was seen. For nonionic groups, such as oligoethylene oxide, on the other hand, no such barrier was predicted to exist. The analysis of this paper is performed only for the case of the 'slow' solubilization; it does not cover the case of the rapid, 'catastrophic' solubilization observed in the other group of experiments; the distinction between the slow and fast mechanisms is discussed and a possible explanation is suggested.

The emergence of order and geometric limit shapes in a three-dimensional (3D) Coulomb phase subject to domain wall boundary conditions (DWBC) is investigated. While the arctic circle phenomenon -- the spatial segregation of frozen and fluctuating degrees of freedom -- is well-established in the two-dimensional six-vertex model (square ice), its extension to 3D remains largely unexplored. A cubic lattice model with Ising degrees of freedom living on the edges, whose ground state manifold is governed by a divergence-free (3-in/3-out) local constraint, is considered. In the bulk, this model realizes a classical spin liquid characterized by algebraic correlations and pinch-point singularities in reciprocal space. It is demonstrated that applying DWBC partially lifts the extensive ground state degeneracy, inducing long-range magnetic order in the thermodynamic limit. Despite this ordering, it is found that the system retains a fluctuating component that exhibits the signature of a Coulomb phase. Finally, by mapping the local vertex polarization density, compelling numerical support is provided for a 3D generalization of the arctic limit shape, bridging the gap between topological constraints and emergent geometry in higher dimensions.

Two dimensional (2D) magnets have emerged as a compelling platform for spin based nanoelectronics, enabling atomic scale control of magnetic order, interfaces, quantum geometry, and symmetry. Here, we highlight recent advances in 2D ferromagnets, antiferromagnets, altermagnets, and related magnetic phases, emphasizing how enhanced Curie temperatures, perpendicular magnetic anisotropy, and unconventional magnetic orders translate into device relevant functionality. Spin dependent transport in vertical magnetic tunnel junctions and lateral spin valves based on 2D heterostructures are discussed, where atomically sharp interfaces enable highly tunable spin injection, propagation, and detection. We further focused on field free energy efficient spin orbit torque magnetization switching in 2D magnetic heterostructures, in which unconventional spin currents originate from an adjacent low symmetry spin orbit layer. Microscopic mechanisms involving symmetry breaking, Berry curvature, and orbital angular momentum transport are discussed, along with key challenges, including switching determinism and torque efficiency. Materials and device design strategies targeting neuromorphic, hybrid quantum spintronic, and multifunctional architectures are outlined. Collectively, these developments position 2D magnets as a promising candidate for tunable, energy efficient integrated spintronic technologies that can harness intertwined spin, charge, orbital, and topological degrees of freedom at the nanoscale.

Fluctuations of integrated currents have attracted considerable interest over the past decades in the context of statistical mechanics. Recently, anomalous current fluctuations, characterized by the M-Wright function, were obtained exactly in a classical automaton [$Ž$. Krajnik et al., Phys. Rev. Lett. 128, 160601 (2022)], and previous studies have shown that the anomalous behavior can arise in a variety of classical systems. Despite the rapidly growing interest in such anomalous behaviors, which capture a universal aspect of one-dimensional many-body transport, the exact derivation of the M-Wright function in quantum many-body systems has remained elusive. In this Letter, we present the first exact microscopic derivation of the M-Wright function in quantum many-body dynamics by analyzing the integrated spin current in a one-dimensional Fermi-Hubbard model with infinitely strong repulsive interactions. Our results lay the groundwork for exploring anomalous integrated currents in a broad class of quantum many-body systems.

Intelligent behavior in life-like systems often arises from the ability to gather, process, and act on information. While active matter provides a framework for studying life-like dynamics, it typically omits internal information-processing and decision-making. Here we introduce an adaptive active particle model that uses minimal information processing capabilities in order to navigate towards a distant target. By combining renewal-based intermittent motion with the Cramér-Rao inequality, we derive a bound on the navigation speed valid for a wide range of information processing strategies. The framework captures hallmark features of cognitive systems, including optimal sensing durations and a speed-accuracy trade-off that balances noise and reliability. Allowing stored information to degrade before action reveals that although deterioration slows navigation, the trade-off remains governed primarily by external orientational noise and is remarkably insensitive to memory decay.

The $p$-wave magnet has emerged as a new type of magnetism exhibiting odd-parity, time-reversal-symmetric spin splitting in momentum space, and has attracted considerable interest as a promising platform for spintronic applications. However, the theoretical understanding of the fundamental mechanism responsible for stabilizing this phase remains limited. In this work, we identify a microscopic interacting model that realizes the $p$-wave magnet as its ground state. We first introduce a Hubbard model and derive the corresponding low-energy spin Hamiltonian. At the classical level, we find that the $p$-wave magnet is stabilized but remains energetically degenerate with competing noncoplanar states. Quantum fluctuations lift this degeneracy, selecting the $p$-wave magnet as the unique ground state. The resulting electronic structure exhibits finite spin accumulation via the Edelstein effect, highlighting the potential of $p$-wave magnetism for spintronic applications. We further discuss the relevance of our theory to quasi-two-dimensional honeycomb magnets such as Ni$_2$Mo$_3$O$_8$. Our findings establish the possibility of spontaneous $p$-wave magnetism.

Electronic interferometers have been identified as possible candidates for building electronic flying qubits. Such a regime requires ultrafast voltage pulses whose duration is shorter than the time of flight through the device. Understanding the corresponding physics in the presence of such short excitations requires a proper treatment of electron-electron interactions. In this article, we take a step in this direction by performing time-resolved simulations of a Mach-Zehnder interferometer treating the interactions at the time-dependent mean-field level. We find that the main effect of the interaction is the renormalization of the pulse velocity. Very importantly, the interference effects appear to be robust to the presence of interactions.

The transient picture for a Bloch electron accelerating in an arbitrarily time-dependent homogeneous electric field is developed. The temporal sequence for the analysis includes the instant after electron injection, followed by the time required for a small change in electron wavenumber away from initial injection, leading to the final time evolution over many Bloch periods. The time-dependent behavior is studied using the properties of the Schrödinger equation. The electric field is described through the vector potential gauge, and the instantaneous eigenstates of the Bloch, electric-field-dependent Hamiltonian are used as basis states in describing the Bloch dynamics in the electric field. For each temporal sequence considered, the solution to the Schrödinger equation is established and comparatively discussed. The expectation value of the momentum is obtained for the special case of first order in a constant electric field; the resulting velocity derived is a field-dependent generalization of the natural Zitterbewegung-like behavior discussed in the recent literature. The early-time and long-time limits of the momentum expectation value and its time derivative demonstrate that the resistance to Bloch acceleration after initial band injection varies from real mass to effective mass dynamics as the electron accelerates through the band under the influence of electric field. This changing inertia from early injection of a free-mass electron is the result of the {\it real mass} electron {\it dressing-up} into the states of the crystal to become {\it an effective mass} electron. The ramifications of this temporal {\it dressing} behavior are discussed in considering the general dynamics of Bloch electrons subject to ultrastrong electric fields.

Thin-film dewetting is classically governed by an adhesion-mediated spinodal instability in which curvature-driven diffusion controls post-rupture coarsening. We show that internal activity fundamentally restructures this instability. Using a minimal microscopic model of an active liquid film on a solid substrate, we identify a competition between active stresses and film-substrate adhesion that produces two independently regulated dynamical length scales: vertical liquid accumulation and lateral rupture propagation. While passive films exhibit universal diffusion-limited growth, $\ell_z(t)\sim t^{1/3}$, activity converts transport from curvature-controlled diffusion to persistence-driven motion, yielding a continuous increase of the coarsening exponent from $\approx 0.33$ to $\approx 0.6$. The growth law analysis shows that persistent self-propulsion introduces an advective flux that competes with curvature-induced chemical potential gradients, enhancing growth when the persistence length becomes comparable to the evolving domain size. Simultaneously, the rupture front transitions from dissipative spreading to strongly accelerated propagation approaching ballistic scaling. This decoupling shows that activity does not simply renormalize effective surface forces but generates a distinct nonequilibrium interfacial instability governed by the balance between persistence length and adhesion. The results provide a minimal physical mechanism linking classical thin-film dewetting to dewetting-like rupture observed in active and biological materials.

Atomically confined spins are emerging as active components in quantum optoelectronic devices such as quantum bits and sensors. However, interrogating single spins at atomic length-scales remains a sizeable challenge, limited by diffraction in conventional optics. Here we show that the highly-local excitation provided by injecting energetic charge carriers from the atomically sharp probe of a scanning tunneling microscope can trigger single-photon emission from individual atomic vacancy centers in a layered semiconductor. With an effective spatial resolution of <1 nm, we show that the captured light closely mirrors the orbital symmetry of the bound-state wavefunction of the vacancy center while photon correlation measurements confirm single-photon emission, as reflected in clear photon anti-bunching signatures. Our results constitute an important step toward the realization of an electrically addressable single-atom quantum light source and solid-state spinphoton interface, addressed at the atomic-scale.

The low-temperature metallic phase of 1T-TaS2 may originate from current- and voltage-driven destabilization of the commensurate charge density wave (CDW) in a strongly correlated Mott insulator, alongside the robust yet rarely realized influence of intrinsic electronic distortions. Electrical pulse-driven transport, combined with second harmonic response, reveals abrupt switching, negative differential resistance (NDR), and multiscale domain-wall reorganization. The free energy analysis identifies a critical order parameter threshold for the Mott-metal transition, with scaling exponents (β approx 1.3) consistent with 2D percolation. The sparse limit fractal dimension D_{f} approx 0.3 at 10 K, rising to approx 0.9 at 300 K, reflects the hierarchical evolution of the conductive pathways throughout the temperature. These findings establish a direct connection between fractal percolation, pulse-induced instabilities, and correlated electron transport, offering a framework for controlled access to non-equilibrium phase transitions in low-dimensional quantum materials.

Water-in-Salt-Electrolytes (WiSEs) are a promising class of next-generation electrolytes. Unlike classical dilute electrolytes or more conventional battery electrolytes, WiSEs are characterised by their super-concentrated salt concentration with only a small amount of water, which gives rise to their expanded electrochemical stability window (ESW). The expansion of the ESW is, in part, due to the formation of an inorganic solid electrolyte interphase (SEI) that passivates the anode; this principle is also important in graphite and Li-metal anodes, and beyond Li-ion technologies. The solvation and ionic associations are key descriptors in understanding the expansion of the ESW. Specifically, as reactions which lead to the SEI (or cathode electrolyte interphase, CEI) must occur at the electrode-electrolyte interface, the distribution of reactants and their various solvation environments are critical. This distribution near the interface is referred to as the electrical double layer (EDL), in the absence of reactions. Here we further develop and analyse a recently proposed thermodynamic theory of hydration and ionic associations in the EDL of WiSEs. We parameterize this theory from bulk molecular dynamics simulations and benchmark it against EDL simulations, finding good qualitative agreement. Using this thermodynamic theory, we rationalise changes in the ESW through: changes in the activity in the bulk electrolyte through the Nernst equation, which directly changes the stability of the electrolytes; and thermodynamic changes to the kinetics of these reactions, from the Butler-Volmer equation and coupled ion electron transfer kinetics, through the concentration of reactant species in the Helmholtz layer.

We consider a probabilistic cellular automaton (PCA) of evaporation-deposition on the one-dimensional lattice having $n$ sites with periodic boundary conditions, in which each site, during each epoch, can be in one of two states: $0$ and $1$. Fix a positive integer $m\geqslant 2$. There are two types of transitions at each discrete time, which are as follows: (i) the first site in every contiguous block of $m$ $0$s becomes a $1$ with probability $p_1$, and (ii) the first site in every contiguous block of $(m-1)$ $0$s followed immediately by a $1$ also becomes a $1$ with probability $(1-p_2)$. As in a PCA, all of these transitions occur simultaneously. We show that the resulting discrete-time Markov chain is ergodic, and we give an explicit formula for its limiting distribution, the partition function and the density. We also propose necessary and sufficient conditions for this Markov chain to be reversible. For $m=2$, we provide a fully analytical expression for the free energy of this model.

We investigate the buckling of hollow cylindrical tubes subject to their own weight and internal pressure, inspired by the columnar cells of the palisade mesophyll in dicotyledon leaves which resemble pressurized cylindrical tubes. When the internal pressure in the cylinder is equal to the outside pressure, the problem is usually termed self-buckling, which has been studied extensively for solid rods, hollow cylinders, and thin cylindrical shells. Specifically, we perform FEM simulations and desktop-scale experiments to determine the instability thresholds for different geometrical parameters. We first test our models against self-buckling results without pressure for solid rods and hollow cylindrical tubes, and then proceed to determine the critical buckling pressure for a set of material and geometrical parameters. We find that positive internal pressures can stiffen cylinders that are unstable under their own weight, leading to an effective Young's modulus that we show scales linearly with the applied pressure. On the contrary, cylinders that are stable under self-weight, buckle under a negative pressure, resembling classical results on pressure-induced ring buckling. Our findings offer new insights on the interplay between gravity and pressure for the mechanical instability of hollow cylindrical tubes, which we hope will be useful for the study of both engineering and biological structures under similar loads.

Topological bound states emerging at domain walls of dimerized chains provide a robust platform for exploring correlation effects beyond single-particle physics. When such a soliton state is coupled to a metallic substrate, local Coulomb interactions can give rise to Kondo screening. Here we demonstrate analytically and numerically that, in an interacting Su-Schrieffer-Heeger (SSH) chain, the Kondo temperature is directly controlled by the topological mass that governs the bulk gap. Near the topological transition, the Kondo scale collapses linearly with the mass parameter while retaining its exponential sensitivity to hybridization. This establishes a minimal mechanism by which a bulk topological parameter quantitatively determines an emergent many-body energy scale. Our results clarify the strong configuration dependence of soliton-induced Kondo signatures observed in graphene nanoribbon systems on Au(111) and provide experimentally testable predictions for scanning tunneling spectroscopy.

In the present paper we perfomrm molecular dynamics simulation of liquid carbon with a machine-learning potential GAP-20. We show that within the framework of this model carbon demonstrates a relatively low critical temperature, which can affect the results of experimental measurements of melting point of graphite.

A physics-based analytical model describing the phase transition from crystalline to conformationally disordered (condis) crystalline phase is developed. In the model, the free energy is written as a function of temperature and the lattice parameter (mean distance between neighboring chains). It consists of two contributions: elastic and conformational. The elastic contribution describes the interaction between neighboring chains, while the conformational part takes into account the conformation of one chain inside the potential tube, formed by the neighboring chains. To verify this approach, polyethylene - the simplest polymer possessing the condis phase - was chosen as a modeling object. Previous experiments and molecular dynamics simulations show that the typical conformation of a polymer chain in a crystalline phase consists mainly of trans dihedrals and a small fraction of gauche dihedrals, which can be considered as defects of the crystalline lattice. These defects displace the chain inside the tube thus increasing the potential energy. The energy required to form such a defect decreases rapidly with increasing distance between neighboring chains. This leads to a first-order phase transition at a certain temperature to the condis phase, in which distance between neighboring chains is large and a fraction of gauche dihedrals is high. This physical picture of the phase transition is described by the proposed analytical model, the parameters of which were calibrated against the results of molecular dynamics simulations for atmospheric pressure. The model predictions for the pressure of 500 atm and 1000 atm are in perfect agreement with the results of molecular dynamics simulations.

We use a quantum formalism for the partition function of the classical $XY$ model to identify a resilience phase transition in a noisy toric-rotor code. Specifically, we consider the toric-rotor code under phase-shift noise described by a von Mises probability distribution and show that the fidelity between the final state after noise and the initial state is proportional to the partition function of the $XY$ model. We map the temperature of the $XY$ model to the width of the noise in the toric-rotor code, such that a Kosterlitz--Thouless phase transition at a critical temperature $T_{c}$ corresponds to a mixed-state phase transition at a critical width $σ_c$. To characterize this phase transition, we develop a quantum formalism for the spin stiffness in the $XY$ model and show that it is mapped to the gate fidelity in the logical subspace of the toric-rotor code. In particular, we introduce a topological order parameter that characterizes the resilience of the toric-rotor code to decoherence within the logical subspace. We show that the logical subspace does not exhibit complete resilience to noise, which is a necessary condition for correctability. However, it exhibits partial resilience to noise for widths less than $σ_c\approx 0.89$, where the resilience order parameter takes values near $1$ and then drops to zero at $σ_c$. We also use our results to shed light on the correctability of toric-rotor codes in higher dimensions $d > 2$. Our work shows that the quantum formalism for partition functions provides a mathematically rigorous framework for studying correctability in continuous-variable quantum codes.

Strong coupling between molecular excitations and quantized electromagnetic fields in optical cavities provides a powerful means to control the physical and chemical properties of molecular systems. Here, we study electron transfer (ET) dynamics in cavity-coupled molecules using the numerically exact hierarchical equations of motion (HEOM) method, which captures nonperturbative and non-Markovian effects beyond standard perturbative theories. We identify distinct resonance and collective effects associated with polariton formation and show that the ET rate saturates in the strong-coupling regime, a feature not captured by perturbative approaches. We further extend the cavity-modified ET model by incorporating the nuclear-coordinate dependence of molecular electric dipole moments, which gives rise to a three-body interaction involving molecular electronic and vibrational degrees of freedom and cavity photons. This vibronic polariton formation leads to non-monotonic, oscillatory dependencies of the ET rate on the light-matter coupling strength and cavity frequency, which we attribute to quantum interference among multiple transfer pathways. These findings establish cavity-modified electron transfer as a multichannel quantum process governed by the interplay of electronic, vibrational, and photonic degrees of freedom.

Nitrogen-vacancy centers in diamond have been shown to be capable of detecting AC magnetic fields with high sensitivity, spectral resolution, and spatial resolution. However, most studies so far have focused on the regime of time-averaged or time-correlated measurements, while little attention has been paid to the single-shot regime. Here we show that the amplitude and phase of an AC field can be retrieved from a single pair of two consecutive measurements. We demonstrate this concept by measuring a 4 MHz AC field with a per-shot amplitude and phase sensitivity of 78 nT and 63 mrad, respectively, at a temporal resolution of 320 us. We also investigate the effects and quantify the errors resulting from probe frequency detunings, as well as operating in the strong field regime. Moreover, we showcase the ability of the measurement protocol to dynamically change the probe frequency in real-time. This work advances the use of NV centers for real-time measurements of AC magnetic fields.

We propose that the symmetry-breaking bifurcation of coupled topological edge states (CTESs) can be used as a general principle for achieving spontaneous symmetry breaking (SSB) in a nonlinear topological lattice. Using an optical resonator array composed of two Su-Schrieffer-Heeger (SSH) chains as an example, we find that as the nonlinearity strength increases, the symmetric CTESs undergo a supercritical bifurcation. Beyond the critical threshold, the originally stable symmetric state becomes unstable, leading to the formation of a pair of stable asymmetric states. Both sides of the symmetric CTESs exhibit sublattice polarization, while the side of the asymmetric CTESs that is predominantly occupied demonstrates stronger sublattice polarization. We further find that as interchain coupling increases, the frequency range for stable CTESs expands, while the frequency range for stable asymmetric CTESs decreases. Our work provides a universal mechanism for realizing SSB in nonlinear topological lattices.

Liquid foams are archetypal athermal amorphous solids whose elasticity arises from the jamming of densely packed bubbles. We investigate the stress relaxation of non-ageing liquid foams following flow cessation, using fast X-ray tomo-rheoscopy. Thanks to in situ, time-resolved measurements, we uncover robust linear affine relationships between shear stress, plastic activity, and coordination number throughout the relaxation toward a residual stress state below the yield value. In contrast to previous studies on amorphous solids, we observe an exponential relaxation governed by the duration of individual plastic events, rather than by cascades of correlated ones associated with much longer, shear-rate-dependent timescales or power-law relaxations. Our results are consistent with a recent theoretical framework proposed by Cuny et al., suggesting that residual stress originates from the orientation of the stress tensor.

This study demonstrates that the apparent complexity of fracture in phantom-chain polymer networks is fully decoupled into two universal master curves: (i) macroscopic softening governed by the absolute stretch, and (ii) microscopic scission governed solely by the relative stretch. Using the previously proposed network mechanics model, an analytical expression has been derived to quantitatively capture the nonlinear growth of microscopic damage. Combining the softening exponent with polymer-solution scaling yields a simple novel relationship, $σ_{nb} / G \propto (c / c^* )^{(-1/3)}$, where $σ_{nb}$ is the nominal broken strength, $G$ is the initial shear modulus, $c$ is the prepolymer concentration, and $c^*$ is its overlapping threshold.

Dirac particles can undergo perfect transmission through a sufficiently high potential barrier in the Klein zone. Although the perfect Klein tunneling (often referred to as the Klein paradox) is similar to the non-relativistic resonant transmission which occurs only when the kinetic energy exceeds the barrier, the underlying mechanism is believed to be fundamentally distinct. In this work, we show that for the relativistic double-barrier model the perfect-transmission curve can pass continuously from the above-barrier zone to the Klein zone. Additionally, in the Klein zone, perfect transmission occurs even for subcritical barrier heights, supported by both bound-state analysis and wave-packet dynamics. These findings suggest a connection between perfect Klein tunneling and resonant transmission, and provide new insights into the physical nature of the Klein paradox.

Hall effect is an important phenomenon when a magnetic field is applied to materials. From the curve depicting the Hall resistance versus the magnetic field, crucial information such as carrier concentration can be extracted. If the curve exhibits a linear dependence up to rather high magnetic fields, it indicates that charge transport involves only a single type of carrier, and if a non-linear curve is measured, then the double-carrier model should be considered for fitting. However, this model involves four unknown parameters, including the concentration and mobility of the two carriers, resulting in that such fitting is usually non-unique, which significantly reduces the reliability and accuracy. In this work, a double-carrier platform was constructed on a probable excitonic insulator Ta2Pd3Te5, and the four-parameter fitting based on the double-carrier model was simplified to a single-parameter fitting by employing methods such as analyzing the shape of the Hall resistance curve and generating gate-induced Shubnikov-de Haas oscillations. Thus, we provide a reliable method for double-carrier fitting of Hall resistance and a new evidence for the existence of excitonic-insulator state in Ta2Pd3Te5.

We present a microscopic, parameter-free approach for computing the photoluminescence spectra of a single semiconductor nanocrystal. The method derives exciton-phonon coupling directly from the semi-empirical pseudopotential framework and systematically incorporates both diagonal and off-diagonal interactions, expanded to second-order in the phonon modes. The dipole-dipole correlation function was calculated using a Dyson expansion within the Kubo-Toyozawa formalism, enabling a consistent description of the role of pure dephasing and population-transfer on the photoluminescence spectral features. Applied to CdSe/CdS core-shell nanocrystals, the approach quantitatively reproduces experimental photoluminescence spectra over a wide temperature range, revealing that quadratic phonon couplings account for nearly half of the homogeneous linewidth above 100-150 K, while off-diagonal couplings leading to exciton thermalization play only a minor role and only as T approaches 300K.

We give a recursion relation for the second-quantized fermionic (bosonic) Halperin state, which avoids exact diagonalization of its two-component first-quantized parent Hamiltonian. We validate this formula by proving that the second-quantized Halperin state, as recursively defined in this formula, is indeed a zero mode of the corresponding second-quantized parent Hamiltonian and that it has the correct filling factor.

Moiré network formation in graphene bilayers breaks stacking symmetry, giving rise to domain walls that host topologically protected one-dimensional states. Here we show that these systems undergo an additional symmetry breaking at the level of the domain-wall network geometry, leading to the spontaneous emergence of chiral network configurations that are not determined by topology alone. Using atomistic structural relaxation and electronic-structure calculations, we show that TDW networks adopt chiral geometries through lattice relaxation. Via developing a comprehensive phase diagram defined by strain and interlayer flexibility, we discover three equilibrium network morphologies: straight, mono-chiral, and dual-chiral. Chiral networks arise from the global minimization of TDW energy under moiré geometric constraints. Tight-binding calculations show that straight networks host junction-centred states, whereas chiral networks shift spectral weight toward asymmetric edge modes. While topologically protected states naturally emerge at AB/BA domain boundaries in moiré bilayers, we demonstrated that the localization of boundary states is network-symmetry dependent. Our results show that symmetry breaking at both the stacking and network levels provides a new way to understand and control low-energy electronic states in moiré bilayers.

In this work, we investigate the differential voltage generation arising from the direct magnetoelectric (ME) effect in nanoscale composite devices upon magnetization rotation from the magnetic ground state to an out-of-plane (OOP) configuration. These composite devices comprise a magnetostrictive ferromagnetic layer and a piezoelectric layer, mechanically coupled through strain. Using a finite element method (FEM) model, developed in COMSOL Multiphysics, we provide a comprehensive analysis of strain transfer mechanisms and resulting voltage generations. Here, the influence of dimensional and material parameters on the device performance is systematically examined. Our results indicate the presence of two distinct strain transfer mechanisms at scaled dimensions, where the device aspect ratio and the magnetic state both determine the dominant mechanism influencing the strain transfer to the piezoelectric layer. Moreover, we observed that the influence of surface clamping diminished as the pillar area was reduced. We also saw that the strain transfer to the piezoelectric layer can be enhanced by using stiffer electrodes or clamping layers. Lastly, we concluded that magnetostrictive materials with large magnetoelastic coupling constants or large Poisson ratios may strongly increase the output voltage at small dimensions. This study provides insight in the dimension and material selection when designing scaled ME pillars, with the aim of generating large output voltages. We showed that output voltages exceeding 200 mV can be achieved in scaled devices, underscoring the potential of these structures for integration into microelectronic applications.

Topology without quasiparticles has emerged as a key framework for understanding Mott insulators, where Green's-function zeros encode nontrivial topological structure. Yet, experimental detection of these zeros represents a challenge. Using exact diagonalization of the one-dimensional Hubbard model with an impurity and Zeeman field, supported by exact analytic results, we show that Green's-function zeros manifest as an in-gap spectral weight in the unitary scattering regime. In this limit, we map the impurity problem onto a doped Mott insulator and identify the resulting in-gap state as a "zeron" excitation which is a localized doublon (holon) for an attractive (repulsive) potential. The zeron spectral weight and its associated zero vanish above a critical Zeeman field. Our results imply that Green's function zeros have in fact already been observed in experiments, and establish impurity and magnetic-field tuning as practical tools for controlling their topology.

The shear viscosities, long-time self-diffusion coefficients, and sedimentation velocities in monodisperse and bidisperse hard-sphere colloidal suspensions are simulated for volume fractions up to 0.40 using multiparticle collision dynamics with a discrete particle model. The bidisperse suspensions have diameter ratios of 2 and 4 and equal amounts of each particle by volume. All measured properties for monodisperse suspensions are found to be in good agreement with prior literature; however, they highlight the sensitivity of the simulation method to discretization effects. The sedimentation velocities for the bidisperse suspensions are also in reasonable agreement with prior literature, including direction reversal for the smaller particles when the diameter ratio is 4. This work provides reference data for transport properties of colloidal suspensions and establishes the suitability of multiparticle collision dynamics for modeling suspensions of particles with different sizes.

Two-dimensional conjugated metal-organic frameworks hold great promise for applications in chemiresistive sensing, electrocatalysis, and energy storage. Their interfacial interaction with metal electrodes, which has been rarely investigated, exerts a critical influence on the electronic properties and device performance. As a representative material, M3(HITP)2 (M = Ni, Cu; HITP = 2,3,6,7,10,11-hexaiminotriphenylene) exhibits excellent performance in various electronic devices, yet the microscopic mechanism of the interfacial interaction in M3(HITP)2/metal heterostructures remains unclear. Here, we report the synthesis, scanning tunneling microscopic characterization, and tight-binding analysis of monolayer M3(HITP)2 epitaxially grown on Au(111). Scanning tunneling spectroscopic mapping reveals a commensurate kagome-hexagonal-honeycomb triple-lattice architecture. The Au(111) substrate renormalizes the electronic band structure of M3(HITP)2, pinning the Fermi level and generating a ligand-derived flat band at 0.4 eV that corrects prior misassignment of orbital character. Meanwhile, the periodic and microporous M3(HITP)2 lattice strongly modulates the surface electronic state of Au(111) via electron-phonon coupling and quantum confinement, the latter of which gives rise to a quantum corral network exhibiting two resonant states within each pore. The formation of fully dispersive electronic bands and the robust quantum corral network requires crystallites comprising at least ten pores. The atomic-scale investigation of M3(HITP)2/Au(111) epitaxial heterostructures elucidates interlayer coupling mechanisms and advances the understanding of metal-organic framework/metal interfaces that are integral to electronic and energy-storage devices.

Defect engineering is crucial for tuning 2D transition metal dichalcogenide properties for quantum and optoelectronic applications. While conventional photoluminescence (PL) and Raman spectroscopies are important characterization tools, their mapping in large area samples can be time-consuming and lacks direct sensitivity for comprehensive defect characterization. Here, we introduce resonance-enhanced four-wave mixing (FWM) imaging for precise imaging and characterization of vanadium-induced defect states in WS2 monolayers. Our multi-modal investigation, integrating hyperspectral PL, Raman, and supported by density functional calculations, reveals nanoscale doping inhomogeneities, their influence on excitonic and vibrational properties. We observe resonance-enhanced FWM signals correlating with vanadium-induced defect regions, evidencing their unique nonlinear optical response. This work establishes FWM as an essential platform for high-resolution, defect-sensitive imaging, advancing defect-engineered excitonic devices and enabling novel nonlinear quantum photonics.

The physics in flat bands has emerged as an essential field in condensed matter physics where a plethora of phenomena can be unveiled, such as anomalous transport properties, superconductivity dominated by quantum geometry or exotic topological phases. Our goal here is to show that even in magnetic systems, the presence of flat bands can give rise to unexpected features. More precisely, we address the impact of an Aharonov-Bohm (AB) flux on the exchange couplings in magnetic diamond chains. The most remarkable result is the significant amplification of magnetic couplings at short distances induced by the AB flux, leading to a considerable increase in the thermal conductivity of the magnons. We have also shown that the flux-dependent decaying length of the couplings is connected to the quantum metric of the flat bands. Our results could be of interest for the control of magnetic properties in spintronic devices and relevant for the heat transport by magnons at the nanoscale in quantum technologies.

We consider the Fradkin-Shenker ${\mathbb Z}_2$ gauge-Higgs lattice model in 2+1 dimensions, i.e. the toric code deformed by an in-plane magnetic field. Its phase diagram contains a multicritical CFT with gapless, mutually non-local electric and magnetic particles, exchanged by a ${\mathbb Z}_2^{\mathsf{D}}$ self-duality symmetry. We introduce a staggered generalization of the model in which these particles carry global $U(1)_e$ and $U(1)_m$ charges, respectively, and we propose a continuum QFT description in terms of QED$_3$ with $N_f = 2$ Dirac fermion flavors and a charge-two Higgs field with Yukawa couplings. The conjectured phase diagram harbors a multicritical CFT with $(O(2)_e \times O(2)_m)\rtimes\mathbb{Z}_2^\mathsf{D}$ symmetry, some of which is emergent in the QFT description. We compute the scaling dimensions of some operators using a large-$N_f$ expansion and find agreement with the emergent selection rules. The staggered model admits a deformation to the original Fradkin-Shenker model, which maps to unit-charge monopole operators in Higgs-Yukawa-QED$_3$ that break the $U(1)_e \times U(1)_m$ symmetry. We show explicitly that this deformation reproduces all features of the Fradkin-Shenker phase diagram. Finally, we propose a multicritical duality between Higgs-Yukawa-QED$_3$ and the easy-plane $\mathbb{ CP}^1$ model (i.e. two-flavor scalar QED$_3$ with a suitable potential), which describes spin-1/2 anti-ferromagnets on a square lattice. This duality implies a first-order line of Néel-VBS transitions ending in a deconfined quantum multicritical point, described by the same $O(2)_e \times O(2)_m$ symmetric CFT that arises in the staggered Fradkin-Shenker model, which separates it from a gapped ${\mathbb Z}_2$ spin liquid phase.

We investigate Coulomb spin liquids in classical spin-3/2 ice and show that the enlarged on-site Hilbert space gives rise to a qualitatively new class of such phases. Beyond the conventional magnetic monopoles of spin-1/2 ice, the system hosts additional low-energy crystal-field excitations, whose interplay with monopoles significantly modifies both equilibrium and non-equilibrium properties. Following a thermal quench, we find a pronounced dynamical arrest manifested in an exponentially long-lived {athermal} plateau in spin autocorrelations. This constitutes a rare example of dynamical arrest in a short-range interacting, disorder-free system. We demonstrate that the arrested dynamics originate from novel composite excitation structures unique to spin-3/2 ice and from kinetically constrained relaxation pathways that require activated processes. Our results establish higher-spin ice as a fertile platform for realising unconventional Coulomb spin liquids and dynamical arrest without quenched disorder.

We consider N interacting quantum particles on a one-dimensional lattice, and subjected to an external linear potential. For N = 1, the corresponding Hamiltonian is explicitly diagonalizable, with superexponentially localized eigenstates. This is called Stark localization. We prove that superexponential spectral localization persists for arbitrary N and every interaction strength.

We analyse a continuous-time random walk model with stochastic reversals of direction. There is no external potential but the reorientation mechanism generates a non-zero current from asymmetry in the forward and backward waiting-time distributions (even when they have the same mean); the system can therefore can be considered as a type of active particle ratchet. We derive an explicit expression for the mean ratchet current with exponentially distributed reorientation times and also develop a general renewal-theory framework to obtain the full large deviations, using this to comment on the possibility of dynamical phase transitions.

Spin glass theory studies the structure of sublevel sets and minima (or near-minima) of certain classes of random functions in high dimension. Near-minima of random functions also play an important role in high-dimensional statistics and statistical learning, where minimizing the empirical risk (which is a random function of the model parameters) is the method of choice for learning a statistical model from noisy data. Finally, near-minima of random functions are obviously central to average-case analysis of optimization algorithms. Computer science, statistics, and machine learning naturally lead to questions that are traditionally not addressed within physics and mathematical physics. I will try to explain how ideas from spin glass theory have seeded recent developments in these fields. (This article was written on the occasion of the 2024 Abel Prize to Michel Talagrand.)

Navigation up a sensory gradient is one of the simplest behaviours, and the simplest strategy is run and tumble. But some organisms use other strategies, such as reversing direction or turning by some angle. Here we ask what drives the choice of strategy, which we frame as maximising up-gradient speed using a given amount of sensory information per unit time. We find that, without directional information on which way to turn, behavioural strategies which make sudden turns perform better than gradual steering. We see various transitions where a different strategy becomes optimal, such as a switch from reversing direction to fully re-orienting tumbles as more information becomes available. And, among more complex re-orientation strategies, we show that discrete turn angles are best, and see transitions in how many such angles the optimal strategy employs.

Developing intrinsic homojunctions without chemical heterogeneity remains a key challenge in future two - dimensional devices. Here, we report a geometry - defined metal--semiconductor--metal homojunction in bilayer blue phosphorus (BlueP) created by a localized bubble corrugation, without chemical doping or foreign - material interfaces. First - principles calculations show that enlarging the interlayer separation in the metallic A\(_1\)B\(_1\) - stacked BlueP bilayer opens a band gap, enabling a semiconducting barrier embedded between metallic segments. First - principles quantum - transport simulations reveal a crossover from ballistic to tunneling transport upon bubble formation. In the tunneling regime, transmission decreases exponentially with bubble width while remaining weakly sensitive to bubble height and bulging direction. The junction acts as an orientation - dependent \(k\) - space filter, producing transport anisotropy and momentum selectivity. Orbital - resolved scattering analysis shows that intralayer - bonding channels persist under deformation whereas interlayer - hybridized channels are quenched, and that σ- type bonding yields higher conductance than π- type bonding. These insights motivate two electromechanical device concepts: a mechanically switchable memory element with ON/OFF ratios up to 30 and a nanoscale sliding rheostat with reproducible exponential resistance tuning for Å- scale displacement sensing.

Altermagnets exhibit momentum-dependent spin splitting despite having zero net magnetization. This enables a spin-splitter effect in which an external electric field generates transverse spin currents by separating oppositely polarized carriers. Here, we develop a unified semiclassical theory of linear extrinsic spin-splitter currents, incorporating impurity-induced side-jump and skew-scattering contributions, and apply it to the $d$-wave altermagnet \ch{FeSb2}. We demonstrate that asymmetric impurity scattering provides a dominant channel for spin-splitter currents. Remarkably, the resulting extrinsic spin conductivity is time-reversal even, in contrast to previously studied spin-splitter responses arising from symmetric scattering.

Macroscopic quantum electrodynamics (MQED) provides a unified framework to describe quantum electromagnetic fields in the presence of arbitrary macroscopic environments. Central to this theory is the field correlation, which governs both radiative (e.g., Lamb shifts and the Purcell effect) and mechanical phenomena, such as van der Waals and Casimir forces. In this tutorial, we provide an overview of MQED and its extension to active media, highlighting fluctuation-induced forces as manifestations of gain-modified field correlations.

We report the chaotic switching phenomenon in the minimal $N = 3$ pendula network with global coupling. Analyzing the stability conditions of the chimera states and their dependence on the parameters, three scenarios of chaotic switchings are identified: 1) a riddling bifurcation scenario, where an unstable periodic orbit inside the chimera manifold becomes transversally unstable, 2) a blowout bifurcation scenario, where the switching is caused by the transverse destabilization of the chaotic chimera with respect to its manifold, and 3) switchings between "laminar" saddle chimeras within a global "turbulent" attractor. The results are obtained based on the detailed examination of the existing regimes including chimera states, limit cycles and fixed points, their multistability and switching regime. In the parameter regions where the chaotic chimeras coexist with stable non-chaotic solutions, the switching trajectory can eventually escape to a stable solution, causing an additional unpredictability in the system behavior, as it is difficult to predict the escaping moment.

We present a renormalization-group perspective on spontaneous stochasticity in hydrodynamic turbulence, viewed through the lens of multiscale dynamical systems. Building on previously established results for a solvable multiscale Arnold's cat model, we show that spontaneous stochasticity emerges as a universal fixed point of an RG transformation acting on Markov kernels, independent of the microscopic regularization. Classical examples - including the Feigenbaum equation, the central limit theorem, and hierarchical spin models - are reinterpreted within the same framework, placing spontaneous stochasticity alongside other universality phenomena.

For fully connected populations of diffusively coupled bistable elements, we identified three qualitatively distinct mechanisms of noise-induced escape as coupling strength varies [H. Ishii and H. Kori, arXiv:2512.01388 (2025)]. Here we generalize these results to a class of networked systems and demonstrate that degree heterogeneity (i.e., variability in node degree) shapes escape mechanisms alongside coupling strength. In applied contexts, networks of noisy bistable elements provide a minimal conceptual framework for understanding abrupt state transitions in complex systems. Theoretically, a quantitative approach to escape is challenging because nonlinearity, network interactions, and dynamical noise jointly shape the collective dynamics. We extend the analytical framework developed for the fully connected model to a class of networked systems based on the annealed network approximation. We derive three effective one-dimensional descriptions of collective escape dynamics. We validate our theoretical predictions for mean escape times by direct numerical simulations. Our analysis reveals that the validity and quantitative behavior of the reduced descriptions depend on degree heterogeneity in addition to coupling strength. This work extends the classification of escape mechanisms to networked bistable elements. Furthermore, our analytical framework provides tools for understanding synergistic phenomena arising from the interplay of nonlinearity, diffusive coupling, and dynamical noise.

We propose a coupled system for the nonlinear interaction between high-frequency, circularly polarized, intense electromagnetic (EM) waves and low-frequency electron-density perturbations, driven by the EM-wave ponderomotive force, in an unmagnetized plasma composed of fully degenerate relativistic electrons and stationary positive ions, including a higher-order correction to the nonlocal nonlinearity. We show that the modulational instability (MI) growth rate associated with the generation of EM envelope solitons gets significantly reduced with a slight increase in either the nonlocal nonlinear correction or the degeneracy parameter. Furthermore, a three-wave temporal model predicts the existence of quasiperiodic and chaotic states of EM solitons while interacting with longitudinal electron density perturbations. We show that the greater the degeneracy (or the higher the contribution from the nonlocal correction), the smaller the instability domain of modulation wave numbers; thus, degeneracy favors the stability of EM soliton evolution. The existence of temporal chaos in a low-dimensional model could be a signature of the development of spatiotemporal chaos in the complete nonlinear model, in which many electromagnetic solitons can be excited and saturated as they interact with electron plasma waves.

Online information is increasingly linked to real-world instability, especially as automated accounts and LLM-based agents help spread and amplify news. In this work, we study how information spreads on networks of Large Language Models (LLMs) using mathematical models. We investigate how different types of offline events, along with the "personalities" assigned to the LLMs, affect the network dynamics of online information spread of the events among the LLMs. We introduce two models: 1) a stochastic agent-based network model and 2) a system of differential equations arising from a mean-field approximation to the agent-based model. We fit these models to simulations of the spread of armed-conflict news on social media, using LLM agents each with one of 32 personality trait profiles on k-regular random networks. Our results indicate that, despite the complexity of the news events, personalities, and LLM behaviors, the overall dynamics of the system are well described by a Susceptible-Infected (SI) type model with two transmission rates.

While very powerful, integrability in semiclassical string solutions is known to be a rare property. Motivated by the need to understand and characterise the large landscape of non-integrable string dynamics, we extend Krylov methods for probing chaos to classical systems. We introduce a Koopman-Krylov framework, formulated in the Koopman-von Neumann description of classical mechanics and implemented via a generator extended dynamic mode decomposition (gEDMD) approximation of the Koopman generator acting on observables. Using this framework, we study how integrability-breaking deformations of integrable string dynamics induce characteristic redistributions of spectral weight, leading to observable-dependent delocalisation and spreading in Krylov space. We illustrate the Koopman-Krylov diagnostics across three classes of non-integrable semiclassical string solutions.

The stochastic dynamics of orthogonal metal cutting with both regenerative and nonsmooth frictional effects are investigated numerically in this paper. The shortcomings of neglecting nonsmoothness in frictional and stochastic effects in modeling the dynamics of such a machining process are demonstrated. Dynamics of the tool motion is observed to exhibit rich nonlinear phenomena such as stick-slip during chatter, with stochastic perturbations in cutting forces adding further complexity, leading to the occurrence of stochastic P and D bifurcations. Measures of entropy are found to be effective in quantifying the dynamical transitions occurring in the dynamics of the tool. Subsequently, basin stability analyses, modified to account for stochasticity and time-delays, are carried out to systematically investigate the dynamics of the cutting tool across multiple surface roughness profiles of the workpiece. Basin stability analyses indicate that chatter can be controlled by restricting initial tool displacement and controlling initial workpiece surface roughness, suggesting practical strategies to improve machining outcomes for precision manufacturing.

Correcting measurements for detector effects and constructing appropriate public data representations is a pressing problem in LHC physics. Current methods solve this inverse problem by relying on iterations, minimax optimization, or a surrogate forward mapping. We introduce Adversary-free Unfolding SanS Iteration or Emulation (AUSSIE), which dispenses with these mechanisms while remaining asymptotically correct. AUSSIE replaces the second OmniFold step with a new loss function that directly yields solutions with minimal dependence on the reference simulation. We showcase AUSSIE on various unfolding tasks, including full-phase-space jet substructure.

Silicon is the undisputed cornerstone of modern technology, with applications ranging from micro- and opto-electronics to quantum technologies. Recently, the exploration of its allotropes has emerged as a pivotal frontier for engineering materials with tailored optical and electronic functionalities. High-pressure experiments have revealed several metastable silicon phases, among which is Si-XIII. First observed more than 20 years ago, this phase has remained structurally unidentified, representing a significant gap in our understanding of elemental silicon allotropy. In this work, a convergent methodology is employed combining advanced theoretical modeling with experimental characterization to finally resolve the long-standing structural assignment of Si-XIII. Guided by careful experimental observations, a structural model validated through first-principles optimization and systematically tested against multiple experimental signatures is constructed. All the fingerprints of this phase are rationalized by our proposed crystal structure: interplanar spacings, Raman frequencies, thermodynamic stability, and kinetic pathways. These findings provide a crucial missing piece in the high-pressure phase diagram of silicon and demonstrate the power of integrating computational predictions with experimental validation to resolve complex structural problems in materials science.

Multi-wavelength digital holography enables surface-shape measurements with an exceptional dynamic range by combining interferometric resolution with synthetic wavelengths spanning multiple length scales. Although the concept promises measurement ranges of many orders of magnitude, its practical implementation is limited by the lack of light sources that allow fast, reliable, and calibration-free switching between synthetic wavelengths over a wide frequency range. Here, we present a synthetic-wavelength generator based on an electro-optic frequency comb with electronically tunable modulation frequency and a set of switchable band-pass filters. By combining discrete selection of comb-lines with continuous radio-frequency tuning, the proposed scheme merges the advantages of single-sideband modulation and filter-based comb extraction. Using only off-the-shelf components, the system provides synthetic frequencies from 0.1-220GHz, corresponding to synthetic wavelengths from meters down to millimeters in the visible. The generator achieves MHz-level frequency accuracy, side-mode suppression exceeding 40dB, and switching times below 25ms, even without active stabilization. We characterize the spectral purity and frequency agility of the source and demonstrate rapid tuning of synthetic wavelengths over 3 orders of magnitude. We apply the light source to multi-wavelength digital holography and reconstruct the surface of an industrially machined metal part featuring height variations from 0.1-100mm. The measurements achieve ten-mum-level precision using 7 single wavelengths covering synthetic wavelengths from 1.36mm to 1.874m within an acquisition time < 2s. The presented architecture combines high dynamic measurement range of 50dB, fast electronic reconfigurability, and intrinsic frequency calibration, making it a promising light source for high-speed interferometric surface metrology.

A wide range of numerical and semi-analytical approaches has been developed for optical modeling of photonic-crystal surface-emitting lasers (PCSELs). However, a systematic framework for comparing their predictive capabilities and identifying their respective validity limits remains largely unexplored. In this work, we introduce a comparative methodology in which four representative methods - including rigorous numerical and effective-index-based approaches - are analyzed and partially hybridized within a coupled-wave-theory framework. Using single-lattice and double-lattice PCSELs as representative models, we demonstrate that this approach not only reveals fundamental differences between predictions of rigorous numerical methods and the coupled-wave-theory framework, but also captures a qualitative phase transition and a symmetry-broken phase of quasi-bound states in the continuum (quasi-BICs) relevant for laser operation.

Preterm birth is associated with premature cervical remodeling, yet current clinical assessments cannot detect the underlying microstructural changes in collagen organization. We apply imaging Mueller polarimetry to murine cervical tissue at three gestational stages (early, mid, late) and develop classification methods to predict gestational stage from polarimetric maps. Using Lu-Chipman decomposition, we extract orientation and azimuth local variability maps that capture collagen fiber alignment and disorder. We evaluate two approaches under 20-fold leave-one-out cross-validation: an analytical threshold classifier on mean azimuth local variability, and a lightweight CNN ensemble (approximately 76k parameters) operating on spatially resolved maps. The ensemble achieves 70..0% sample-level accuracy, outperforming the analytical baseline (55.0%), with strong performance on early (71.0%) and late (86.0%) gestation. Spatial prediction maps confirm that classification accuracy is highest in the stroma, where collagen remodeling is most prominent. These results demonstrate that Mueller polarimetry combined with deep learning models can detect gestational collagen remodeling noninvasively, offering a potential pathway toward objective cervical assessment for preterm birth risk.

We report on the experimental detection of shockwave dynamics produced in a spark discharge, using a nonresonant four-wave mixing optical technique. In particular, we observe the spark-induced local density perturbation across a millimeter-range probe volume, centered on the discharge, via single-shot coherent Rayleigh-Brillouin scattering. We detect the emergence of shock-induced flow velocities, which appear as distinct features in the spectrum, and monitor their dynamic evolution from a few hundred nanoseconds to microseconds after the spark. Finally, we benchmark our measurements against simulations based on a one-dimensional compressible flow model. Our results pave the way for quantitative measurements of highly non-uniform transient flows in challenging environments featuring non-equilibrium gas kinetics.

This workshop report from "Shaping the Digital Future of ErUM Research: Sustainability & Ethics" (Aachen, 2025) reviews progress on sustainability measures in data-intensive ErUM-Data research since the 2023 call-to-action on resource-aware research. It evaluates short-, medium-, and long-term actions around monitoring and reducing CO2 emissions, improving data and software FAIRness, optimizing workflows and computing infrastructures, and aligning operations with low-carbon energy availability, including concepts such as "breathing" computing centers, long-term data storage strategies, and software efficiency certification. The report stresses the need for systematic teaching, training, mentoring, and new support formats to establish sustainable coding and computing practices, particularly among students and early-career researchers, and highlights the importance of dedicated steering and funding instruments to embed sustainability in project planning. Ethical discussions focus on the transformative use of AI in ErUM-Data, addressing autonomy, bias, transparency, explainability, attribution of responsibility, and the risk of deskilling, while reaffirming that accountability for scientific outcomes remains with human researchers. Finally, the report emphasizes that sustainable transformation requires not only technical measures but also targeted awareness-building, communication strategies, incentives, and community-driven initiatives to move from awareness to action and to integrate sustainability and ethics into everyday scientific practice.

Microorganisms live in inherently dynamic environments where fluctuations in biotic and abiotic factors shape their behaviour, physiology, and fitness. The concept of ecological memory: the lasting imprint of prior environmental cues, suggests that past exposures can exert prolonged effects on microbial growth, resilience, and phenotypic expressions. For motile microbes in aquatic ecosystems, environmental variability is mediated by fluid motion, which may engender a form of hydrodynamic memory, whereby prior exposure to specific spatio-temporal cues influence future growth and migratory behaviour. Yet, the emergence of such flow-induced memory, or its long-term consequences for trait evolution and population dynamics, remain unexplored. We integrate millifluidic flow control, high-resolution cell tracking, and tunable hydrodynamic cues to quantify growth and migration of Heterosigma akashiwo, a model microbe, across growth stages. Using two complementary perturbation scenarios: standard (flow after static conditions) and reverse (flow before static growth), we test how the temporal structure of forcing shapes multigenerational responses. This combinatorial design disentangles exposure history from its duration, and reveals how prior flow modulates sensitivity, generating legacy effects. Compared with static controls, repeated hydrodynamic exposure alters doubling time, carrying capacity, gravitactic stability, and swimming speed distributions; shifting growth phase progression and tolerance to subsequent perturbations. These results establish a mechanistic framework for flow-induced memory in motile microbes, revealing how past fluidic cues shape future growth and migration. Our study advances predictive understanding of motile microbes in natural and engineered hydrodynamic systems experiencing increasing variability under global environmental changes.

Dynamical control of the nonlinear optical properties of solids -- with light itself -- will be essential for future ultrafast photonic technologies. Previously, methods to modulate nonlinear processes including second-harmonic generation (SHG) have relied primarily on non-resonant light-matter interaction or photo-generation of hot electrons in nanoscale materials. However, these approaches are typically constrained by limited interaction lengths and the initial frequency conversion is relatively weak under equilibrium conditions. Here, an approximately 30\% modulation of efficient phase-matched SHG in bulk beta-barium borate (beta-BaB2O4) is achieved through transient lattice deformation by intense terahertz (THz) pulses that are tuned to resonance with an infrared-active phonon mode. The effect originates from modification of the index of refraction ellipsiod and the corresponding nonlinear phase-matching conditions, rather than from direct modulation of the nonlinear susceptibility through THz-mediated chi^(3) processes. This mechanism, of resonant selective lattice excitation, points toward novel THz-control schemes to tune the nonlinear optical response in materials.

Spectral density functions quantify how environmental modes couple to quantum systems and govern their open dynamics. Inferring such frequency-dependent functions from time-domain measurements is an ill-conditioned inverse problem. Here, we use exactly solvable spin-boson models with pure-dephasing and amplitude-damping channels to reconstruct spectral density functions from noisy simulated data. First, we introduce a parameter estimation approach based on machine learning regressors to infer Lorentzian and Ohmic-like spectral density parameters, quantifying robustness to noise. Second, we show that a cosine transform inversion yields a physics-consistent spectral prior estimation, which is refined by a constrained neural network enforcing positivity and correct asymptotic behaviour. Our neural network framework robustly reconstructs structured spectral densities by filtering simulated noisy signals and learning general functional dependencies.

We investigate the charging of a nonlinear quantum battery consisting of a single bosonic mode subject to a saturable nonlinearity, coherent driving, and dissipation. In contrast to Kerr-type anharmonicities, the saturable interaction induces a bounded and nonlinear distortion of the energy spectrum, leading to a progressive increase in the density of energy levels. We analyze the time evolution of the energy and ergotropy of the battery by solving a Lindblad master equation and show that the nonlinear spectral structure significantly affects both transient charging behavior and steady-state properties. Our results reveal that, for a broad range of parameters, the saturable nonlinearity enhances the maximum stored energy and modifies the ergotropy generation in the presence of losses. The interplay between dissipation and bounded spectral nonlinearity provides a controllable mechanism to tune energy storage and work extraction in bosonic quantum batteries.

As terahertz (THz) frequencies emerge as promising candidates for next-generation wireless networks, accurate characterization of propagation mechanisms in indoor/outdoor environments becomes essential for system design and performance optimization. This article presents an experimental and theoretical investigation of structure-induced indoor surface scattering on THz channels, examining how material properties and structural configurations jointly govern channel power and angular distribution. Six representative indoor surfaces are characterized, revealing that intrinsic structural inhomogeneity -- particularly the quasi-periodic earlywood-latewood arrangement in pine wood -- induces measurable angular scattering whose dominant lobes and angular shifts are reproduced by a beam-propagation modeling (BPM) framework. Material-covered surface configurations are further investigated, demonstrating that thin dielectric covering layers can substantially modify reflection characteristics through thickness- and frequency- dependent thin-film interference effects. Wide-angle bistatic measurements conducted in a conference-room environment reveal that structured indoor elements, such as folded curtains, can enhance angular scattering and extend spatial coverage. These findings establish that structure-induced surface scattering mechanisms offer potential for constructing non-line-of-sight THz links in indoor environments.

Recent studies have shown that novel collective behaviors emerge in complex systems due to higher-order interactions. However, the way in which the structural correlations of these interactions shape such behaviors remains a significant gap in current research. To address this, we use signatures of higher-order behaviors (HOBs) to identify the underlying dynamical rules, or higher-order mechanisms (HOMs). In this work, we compare several HOB measures derived from information theory. Utilizing a simplicial SIS contagion model, we demonstrate that simpler, computationally efficient measures can serve as robust indicators of HOMs. We uncover the novel phenomenon of cross-order induced behaviors, where behavioral signatures emerge at interaction orders where no direct mechanism is present. Crucially, these cross-order HOBs are not simply induced by structural correlations -- such as nestedness and hyperedge overlap -- but they appear in the neighborhood of any HOM. Among the information-theoretic measures we tested, synergy is the most reliable indicator of the true order where the underlying mechanism is at play. These findings offer new insights into the relationship between the network structure and observed dynamics of higher-order systems.

Speckled illumination enhances widefield fluorescence microscopy by enabling optical sectioning and super resolution. In random illumination microscopy, sequences of speckled illumination patterns are used to excite fluorescent samples and images are reconstructed based on a statistical analysis of the intensity fluctuations. Although random illumination microscopy has been shown to give excellent performance, its widespread implementation is hindered by the high cost and complexity of the generation of suitable speckled illumination patterns, which is achieved using digital micro-mirror devices or spatial light modulators. Here, we present a zwitterion-doped liquid crystal (LC) device capable of generating independent, high-contrast speckle patterns with a tunable decorrelation time in the 0.1 s to 0.1 ms range under visible laser illumination. This LC-based dynamic speckle generator is applied to widefield random illumination fluorescence microscopy of tissue and cell samples, where it enables optical sectioning with a 2 micron axial resolution, and a 1.5-fold improvement in lateral spatial resolution. Owing to its low cost and simplicity, this LC speckle generator offers an attractive alternative to digital micro-mirror and spatial light modulator devices for implementing widefield random illumination microscopy.

We report an efficient experimental method for generating high-order Laguerre-Gauss (HOLG) modes by simply coupling a Gaussian beam into the cladding of a multimode fiber (MMF). In particular, the order of the HOLG mode remains invariant with respect to input power, propagation distance, and pulse duration. Furthermore, spectral and power measurements confirm that the beam-shaping mechanism is predominantly linear, whereas Kerr nonlinearity primarily affects the longitudinal phase-matching condition and conversion efficiency, without altering the generated mode order. Altogether, these findings establish our approach as a highly robust and scalable platform for generating tailored optical beams.

Photoacoustic tomography (PAT) and thermoacoustic tomography (TAT) both leverage acoustic signals generated by electromagnetic absorption to noninvasively image deep tissues. PAT operates by detecting optical absorption, whereas TAT targets radiofrequency (RF) absorption, providing complementary information on tissue composition and structure. Combining these modalities into a single system promises richer contrast but remains difficult due to the expense and complexity of the RF source. Here, we show that PAT can be integrated with a low-cost RF heater and used to image both optical and RF absorption in tissue phantoms. RF Heating-Enhanced Photoacoustic Tomography (HEPAT) maps RF absorption via temperature-dependent changes in thermomechanical properties, which enables the use of slow, inexpensive RF subsystems and provides an additional layer of contrast. HEPAT therefore provides distinct, complementary contrast relative to existing photoacoustic imaging systems, expanding specificity and diagnostic power while opening new avenues for studying temperature-related tissue phenomena.

Stroboscopic nanoscale imaging with free electron laser light is revolutionizing our understanding of fast dynamics in heterogeneous systems. The short wavelength of X-ray and extreme ultraviolet radiation makes it possible to achieve nanoscale resolution, while resonance with atomic transitions gives access to electronic and magnetic degrees of freedom. Here, we report on our implementation of a recently developed imaging method, randomized probe imaging, at a free electron laser. The advantage of randomized probe imaging over existing methods is its compatibility both with extended and strongly scattering samples. Our implementation delivers robust single-shot reconstructions at up to a full-pitch resolution of 400 nm over a field of view with a 40 μm diameter. We also demonstrate single-shot imaging of magnetic domain structures using circular dichroism at resonance, paving the way to future time-resolved studies of magnetic dynamics, shock physics, and the dynamics of collective electronic phases.

We report on a comb-locked cavity ring-down spectrometer developed for high-precision, SI-traceable, molecular spectroscopy of air-broadened CO2 gas samples. The experimental setup relies on the use of a singly-resonant optical parametric oscillator that acts as an intermediate link between a 2 micron external-cavity diode laser and an optical frequency comb stabilized against a GPS-disciplined Rb-clock. Absorption spectra of the R(50) ro-vibrational component of the CO2 20012-00001 band have been recorded with high precision and fidelity. As a result of a refined spectral analysis, based on the implementation of the modified Hartmann-Tran profile, line center frequencies, pressure broadening and pressure shifting coefficients have been determined. Finally, we demonstrate the measurement of CO2 mole fractions with a subpromille statistical uncertainty

Optical-field sampling using second-harmonic generation (SHG) from strong-field ionization enables ultrabroadband terahertz detection, but the microscopic origin of the SHG signal and its ultrabroadband response have been unclear. Here we show that the target field lifts the half-cycle cancellation of photoelectron dipole emission, generating the SHG signal used for field sampling. Time-dependent Schrodinger-equation simulations, supported by classical-trajectory Monte Carlo analysis, demonstrate that the SHG yield directly encodes the instantaneous target electric field at the ionization time, enabling waveform retrieval by scanning the probe-target delay. Because the SHG response is gated by a subcycle ionization window rather than the probe envelope, the detection bandwidth can extend far beyond the probe duration. We further quantify practical constraints on retrieval, including intrinsic probe asymmetry and SHG back-action, providing a predictive framework to optimize sensitivity, temporal resolution, and fidelity through controlled electron dynamics.

Smoking behavior and awareness co-spread through social interactions, giving rise to coupled contagion processes on social contact networks. In addition to initiation and cessation, awareness of the harmful effects of smoking plays an important role in shaping individual behavior and population-level outcomes. In this work, we develop a mathematical model to study the coupled dynamics of smoking behavior, quitting, and awareness in a population. A deterministic framework based on ordinary differential equations is first formulated to capture the interplay between social influence and awareness-driven behavioral change. Analysis of the model reveals the existence of smoking-free and smoking-endemic steady states, and identifies conditions under which awareness can reduce or suppress the persistence of smoking. Since social interactions are often localized rather than well mixed, the mean-field description is then extended to a network-based model that incorporates structured contact patterns. Numerical simulations performed on empirical social networks indicate that contact heterogeneity and localized awareness spreading can influence the effectiveness of interventions. Our findings underscore the importance of population structure when devising awareness-based intervention strategies for smoking cessation.

In modern correlation imaging systems, also known as ghost imaging (GI), particularly under low-light or noisy conditions, preserving high image fidelity presents a significant challenge. This paper introduces an innovative approach by integrating Bose-Chaudhuri-Hocquenghem (BCH) error control coding (ECC) into CGI systems to assist imaging. By encoding target image with BCH codes and using order-statistic decoding (OSD) for error correction during reconstruction, this approach significantly improves image quality across various signal-to-noise ratio (SNR) conditions. Simulation and experiment results validate that BCH coding assisted imaging achieves significantly enhanced robustness against additive white Gaussian noise (AWGN) and improved image reconstruction quality. In addition, the imaging performance of different BCH codes varies, with each code exhibiting distinct advantages based on factors such as code length and coding efficiency.

Imaging the full-field microvibration of extended targets remains a formidable challenge for conventional remote sensing. Traditional array-based sensors are often severely constrained by data throughput and sensitivity limits when scaling to high spatial resolutions, while point-scanning interferometric systems lack the instantaneous full-field capability required to capture transient, spatially coupled vibration modes. To overcome these limitations, we propose a Coherent Detection-Ghost Imaging (CD-GI) framework that synergizes the spatial multiplexing capability of single-pixel imaging with the high-dimensional sensitivity of coherent detection. We establish a comprehensive mathematical model that describes the coupling mechanism of the target's spatial distribution and temporal micro-dynamics within a 1D bucket detector signal. To resolve the resulting inverse problem, we develop a frequency-channel self-calibration scheme. This approach effectively decouples the micro-Doppler signatures from spatial speckle patterns without requiring prior knowledge of the vibration frequency. Experimental results demonstrate that our system successfully reconstructs the spatially resolved microvibration patterns of adjacent targets with a frequency difference as small as 1 Hz, achieving sub-wavelength vibration sensitivity. This work bridges the gap between computational imaging and coherent metrology, offering a robust solution for non-invasive, high-precision structural health monitoring.

Structured x-rays carrying an orbital angular momentum break spatial inversion symmetry and have been proposed as a means to probe chirality. We theoretically investigate twisted non-resonant x-ray diffraction from chiral molecules and demonstrate that no dichroic signal can arise from randomly oriented molecules, irrespective of the beam spatial profile. However, a dichroic response is found to emerge if the molecule is oriented. Our results establish the beam and sample conditions for which a measurable dichroic scattering signal survives axial and focal averaging.

In this work, we develop a random batch sum-of-Gaussians (RBSOG) method for molecular dynamics simulations of charged systems in the isothermal-isobaric (NPT) ensemble. We introduce an SOG splitting of the pressure-related $1/r^3$ kernel, yielding a smooth short-/long-range decomposition for instantaneous pressure evaluation. The long-range part is treated in Fourier space by random-batch importance sampling. Because the radial and non-radial pressure components favor different proposals, direct sampling either increases structure-factor evaluations and communication or leads to substantial variance inflation. To address this tradeoff, we introduce a measure-recalibration strategy that reuses Fourier modes drawn from the radial proposal and corrects them for the non-radial target, producing an unbiased pressure estimator with significantly reduced variance and negligible extra cost. The resulting method mitigates pressure artifacts caused by cutoff discontinuities in traditional Ewald-based treatments while preserving near-optimal $O(N)$ complexity. We provide theoretical evidence on pressure decomposition error, consistency of stochastic approximation, and convergence of RBSOG-based MD. Numerical experiments on bulk water, LiTFSI ionic liquids, and DPPC membranes show that RBSOG accurately reproduces key structural and dynamical observables with small batch sizes ($P\sim 100$). In large-scale benchmarks up to $10^7$ atoms on $2048$ CPU cores, RBSOG achieves about an order-of-magnitude speedup over particle-particle particle-mesh in electrostatic calculations for NPT simulations, together with a consistent $4\times$ variance reduction relative to random batch Ewald and excellent weak/strong scalability. Overall, RBSOG provides a practical and scalable route to reduce time-to-solution and communication cost in large-scale NPT simulations.

Hyper-reduction methods have gained increasing attention for their potential to accelerate reduced order models for nonlinear systems, yet their comparative accuracy and computational efficiency are not well understood. Motivated by this gap, we evaluate a range of hyper-reduction techniques for nonlinear finite element models across benchmark problems of varying complexity, assessing the inevitable tradeoff between accuracy and speedup. More specifically, we consider interpolation methods based on the gappy proper orthogonal decomposition as well as the empirical quadrature procedure (EQP), and apply them to the hyper-reduction of problems in nonlinear diffusion, nonlinear elasticity and Lagrangian hydrodynamics. Our numerical results are generated using the open source libROM, Laghos and MFEM numerical libraries. Our findings reveal that the comparative performance between hyper-reduction methods depends on both the problem and the choice of time integration method. The EQP method generally achieves lower relative errors than interpolation methods and is more efficient in terms of quadrature point usage, resulting in a lower wall time for the nonlinear diffusion and elasticity problems. However, its online computational cost is observed to be relatively high for Lagrangian hydrodynamics problems. Conversely, interpolation methods exhibit greater variability, especially with respect to the use of different time integration methods in the Lagrangian hydrodynamics problems. The presented results underscore the need for problem specific method selection to balance accuracy and efficiency, while also offering useful guidance for future comparisons and refinements of hyper-reduction techniques.

A recent study [Science 2025, eaea6425] proposes that magnetoreception in pigeons may arise from electromagnetic induction within the semicircular canals of the inner ear. In this framework, motion through the geomagnetic field is suggested to generate an induced electromotive force that leads to ion redistribution in the endolymph, activation of voltage-gated calcium channels, and subsequent engagement of downstream neural circuits. In this work, we examine the physical plausibility of this mechanism using a toy model of the induction process combined with an information-theoretic analysis. We find that, under idealised assumptions, Faraday induction in the semicircular canals would not generate a signal of sufficient informational content to support the extraction of directional magnetic field information from the geomagnetic field. However, the model supports the possibility of inferences due to radio-frequency (RF) electromagnetic waves of a miniscule amplitude, thereby providing a potential rationalisation of their disruptive effect on avian compass navigation. We stress that our analysis does not call into question the experimental evidence for magnetically responsive pathways within the vestibular system of pigeons. Rather, it constrains the class of viable physical mechanisms, indicating that a functionally competent magnetosensory system likely relies on different sensing principles or, if induction-based, on a different sensing architecture, while highlighting induction as a potential interference pathway of RF electromagnetic fields.

Ultrahigh continuous-wave intensities (>300 GW/cm$^2$) in high-NA optical cavities enable applications from phase-contrast electron microscopy to ultradeep dipole traps for molecules. However, the intensity can be limited by the parametric oscillatory instability (PI), where mirror vibrations scatter light from one cavity mode into another. We observe PI in a table-top Fabry-Pérot cavity, show that the mechanical modes are MHz-frequency bulk acoustic modes inside the mirrors, and measure their $Q$ factor. By using low-$Q$ mirrors, we achieve >500 GW/cm$^2$ intensities in an open, free-space cavity.

Departure from idealised plane waves gives rise to intricate geometric structures in wave fields. One such structure is the polarisation singularity, which emerges when multiple monochromatic waves interfere (such as would be the case for stochastic backgrounds), producing loci of purely circular or linear polarisation. In this work, we extend the theory of polarisation singularities to gravitational waves and higher spin fields. Building on the electromagnetic description, we formulate the gravitational analogue of polarisation singularities and show that they are generic features of gravitational waves. Their dimension, however, depends on the spin of the field. We illustrate these results with simulations of plane-wave interference and analyse the resulting singularity densities.

Recent experiments on temporal reflection in transmission line metamaterials and theoretical treatments of dispersive time-varying media have unearthed the fundamental role of modulation mechanisms on the interface conditions, underpinning the introduction of passive photonic time crystals with stable momentum band gaps. Drawing from these concepts, it is shown that temporal metamaterials with simultaneous passive permittivity and permeability switching exhibit wideband absorption with impedance-matching, effectively behaving as one-dimensional perfectly matched layers. Under the effective medium theory, the loss mechanism is attributed to the emergent effective electric and magnetic conductivities, which are used to derive an approximate matching condition for asynchronous modulation and to engineer lossy material properties. The proposed approach and its performance beyond the Rozanov bound are verified with semi-analytical calculations as well as full-wave simulations, and the possibility of realizing a two-dimensional temporal perfectly matched layer is discussed.

We present mrfmsim, an open-source package that facilitates the design, simulation, and signal validation of magnetic resonance force microscopy experiments. The mrfmsim package uses directed acyclic graphs (DAGs) to model experiments and employs a plugin system that enables adding custom experiments and functionalities. Unlike common DAG-powered workflow packages, mrfmsim allows flexible customization of experiments post-definition, such as optimized looping, without requiring rewriting the internal model. In this paper, we highlight the challenges of building simulation packages for experiments that undergo continuous development in a graduate research setting. We demonstrate how a one-off approach to experimental simulation yielded erroneous results, and how the modularity, extendibility, and readability of the new platform enabled correct results and a significantly accelerated development cycle.

Through temporal shaping of the excitation signal, the complex-frequency scattering zeros of a lossless structure can be accessed, enabling a storage-release mechanism referred to as coherent virtual absorption. Practical demonstrations of this mechanism, however, have been limited to simple configurations such as slabs and spheres, where analytical solutions allow accurate prediction of the complex-frequency scattering zeros. Here, we extend this concept into the realm of metasurfaces and demonstrate coherent virtual absorption in realistic and dispersive metasurface configurations. Through a combination of full-wave analysis and rational approximation, we present a practical scheme to identify suitable complex-frequency zeros and achieve coherent virtual absorption successfully. Our approach can be implemented in arbitrary metasurface configurations with any number of ports, providing a robust framework for optimized energy storage, memories, optical sensing, and modulation in practical photonic systems.

Atomistic simulations are widely used to investigate reactive processes but are often limited by the rare event problem due to kinetic bottlenecks. We recently introduced an enhanced sampling approach based on the committor function, machine-learned following a variational principle. This method combines a transition-state-oriented bias potential, expressed as a functional of the committor, with a metadynamics-like bias along a committor-based collective variable, enabling uniform exploration of reaction pathways. In its original formulation, the committor is represented by a neural network that takes physical descriptors as input and is trained by minimizing a functional involving gradients with respect to atomic coordinates, which can be computationally demanding in some cases. Here, we propose a simplified learning criterion formulated entirely in the descriptor space, which bypasses the need for explicit and costly coordinate gradients and provides a relaxed upper bound to the original variational principle. Although this approach does not formally target the exact committor, we show that it retains robust sampling performance while significantly reducing computational costs, thus enabling the study of processes that would be practically unfeasible using the original formulation.

This paper presents a detailed description and characterization of a new open-source vehicle trajectory dataset, namely SWIFTraj, constructed from videos recorded by a swarm of 16 drones equipped with 5.4K-resolution cameras. The dataset is distinguished from existing open-source trajectory datasets in several aspects. First, it provides long-distance continuous trajectories of up to 4.5 km on a freeway, enabling in-depth investigation of traffic phenomena and their spatial and temporal evolution. Second, the data collection site covers an integrated network consisting of a long freeway corridor and parts of its connected urban network, facilitating traffic analysis and modeling from a network perspective. The potential applications of the dataset for transportation research, including traffic flow analysis, modeling, and control at multiple scales, as well as topics related to autonomous driving, are thoroughly discussed. Finally, SWIFTraj is released as a freely accessible open-source dataset to support and accelerate future research in the transportation community. The dataset is publicly available at the SWIFTraj website (https://www.swiftraj.com).

Tuberculosis (TB) is an airborne disease caused by the bacterium Mycobacterium tuberculosis (M. tb). Prior to the COVID-19 pandemic, TB was the leading cause of death from an infectious agent globally. However, most people exposed to M. tb do not develop active TB and go on to display symptoms. Instead, in the majority of cases, the bacteria are contained within a granuloma (an aggregation of immune cells) without being eliminated; this is called latent TB. The spatial organisation of the bacteria and immune cells is important in determining whether an individual exposed to M. tb will develop latent or active TB. In this paper, we present a multi-cell, multiscale model of TB progression to investigate the importance of the spatial organisation. This is a novel TB within-host dynamics modelling framework, having been developed using CompuCell3D (CC3D), an open-source computer software used for simulating cellular biological processes both within and between cells. We used this model to compare the generated results with those from a previously developed within-host infectious disease model. We found that, although the results of our CC3D model mostly agree qualitatively with those from the previously developed model, there are quantitative differences. Additionally, we conducted a robustness analysis of key model parameters from the CC3D model to determine their importance to the CC3D model output, using a methodology specifically designed for agent-based models. The model output appears to be robust in response to perturbations in parameters controlling chemotactic movement, but less so in response to perturbations in parameters controlling persistence of movement in cells, cell adhesion and volume constraints. This work compares our CC3D model of TB progression with another agent-based modelling approach to the same problem.

In this paper we propose a Moran model that describes the population dynamics of two types: While the first type has a selective advantage during reproduction, the second type can avoid replacement during reproduction with some positive probability by switching temporarily into a dormant state. We investigate the interplay of both evolutionary strategies by studying the invasion dynamics of the dormant type into the resident (selectively advantageous) population in the large population limit of the system. It turns out that the dormancy trait can not only invade and subsequently fixate under suitable parameter assumptions despite its selective disadvantage (a phenomenon that has already been observed in a related context in Blath and Tóbiás (2020)), but that there is also a novel regime of stable coexistence of both types due to a frequency-dependent balancing effect that did not arise in the previous setup with Lotka--Volterra type symmetric competition. The emergence of a coexistence regime here rests in part on specific properties of the Moran modelling framework, in particular its fixed overall population size that enforces instant re-colonization after death events, as well as on the (positive) mortality and resuscitation rates of the dormant state. We provide heuristic explanations for the observed types of behaviour and the corresponding proofs, which involve comparisons to suitable branching processes, approximations by dynamical systems, and an analysis of asymptotic behaviour of the latter.

We study a multitype SIR epidemic model where individuals are categorized into different types, and where infection spread is characterized by a next-generation matrix $M=\{m_{ij}\}$ with community fractions $\{π_j\}$ for the different types of individuals. We analyse two key quantities: the basic reproduction number $R_0$ and the final epidemic outcome of the different types $\{τ_i\}$. We consider the situation where $M$ is only partly known, through the row sums $\{r_i\}$ or the column sums $\{c_j\}$, and treat both a general $M$ and the special but common situation where $M$ is proportional to a contact matrix satisfying detailed balance. For a general $M$, which is partially observed through $\{r_i\}$ or $\{c_j\}$, we obtain sharp upper and lower bounds of $R_0$ and $\{τ_i\}$, but for the case where $M$ satisfies detailed balance the problem is harder: our obtained bounds for $R_0$ are narrower than the general case but still not sharp, and bounds for the final size are only obtained when there are two types of individual.

Sex chromosomes have evolved repeatedly across the Tree of Life, yet their evolutionary fates differ strikingly among lineages. In mammals and birds, highly degenerated Y/W chromosomes have remained stable for >100 Myr, whereas in most amphibians, many teleosts, non-avian reptiles and flowering plants, sex chromosomes remain largely homomorphic and undergo frequent turnover. Classical explanations, including the evolutionary trap hypothesis, sexually antagonistic selection, mutation load accumulation, genetic drift and selfish genetic elements, do not fully explain observed patterns. Here we develop a complementary developmental perspective and propose the sex determination developmental gene regulatory network (GRN) lock-in hypothesis. In mammals and birds, sex is determined by an early, initiation by somatic cells, fully penetrant genetic master signal acting within a narrow, thermally buffered embryonic window. This signal operates within highly canalised GRNs, coupled to chromosome-scale dosage compensation, with alternative splicing events playing no causal role in primary sex determination. This configuration makes it difficult for new sex-determining loci to invade without creating deleterious intermediate states. By contrast, many ectothermic vertebrates possess more flexible, integrative threshold GRNs in which genetic, epigenetic, germ-cell and environmental inputs interact over a prolonged sensitive period, with absent or largely gene-by-gene based dosage compensation and environmentally responsive splicing near key regulatory nodes, providing many entry points for new master sex-determining genes to evolve. We outline empirical predictions of this framework and highlight how integrating developmental biology, molecular mechanisms and population genetics can yield testable models for when sex chromosomes become evolutionarily locked-in versus when they turnover repeatedly.

In this paper, we develop a mean-field game model for SEIR epidemics on heterogeneous contact networks, where individuals choose state-dependent contact effort to balance infection losses against the social and economic costs of isolation. The Nash equilibrium is characterized by a coupled Hamilton--Jacobi--Bellman/Kolmogorov system across degree classes. An important feature of the SEIR setting is the exposed compartment: the incubation period separates infection from infectiousness and changes incentives after infection occurs. In the baseline formulation, exposed agents optimally maintain full contact, while susceptible agents reduce contact according to an explicit best-response rule driven by infection pressure and the value gap. We also discuss extensions that yield nontrivial exposed precaution by introducing responsibility or compliance incentives. We establish existence of equilibrium via a fixed-point argument and prove the uniqueness under a suitable monotonicity condition. The analysis identifies a delay in the onset of precaution under longer incubation, which can lead to weaker behavioral responses and larger outbreaks. Numerical experiments illustrate how network degree and the cost exponent shape equilibrium policies and epidemic outcomes.

Generalizable protein function prediction is increasingly constrained by the growing mismatch between exponentially expanding sequences of environmental proteins and the comparatively slow accumulation of experimentally verified functional data. Active learning offers a promising path forward for accelerating biological function prediction, by selecting the most informative proteins to experimentally annotate for data-efficient training, yet its potential remains largely unexplored. We introduce HATTER (Human-in-the-loop Adaptive Toolkit for Transferable Enzyme Representations), a modular framework that integrates multiple active learning strategies with human-in-the-loop experimental annotation to efficiently fine tune function prediction models. We compare active learning training to standard supervised training for biological enzyme function prediction, demonstrating that active learning achieves performance comparable to standard training across diverse protein sequence evaluation datasets while requiring fewer model updates, processing less data, and substantially reducing computational cost. Interestingly, point-based uncertainty sampling methods like entropy or margin sampling perform as well or better than more complex acquisition functions such as bayesian sampling or BALD, highlighting the relative importance of sequence diversity in training datasets and model architecture design. These results demonstrate that human-in-the-loop active learning can efficiently accelerate enzyme discovery, providing a flexible platform for adaptive, scalable, and expert-guided protein function prediction.

The ribosome is responsible for protein synthesis in all cells, and is the largest energy consumer in the cell. We propose that the ribosome originated as a mutualistic symbiont of an RNA-dependent RNA polymerase ribozyme, supplying peptides that enhanced replication. As life transitioned from the RNA to the RNA-protein world, autonomous replicators became irreversibly addicted to the ribosome for producing replication proteins. Subsequent evolution is construed as a ribosomal takeover, whereby the ribosome evolved to consume most of the resources of the cell, while other cellular componentry ensured the propagation of the ribosome. Under this perspective, the ribosome is the ultimate biological selfish element.

We propose a low-complexity phase recovery scheme that simultaneously mitigates laser phase noise and fiber nonlinearity across several subcarriers. In a long single-span link with Raman amplification, the scheme achieves 0.9 dB gain with 99 real multiplications per complex symbol.

Beyond diagonal reconfigurable intelligent surface (BD-RIS)s enhance wave manipulation through inter-element couplings but pose significant channel estimation challenges due to cascaded channels and block-Kronecker structures. This paper proposes a compressive sensing framework exploiting sparse Tucker decomposition of the measurement tensor and the Kronecker rank-one structure of channel components. Two algorithms are developed: Sparse Tensor Orthogonal Recovery Method (STORM), which uses orthogonal matching pursuit (OMP) for greedy support recovery, and Sparse Tensor subspace- Aided Recovery (STAR), which leverages subspace-based projection for enhanced noise robustness. Both perform joint sparse support identification, followed by a Kronecker rank-one factorization via singular value decomposition (SVD) to recover the channel parameters. Simulations show that STAR achieves oracle-assisted least squares (LS) performance at moderate-to-high signal-to-noise ratio (SNR) with significantly fewer measurements than baseline methods, enabling practical BD-RIS deployment in next-generation millimeter wave (mmWave)/sub-terahertz (sub-THz) networks.

Cell-free massive-multiple-input-multiple-output (CFmMIMO) is a key enabler for sixth-generation (6G) wireless communication networks, where distributed access points (APs) jointly serve user equipments (UEs). In commonly adopted channel models for CFmMIMO networks, inter-AP channel correlation is assumed to be absent, thereby eliminating the potential benefits of centralized processing. However, by carefully designing the pilot transmission phase, the AP received signals during pilot transmission can become correlated, and thus, centralization can improve channel estimation performance, despite the absence of inter-AP channel correlation. In this paper, we propose a channel estimation scheme, termed master-assisted channel estimation (MACE), that aims to leverage inter-AP signal correlation by means of partially centralized processing and hence improve channel estimation performance. In MACE, a subset of APs fuse and forward their received pilot signals to a master AP, which then performs channel estimation using the fused signals together with its locally received signals. This scheme strikes a balance between local and fully centralized processing by leveraging inter-AP signal correlation, while reducing fronthaul signaling and computational complexity. Numerical experiments demonstrate that MACE consistently outperforms local channel estimation, where inter-AP signal correlation is neglected.

Cooperative sensing with uncrewed aerial vehicles (UAVs) is a key enabler for low-altitude wireless networks (LAWNs), where sensing accuracy critically depends on the spatial configuration of the UAV formation. In this paper, we study formation design and control for Cramer-Rao lower bound (CRLB)-optimal cooperative target sensing. We first establish a sensing performance model based on range measurements and derive the Fisher information matrix (FIM) of the target location. By adopting the A-optimality criterion, we analytically characterize the formation geometry that minimizes the CRLB of the estimation error. The optimal formation is shown to exhibit isotropic Fisher information in the horizontal plane, leading to a regular polygon geometry with an elevation angle determined by the tradeoff between path loss and geometric diversity. Building on this result, we further develop a distributed formation control strategy that steers UAVs from arbitrary initial deployments toward the sensing-optimal configuration while maintaining formation motion and obstacle avoidance. Numerical results demonstrate that the proposed scheme consistently outperforms benchmark formations in terms of CRLB and achieves reliable convergence under practical constraints.

Learning-based signal processing systems increasingly support high-stakes medical decisions using heterogeneous biomedical signals, including medical images, physiological time series, and clinical records. Despite strong predictive performance, many models rely on statistical correlations that are unstable across acquisition settings, patient populations, and institutional practices, limiting robustness, interpretability, and clinical trust. We advocate a causal signal processing perspective in which biomedical signals are treated as effects of latent generative mechanisms rather than as isolated predictive inputs. Using clinical risk prediction as a motivating example, we show how disease-related factors generate observable biomarkers, while acquisition processes act as confounders influencing signal appearance. In clinical disease risk prediction from chest CT scans and patient risk factors, correlational models may fail under scanner changes, whereas causal abstractions remain invariant. Building on this view, we propose a unifying conceptual framework integrating causal modeling with learning-based signal processing and neuro-symbolic reasoning. Statistical models extract multimodal representations that are mapped to interpretable causal abstractions and combined with symbolic knowledge encoding clinical risk factors and guidelines. This structure enables clinically grounded explanations, counterfactual reasoning about hypothetical interventions, and improved robustness to distribution shifts arising from changes in acquisition conditions or screening policies. Rather than introducing a specific algorithm, this article presents schematic causal structures and a comparative analysis of correlation-based, causal, and neuro-symbolic approaches to guide the design of robust and interpretable medical decision-support systems.

Hypercomplex signal processing (HSP) offers powerful tools for analyzing and processing multidimensional signals by explicitly exploiting inter-dimensional correlations through Clifford algebra. In recent years, hypercomplex formulations of the phase retrieval (PR) problem, wheren a complex-valued signal is recovered from intensity-only measurements, have attracted growing interest. Hypercomplex phase retrieval (HPR) naturally arises in a range of optical imaging and computational sensing applications, where signals are often modeled using quaternion- or octonion-valued representations. Similar to classical PR, HPR problems may involve measurements obtained via complex, hypercomplex, Fourier, or other structured sensing operators. These formulations open new avenues for the development of advanced HSP-based algorithms and theoretical frameworks. This chapter surveys emerging methodologies and applications of HPR, with particular emphasis on optical imaging systems.

Short-range reliable and secure communication is a major priority in the tactical, military and disaster response settings where the traditional communication infrastructure is either off-line or prone to interception. Current VHF/UHF radios and software-defined radios are popular but large-sized devices and require lots of power, making them not suitable to be used as lightweight wearable devices with seamless hand-free use. In this paper, the design and theoretical framework of a miniature, LoRa based encrypted intercommunication device that can be used in secure field communication over a range of 1-1.5km and under line-of-sight conditions is provided. The suggested system consists of a voice-activated acquisition block, digital audio compression, an embedded microcontroller processor, and AES-128 encryption followed by a low-power transmission via the LoRa protocol. Through the ability of chirp spread spectrum modulation to utilize the long-range and low-energy properties, the system is guaranteed reliable communications coupled with low power consumption and low electromagnetic footprint. The theoretical analysis of the proposed communication range is justified using a link-budget that justifies the practicability of the communication range in the real propagation conditions. This architecture focuses on infrastructural agnosticism, peer-to-peer security as well as wearable ergonomics. The given scheme shows the possibilities of LoRa technology in the scope of other traditional IoT telemetry, and it can be further extended to include secure tactical voice communication platforms.

This paper presents an optimization-based behavioral model for mixers driven by multi-tone local oscillator (LO) signals, considered specifically for frequency comb orthogonal frequency-division multiplexing radar applications. Unlike traditional models, the proposed approach is designed and tested for multi-tone LO excitations. The model uses polynomial nonlinearities for both intermediate frequency and LO ports, supported by spectrum-domain fitting that selectively emphasizes strong intermodulation products. In addition, a polynomial block is introduced to capture input power-dependent phase nonlinearity. The approach is validated using circuit-level simulations and supported by measurements. Radar processing results show the model replicates distortive effects in simulations. The proposed model enables rapid system-level performance estimations and waveform optimization, replacing computationally expensive circuit-level simulations.

This paper proposes a novel low probability of intercept (LPI) waveform design approach for orthogonal frequency-division multiplexing (OFDM)-based integrated sensing and communication systems by introducing artificial phase and Doppler shifts. These controlled impairments, unknown to eavesdroppers, effectively disrupt passive radar processing and intercept attempts. At legitimate receivers, they can be fully compensated, so that standard OFDM communication and sensing performance are preserved. To support the effectiveness of the proposed LPI waveform design for OFDM-based ISAC, measurement results with 1 GHz bandwidth at 27 GHz are presented considering different impairment introduction approaches, all with no impact on cooperative system performance, and compensation capabilities at the eavesdropper.

Pixel antenna is a promising antenna technology that enables flexible adjustment of radiation characteristics and enhancement of wireless systems through antenna coding. This work proposes a novel deep learning-based antenna coding optimization algorithm. Specifically, the proposed algorithm is supported by a heterogeneous multi-head selection mechanism, whose main idea is to train multiple neural networks based on various coding schemes and select the one that leads to the best system performance. Unlike traditional heuristic searching-based algorithms that require high computational complexity to achieve satisfactory performance, the proposed data-driven deep learning approach can achieve 98\% of the performance achieved by the searching-based algorithms with significantly reduced computational complexity. Results demonstrate that in pixel antenna empowered single-input single-output systems, the proposed algorithm achieves a computational speed 81 times faster than the searching-based algorithm. For more complex pixel antenna empowered multiple-input multiple-output systems, the computational speed is 297 times faster than the existing searching-based algorithm. Benefiting from the high performance and low computational complexity, this algorithm demonstrates the significant potential of pixel antennas as a novel and practical technology to enhance wireless systems.

Convolution serves as a powerful operation for the regularization of functions. While polynomials inherently possess smoothness, it is particularly interesting to investigate their behavior under convolution. This interest stems from the fact that numerous engineering and physical phenomena can be modeled through such operations, including weighted averages, blurring effects and convolutional integral equations. In this work, we show that under certain mild conditions, the convolution with any even Schwartz function acts as an automorphism on the vector space of finite-order polynomials. We derive explicit equations for the inverse operation of this convolution, which are numerically simple to implement. In addition, we extend the deconvolution with (not necessarily even) Schwartz functions to a broader class of functions, including $L^1(\mathbb{R})$, $L^2(\mathbb{R})$, the Schwartz space and tempered distributions. Specifically, we establish a explicit rigorous formula for the deconvolution of a function or distribution that has been convolved with a Schwartz function, being a particular example the Weierstrass Transform. For the latter, we show that any Schwartz function and tempered distribution that has been transformed, can be recovered, in their respective topologies, by the limit of a sequence of linear combination of recursive convolutions. This provides a new formula for the inverse of the Weierstrass Transform that can be numerically implemented.

Flexible antenna arrays (FAAs) can physically reshape their geometry to add new spatial degrees of freedom, whereas transmit beamforming adjusts the complex element weights to electronically steer and shape the array's radiation pattern, thereby significantly improving communication performance. This paper is the first to explore the integration of FAA geometry control and beamforming for physical layer security enhancement, where a base station equipped with an FAA communicates with a legitimate user in the presence of passive eavesdroppers. To safeguard confidential transmissions, we formulate a new secrecy rate maximization problem that jointly optimizes the transmit beamforming vector and a continuous FAA shape control parameter. Due to the non convex nature of the problem, an alternating optimization algorithm is developed to decompose the joint design into tractable subproblems, which are solved iteratively to refine both the FAA geometry and beamforming strategy. Simulation results confirm that the proposed joint optimization framework significantly outperforms conventional fixed shape or beamforming only schemes, demonstrating the potential of FAA enabled reconfigurability for secure wireless communications.

Surface electromyography (sEMG) signals exhibit substantial inter-subject variability and are highly susceptible to noise, posing challenges for robust and interpretable decoding. To address these limitations, we propose a discrete representation of sEMG signals based on a physiology-informed tokenization framework. The method employs a sliding window aligned with the minimal muscle contraction cycle to isolate individual muscle activation events. From each window, ten time-frequency features, including root mean square (RMS) and median frequency (MDF), are extracted, and K-means clustering is applied to group segments into representative muscle-state tokens. We also introduce a large-scale benchmark dataset, ActionEMG-43, comprising 43 diverse actions and sEMG recordings from 16 major muscle groups across the body. Based on this dataset, we conduct extensive evaluations to assess the inter-subject consistency, representation capacity, and interpretability of the proposed sEMG tokens. Our results show that the token representation exhibits high inter-subject consistency (Cohen's Kappa = 0.82+-0.09), indicating that the learned tokens capture consistent and subject-independent muscle activation patterns. In action recognition tasks, models using sEMG tokens achieve Top-1 accuracies of 75.5% with ViT and 67.9% with SVM, outperforming raw-signal baselines (72.8% and 64.4%, respectively), despite a 96% reduction in input dimensionality. In movement quality assessment, the tokens intuitively reveal patterns of muscle underactivation and compensatory activation, offering interpretable insights into neuromuscular control. Together, these findings highlight the effectiveness of tokenized sEMG representations as a compact, generalizable, and physiologically meaningful feature space for applications in rehabilitation, human-machine interaction, and motor function analysis.

This paper investigates channel-aware decision fusion empowered by massive MIMO systems and reconfigurable intelligent surfaces (RIS). By integrating both, we aim to improve goal-oriented (fusion) performance despite the unique propagation challenges introduced. Specifically, we investigate traditional favorable propagation properties in the context of RIS-aided Massive MIMO decision fusion. The above analysis is then leveraged (i) to design three sub-optimal simple fusion rules suited for the large-array regime and (ii) to devise an optimization criterion for RIS reflection coefficients based on long-term channel statistics. Simulation results confirm the appeal of the presented design.

Multi-band sensing has emerged as a key enabler of integrated sensing and communication (ISAC), one of the six primary usage scenarios defined for IMT-2030 (6G). The introduction of frequency range 3 (FR3, 7-24 GHz), comprising non-contiguous sub-bands across a wide frequency span, further reinforces the importance of multi-band operation. In such scenarios, frequency-dependent propagation effects that are collectively referred to as dense multipath components (DMC), including clutter, diffraction, and diffuse scattering, must be carefully considered. Building on prior literature and our experimental observations, this paper proposes a novel ISAC channel analysis tailored to multi-band sensing, based on a channel model with background DMCs. It also assesses the sensing trade-offs among sub-bands by analyzing Cramér-Rao bound (CRB)-based fundamental limits. Furthermore, a scalable multi-band estimator is proposed that resolves angular ambiguities arising from the grating lobes effect. Simulation results of the multi-band estimator demonstrate substantial gains in estimation accuracy and reductions in false alarm rate over single-band estimators operating on each constituent sub-band within the CRB-achieving regime. In a representative test case, the proposed estimator achieves reductions of 37.41% and 17.04% in the root mean squared error of delay estimation compared to single-band estimators operating at 8.75 GHz and 21.7 GHz, respectively.

Quantum-enhanced LiDAR, exploiting squeezed states of light, promises significant sensitivity gains over classical protocols. However, in realistic scenarios characterized by high optical losses and atmospheric turbulence, standard figures of merit, such as quantum fidelity or the quantum Chernoff bound, saturate rapidly, failing to provide a usable gradient for system optimization. In this work, we propose the Quantum Wasserstein Distance of order 2 ($W_2$) as a robust geometric metric for lossy quantum sensing. Unlike overlap-based measures, $W_2$ quantifies the transport cost in phase space and maintains a linear response to channel transmissivity, even in regimes where the quantum state is virtually indistinguishable from thermal noise. We derive an analytical threshold for the quantum advantage, demonstrating that squeezing is only beneficial when the transmissivity exceeds a critical value determined by the environmental noise-to-signal ratio. Furthermore, using Monte-Carlo simulations of a fading channel, we show that $W_2$ acts as a high-fidelity estimator of instantaneous link quality, exhibiting a wide dynamic range immune to the numerical instabilities of fidelity-based metrics. This geometric framework bridges the gap between quantum optimal transport and practical receiver design, paving the way for adaptive sensing in scattering media.

In this work, we study the dynamics of a spinning particle in pp-waves spacetimes; in particular, plane gravitational waves and impulsive shockwaves. W pay special attention to analytical considerations; this is possible due to an appropriate choice of the spin supplementary condition, various Hamiltonian formalisms (including a non-minimal one) and constants of motion associated with conformal fields. Based on these results, we establish a relation between the motions of a spinning particle in pp-waves and electromagnetic fields suggested by a gauge-gravity duality.

According to general relativity, clocks are the basic measuring devices needed to probe spacetime geometry. However, it is generally accepted that the mass of clocks capable of measuring small time intervals must be bounded from below. In this article, we consider two gravitationally induced phenomena: first, the extent to which such a mass disturbs the geometry that the clocks intended to probe; second, the magnitude of the gravitational self-interaction. We adopt the semiclassical coupling between gravity and quantum matter in the non-relativistic regime to obtain upper bounds on the mass of the clocks for a given time resolution and running time.

The Kochen-Specker no-go theorem established that hidden-variable theories in quantum mechanics necessarily admit contextuality. This theorem is formally stated in terms of the partial Boolean algebra structure of projectors on a Hilbert space. Each partial Boolean algebra provides a semantics for interpreting propositional logic. In this paper, we examine the complexity of propositional satisfiablity for various classes of partial Boolean algebras. We first show that the satisfiability problem for the class of non-trivial partial Boolean algebras is NP-complete. Next, we consider the satisfiability problem for the class of partial Boolean algebras arising from projectors on finite dimensional Hilbert spaces. For real Hilbert spaces of dimension greater 2 and any complex Hilbert spaces of dimension greater than 3, we demonstrate that the satisfiablity problem is complete for the existential theory of the reals. Interestingly, the proofs of these results make use of Kochen-Specker sets as gadgets. As a corollary, we conclude that deciding quantum homomorphism in these fixed dimensions are also complete for the existential theory of the reals. Finally, we show that the satisfiability problems for the class of all Hilbert spaces and all finite-dimensional Hilbert spaces is undecidable.

Bosonic quantum error-correcting codes encode logical information in a harmonic oscillator, with the Gottesman-Kitaev-Preskill (GKP) and number-phase (NP) codes representing two fundamentally different encoding paradigms. Although both have been extensively studied, it remains unclear under what physical noise conditions (including photon loss and dephasing) one encoding intrinsically outperforms the other. Here we estimate a quantitative performance boundary between GKP and NP codes under general photon loss-dephasing noise. By optimizing code parameters within each encoding family, we identify the noise regimes in which each code exhibits a fundamental advantage. In particular, we find that the crossover occurs when the dephasing strength is approximately two orders of magnitude smaller than the loss strength, revealing a sharp separation between operational regimes. Beyond this specific comparison, our work establishes a practical and extensible methodology for benchmarking bosonic codes and optimizing their parameters, providing concrete guidance for the experimental selection and deployment of bosonic encodings in realistic noise environments.

Superconducting quantum processors operate at microwave frequencies in millikelvin environments, making it challenging to interconnect distant nodes using conventional microwave wiring. Coherent microwave-to-optical (M2O) transduction enables superconducting quantum networks by interfacing itinerant microwave photons with low-loss optical fiber. However, many state-of-the-art transducers provide efficient conversion only over a narrow frequency span, complicating deployment with heterogeneous superconducting devices that are detuned by gigahertz-scale offsets. Here we demonstrate a frequency-agile microwave-optical interface that overcomes this bandwidth mismatch by cascading an electro-optic M2O transducer with a multimode microwave-to-microwave (M2M) frequency converter, with in situ tunability of the microwave resonances in both stages. Using this architecture, we realize continuous frequency coverage from 5.0 to 8.5 GHz within a single system. As an application relevant to superconducting-qubit networking, we use the cascaded M2M-M2O interface to optically read out a superconducting qubit whose readout resonator is detuned by 1.7 GHz from the native M2O microwave resonance, demonstrating a scalable route toward fiber-linked superconducting quantum nodes.

The gravitational-wave (GW) detections reported by the LIGO-Virgo-KAGRA (LVK) collaboration have so far been consistent with quasi-circular compact binary coalescences (CBCs). Nevertheless, a small fraction of binaries driven to merge through dynamical interactions in dense stellar environments or in field triples may retain measurable orbital eccentricity when entering the sensitive frequency band of LVK detectors. Confident measurement of eccentricity in the LVK band would provide strong evidence for such dynamically driven mergers; however, eccentric gravitational-waveform models are computationally expensive, and performing production-level inference on all detected signals is not an efficient use of resources when eccentric signals are expected to be rare. An intermediate step between detection and analysis, in which the signal is assessed for the potential presence of eccentricity, could provide quick recommendations for which signals should undergo full eccentric inference. We apply the wavelet scattering transform (WST) to a large set of synthetic waveforms in realistic noise and assess its discriminatory power using simple linear and shallow neural-network classifiers. We find that the WST representation enables effective discrimination between eccentric and quasi-circular binaries and provides a compact multi-scale representation of GW signals. Our approach achieves ~64% percent detection accuracy at a false alarm rate of 10%, with an AUC of 0.844 and an average precision of 0.876. We also examine the ability of our classifiers to distinguish eccentricity from spin-induced precession and find robust performance across a range of spin-precession magnitudes.

We propose a heralded entanglement generation scheme based on Gaussian sources augmented with photon addition and subtraction operations. By combining single-mode squeezing, linear interferometers, and conditional photon-number measurements on ancillary modes, our model probabilistically generates dual-rail encoded Bell, GHZ, and W states. We systematically optimize the squeezing parameters and interferometer settings to maximize both the heralding success probability and the fidelity with the target states. Our results show that the inclusion of photon addition and subtraction significantly enhances the non-classicality of the output states, leading to improved generation performance, while maintaining computational efficiency comparable to single-photon source models. We further analyze the robustness of the scheme under parameter perturbations, demonstrating stable performance against realistic experimental imperfections. This work provides a versatile and experimentally feasible framework for scalable heralded entanglement generation using Gaussian resources with non-Gaussian operations.

Superconducting quantum interference devices (SQUIDs) are state-of-the-art in ultra-sensitive magnetometry; however, conventional SQUID devices are fundamentally limited by the inherently nonlinear and periodic nature of their transfer function. Although flux-locked loop (FLL) configurations can mitigate this issue, they introduce electronic complexity and bandwidth constraints that hinder scalability in quantum circuits. In this work, we present an experimental demonstration of the bi-SQUIPT, a flux transducer that modulates the density of states in a proximitized superconducting weak link. The device employs a dual-loop architecture with differential readout, which enables cancellation of non-linearities typical of individual elements, achieving a voltage swing of approximately 120 $μ$V. Measurements yield a spurious-free dynamic range (SFDR) of up to 60 dB, consistent with theoretical predictions and comparable to that of SQUID arrays, while maintaining power dissipation in the femtowatt range. The results further highlight a remarkable operational stability up to 600 mK, positioning the bi-SQUIPT as an enabling technology for high-density cryogenic quantum electronics.

In many practical applications, quantum algorithms require several qubits, significantly more than those available with current noisy intermediate-scale quantum processors. Distributed quantum computing (DQC) is considered a scalable approach to increasing the number of available qubits for computational tasks. In the DQC setting, a quantum compiler must find the best partitioning for the quantum algorithm and then perform smart non-local operations scheduling to optimize the consumption of Einstein-Podolsky-Rosen (EPR) pairs. In this work, the focus is on minimizing the use of EPR pairs when the circuit structure allows for multiple non-local gates to utilize a single TeleGate operation. This is achieved by using a greedy algorithm that explores the circuit and groups together the gates that could share an EPR pair while also changing the order of commutative gates when necessary. With this preliminary pass, the compiled circuits show reduced depth and EPR usage. Since the quality of each EPR pair quickly deteriorates, the number of non-local gates using the same EPR pair should also be bounded. This means that, depending on the features of the target quantum network, the user can achieve different levels of optimization. Here, it is shown that this approach brings benefits even while assuming a low EPR pair lifetime.

Recent high-cadence observations by Subaru-HSC have identified a population of ultrashort-timescale microlensing events, providing a compelling window for planet-mass primordial black holes (PBHs) to constitute the entirety of dark matter. In this Letter, we demonstrate that this PBH population and the nanohertz stochastic gravitational-wave (GW) background reported by pulsar timing arrays (PTAs) can be naturally unified by a single primordial origin: a broad, nearly-flat enhancement of the curvature power spectrum with an amplitude of $O(10^{-2})$. The resulting PBH mass function spans the planet-to-solar mass range, while remaining consistent with all current observational constraints. This unified PBH--induced-GW framework makes concrete multi-messenger predictions, which can be decisively scrutinized by forthcoming microlensing surveys, next-generation PTAs, space-borne interferometers, precision astrometry, and laser ranging experiments.

Quantum walks provide a versatile framework for probing the structural and dynamical properties of complex systems ranging from biological networks to synthetic materials. However, their realization on current noisy pre-fault-tolerant quantum computers is fundamentally limited by decoherence. Conventional dense encodings of graph structures require prohibitively deep circuits, making them incompatible with existing hardware. Here we introduce an algorithm that leverages symmetry-sector encoding and trades circuit depth for qubits, while integrating symmetry-respecting postselection as an effective noise-mitigation strategy. This combination enables us to execute practical quantum-walk circuits for biological networks on actual quantum hardware. We benchmark the proposed methodology against known state-of-the-art circuit architectures, highlighting significant reduction of circuit depth in our approach at the cost of moderate qubit overhead. Utilizing 40 qubits, we implement quantum walks on complex graphs containing up to 17 nodes and 20 edges -- the largest experiment on superconducting hardware to date, with the Hellinger fidelity exceeding 87% throughout 7 steps. We present a case study that illustrates how experimentally obtained quantum-walk dynamics on a protein-protein-interaction network can be applied to prioritizing disease-associated genes. We discuss the framework scalability in the pre-fault-tolerant era and its potential for studying larger biological networks.

We extend previous conduction-based analogies between ice growth in a lake and cosmological expansion by incorporating buoyancy-driven heat transport. Reformulating the Stefan problem with both conductive and convective fluxes yields an evolution equation for the ice thickness $s(t)$ that is structurally analogous to the Friedmann equations for the cosmological scale factor $a(t)$. Beyond reproducing radiation-, matter-, and curvature-like behaviors, we introduce a reduced description of convection in which the vertically integrated heat flux reaching the moving ice-water interface is modeled as a power-law function of the instantaneous liquid-layer thickness, generating two additional effective contributions. The first is a constant term, directly analogous to a cosmological constant, arising from the persistence of buoyancy-driven transport under geometric confinement. The second is a $s^{-1}$ contribution originating from the coupling between the moving ice boundary and the convective boundary layer. This term reflects the specific reduced flux-height Ansatz adopted, rather than a universal physical prediction. When expressed in Friedmann-like cosmological form, this term entails a fluid with negative energy density and equation-of-state parameter $w=-2/3$. In cosmology this term may be an effective one associated to a network of domain walls made of exotic energy/matter, but it might also arise from an energy exchange between cosmological components. Overall, the results should be interpreted as a structural analogy between evolution equations, showing how nonlinear transport mechanisms in a classical moving-boundary problem can reproduce the hierarchy of scaling terms familiar from cosmology within a reduced and analytically tractable framework.

An asymmetric three-component Fermi gas, featuring Raman-induced spin-orbit coupling between the first and second components and contact interaction only between the first and third components, introduces both spin-orbit coupling and population imbalance-two mechanisms known to stabilize the Fulde-Ferrell superfluids.We systematically study Fulde-Ferrell superfluids in an asymmetric three-component Fermi gas by finding the global minima of the thermodynamic potential. We reveal a new class of composite Fulde-Ferrell superfluids that emerges when strong spin-orbit coupling generates a double-well structure in momentum space within the lower spin-orbit-coupled band. The key features of these composite superfluids are identified.

Microwave filtering for superconducting qubits is a key element of quantum computing technology, enabling high coherence and fast state detection. This work presents the design and implementation of novel microwave Purcell filters for superconducting quantum circuits, integrated within a multilayer printed circuit board (PCB). The off-chip design removes all filter components from the qubit substrate, reducing device complexity, improving layout footprint and allowing better scalability to large qubit counts. Each embedded filter can couple up to nine readout resonators, enabling efficient multiplexed readout. Electromagnetic simulations of the filter predict a thousand-fold improvement in qubit isolation from the readout port. The design was experimentally validated under cryogenic conditions in conjunction with a 35-qubit device, demonstrating compatibility of the PCB-based filter with high-coherence superconducting qubits. The comparison of the measured qubit median T1 of 84 $μ$s with the expected radiative limit from electromagnetic simulations validated the presence of Purcell filtering in the system.

We present an end-to-end pipeline for large-scale portfolio selection with cardinality constraints and experimentally demonstrate it on trapped-ion quantum processors using hardware-aware decomposition. Building on RMT-based correlation-matrix denoising and community detection, we identify correlated asset groups and introduce a correlation-guided greedy splitting scheme that caps each cluster by the executable qubit budget. Each cluster defines a hardware-embeddable QUBO subproblem that we solve using bias-field digitized counterdiabatic quantum optimization (BF-DCQO), a non-variational method that avoids classical parameter-training loops. We recombine low-energy candidates into global portfolios and enforce feasibility with a two-stage post-processing routine: fast repair followed by a cardinality-preserving swap local search. We benchmark the workflow on a 250-asset universe taken from the S&P 500 and execute subproblems on a 64-qubit Barium development system similar to the forthcoming IonQ Tempo line. We observe that larger executable subproblem sizes reduce decomposition error and systematically improve final objective values and risk-return trade-offs relative to randomized baselines under identical post-processing. Overall, the results establish a hardware-tested route for scaling financial optimization problems, defined by a trade space in which executable problem size and circuit cost are balanced against the resulting solution quality.

Quantum mechanics, with its counterintuitive principles and probabilistic nature, has long been confined to the microscopic realm of atoms and photons. Yet, recent breakthroughs have pushed the boundaries of quantum behavior into the macroscopic world, where objects are visible to the naked eye and governed by classical physics. This review article traces the extraordinary progress toward achieving coherent control of population distributions among multiple quantum levels, as well as manipulation of absorption and refractive index, in such large-scale quantum systems, a feat once considered beyond reach.

Time-reversal symmetry is of fundamental importance to physics. In the classical theory of time-reversal symmetry, the time-reversal symmetry of a quantum system is described by an anti-unitary operator, which is known as the time-reversal operator of the system. In this paper, we introduce and study a model of continuous-time quantum walk on a special locally infinite graph. After examining its spectral property, we investigate the time-reversal symmetry of the model. To our surprise, we find that its time-reversal symmetry can be described directly by a unitary operator, which contrasts sharply with that in the classical theory of time-reversal symmetry. Some other related results are also proven.

Understanding the aging behavior of Josephson junctions and the effect of annealing on junction resistances is important in building large-scale superconducting quantum processors. Here we study the effects of aging of Josephson junctions under different storage conditions from immediately after fabrication up to 2 to 3 months. We find that the aging curve follows a logarithmic curve, with the aging amplitude mainly determined by fabrication conditions and the aging speed determined by storage conditions. Junctions stored at ambient laboratory conditions aged faster compared to junctions stored in a nitrogen atmosphere or vacuum, with the aging speed appreciably changes when the storage condition changed. We also compared the effect of thermal annealing under nitrogen environment with annealing under ambient conditions up to 250$^\circ$ C. We find that under nitrogen environment, the resistances decreased at all temperatures tested, while under ambient environment the resistances increased at 200$^\circ$ C and decreased at 250$^\circ$ C instead. We were unable to decrease the resistance below the initial-time resistance, suggesting a lower limit on the range of resistance tuning.

The Kochen Specker theorem revealed contextuality as a fundamental nonclassical feature of nature. Nonlocality arises as a special case of contextuality, where entangled states shared by space like separated parties exhibit nonlocal correlations. The notion of maximality in correlations, analogous to maximal entanglement, is less explored in multipartite systems. In our work, we have defined maximal correlations in terms of contextual models, which are analogous to absolutely maximally entangled (AME) states. Employing the sheaf theoretic framework, we introduce maximal contextual correlations associated with the corresponding maximal contextual model. The formalism introduces the contextual fraction CF as a measure of contextuality, taking values from 0 (noncontextual) to 1 (fully contextual). This enables the formulation of a new class of correlations termed absolutely maximal contextual correlations (AMCC), which are both maximally contextual and maximal marginals. In the bipartite setting, the canonical example is the Popescu Rohrlich (PR) box, while in the tripartite case, it includes Greenberger Horne Zeilinger (GHZ) correlations and three way nonlocal correlations. In this work, we extend these findings to four party correlations. Notably, no AME state exists for four qubits, which introduces a subtle difference between AMCC and AME. The construction follows the constraint satisfaction problem (CSP) and parity check methods. In particular, the explicit realization of a non AMCC correlation that is maximally contextual yet not maximal marginal is obtained within the CSP framework.

The scalar-tensor representation of $f(R,T)$ gravity is extended to incorporate the Herglotz variational principle. The field equations are derived in both the geometric and scalar-tensor frameworks. Although the divergence of the energy-momentum tensor in matter-geometry coupling theories is generally nonvanishing, conservation can be achieved through the introduction of the Herglotz vector. The generalized Friedmann equations in scalar-tensor Herglotz $f(R,T)$ theory are obtained, and a conservative cosmological model is shown to be consistent with late-time observational data. Comparisons with analogous nonconservative models and with the standard $Λ$CDM model are also provided.

Measurement-only models offer an ideal platform for exploring entanglement dynamics in the absence of unitary evolution. Despite extensive numerical evidence for entanglement phase transitions in measurement-only dynamics, the underlying mechanism attributed to non-commutativity among multi-site projective measurements has remained qualitative and coarse-grained. In this work, we identify a quantitative non-commutative index. By applying this index into three representative measurement-only models, we elucidate the role of non-commutativity in measurement-only dynamics: the emergence of a volume-law phase is governed by the non-commutative structure of the measurement ensemble, while the transition point is quantitatively determined by the amount of critical non-commutativity. More strikingly, the critical non-commutativity exhibits a universal linear scaling with the measurement range, independent of the microscopic details of the measurement ensembles. Our findings deepen the understanding of the fundamental mechanism behind the measurement-only entanglement phase transition.

Categorical supermaps generalise higher-order quantum operations from finite-dimensional quantum theory to arbitrary circuit theories. In this paper, we establish the Yoneda lemma for categorical supermaps, which states that whenever a physical theory has a suitable notion of channel-state duality, then categorical supermaps on that theory can be concretely represented in terms of that duality. This lemma eliminates any guesswork or ambiguity when defining the appropriate notion of supermap for these theories. As a concrete application, we show that the recently proposed higher-order processes on boxworld can be obtained as a particular instance of categorical supermaps, and put forward a stable definition of higher-order real quantum theory.

We show that a universal propagation scale can emerge from purely relational, scale-invariant $N$-body dynamics formulated on shape space, i.e. the space of configurations modulo translations, rotations, and dilatations. Although pure shape dynamics treats only unparametrized curves as fundamental, we adopt an affine parametrization as a gauge choice to perform a perturbative analysis, with all physically meaningful results expressed in parametrization-independent terms. Linear perturbations around central configurations satisfy second-order equations on shape space whose high-frequency spectrum defines a dimensionless constant $c_{\mathrm{rel}}$. Under general conditions of reparametrization invariance, spectral universality, and strict hyperbolicity, $c_{\mathrm{rel}}$ functions as an emergent light-cone velocity, endowing the product space $\mathbb{R} \times \mathcal{S}$ with an effective Lorentzian structure. These results suggest that causal structure and a maximal signal speed may arise dynamically from the geometry and spectral properties of relational configuration space, providing a novel perspective on the origin of relativistic kinematics.

Multiple sequence alignment (MSA) is a core operation for comparing genome sequences and is widely used in bio-informatics. MAFFT, a practical MSA tool, repeatedly shifts a pair of sequences and computes a distance. Because the number of sequence pairs grows quadratically with the number of sequences, this procedure can become a bottleneck. We propose Quantum Shift-based Sequence Alignment (QShift-SA), which implements this ``shift-wise score computation'' as a gate-based quantum circuit and searches over shift amounts and sequence pairs using Grover algorithm. QShift-SA constructs an oracle circuit that compute the Hamming distance (the number of mismatches) between two sequences with data encoding, controlled shift, comparison, and addition. This oracle can search for candidates with small distances. QShift-SA does not aim to replace the full MSA workflow; instead, it targets the screening steps that often dominate the runtime in classical MAFFT as stated above. We evaluate circuit resources (number of qubits, gate count, and depth) and benchmark simulation time across multiple quantum circuit simulators. We find that a decision diagram (DD)-based quantum circuit simulator runs more than 1,000$\times$ faster than state-vector and MPS simulators and can handle larger circuits.

In a previous work we constructed different families of stationary electrically charged Proca stars characterized by a charge parameter $q$, by solving the Einstein--Maxwell--Proca system in spherical symmetry, and imposing a harmonic time dependence ansatz for the Proca field (Mio and Alcubierre, 2025). We showed that there is a critical value for the charge $q_c$ that corresponds to the value for which the Coulomb repulsion of the charged Proca field exactly cancels the Newtonian gravitational attraction, and we found that supercritical solutions can only exist for a limited range of charges above this critical value $q>q_c$. Here we study the dynamical stability properties of these charged Proca stars by adding a small but finite perturbation to the original stationary configurations, and then performing numerical evolutions while keeping the spherical symmetry. We show that, for any given family, the parameter space can be separated into three regions corresponding to gravitationally bound stable configurations, gravitationally bound unstable configurations, and gravitationally unbound unstable configurations. For the unstable configurations we follow the evolution in time in order to determine their final state, and find that this final state can be collapse to a charged Reissner--Nordstrom black hole, migration to a new state in the stable branch, or dispersion to infinity, depending on the value of the binding energy and the specific form of the perturbation.

Local complement is a graph operation formalized by Bouchet which replaces the neighborhood of a chosen vertex with its edge-complement. This operation induces an equivalence relation on graphs; determining the size of the resulting equivalence classes is a challenging problem in general. Bouchet obtained formulas only for paths and cycles, and brute-force methods are limited to very small graphs. In this work, we extend these results by deriving explicit formulas for several broad families of distance-hereditary graphs, including complete multipartite graphs, clique-stars, and repeater graphs. Our approach uses a technique known as split decomposition to establish upper bounds on equivalence class sizes, and we prove these bounds are tight through a combinatorial enumeration of the graphs' decomposed structure up to symmetry.

We investigate how environment-induced interactions influence the entanglement dynamics of two static atoms placed near a perfectly reflecting boundary. In this setting, the environment-induced interactions include both atom-boundary contributions (position-dependent Lamb shifts) and the induced atom-atom interaction mediated by the field. We show that, regardless of the initial two-atom state, the entanglement dynamics differs qualitatively and quantitatively from predictions that neglect these energy-shift effects. Depending on the geometry and parameter regime, the environment-induced interactions can either enhance entanglement generation -- yielding a larger maximum concurrence and a longer entanglement lifetime -- or suppress it, reducing both the peak concurrence and the survival time. This behavior contrasts sharply with the free-space case, where the environment-induced atom-atom interaction affects entanglement generation only for a restricted class of initial states and does so in an exclusively assisting manner.

In this work, an extensive class of coherent states is introduced by taking the Fox Wright function as the normalization function. It is demonstrated that these states satisfy the key requirements of continuity, normalizability and resolution of unity. Furthermore, coherent states associated with the continuous spectrum are obtained through a discrete to continuous limiting procedure. Moreover, FW generalized multi parameter nu function is introduced and shown to act as the normalization function for the Fox Wright coherent states in the continuous spectrum. Later the Fox Wright function with bicomplex arguments has been introduced and its existence has been investigated. Bicomplex Fox Wright coherent states are also developed for the discrete spectrum based on this new function and their properties are analyzed. Subsequently, the results regarding Fox Wright coherent states are generalized to the bicomplex setting. In addition, a bicomplex FW generalized multi-parameter nu function is defined to demonstrate that it provides the normalization for these states in the continuous spectrum.

In this paper, we study an acoustic black hole in Hayward spacetime from the relativistic Gross-Pitaevskii theory. By examining the critical null geodesics, the shadow of the acoustic horizon is sketched. Then the quasinormal mode (QNM) frequencies of the acoustic Hayward black hole are computed numerically using the WKB method, which are shown to be more stable than those of the Hayward black hole, and the variations in the QNM frequencies are shown to correlate with the behavior of the effective potential. Moreover, the WKB method is also employed to calculate the grey-body factor and energy emission rate of the analogue Hawking radiation. It is shown that, as the tuning parameter increases, both the grey-body factor and the energy emission rate are enhanced, which can likewise be attributed to changes in the effective potential. Besides, the radius of acoustic shadow increases with the tuning parameter as well. Our results not only construct an acoustic black hole in regular black hole spacetime, but may also provide potential applications in future observations of astrophysical black holes.

We propose the protocol for preparing the maximally entangled Bell state between remote qubits at the ends of the spin-1/2 chain governed by the specially engineered nearest-neighbor XX-Hamiltonian with excited central spin as the initial state. This method does not require including optical constituent in the teleportation protocol and can be implemented in the quantum devices with solid-state architecture for teleporting unknown states or organizing quantum gates between remote qubits. A superconducting flux-qubit chain is an example of such devises.

Nonstabilizerness, also known as magic, plays a central role in universal quantum computation. Hypergraph states are nonstabilizer generalizations of graph states and constitute a key class of quantum states in various areas of quantum physics, such as the demonstration of quantum advantage, measurement-based quantum computation, and the study of topological phases. In this work, we investigate nonstabilizerness of 3-uniform hypergraph states, which are solely generated by controlled-controlled-Z gates, in terms of the stabilizer Rényi entropy (SRE). We find that the SRE of 3-uniform hypergraph states can be expressed using the matrix rank, which reduces computational cost from $\mathcal{O}(2^{3N})$ to $\mathcal{O}(N^3 2^{N})$ for $N$-qubit states. Based on this result, we exactly evaluate SREs of one-dimensional hypergraph states. We also present numerical results of SREs of several large-scale 3-uniform hypergraph states. Our results would contribute to an understanding of the role of nonstabilizerness in a wide range of physical settings where hypergraph states are employed.

We demonstrate an efficient circuit to prepare a quantum state with amplitudes proportional to a harmonic sequence. We do this by first preparing a large quantum state with linearly related amplitudes and then applying a quantum Fourier transform; this has a direct analogy to the fact that the Fourier coefficients of a sawtooth wave follow a harmonic sequence. We then consider an extension of this problem by block-encoding a matrix with a harmonic sequence along its diagonal. The cost of both circuits is dominated by the costs associated with the quantum Fourier transform.

We develop an explicit model of quantum gravity coupled to the matter fields of the Standard Model, based on the 3-group structure and the 3BF action, within the framework of higher gauge theory. The model is constructed by providing a rigorous definition for the path integral of the theory, achieved by defining the whole theory on a piecewise-flat spacetime manifold. To that end, we develop a method to systematically discretize both the action and the path integral measure by passing from a smooth manifold to a piecewise-flat manifold. Finally, we discuss in some detail the structure of the resulting quantum gravity model, and provide a preliminary analysis of its semiclassical limit.

In Einstein-scalar-Maxwell theories with a coupling between the scalar field $φ$ and the electromagnetic field strength $F$ of the form $μ(φ) F$, we investigate the existence of exotic compact objects (ECOs) and their observational signatures in photon and massive-particle dynamics. For $μ(φ)$ diverging at the origin while all physical quantities remain finite, we demonstrate the existence of electrically charged ECOs with a shell-like structure whose density peaks at an intermediate radius. We compute their mass and radius, together with the scalar and vector field profiles, on a static and spherically symmetric background. We then examine the existence of photon rings and place bounds on a model parameter by requiring the absence of a linearly stable photon ring. Under this condition, photon echoes from ECOs are absent. We also compute the gravitational-lensing deflection angle $Ψ$ and show that it attains a maximum for an impact parameter of the same order as the ECO radius. Finally, we study the parameter space in which innermost stable circular orbits of massive particles exist.

Binary neutron star mergers are thought to produce hot, rapidly rotating neutron stars with masses that can far exceed their Tolman-Oppenheimer-Volkoff mass. The gravitational-wave emission from such remnants provides a unique opportunity to measure the nuclear equation of state at densities and temperatures not available to terrestrial experiments. Current detector design is informed by gravitational-wave signals from general relativistic hydrodynamics simulations of neutron star mergers, typically with hybrid thermal treatments for the equation of state, where a cold equation of state is modified by adding a thermal component. We use realistic equations of state based on the relativistic mean field model with consistent treatment of thermal effects to compute the distribution of expected peak gravitational-wave frequencies. Marginalising over equation of state and progenitor neutron star masses, we show the peak frequency of emission ranges from $\sim2.5$ to 4 kHz. The width of this distribution suggests the need for broadband observatories with kHz sensitivity, and calls into question some of the so-called post-merger optimised configurations. We show the proposed KAGRA high-frequency design is well-suited to measuring post-merger remnants when compared to the KAGRA broadband design.

In quantum time (QT) schemes, time is promoted to a degree of freedom, allowing Lorentz covariance to be made explicit for single particles. We ask whether this can be lifted to QFT, so that Lorentz covariance becomes manifest at the Hilbert-space level, rather than being hidden as in the standard canonical formulation. We address this question by proposing a second-quantized approach in which the elementary particle is the QT particle itself, leading naturally to the notion of spacetime field algebras and of quantum action. We show, however, that a naive many-body construction runs into inconsistencies. To pinpoint their origin we introduce a classical counterpart of the second-quantized formalism, spacetime classical mechanics (SCM), and prove a no-go theorem: Dirac quantization of SCM collapses back to standard QFT and therefore hides covariance. We circumvent this problem by presenting a quantum-action--based quantization that yields a spacetime version of quantum mechanics (SQM), making covariance manifest for (interacting) QFTs. Finally, we show that this resolution is tied to a genuine spacetime generalization of the notion of quantum state, required by causality and closely connected to recent ``states over time'' proposals and, in dS/CFT-motivated settings, to microscopic notions of timelike entanglement and emergent time.

We measure the effect of ionizing radiation on superconducting qubits with a timing resolution of 1 $μs$ using microwave kinetic inductance detectors (MKIDs) fabricated on the same substrate. We observe no correlation between two-level system (TLS) scrambling events and ionizing radiation events detected with the MKIDs, suggesting TLS scrambling events may not arise from ionizing radiation and instead the previously reported apparent correlation may be due to events without sufficient energy to trigger our MKIDs. We characterize the fast-time system recovery of transmons following a radiation event, where we observe the recovery of the enhanced qubit relaxation and excitation to be well-described by an exponential recovery to the baseline quasiparticle density, with a characteristic time of $13\pm1\ μ$s, and a peak quasiparticle density at the junction per deposited energy of $240/μm^3/MeV$. The fast recovery is consistent with literature reported values for Nb-based devices with direct injection of 2$Δ_{\text{Al}}$ phonons, demonstrating the recovery is strongly dependent on the proximity of niobium to the junction.

Starting from the most general second-order in derivatives theories describing extended objects of arbitrary dimension evolving geodetically in a codimension-one flat ambient space-time, we determine the subset of models yielding second-order equations of motion, forming an intriguing theory known as Lovelock-type brane gravity (LBG). These models further extend the so-called geodetic brane gravity (GBG) approach, thereby naturally promoting the GBG geometric properties, allowing LBG to be reformulated as a mimetic embedding gravity and, in turn, the possibility of introducing fictional matter through a peculiar current $\mathcal{T}^{a\,μ}$. Grounded in the elasticity theory, we provide a possible origin of such a current. Finally, variational techniques are employed to elucidate the mechanical function of both the dark current $\mathcal{T}^{a\,μ}$ and its tangential components $\mathcal{T}^{ab}$; these serve as the constituents of a fictional energy-momentum tensor that shares characteristics with a perfect fluid.

This work studies the relativistic quantum dynamics of scalar bosons in a spacetime containing both a cosmic string and a global monopole within the framework of Rainbow Gravity. An effective metric is constructed to describe the combined topological defects together with the energy-dependent deformation of spacetime. The Klein-Gordon equation is formulated in this background, including scalar, vector, and nonminimal couplings, and its solutions are obtained by separation of variables. Generalized Coulomb-type interactions are considered, allowing a unified analysis of scattering and bound states. The bound-state spectrum is determined from the poles of the corresponding $S$-matrix. Two specific choices of rainbow functions are examined, and their influence on the energy spectrum is analyzed through numerical calculations and, in suitable limits, analytical approximations. The results show how the interplay between topological defects and rainbow gravity corrections affects the spectral properties of scalar bosons, while known results are consistently recovered in appropriate limits.

In a free space optical (FSO) communication link from satellite to ground, the losses in the channel will be dynamic. Thus, the characterization of the FSO channel is of great importance and this can be emulated in the lab to evaluate the realistic performance of a satellite payload. In this work, we introduce a novel optical channel emulator capable of replicating these dynamics, especially for Low Earth Orbit based CubeSats. We demonstrate its ability to accurately emulate a satellite-to-ground optical communications channel under various atmospheric turbulence strengths, satellite trajectories, and optical ground station parameters at a given optical wavelength of interest. Our satellite channel emulator was designed to test and benchmark the performance of the continuous variable quantum key distribution payload for the Satellite Platform for Optical Quantum Communications mission - an in-orbit demonstrator for the UK's Quantum Communication Hub, to be launched in early 2026.

Quantum machine learning seeks a computational advantage in data processing by evaluating functions of quantum states, such as their similarity, that can be classically intractable to compute. For quantum advantage to be possible, however, it is essential to bypass costly characterisation of individual data instances in favour of efficient, direct similarity evaluation. Here we demonstrate a sample-optimal, hardware-efficient protocol for estimating quantum similarity -- the state overlap -- using bosonic quantum interference. The sample complexity of this approach is independent of the system dimension and is information-theoretically optimal up to a constant factor. Experimentally, we implement the scheme on \emph{Prakash-1}, a quantum computing platform based on a fully programmable integrated photonic processor. By preparing and interfering qudit states on the chip to directly extract their overlap, we demonstrate classification and online learning of quantum data with high accuracy in realistic noisy experiments. Our results establish joint overlap measurements as a scalable pathway to efficient quantum data analysis and a practical building block for network-integrated quantum machine learning.

The probabilistic description of the time evolution of a physical system can take two conceptually distinct forms: a trajectory of probabilities, which specifies how probabilities evolve over time, and a probability on trajectories, which assigns probabilities to possible histories. A lack of a clear distinction between these two probabilistic descriptions has given rise to a number of conceptual difficulties, particularly in recent analyses of stochastic-quantum correspondence. This paper provides a systematic account of their relationship. We define probability dynamics and stochastic process families together with a precise notion of implementation that connects the two descriptions. We show that implementations are generically non-unique, that every probability dynamics admits a Markovian implementation, and characterize when non-Markovian implementations are possible. We expose fallacies in common arguments for the linearity of probability dynamics based on the law of total probability and clarify the proper interpretation of ``transition matrices'' by distinguishing dynamics-level maps from the conditional probability matrices of implementing processes. We further introduce decomposability as the appropriate general notion of stepwise evolution for (possibly nonlinear) probability dynamics, relate it to divisibility in the linear case -- showing that the two can come apart -- and disentangle both notions from Markovianity and time-homogeneity. Finally, we connect these results to what we call statistical dynamics, in which linearity is indeed physically motivated, and contrast the framework with quantum mechanics.

Gravitational-wave (GW) astronomy has advanced our understanding of compact mergers through instruments like the Laser Interferometer Gravitational-Wave Observatory (LIGO). However, the extreme sensitivity required for these detections makes the instruments susceptible to short-duration transient noise, or glitches, which obscure GW data. Current tools such as Omega Scan and GravitySpy assist in identifying and classifying such noise, but are limited by manual inspection or dependence on large training sets. To address these challenges, we present \textit{OmegaNeuron}, a machine-learning tool that integrates GravitySpy's image similarity methods with Omega Scan's transient analysis to automate the identification of auxiliary channels that witness glitches. Applied to multiple glitch examples, OmegaNeuron consistently highlighted plausible witness channels and showed strong agreement with existing correlation tools, while providing clearer ranking through a quantitative similarity metric. Integrated into the \texttt{gwdetchar} package, OmegaNeuron enables faster analysis that improves glitch witness identification, enhancing both detector sensitivity and the reliability of gravitational-wave observations.

We analyse a variety of Euclidean saddles in the gravitational path integral, with asymptotic AdS boundary conditions, in a class of Einstein-Scalar-Maxwell models. These include single boundary solutions, usual and wineglass wormholes, as well as more exotic (quasi)-oscillatory saddles. We find several interesting phase transitions between these solutions. The Euclidean wormhole backgrounds can be analytically continued to Lorentzian FLRW universes. Some of them contain an early period of inflation. We delineate the conditions under which they can be the dominant saddles in the gravitational path integral and use them to estimate ratios of probabilities for different cosmological outcomes.

For the 1D quantum East model with open boundaries, we show that in the limit $s \to -\infty$, the ground state is accurately captured by a simple spin-coherent product state. We further identify a low-entanglement excited eigenstate that differs from the ground state only by a $π$-rotation of the boundary spin, remaining well approximated by a spin-coherent state. For a range of $-\infty<s<0$, the edge-coherent product state overlaps with two eigenstates separated by a size-independent energy gap, leading to persistent coherent oscillations of both global and local observables in the thermodynamic limit. These oscillations originate from boundary physics and are distinct from quantum many-body scars or hypercube-like Fock-space mechanisms.

The Blandford-Znajek (BZ) mechanism is widely recognised as the most compelling process to extract rotational energy from an accreting black hole and power the emission of relativistic jets. We explore the universality of this process for generic black-hole spacetimes within the Konoplya-Rezzolla-Zhidenko formalism and find that the lowest-order contribution to the BZ power is invariant across different black-hole spacetimes. We also show that at the next-leading-order, different black-hole spacetimes will lead to different BZ luminosities. As a result, while slowly rotating black holes cannot be distinguished via measurements of their jet power, rapidly rotating ones have the potential of providing information on the strong-field properties of the spacetime when independent measurements of the BZ luminosity and of the black-hole angular velocity are available.

Identifying the boundary between classical and quantum computation is a central challenge in quantum information. In multi-qubit systems, entanglement and magic are the key resources underlying genuinely quantum behaviour. While entanglement is well understood, magic - essential for universal quantum computation - remains relatively poorly characterised. Here we show that determining membership in the stabilizer polytope, which defines the free states of magic-state resource theory, requires super-exponential time $\class{exp} ( n^2)$ in the number of qubits $n$, even approximately. We reduce the problem to solving a $3$-\class{SAT} instance on $n^2$ variables and, by invoking the exponential time hypothesis, the result follows. As a consequence, both quantifying and certifying magic are fundamentally intractable: any magic monotone for general states must be super-exponentially hard to compute, and deciding whether an operator is a valid magic witness is equally difficult. As a corollary, we establish the robustness of magic as computationally optimal among monotones. This barrier extends even to classically simulable regimes: deciding whether a state lies in the convex hull of states generated by a logarithmic number of non-Clifford gates is also super-exponentially hard. Together, these results reveal intrinsic computational limits on assessing classical simulability, distilling pathological magic states, and ultimately probing and exploiting magic as a quantum resource.

As quantum information systems mature, efficient and coherent transfer of quantum information through noisy channels becomes increasingly important. We examine how passive environment-assisted quantum communication enhances direct quantum information transfer efficiency. A bosonic pure-loss channel, modeled as transmission through a beam splitter with a vacuum input state at the dark port, has zero quantum capacity when transmissivity is below 50%. Quantum communication through the channel can be enhanced by passive environment assistance, achieved via the selection of an appropriate input state for the ancilla port. Although ideal Gottesman-Kitaev-Preskill (GKP) states enable perfect quantum information transmission at arbitrarily small transmissivity, they are challenging to realize experimentally. We therefore explore more experimentally accessible non-Gaussian ancilla states, such as Fock, cat, and squeezed cat states, and numerically determine the optimal encoding and decoding strategies. We also construct analytical schemes that yield high-fidelity transmission and good information rates.

This work extends the CoLoRFulNNLO subtraction method to address soft and collinear divergences in the computation of higher-order corrections for hadronic collisions. By utilizing universal local counterterms which can be integrated analytically over the unresolved phase space, we achieve numerically stable, fully-differential predictions. Our publicly available NNLOCAL code serves as a proof-of-concept implementation, validated by calculating the NNLO cross-section for Higgs boson production in gluon-gluon fusion with no light quarks.

We present a class of models based on a gauged $U(1)_F$ flavor symmetry that explains the hierarchical structure of fermion masses and mixings via the Froggatt-Nielsen (FN) mechanism, while also solving the strong CP problem by the Peccei-Quinn (PQ) mechanism. A global $U(1)_{\rm PQ}$ symmetry with a nonzero QCD anomaly emerges accidentally in this setup as a byproduct of the gauged $U(1)_F$ symmetry. The resulting axion is shown to be of high quality, with the axion potential safeguarded against quantum gravity corrections by the gauge symmetry. Three models, which are generalizations of the Dine-Fischler-Srednicki-Zhitnitsky (DFSZ) axion model, are presented realizing this idea. The right-handed neutrino mass scale is identified as the Froggatt-Nielsen scale in these models. We present explicit UV completions of the FN sectors of these models and show that they preserve the high quality of the axion. In these models, the axion acts as a flavon field, leading to testable predictions in flavor-changing decays of neutral mesons. The axion also serves as the dark matter of the universe with the right amount of relic abundance without causing cosmological domain wall problems. Baryon asymmetry of the universe is realized via leptogenesis which is calculable in these models and found to be of the right order of magnitude.

We perform a comprehensive study of the properties of Dirac eigenvalue spectrum in QCD as a function of temperature on the lattice. In addition to effects due to interplay between interactions and disorder inherently present in a many-body system, the Dirac spectrum also contains crucial information about the effective restoration of different subgroups of almost exact two flavor chiral symmetry in QCD. We calculate the infrared eigenvalues of the overlap Dirac operator on 2+1 flavor QCD ensembles generated using domain wall fermion discretization, on a large volume lattice. From the normalized level spacing ratios we identify those eigenvalues which have intermediate level statistics, distinctly different from the majority in the bulk spectrum that follow universal level fluctuations similar to a random matrix of Gaussian unitary type. We provide an explanation of these intermediate level ratios in terms of a specific random matrix model and quantify correlation between these eigenstates and disorder in the gauge fields manifested in the renormalized Polyakov loop values. Whereas existence of these intermediate eigenmodes are intimately connected to the effective restoration of different subgroups of chiral symmetry close to chiral crossover transition, these arise due to effects of random uncorrelated disorder at higher temperatures when the $U_A(1)$ is effectively restored. We also, for the first time, calculate the Thouless conductance for the Dirac spectrum that quantifies the structural rigidity of the eigenvectors, and use it as a diagnostic tool to understand the restoration of the anomalous $U_A(1)$ subgroup of chiral symmetry and localization driven due to disorder.

Texture-zero ansatze offer an economical description of neutrino masses, with current data allowing only seven inequivalent two-zero Majorana textures in the charged-lepton mass basis. We investigate how such textures can arise from non-invertible symmetries realized through $\mathbb{Z}_3$ gauging of $\mathbb{Z}_N$. In contrast to $\mathbb{Z}_2$ gauging, which necessarily induces diagonal neutrino mass terms via the Weinberg operator, $\mathbb{Z}_3$ gauging admits complex representations and allows a richer class of neutrino mass textures. If the light neutrino mass is described by the Weinberg operator, we find that the textures $\mathbf{A}_{1,2}$, $\mathbf{B}_{3,4}$, and $\mathbf{C}$ can be realized from the $\mathbb{Z}_{3}$ gauging of $\mathbb{Z}_{13}$ symmetry, while all the seven phenomenologically viable two-zero textures can emerge from $\mathbb{Z}_{3}$ gauging of $\mathbb{Z}_{19}$ symmetry without requiring supersymmetry. When the neutrino mass is generated by the type-I seesaw mechanism, the structure of the non-invertible symmetry is more restrictive, yielding only texture $\mathbf{C}$ for $N\neq7$. These results demonstrate the strong predictive power of non-invertible symmetries for neutrino mass textures. Furthermore, the more general $\mathbb{Z}_{n}$ gauging of the $\mathbb{Z}_{N}$ symmetry with $n>3$ is analyzed, which results in novel fusion rules.

We compute the twist-2 unpolarized quark parton distribution functions (PDFs) of the pion and the $ρ$ meson within the Dyson-Schwinger equations (DSEs) framework using the rainbow-ladder (RL) truncation. A new DSE for the spin-1 hadron's quark-quark correlation matrix is derived, from which the PDFs can be extracted. For the $ρ$ meson, we obtain for the first time within RL-DSEs the unpolarized PDFs corresponding to different helicity states. A pronounced difference is observed between the $|Λ|=1$ and $Λ=0$ cases, where $Λ$ denotes the meson helicity, leading to a nonvanishing and numerically sizable tensor-polarized PDF $f_{1LL}(x)$. We further compare these results with those obtained under a leading Fock-state ($|q\bar{q}\rangle$) truncation and find substantial deviations. This comparison demonstrates that the present RL-DSEs framework incorporates higher Fock-state contributions associated with gluonic degrees of freedom, which has a significant impact on the quark distributions.

We present a systematic QCD sum-rule analysis of the in-medium properties of the charged kaon doublet $K^{\pm}$ over the full $(T,ρ)$ plane relevant to current and forthcoming heavy-ion experiments. Working within the QCD sum-rule framework and incorporating temperature-and density-dependent quark, gluon, and mixed condensates, we derive Borel-transformed sum rules for the effective masses $m_{K^{\pm}}$, the pseudoscalar decay constants $f_{K^{\pm}}$, and the vector self-energy $Σ_{v}$ of both charged states simultaneously. Our vacuum results, $m_{K^{-}} = 494.6^{+4.9}_{-6.9}$~MeV and $f_{K^{-}} = 157.3^{+4.1}_{-2.9}$~MeV (with near-degenerate $K^{+}$ values), are in excellent agreement with Particle Data Group values at the sub-percent level. In the medium, $m_{K^{\pm}}$ decreases monotonically with increasing baryon density and temperature, signalling progressive partial restoration of chiral symmetry. A pronounced mass splitting $Δm = m_{K^{-}} - m_{K^{+}}$ develops in baryonic matter, driven by the opposite sign of the Weinberg--Tomozawa vector interaction for the two charge states; it reaches $|Δm| \sim 0.35$~GeV near $ρ\simeq 3.2\,ρ_{\rm sat}$ at $T = 0$ and is partially quenched by thermal fluctuations. A central outcome of this study is the extraction of the critical onset density $ρ_c$, defined as the threshold beyond which the in-medium modifications of $K^{-}$ properties signal the onset of the transition toward the chirally restored phase. We stress that $ρ_c(T)$ should not be interpreted as a precise determination of the QCD critical point-a task beyond the reach of any current effective framework-but rather as an indicator ....

This paper presents a radiation-hardened current-mode delta-sigma ADC fabricated in a standard 130~nm CMOS technology and qualified for total ionizing doses up to 100~Mrad. The converter is designed for beam loss monitoring applications in high-energy physics, where it must handle input currents spanning nine decades, from 1~mA down to 1~pA, while providing a fast 10~\textmu s response time for machine protection. To meet these conflicting requirements, the architecture exploits the inherent trade-off between resolution and acquisition time: a first-order modulator sampled at 20~MHz delivers 11-bit effective resolution within the critical 10~\textmu s window for the mA current range. Extended integration times of up to 100~s enable the sub-picoampere resolution required for beam alignment and background monitoring and provides an operational dynamic range exceeding 200~dB. The chip integrates two independent channels, consumes 25~mW from a 1.2~V supply, and includes radiation-hardening techniques such as triple-redundant digital logic and SEU-tolerant comparator banks. Post-irradiation measurements up to 100~Mrad show no performance degradation, and the uncalibrated integral nonlinearity remains within [+0.2\%, --0.3\%] of full scale over the 1~mA to 5~\textmu A range. The converter's flexibility and radiation tolerance make it suitable not only for the HL-LHC beam loss monitoring upgrade but also for other precision current measurement applications in harsh environments.

Loops of portal matter (PM) fields carrying both dark and Standard Model (SM) quantum numbers can lead to the kinetic mixing (KM) of the SM photon and the analogous dark photon (DP) of a phenomenologically interesting magnitude. However, in specific frameworks, different loops of these same PM fields can also lead to other new types of interactions between some of the SM fields and the DP via the generation of `dark moment'-like couplings, even though the SM fields carry a zero dark charge at tree-level. In recent work, this possibility has been explored for the case when these SM fields are fermionic. In this paper, we extend this idea to the case of the SM $W$ boson employing a previously examined model wherein PM consists of a complex scalar triplet plus a complex singlet, both of which obtain vevs and thus also generate the DP mass. While this setup results in a somewhat stronger interaction between the DP and the $W$ than in the familiar KM setup, the rate for $W+$DP production at the LHC is shown to still be rather small and is very difficult to observe due to large SM backgrounds. The direct production of the scalar PM itself, however, is shown to lead to similar new physics signatures but with much larger rates and are considered here in their own right. The properties of such new PM states may already be constrained by LHC searches in both the $W^\pm+$MET and $W^+W^-+$MET channels.

Based on $e^+e^-$ annihilation data collected with the BESIII detector at center-of-mass energies from 4.600 to 4.699 GeV, corresponding to an integrated luminosity of 4.5 fb$^{-1}$, we present the first measurement of the longitudinal polarization of $Λ$ hyperons produced in the inclusive decay $Λ_c^+ \to ΛX$, where $X$ denotes any allowed final state. The polarizations are determined to be $\mathcal{P}_Λ = -0.393 \pm 0.055_{\mathrm{sta.}} \pm 0.020_{\mathrm{sys.}}$ and $\mathcal{P}_{\barΛ} = 0.288 \pm 0.056_{\mathrm{sta.}} \pm 0.017_{\mathrm{sys.}}$. We then search for CP violation using an asymmetry constructed from the $Λ$ polarization and the $Λ\to p π^-$ decay asymmetry parameters, and obtain $\mathcal{A}_{\mathrm{CP}}^{\mathrm{pol}} = 0.15 \pm 0.12_{\mathrm{sta.}} \pm 0.04_{\mathrm{sys.}}$. We also perform an updated measurement of the absolute branching fraction, resulted as $\mathcal{B}(Λ_c^+ \to ΛX) = (38.07 \pm 0.38_{\mathrm{sta.}} \pm 0.49_{\mathrm{sys.}})\%$, with precision improved by a factor of four relative to the current world average. A search for direct CP violation yields $\mathcal{A}_{\mathrm{CP}}^{\mathrm{dir}} = (1.5 \pm 1.0_{\mathrm{sta.}} \pm 1.0_{\mathrm{sys.}})\%$. No evidence for CP violation in inclusive charm baryon decays is observed.

We investigate several possible generalisations of $T\overline{T}$ deformations to three- and higher-dimensional field theories. Starting from the two-dimensional $T\overline{T}$ flow, we work out its higher-dimensional uplift, which results in a non-local and non-isotropic three-dimensional theory. Starting instead from the relation between the Nambu-Goto action and $T\overline{T}$ in $d=2$, we study the flow equation obeyed by the Dirac-Nambu-Goto actions in $d>2$ dimensions, written in terms of the stress-energy tensor only. Similarly, we derive the stress-tensor flow obeyed by the Born-Infeld actions in $d$ dimensions and by the Dirac-Born-Infeld actions in $d=2$ and $d=3$.

We propose a correction to one of the elliptic blocks in the NS sector of 2d $\mathcal N = 1$ superconformal field theories. We analyze the 4-point block in the pillow geometry to demonstrate the necessity of the correction and verify the formula by numerically checking the crossing symmetries in the $\mathcal N =1 $ super Liouville theory, as well as directly comparing the $c$-recursion and $h$-recursion results.

Parton Distribution Functions (PDFs) are a key ingredient in theoretical predictions for Large Hadron Collider (LHC) observables and play a central role in the extraction of precision Standard Model (SM) and Beyond the SM (BSM) parameters from LHC data. Recent analyses demonstrate that the determination of fundamental SM parameters such as $α_s(m_Z)$, $m_W$, $m_t$, and $\sin^2θ_W$ is strongly influenced by the choice of input PDFs. In this contribution, we present the status and challenges of PDF determination from the NNPDF perspective, both in stand-alone fits and in joint extractions with (B)SM parameters. We place particular emphasis on results for $α_s(m_Z)$, $m_t$, and Wilson coefficients in the SM Effective Field Theory (SMEFT) framework.

This proceedings contribution summarises projected constraints from the QUEST-DMC concept, a surface-based direct-detection experiment using superfluid $^3$He operated below the millikelvin regime and instrumented with nanomechanical resonators read out by SQUIDs. The low recoil-energy threshold (down to sub-eV for the SQUID configuration) enables sensitivity to sub-GeV dark matter across a wide set of interaction structures beyond the canonical spin-independent and spin-dependent limits. We present projections in the non-relativistic Effective Field Theory (EFT) framework, scanning the standard set of fourteen Galilean-invariant operators and expressing reach in terms of effective dark matter-nucleon (or dark matter-neutron) cross sections. Because QUEST-DMC operates at the surface, we also account for suppression of the incident flux due to scattering in the atmosphere and Earth, which produces an interaction-dependent sensitivity ceiling at large couplings. Finally, we outline how the non-relativistic results map onto representative relativistic EFT dark matter-nucleon bilinears, enabling a compact interpretation of the projected reach in terms of UV-motivated coupling structures.

This Letter presents a combination of searches for Higgs boson pair (HH) production performed by the ATLAS and CMS Collaborations using proton-proton collision data sets recorded at $\sqrt{s} = 13$ TeV during the Large Hadron Collider Run 2, corresponding to integrated luminosities ranging between 126 and 140 $\mathrm{fb^{-1}}$. The upper limit at the 95% confidence level on the total HH signal strength, defined as the ratio of the measured cross section to the SM prediction, corresponds to 2.5, with an expected value of 1.7 (2.8) assuming the absence (presence) of the standard model (SM) HH signal. The strength of the HH signal is measured to be $0.8^{+0.9}_{-0.7}$ relative to the SM prediction. The observed significance is found to be 1.1 standard deviations whereas 1.3 are expected for the SM HH signal. Constraints are set on the Higgs boson trilinear self-coupling and on the couplings of two Higgs bosons to two vector bosons, both normalized to the SM predictions and denoted as $κ_λ$ and $κ_{2\mathrm{V}}$, respectively. The observed individual constraints at the 95% confidence level are $-0.71 < κ_λ< 6.1$ and $0.73 < κ_{2\mathrm{V}} < 1.3$, while the expected constraints assuming the presence of the SM HH signal are $-1.3 < κ_λ< 6.7$ and $0.66 < κ_{2\mathrm{V}} < 1.4$.

Neutrino non-standard interactions (NSI) with electrons, predicted in many extended theoretical models of particle physics, are known to alter the picture of neutrino decoupling from the cosmic plasma. We update previous analyses of neutrino decoupling in presence of NSI with electrons, extending the parameter space in order to provide, for the first time, a full study of their effect on the production of light elements during Big Bang Nucleosynthesis (BBN). We compare the BBN bounds on non-universal and flavour-changing NSI parameters with the constraints from terrestrial experiments. Our results show that the limits from BBN are significantly less stringent than the experimental bounds, but they are complementary and can provide a test of neutrino physics at different temperature scales and epochs.

The Hubble tension and the recently reported anomaly in data from the Dark Energy Spectroscopic Instrument (DESI) are considered to pose serious challenges to the standard $Λ$CDM model of cosmology. In this work, we show that resolving the Hubble tension with a scenario featuring dark radiation-matter decoupling (DRMD) predicts the presence of dark acoustic oscillations (DAO) similar in scale to baryon acoustic oscillations (BAO). Using an inference independent of large-scale structure data, relying only on Planck measurements of the cosmic microwave background and SH$0$ES-calibrated supernova data, we find evidence for a DAO signal with drag-horizon scale $r_{d,\mathrm{DAO}} \in[54,65]\,\mathrm{Mpc}/h$ ($68\%\,\mathrm{C.I.}$) and amplitude $A_\mathrm{DAO} \in [0.02,0.05]$ ($68\%\,\mathrm{C.I.}$). These predictions provide a concrete target for current and upcoming large-scale structure surveys, including DESI, Euclid, and the Roman Space Telescope. Remarkably, the predicted DAO properties are consistent with those required to explain the DESI anomaly, offering both an alternative to evolving dark energy and a preliminary validation of the relevance of a dark radiation-matter decoupling scenario for addressing the Hubble tension.

The possible anomalous New Physics contributions to magnetic and electric dipole moments of the $τ$ lepton have brought renewed interest in studying $τ$-pair production at energies of the LHC and future colliders. We discuss effects of electromagnetic and weak dipole moment contributions to the $τ$-lepton polarization and $ττ$ spin correlations in the $γγ\to τ^-τ^+$ and $q \bar{q} \to τ^- τ^+$ processes. Such processes have been observed in $pp$ and PbPb collisions in the LHC experiments. Extensions of the Standard Model amplitudes for $γγ\to τ^-τ^+$ and $q \bar{q} \to τ^- τ^+$ processes, which include dipole moments of the $τ$ lepton, are implemented in the TauSpinner Monte Carlo program. A few examples of signatures of $ττ$ spin correlations and $τ$-lepton dipole moments in observables are presented.

Using $e^+e^-$ collision data corresponding to an integrated luminosity of 44 fb$^{-1}$ collected with the BESIII detector at the BEPCII collider, we report the first measurement of Born cross sections and effective form factors for $e^+e^-\toΣ^-\barΣ^+$ at centre-of-mass energies between 3.51 and 4.95 GeV. With a fit to the $\sqrt{s}$-dependent dressed cross sections, the decay $ψ(3770)\toΣ^-\barΣ^+$ is observed for the first time with a significance of 5.5$σ$, including systematic uncertainties. Upper limits at the 90\% confidence level on the product of the branching fraction and the electronic partial width are provided for other possible charmonium(-like) states. The ratios of Born cross sections for the $Σ$ isospin-triplet states are determined and can be used to test the vector-meson-dominance model.

The $^{13}$C($α$,$n$)$^{16}$O reaction at low energies is studied by constructing an effective field theory. We choose a separation scale at $E=1$~MeV, where $E$ is the initial $α$-$^{13}$C energy in center-of-mass frame, just below the sharp resonant $5/2^+$ state of $^{17}$O, and include two open channels, $α$-$^{13}$C and $n$-$^{16}$O, and resonant $1/2^+$, $5/2^-$, $3/2^+$ states of $^{17}$O in the study. Parameters of the theory are fitted to experimental data, $S$ factor of $^{13}$C($α$,$n$)$^{16}$O at the energies below $E=1$~MeV, including the data sets recently reported by the LUNA and JUNA collaborations, and the $S$ factor of $^{13}$C($α$,$n$)$^{16}$O is extrapolated to the Gamow peak energy $E_G= 0.19$~MeV in the low mass AGB stars. We discuss an uncertainty in the estimate of the $S$ factor and confirm that the main part of the uncertainty emerges from the parameter fit of the near-breakup threshold $1/2^+$ state of $^{17}$O.

We propose that a pair of annihilation of the secluded dark matter may accommodate a source of the halo gamma ray signal reported recently, through the p-wave Sommerfeld enhancement. We show that given a weakly coupling of the dark sector to the Higgs bosons, the dark matter annihilation into W+W-, even though induced at 1-loop level, is desirably amplified. We also argue that the supersymmetric framework may readily embody such a model within rather minimal contents.

We present numerical results obtained in a finite-temperature study of the Sp(4) Yang-Mills theory on the lattice. We study its first-order confinement/deconfinement phase transition, by reconstructing the density of states via the Logarithmic Linear Relaxation (LLR) algorithm. We perform our measurements on lattices with different extents of space and time (and aspect ratios). We estimate the size of discretisation and finite-volume artefacts. We find clear signatures of a first-order transition. We determine the critical coupling, the specific heat, and the surface tension, for finite extents of the thermal circle, and use the results to set bounds for the continuum theory.

Dark matter much heavier than the weak scale remains a comparatively unexplored frontier. This review surveys theoretical and experimental developments on very heavy dark matter, including composite and dissipative formation mechanisms, multiscatter detection, and astrophysical searches.

We investigate the geometric structure associated with CP-violating dynamics in entangled neutral meson systems. We formulate the time-dependent geometric phase for the correlated two-meson state and analyze its system-dependent behavior across different neutral meson mixing scenarios. We demonstrate that the geometric phase associated with the entangled state cannot, in general, be expressed as a sum of independent single-meson contributions, indicating the presence of genuinely entanglement-induced geometric effects. Furthermore, we show how CP violation enters the geometric phase through mixing, leading to a rephasing-invariant geometric structure of correlated evolution that remains indirectly sensitive to the mixing parameters. These results clarify the geometric characterization of entanglement and CP violation in neutral meson systems.

We present a systematic study of the hadronic contributions to the muon anomalous magnetic moment within several infrared-improved AdS/QCD models. The models are constrained by the pion decay constant and the $ρ$-meson mass and are shown to reproduce phenomenologically reasonable low-energy hadron spectra. Within a unified holographic framework, we evaluate both the leading-order hadronic vacuum polarization contribution and the pseudoscalar-pole contribution to hadronic light-by-light scattering. The holographic predictions for the hadronic vacuum polarization contribution are found to be systematically lower than recent dispersive determinations, and we demonstrate that this discrepancy is closely correlated with an underestimation of the $ρ$-meson decay constant in the models. We further compute the pion transition form factor and the corresponding pseudoscalar-pole hadronic light-by-light contribution. Although the models yield similar meson spectra and reproduce the expected asymptotic behavior, their hadronic light-by-light predictions exhibit sizable differences, reflecting significant variations in the transition form factor at low momentum transfer $Q^{2}$.

Noble liquid time projection chambers (TPCs) are a leading technology in the detection of ionizing radiation, particularly in applications such as accelerator neutrino physics, dark matter detection, and neutrinoless double beta decay. This paper addresses the design considerations for implementing stable high voltage (HV) systems within large noble liquid TPCs, with a focus on the nEXO experiment. Utilizing insights from prior HV research and experimental investigations, we outline factors influencing HV stability and discuss design choices to improve stability and prevent electrical discharges. A novel HV delivery system concept is presented, tailored for the nEXO TPC, which incorporates these design considerations while also meeting the stringent radiopurity requirements of the nEXO neutrinoless double beta decay search. These design considerations and their specific implementation towards a HV delivery system offer guidance to future experiments applying high voltage in noble liquid environments.

Large liquid argon time projection chambers (LArTPCs) require stable and well-characterized delivery of ultraviolet (UV) light for in situ calibration of photosensors at cryogenic temperatures. This article reports bench-top and cryogenic measurements of the optical components used in a UV light calibration system, including multi-mode fused-silica fibers, SMA-to-SMA connectors, optical fiber feedthroughs, and light-diffuser assemblies. Light loss in several fiber types and SMA connectors was measured across wavelengths from \qtyrange{275}{970}{nm}. In addition, light-loss measurements of the tested fibers after several liquid-nitrogen thermal cycles showed no statistically significant degradation relative to baseline measurements, and high-rate pulsed exposure (30-90 million pulses from a \qty{275}{nm} LED) likewise showed no measurable aging in jacketed fibers. A compact, palm-sized, 3D-printed PEEK diffuser housing with stacked UV-grade fused-silica diffusers yields Lambertian emission and the most uniform angular distribution. Optical components exhibiting improved UV transmission were deployed successfully in multiple DUNE small- and large-scale prototypes, demonstrating reliable operation of UV light calibration systems. These findings inform component selection and calibration procedures for achieving reliable, uniform UV light delivery in large-scale cryogenic detectors such as DUNE.

In this work, we investigate some aspects of the Maxwell electrodynamics with the additive Lorentz-violating (LV) CPT-even term. For this model, we derive the energy and momentum conservation laws, highlighting the modifications introduced by Lorentz violation. Furthermore, using the modified dispersion relations, we analyze the correction to Compton effect arising from the presence of the LV vector.

We consider actions that are general functions of the worldsheet/worldvolume metric and the induced metric for extended objects embedded in spacetime as Riemannian manifolds, areal-metric manifolds, and volume-metric manifolds. For strings on a Riemannian spacetime, we consider general actions respecting volume-preserving diffeomorphisms (VPD), general diffeomorphisms, and diffeomorphisms with Weyl symmetry, respectively. Well-known Schild, Nambu-Goto, and Polyakov actions are included as special cases. We reach two main conclusions: (1) When actions are functions of both the worldsheet metric and induced metrics, all nontrivial self-consistent actions are classically equivalent. (2) As a physical constraint on the classical action, VPD symmetry is as strong as the full diffeomorphism symmetry. The discussion is then extended to strings in spacetime manifolds equipped with the areal or volume metrics. Then, we further consider higher-dimensional extended objects in spacetime defined with areal or volume metrics, and show the equivalence between the generalized Schild actions and the generalized Nambu-Goto action. We prove a general theorem on VPD that explains this equivalence. Incidentally, while only the areal metric is needed to define the string worldsheet action, we show that the Polyakov action with an areal-metric perturbation cannot describe critical strings without other interaction terms.

We perform a stability analysis of the homogeneous ground state of nuclear matter against inhomogeneous perturbations of the pion condensate. In $β$-equilibrium, restricting the baryon species to nucleons only, we observe no instability; however, at high densities, the pseudoscalar density-density correlations assume a moat regime, i.e. a damped oscillatory patterned spatial correlation, which in momentum space appears as a non-zero global minimum for some finite three-momentum. When hyperons are permitted to appear, this minimum can cross down to negative values, which configures an instability towards an inhomogeneous pion condensate which ultimately will affect the equation of state.

We initiate the lightcone bootstrap analysis of multipoint correlators in a defect conformal field theory. The setup we consider is the three-point function of two bulk and one defect operator. Requiring consistency of the crossing equation in the lightcone limit, we find constraints on the defect spectrum at large transverse spin. Specifically, to reproduce the exchange of the leading-twist operator in the bulk channel we find two new twist-accumulating families of defect operators at large transverse spin and we compute their defect CFT data in this limit.

The string axiverse predicts a unique connection between the high scales approachable only through theory and the low energies within reach of experimental verification: a multitude of light, feebly interacting axions. In order to capture the collective effects of such an axion ensemble, we model the string axiverse by $N$ coupled axions with a simple assumption: hierarchical axion masses that arise from hierarchical instantons with statistically distributed axion couplings. In this limit, we find that axion field ranges, which determine late-time cosmological abundances, shrink as $1/\sqrt{N}$ as the number of axions grows. Moreover, the heaviest modes tend to align with the smallest kinetic eigenvalues, further reducing their field ranges. Interactions with the Standard Model (SM) are largely set by the kinetic structure and do not grow with $N$, thus suppressing detection prospects relative to the individual-axion expectation. The exceptions are the ensemble's lightest and heaviest states as well as the Quantum Chromodynamics (QCD) axion, which incur no such suppression. We further find that coupled axiverse dark matter has parametrically relaxed tuning on initial conditions when produced via long, low-scale inflation relative to independent axions and high-scale inflation. Taken together, these results sharpen the observational outlook: the most accessible signals typically come from the QCD axion and from heavy axions that make up small dark matter subcomponents. An anthropic plateau of comparable energy density states produces subdominant signals; meanwhile, if light axions have SM interactions independent of QCD, they can also be within reach of future direct-detection experiments.

We investigate near-threshold top-antitop production at the LHC, focusing on the impact of toponium formation on spin correlations and quantum information properties of the final state. Considering the top-antitop system as a mixed two-qubit state, we reconstruct spin density matrices via quantum tomography and evaluate several observables including some inspired by quantum information. We then compare their sensitivity in discriminating toponium effects from top-antitop production without these effects. Our results demonstrate that combining these variables is expected to significantly enhance sensitivity to toponium effects, bringing new ways to explore these subtle features.

We propose the use of the super-binomial variance in the count rate of an atom interferometer as a novel signature of dark matter. We show that the dark matter induced shift in this observable is enhanced by N, the number of atoms used per run of the interferometer, and therefore offers sensitivity that is enhanced by orders of magnitude relative to an independent-atom estimate. As an application, we consider dark matter that interacts with electrons, protons, and/or neutrons, via a long-range Yukawa interaction and new constraints on strongly interacting dark matter that thermalizes in the overburden of conventional direct detection experiments. We find that searches for super-binomial variance extend, and complement, existing atom interferometer observables; they are well suited to search for both short- and long-ranged forces.

Dark matter direct detection experiments can observe solar neutrinos via coherent elastic neutrino-nucleus scattering, making it possible to test new physics in the neutrino sector. In this article, we study the sensitivity of RES-NOVA, a novel cryogenic calorimetric experiment employing PbWO$_4$ crystals grown from archaeological lead, to neutrino non-standard interactions (NSI). We perform a sensitivity study for a benchmark setup with a nominal energy threshold of 1 keV and an exposure of 1 ton$\cdot$y, both for a conservative (only heat readout) and ideal (heat and scintillation) background rejection scenario. We find that, in its nominal configuration, RES-NOVA can reach sensitivities to NSI at the level of current global fits. With moderate or significant improvements of the threshold down to 0.5 keV and 0.1 keV, RES-NOVA will be able to achieve sensitivities beyond NSI global fit results, testing new areas of the parameter space in the electron and tau sectors, $\varepsilon_{ee}$, $\varepsilon_{ττ}$, and $\varepsilon_{eτ}$. A similar improvement in sensitivities is expected when instead increasing the exposure to 10 ton$\cdot$y.

Hypothetical light scalar particles trigger the superradiant instability around spinning black holes (BHs), causing clouds of scalars to grow around the BH. In the presence of sufficiently strong particle self-interactions (characterized by the decay constant $f$), scalars are ejected from BH orbits, resulting in coherent, non-relativistic emissions that continuously carry away the BH's angular momentum. Parameters exist for which cloud growth is much faster, and scalar depletion is much slower, than the age of the Galaxy. This defines a distinct class of astrophysical sources of scalars, which we call \emph{BH scalar sirens} -- BHs that persistently emit scalars effectively forever. We compute the scalar background from the expected population of $N_\text{BH}\sim 10^{8}$ isolated stellar-mass BHs in the Milky Way, which are sirens for scalars in the mass range $10^{-13}$--$10^{-11}\,$eV and $f\lesssim 10^{14}$--$10^{9}\,$GeV. This provides a detection target independent of early-universe scalar production or cosmological initial conditions. The generated observable signals are up to two orders-of-magnitude larger than those expected from a misaligned cosmic scalar in this mass range. The energy spectrum of emitted scalars is distinctly broader and at higher velocities (up to $\sim 10^{-1}c$) than that of virialized dark matter, and encodes the mass and spin distributions of the BH population. While stellar-mass Milky Way BHs are our primary target, our framework extends to supermassive, intermediate-mass and light BHs. Given the difficulty of directly observing populations of isolated BHs, scalar emissions offer a novel probe of these otherwise invisible objects, highlighting the potential for joint discovery between scalars and BHs, and broadly motivating searches for scalars over many orders-of-magnitude in mass.

The SND@LHC experiment investigates neutrinos in the forward pseudorapidity range of $7.2 < η< 8.4$. The detector consists of a veto system, a scintillating fiber (SciFi) tracker interleaved with emulsion cloud chambers (ECCs), and a downstream muon system. Muons originating from collisions at ATLAS (IP1) constitute the primary background for CC neutrino interactions and determine the replacement frequency of the emulsion target. A precise characterization of this flux is therefore essential. In this work, we report the muon flux measured in the central $31 \times 31 \text{ cm}^2$ fiducial area of the detector using data from 2023 through 2025. The measured fluxes for proton collisions are: $(1.90 \pm 0.04) \times 10^{-2} \text{ nb/cm}^2$ (2023), $(3.74 \pm 0.06) \times 10^{-2} \text{ nb/cm}^2$ (2024), and $(2.48 \pm 0.04) \times 10^{-2} \text{ nb/cm}^2$ (2025). The measured fluxes for heavy-ion collisions are $(3.13 \pm 0.11) \times 10^4 \text{ nb/cm}^2$, $(5.54 \pm 0.17) \times 10^4 \text{ nb/cm}^2$, and $(3.60 \pm 0.13) \times 10^4 \text{ nb/cm}^2$ in 2023, 2024, and 2025, respectively. Uncertainties are dominated by systematic effects, with the statistical component contributing $\lesssim 1\%$ to the total uncertainty. These results are in agreement with Monte Carlo predictions.

We investigate the arrival time and the Faraday rotation of extragalactic electromagnetic signals from fast radio bursts (FRBs) propagating through chiral cosmic media within the framework of Maxwell-Carroll-Field-Jackiw (MCFJ) electrodynamics. By treating the interstellar medium as a cold, ionized chiral plasma, we derive the time delay between two traveling signals, expressing it in terms of modified dispersion measures (DMs) containing chiral contributions. The Faraday rotation angle is then written in terms of modified rotation measures (RMs). By combining the DMs and redshift data from a set of FRBs, we obtain constraints on the chiral parameter magnitude at the order of $10^{-26}$--$10^{-24}$ GeV. Using the Faraday rotation formulae and RM measurements, upper bounds as stringent as $10^{-43}$ GeV on the MCFJ parameters are also obtained.

As a characteristic property of all quantum systems, entanglement participates in many important quantum phenomena. In this proceeding, we employ it in the study of quantum field theories at finite density. We incorporate evaluations of entanglement entropy using the replica trick into MC simulations of $O(N)$ models at finite density with the worm algorithm and present some initial results for the nonlinear $O(4)$ model in 3 dimensions.

We report the first unambiguous detection of refractive substructure in an active galactic nucleus (AGN) using ground-based Very Long Baseline Interferometry (VLBI). Our analysis of TXS 2005+403 - observed at 1--5 GHz along a line of sight through the Cygnus region - reveals clear signatures of turbulence-induced substructure on long baselines that cannot be explained by the smooth scatter-broadened profile from diffractive effects alone. This signal persists across multiple observations spanning 2010-2019, demonstrating stable scattering properties along the line of sight. The combination of high flux density, compact intrinsic structure, and strong scattering establishes TXS 2005+403 as an exceptional laboratory for probing Galactic turbulence. This detection demonstrates that AGNs can serve as cosmic lighthouses illuminating interstellar plasma across the sky, complementing pulsar scintillation studies and informing scattering mitigation for millimeter-wavelength imaging of Sagittarius A*.

Being the most massive known young stellar cluster in the Milky Way, Westerlund 1 (Wd1) constitutes an ideal benchmark for understanding the evolution of massive stars. However, the cluster age remains highly controversial (~4-10 Myr), hindering the use of Wd1 as a reference for massive star evolution. One of the main issues is high foreground extinction, which has so far prevented the detection of the main sequence. Using infrared spectroscopy we seek to detect the cluster's main sequence for the first time, to characterise the Hertzsprung-Russell diagram, and to use the cluster's turn-off to obtain a robust age estimate. We obtained multi-epoch, near-infrared VLT/KMOS spectroscopic observations of Wd1 to map its population of massive stars. The spectra of ~110 members were analysed with CMFGEN models to derive stellar parameters, populate the cluster Hertzsprung-Russell diagram, and compare it with isochrones from evolutionary models. Our observations returned 47 new spectroscopically identified cluster members, with spectral types O9-B1 III-V. The cluster turn-off indicates an age of 5.5+/-1.0 Myr at a distance of 4.23+0.23-0.21 kpc, displaying a moderate degree of coevality. We demonstrate that our estimate of the age of Wd1 is robust against reasonable changes in the distance and extinction law, and the adopted rotational velocity and metallicity of the stellar isochrones. We further find that ~65% of the OB stars with multi-epoch coverage exhibit radial-velocity variability. Infrared observations of the unevolved stellar population support a single episode of star formation with an age of ~5.5 Myr, reinforcing its potential as a benchmark for massive star evolution and providing a reference sample for future binary population studies.

We present a broadband study of the WLQ SDSS J101353.45+492758.1, which displays a nearly featureless UV-optical spectrum with only a weak Mg II line alongside an exceptionally low X-ray flux. We model its spectral energy distribution using the relativistic thin-disk model kerrbb with a power law, and the multicomponent AGN model relagn, a physically motivated extension of agnsed incorporating warm and hot Comptonizing regions. Our fits constrain the black hole mass, accretion rate, X-ray loudness, and coronal energetics. Both approaches yield consistent BH masses of M_{BH} \approx 2 \times 10^{9} M_\odot and an Eddington accretion rate of \dot m \approx 0.1. The relagn fit including a warm Comptonizing region provides a significantly improved representation of the UV-soft X-ray continuum. The warm corona, characterized by kTe \simeq 0.20 keV, Γ \simeq 3.8, and an optical depth τ \simeq 7.26, extends to \sim 34 R_g. The hot corona appears compact and energetically suppressed, leading to an intrinsically weak X ray output with log(L_{X}/L_{bol}) \simeq -4.29, among the lowest reported for WLQs. The α_{ox} \sim 2.06 indicates the source to be in high/soft AGN spectral state. The combination of a luminous, standard disk and extremely weak hot corona suggests that this quasar hosts a highly inefficient inner coronal region. This explains its X-ray faintness and extreme deficit of high-ionization emission lines. The source may represent an AGN analog in "ultrasoft" accretion state, or a system in which the ionizing continuum is suppressed by a compact or quenched corona. Our study suggests that the source is not accreting at high Eddington ratio, highlighting the physical diversity of WLQs, and supports the view that geometric and radiative effects jointly shape their extreme spectral properties.

One key piece of evidence for dark matter is the rotation-curve problem: the disagreement between measured galactic rotation curves and their luminous mass. A novel solution to this problem is presented here, in a model that predicts observed Doppler-shifted spectra based only on the luminous matter estimates and one free model parameter. This model is applied to fit the rotation curves of the SPARC sample of 175 galaxies, yielding mass-to-light ratios, goodness of fit measurements, and the free parameter. The model's average reduced chi square compares favorably with the dark matter model for the same data, and more galaxies are successfully fit by this model. The model provides a useful formulation linking luminous matter to the observed rotation curves, with the dark matter contribution to galaxies encoded in two transformation terms of the luminous mass. It also offers a lower-parameter characterization of the rotation curve problem, and a power law relationship between the model's free parameter and galactic photometric quantities is observed, potentially removing the need for the free parameter.

Shock waves generated during cluster mergers offer a powerful probe of how large-scale structure grows and evolves in the Universe. As part of the MeerKAT-South Pole Telescope (SPT) survey, we report the discovery of a single arc-like radio relic in the galaxy cluster Abell 4067 ($z=0.099$), one of the lowest-mass clusters known to host such a structure. MeerKAT UHF-band (0.58--1.09 GHz) observations reveal a relic with a largest linear size of $\sim 1.48 \pm 0.02$ Mpc, located at a projected distance of 0.95 Mpc from the cluster centre. XMM-Newton X-ray data show that the relic's position and orientation relative to the intracluster medium (ICM) elongation are consistent with a merger-driven shock-wave scenario. The relic has an estimated radio power of $3.10 \pm 0.03 \times 10^{24}$ W Hz$^{-1}$ at 150 MHz. When placed in the $P_{150\,\mathrm{MHz}}$--$M_{500}$ scaling relation, the Abell 4067 relic appears less luminous compared to relics in more massive clusters, suggesting an association with weak merger shocks. This finding supports the idea that relics in low-mass clusters may form through less energetic merger events, leading to weak merger shocks. This is further supported by the absence of a detectable central radio halo in Abell 4067, reinforcing the idea that luminous radio halos are not a universal outcome of cluster mergers and highlighting the role of cluster mass, merger energetics, and evolutionary stage in shaping diffuse radio emission in the intracluster medium.

We forecast that the Terahertz Intensity Mapper (TIM) cross-correlated with Euclid's Fornax deep field (EDF-F), TIM$\times$EDF-F, will detect the [CII]-galaxy cross-power spectrum at a median redshift of 1.1 with $\gtrsim 7 σ$ confidence. The Poisson component of the cross-power spectrum at $0.1 \leq k \leq 10$ hMpc$^{-1}$ (i.e. cross-shot noise) will be detected at $\gtrsim 3 σ$ in 4 bins spanning $0.5 < z< 1.7$. This measurement will constrain the mean [CII] specific intensity over half of cosmic history and assess the degree to which Euclid-selected galaxies account for the [CII] intensity observed by TIM. We find that TIM can detect the cross-power spectrum across a wide range of [CII] intensity models.

We present a detailed study of HUDFJ0332.4-2746.6 (HUDF46), a $z \approx 1.84$ overdensity in the Hubble Ultra Deep Field, previously identified with HST as a proto-cluster. JWST/NIRISS spectroscopy expands its confirmed membership from 18 to 37 galaxies, while deep HST/ACS, JWST/NIRCam, and JWST/MIRI imaging provide a comprehensive multiwavelength view from the rest-frame UV to the mid-infrared. This dataset probes the population across three dex in stellar mass ($M_\bigstar \approx 10^{7.5\text{--}10.5}\,M_\odot$), delivering the first direct view of a young cluster down to such low-$M_\bigstar$ at $z\gtrsim1$. Assuming virialization, we derive a velocity dispersion of $σ\approx 670\pm 91\,\mathrm{km\,s^{-1}}$ and a halo mass of $M_{200} \approx (1.2\pm0.2) \times 10^{14}\,M_\odot$, in agreement with X-ray constraints from deep {\it Chandra} data. Despite residing in a massive halo likely in the hot-halo regime, the population is overwhelmingly star-forming, with no established red sequence and no extended X-ray emission from a hot intracluster medium. HUDF46 members have stellar and structural properties nearly indistinguishable from coeval field galaxies, and the structure hosts only one AGN candidate, found in its brightest galaxy, which lies at the cluster center. Overall, HUDF46 appears to be in a transitional phase prior to the onset of environmental quenching, making its galaxy population a key benchmark for tracing the processes that will later build a passive population and shape the assembly of massive clusters at later cosmic times.

Going beyond second order in weak lensing (WL) statistics is known to break degeneracies among cosmological parameters. We take a step further here, investigating whether higher-order statistics (HOS) in weak lensing can disentangle among General Relativity (GR) and modified gravity (MG), also taking into account the presence of massive neutrinos. To this end, we rely on mock convergence maps obtained from GR and $f(R)$ gravity N - body simulations, and we look for MG signatures in a wide set of higher-order WL probes. We rely on different metrics to quantify the discriminatory power of each probe, also varying the measurement setup. We find out that WL HOS can indeed disentangle MG and GR also in the presence of massive neutrinos.

Across the peak of cosmic star formation at $z\sim1-2$, inflow, processing, and feedback drive rapid changes in the spatial distribution and chemical composition of baryons in galaxies and surrounding reservoirs; this baryon cycle can be tomographically mapped by line intensity mapping (LIM) of atomic hydrogen, ionized carbon, and carbon monoxide. We present a simulation-based forecasting framework for detecting auto- and cross-power spectra between spectroscopic surveys of four such tracers at $z\sim0.5-1.7$ mapping the same deep field - TIM, EoRSpec/FYST, MeerKAT, & Euclid. We forward-model 3-D distributions for these tracers from magnetohydrodynamic simulations, directly capturing the two-halo, one-halo, and shot statistics without relying on analytical decompositions. We further detail a signal-to-noise formalism, tailored to LIM surveys with highly anisotropic geometries and Fourier-space coverage. We demonstrate that galaxy cross-correlations will be the dominant discovery channel for current-generation surveys. These instruments will detect the auto-spectra for CO and HI 21 cm and the CO $\times$ 21 cm cross-spectrum at modest S/N $\sim 1-10$, while placing upper limits on the [CII]-158$μ$m signals. [CII], CO, and HI LIM will be $\sim3-30\times$ ($0.5-1.5$ dex) more sensitive to cross-correlation with the Euclid survey, however, than their respective auto-correlations, constraining all three models of line emission at high significance (S/N $\sim 10-40$) within this decade. Finally, we formulate a staged instrumental trajectory with planned or reasonable improvements, including the as-proposed SKA-Mid. We forecast advancing the per-$k$-mode sensitivities of each auto-, galaxy-line, and line-line spectrum by several orders of magnitude, enabling new percent- and sub-percent level constraints on cosmology and the redshift evolution of star formation and the baryon cycle.

The reionization process is driven by ionizing photons from dwarf galaxies in halos with virial temperature $T_{\rm vir} \gtrsim 10^4$ K, while minihalos whose $T_{\rm vir}\lesssim 10^4$ K consume ionizing photons and have negative contributions to reionization. Since ionizing sources and minihalos have different clustering characteristics, not only the reionization history, but also the morphology of the ionization field, is sensitive to the small-scale power spectrum. If the power spectrum at small scales is enhanced compared with the standard six-parameter $Λ$CDM model, then both the sources and sinks of ionizing photons would be boosted and the net impact depends on the competition between them. Therefore, the 21 cm signal that can probe the morphology of the ionization field will be a useful tool for detecting the small-scale power spectrum. Using the power spectrum proposed by Cielo et al. (2025) (C25) as a demonstration, we investigate the influence of small-scale power spectrum enhancement on the ionization field and the 21 cm signal. We find that for the C25 model, even under the constraints of observed UV luminosity functions for high-$z$ galaxies and reionization history, the 21~cm power spectrum and the bubble size distribution could be still significantly different from the regular $Λ$CDM model. The upcoming SKA-low AA* telescope, and a further imaging telescope, have the potential to detect the small-scale power spectrum more deeply.

Relativistic jets from Active Galactic Nuclei (AGN) are highly energetic and emit radiation across a wide range of frequencies. Despite several observational studies, their particle composition still remains a key open question. The detection of high-energy neutrinos from blazar sources such as TXS 0506+056 has highlighted the plausibility of hadronic/lepto-hadronic models for AGN jets. To understand the origin of high-energy neutrinos from such sources, it is imperative to capture the complex interplay between the jet dynamics, their composition, and the mechanism of particle acceleration and cooling in relativistic jets. In this pilot study, we have coupled a numerical multi-zone framework for lepto-hadronic modeling, with 3D relativistic magneto-hydrodynamic simulations of AGN jets, including external photon fields. Our framework provides synthetic multi-wavelength and neutrino flux by spatially sampling the simulated jet into multiple zones. We investigate the implications of such a framework in exploring the different intrinsic and extrinsic pathways for proton-enrichment in jets. Essentially, we find that for low proton-to-electron number density ratios, producing a substantial jet neutrino flux, requires the underlying proton energy distribution to have a relatively flat spectrum with a power-law index of $\simeq 2.0$. We further find that while intrinsic shocks triggered by kink-instabilities in the jet can accelerate electrons to high energies, they may not be sufficient to produce such flat particle energy distributions for the chosen set of parsec-scale jet parameters. Finally, to produce a significant jet neutrino emission, our simulations suggest the need to consider particle acceleration mechanisms through alternative pathways, either internal or external.

The origin of the X-ray plateau phase observed in a large fraction of gamma-ray burst afterglows remains debated. We present a novel analytic framework for the hydrodynamics of ultra-relativistic, radially stratified outflows interacting with an external medium. By explicitly accounting for a continuous distribution of Lorentz factors within the ejecta, we derive analytic expressions describing the evolution of a long-lived, mildly relativistic reverse shock and determine its crossing time. Then, we compute the resulting synchrotron emission from both the forward and reverse shocks. The forward shock naturally produces a shallow, long-lasting X-ray decay consistent with the observed properties of X-ray plateaus (including the Dainotti relation). We further show that reproducing the observed plateau durations requires the stratified ejecta to extend to Lorentz factors of order $\gtrsim 100$, consistent with the ultra-relativistic outflow that powers the prompt $γ$-ray emission. The reverse shock generates a long-lived millimeter emission component that outshines the forward shock emission at these wavelengths. Both the plateau and reverse shock emission terminate smoothly once the slowest ejecta are processed, marking a transition to the standard Blandford-McKee self-similar evolution without requiring late-time energy injection or an additional emission component. Such stratified outflows are expected on physical grounds, as the ultra-relativistic ejecta responsible for the prompt $γ$-ray emission are unlikely to be launched with a single Lorentz factor. This model provides a unified picture in which the same outflow powers the prompt emission, the X-ray plateau, and the subsequent afterglow evolution.

Wide binaries (WBs) offer a unique opportunity to test gravity in the low-acceleration regime, where modifications such as Milgromian dynamics (MOND) predict measurable deviations from Newtonian gravity. We construct a rigorous framework for conducting the wide binary test (WBT), emphasizing high quality sample selection, filtering of poor astrometric solutions, contamination mitigation, and uncertainty propagation. We show that undetected close binaries, chance alignments, and improper treatment of projection effects can mimic MOND-like signals. We introduce a checklist of best practices to identify and avoid these pitfalls. Applying this framework to Gaia DR3 data, we compile a high-purity sample of WBs within 130 pc with projected separations of 1 - 30 kAU, spanning the transition between the Newtonian and MOND regimes. We find that the scaled relative velocity distribution of wide binaries does not exhibit the 20% enhancement expected from MOND and is consistent with Newtonian gravity across all separations. A meta-analysis of previous WBTs shows that apparent MOND signals diminish as methodological rigour improves. We conclude that when stringent quality controls are applied, there is no observational evidence for MOND-induced velocity boosts in wide binaries. Our results place strong empirical constraints on modified gravity theories operating between a0/10 and 200 a0, where a0 is the MOND acceleration scale. Across this range of internal accelerations, Newtonian gravity is up to 1500x more likely than MOND for our cleanest sample.

This work investigates symbolic regression (SR) as an interpretable alternative to black-box machine learning for the classification of stars, galaxies, and quasars in the Sloan Digital Sky Survey Data Release 17 (SDSS DR17). We conduct a systematic comparative study of four state-of-the-art SR frameworks: {\tt PySR}, Exhaustive Symbolic Regression ({\tt ESR}) with MDL-based selection, Physical Symbolic Optimization ({\tt PhySO}) using deep reinforcement learning, and Multi-View Symbolic Regression ({\tt MvSR}). By deriving compact analytic functions (complexity $\leq 10$) on a representative training subset and subsequently evaluating them via a 5-fold stratified cross-validation protocol on 100,000 spectroscopically confirmed objects, we map spectroscopic redshift ($z$) to continuous classification scores. Our results demonstrate that these low-complexity expressions achieve high predictive reliability, with {\tt MvSR} reaching a Cohen's Kappa of 0.8948 and {\tt PhySO} achieving exceptional parametric stability ($σ< 0.002$). We show that these models not only match the performance of traditional baselines but also provide a transparent, mathematically concise characterization of the astrophysical boundaries separating galactic and extragalactic populations.

The ionization mechanisms of low-ionization nuclear emission-line regions (LINERs), which are common in the local Universe, have been debated for decades. Our nearest large neighbor, M31, is classified as a LINER based on its optical emission line properties within the central kpc. In this work, we present a detailed photoionization modeling of the circumnuclear ionized gas in M31, explicitly tailored to its well-constrained physical conditions, including the absence of ongoing star formation and a currently inactive active galactic nucleus (AGN). Using spatially resolved CFHT/SITELLE observations, we find that photoionization by hot, evolved low-mass stars distributed throughout the bulge can roughly reproduce the observed radial intensity profiles of Hα, H\b{eta}, and [NII]. However, these models fail to match the observed [OIII] emission, producing radial profiles and [O III]/H\b{eta} ratios that are significantly steeper than observed. This discrepancy indicates a deficit of high-energy ionizing photons in standard stellar photoionization models, even with extended ionizing sources. We explore whether this tension can be alleviated by invoking either a bulge-filling, low-density ionized medium surrounding a denser Hα-emitting disk, or enhanced AGN activity in the recent past. While both scenarios can partially increase the [O III] emission, neither provides a fully satisfactory explanation under physically plausible conditions. Together with our earlier results for M81, these findings underscore persistent challenges in explaining LINER-like emission solely through conventional photoionization mechanisms.

We use the galaxy ultraviolet luminosity function measurements at $z=6-9$ to constrain modification to standard inflationary power spectrum. These observables are sensitive to the matter power spectrum which itself depends on inflationary initial conditions. We consider specific models where a bump or oscillatory features are introduced to the standard power law inflation spectrum. We find that the galaxy luminosity observables can probe such modifications at wavenumbers $0.5\lesssim k \lesssim 20$ Mpc$^{-1}$. We obtain upper limits on the amplitude of bump-like features at the mentioned wavenumbers. We obtain constraints which are similar to previous constraints on these models using measurements of optical depth of reionization. However, the galaxy luminosity functions are a more direct probe for these type of models and, therefore, can complement indirect constraints coming from measurements of IGM properties.

The Milky Way's nuclear star cluster (NSC) is a unique laboratory to study the formation and evolution of dense stellar systems around a supermassive black hole. Previous work suggests that most stars in the NSC are old; however, the detailed age and metallicity distributions remain uncertain. We constrain the star formation history (SFH) and metallicity of a poorly explored region located $\sim$3 pc from SagittariusA*. We analyse VLT/NACO imaging in an intermediate-band filter centred at 2.24 $μ$m, complemented by $H$-band data. We construct completeness-corrected $K$-band luminosity functions (LFs), clearly identifying the Red Clump and Red Giant Branch Bumps. The SFH is derived by fitting cumulative LFs with MIST, PARSEC, and BaSTI models spanning a wide range of ages and metallicities, using Monte Carlo sampling to estimate uncertainties. Metallicity constraints are refined using spectroscopic measurements from the literature. The stellar population is predominantly old and metal-rich: $75.6 \pm 9.5$% of the stellar mass formed $\gtrsim 10$ Gyr ago, with median [M/H] $\sim +0.35$. An intermediate-age component at 2-3 Gyr contributes $20.8 \pm 8.7$%, while minor populations are present at $\sim$400 Myr ($0.9 \pm 0.8$%) and 20 Myr ($3.6 \pm 1.4$%), the latter representing a small but non-negligible young population. Systematic uncertainties from stellar models, binning, photometric range, unresolved binaries, and filter choice are assessed. These results indicate early dominant formation, a significant 2-3 Gyr episode, and minor recent activity, consistent with spectroscopic measurements and with properties of the inner NSC and nuclear stellar disc.

Over the past decade, runaway stars have been identified, believed to originate either as surviving donors of Type Ia supernovae or as partially deflagrated accretors producing Type Iax supernovae. While the former have been extensively studied recently, the origins of the latter (also called LP 40-365 type stars) remain under-explored and therefore less well understood. So far seven such objects are known. In this paper, we report the discovery of a new LP 40-365 type runaway star, notably hotter than previously studied members of this class. Spectral analysis confirms that its atmosphere is neon- and oxygen-dominated, consistent with earlier analyses of other LP 40-365 type stars. Kinematic analysis indicates that the star has a high probability of being unbound from the Galaxy and was most likely ejected from the Galactic disk approximately 2.8 Myr ago with an ejection velocity exceeding 600 km/s. This result further emphasizes the discrepancy between the abundance yields and kick velocities predicted by white dwarf deflagration models and those observed in stars of LP 40-365 type, underscoring the need for a reassessment of such systems.

We present a timing and scintillation study of the young Galactic halo pulsar PSR J1740+1000 using observations from the Nanshan, FAST, and Parkes radio telescopes. From timing analysis, we measure the pulsar's proper motion for the first time, indicating motion away from the Galactic plane at a position angle of 16.7 +/- 4.8 degrees (Galactic coordinates), with a total proper motion of 56.9 +/- 8.0 mas/yr and a corresponding transverse velocity of 329 +/- 80 km/s. This velocity suggests that PSR J1740+1000 is a typical-velocity young pulsar born within the Galactic halo. In scintillation studies, we detect scintillation arcs, arclets, and double-layered adjacent arcs in the secondary spectra. Under isotropic and anisotropic scattering assumptions, the screen-to-pulsar distance is 370 +/- 72 pc and 1 +/- 12 pc, respectively. The latter closely matches the scale of the pulsar wind nebula associated with PSR J1740+1000 and provides a better fit, suggesting that scattering is likely dominated by the PWN. The double-layered adjacent arcs observed on MJD 60180 imply that the pulsar's scattered image consists of two dominant components (A and B) and multiple weaker components. Component A is located at the pulsar's geometric position (0 uas), while Component B is located 112 +/- 16 uas and 23 +/- 17 uas from the central component under isotropic and anisotropic scattering, respectively. The frequency-independent angular position of Component B hints at refraction by an AU-scale structure within the scattering region, possibly originating from the PWN.

Observations of linear structure connecting pulsars to gamma-ray halos reveal injection of TeV electrons into the interstellar medium (ISM). In some cases, this could be attributed to nearly scattering-free electron transport along large-scale magnetic fields connected to pulsar winds with very slow diffusion across the field lines. In this work we model this process with a magnetic flux tube emerging from the pulsar and attached to the ISM magnetic field. We show that particles in this case have an anisotropic distribution of magnetic pitch angle, such that the overall emission is highly beamed. We apply this model to pulsar tails and filaments, including the extended X-ray and TeV emission associated with PSR J1740+1000 and the misaligned X-ray jet in the Guitar Nebula, to constrain their particle population and magnetic fields.

When a dark matter halo falls into a more massive object and becomes a subhalo, it typically loses much of its mass through tidal stripping. The reverse process is also possible in principle. The subhalo may gravitationally capture material from its host. If sufficiently efficient, this process could make an initially starless subhalo visible. We use high-resolution N-body simulations to estimate the efficiency of capture. We find that after an extended period orbiting within its host, at most $\sim 10^{-4}$ of a subhalo's remaining mass has been acquired since infall. This captured material is less concentrated to subhalo centre than material retained from before infall. It is also very much less abundant than host material that is instantaneously passing through the subhalo on almost unperturbed orbits. Captured stars are not sufficiently spatially concentrated to be distinguished from the dominant background of "field" stars, and their concentration in velocity space is no greater than that of typical stellar streams in the halo. Unfortunately, stellar capture is not efficient enough to allow initially starless low-mass subhaloes to be detected.

Stellar winds interact with the partially ionized interstellar medium (ISM), forming astrospheres. A key feature of this interaction is the hydrogen wall - secondary interstellar atoms produced via charge exchange near the tangential discontinuity separating the stellar wind from the ionized ISM component. This secondary component is decelerated and heated compared to primary interstellar hydrogen, making the hydrogen wall detectable through Lyman-$α$ absorption spectra toward nearby stars. Such structures have been observed by the Hubble Space Telescope around the Sun and other stars. In this paper, we propose that another feature of the stellar wind/partially ionized ISM interaction may also be detectable in Lyman-$α$ spectra: the neutral stellar wind. It forms via charge exchange between supersonic stellar wind protons and interstellar atoms penetrating deep into the astrosphere due to their long mean free paths. We present a parametric numerical analysis of astrospheric structures and their synthetic Lyman-$α$ absorption spectra. Using a 2D kinetic-hydrodynamic model, we vary the terminal wind velocity while maintaining constant dynamic pressure to keep the astrosphere size consistent. For winds slower than the solar wind (terminal velocities $V_0 \lesssim 200$ km/s), charge exchange efficiency in the supersonic region increases dramatically, producing a distinct and observable absorption feature from the neutral wind. This signature is negligible for solar-like winds ($V_0 \approx 400$ km/s) but emerges as a direct spectroscopic diagnostic for winds up to $\sim 200$ km/s. Detecting this neutral wind absorption offers a novel method to directly constrain stellar wind velocities.

Using the ZeroDegree multi-reflection time-of-flight mass spectrograph of the CRISMASS project at RIKEN Radioactive Isotope Beam Factory, we performed high-precision mass measurements of proton-rich nuclei near the doubly magic nucleus 100Sn, achieving uncertainties on the order of 10 keV. The masses of 91Rh, 92Pd, and 96Cd were determined for the first time with high precision, and the accuracy of several additional masses was substantially improved. Incorporating the new data into X-ray burst simulations significantly reduces the abundance uncertainties in the $A$ = 90-100 region, shifting the reaction flow toward $A$ = 90 production and suppressing the synthesis of heavier nuclei. Further investigation of the $νp$-process indicates that 99Rh plays a significant role in the reaction flow within the mass region studied. These high-precision mass measurements refine the mass surface near 100Sn and provide critical constraints on models of proton-rich nucleosynthesis.

Accurate spatially resolved dust corrections are critical for interpreting the structure and evolution of star-forming galaxies (SFGs). We present an empirical model for predicting spatially resolved dust attenuation ($A_V$) in SFGs using integral field spectroscopy from the Mapping Nearby Galaxies at Apache Point Observatory (MaNGA) survey. Using a sample of 5,155 galaxies over $7.20<M_\ast<11.14$ and $0.0002 < z < 0.1444$, we derive $A_V$ maps from the Balmer decrement across more than 1,898,954 star-forming spaxels. Using local star formation rate surface density ($Σ_{\text{SFR}}$) as a predictor, the model achieves $R^2 = 0.69$ and RMSE $=0.22$ mag, with residuals that are approximately Gaussian and centred near zero. It predicts $A_V$ within a factor of $\sim$1.3 on kpc scales. We also demonstrate that the relation can be applied iteratively to recover dust-corrected $Σ_{\mathrm{SFR}}$ from uncorrected values, converging by the fourth iteration with minimal residual bias ($-0.01$ mag) and low RMSE ($0.42$ mag). The model accurately reproduces $A_V$ maps across diverse morphologies and orientations, including edge-on systems. It also recovers the observed radial $A_V$ profiles, capturing their dependence on stellar mass and relative star formation activity, with more massive and more strongly star-forming galaxies showing steeper gradients.

The circumgalactic medium (CGM) surrounding galaxies is believed to be a significant reservoir of baryons, yet its total mass remains poorly constrained. We present a novel approach to probe the CGM of the Andromeda galaxy (M31) using fast radio bursts (FRBs) from the CHIME/FRB Catalog 2. By comparing the dispersion measures (DMs) of 171 FRBs whose sightlines intersect M31's halo (within $r_{\rm vir}= 302\,\mathrm{kpc}$) to a control sample of 684 FRBs, we estimate the DM contribution from M31's CGM. We find evidence for an excess DM of $δ\mathrm{DM} = 5.9$--$59.6\,\mathrm{pc\,cm^{-3}}$ in the inner halo ($\approx 0$--$151\,\mathrm{kpc}$) and $δ\mathrm{DM} = 25.7$--$64.6\,\mathrm{pc\,cm^{-3}}$ in the outer halo ($\approx 151$--$302\,\mathrm{kpc}$). Using a generalized halo model parameterized by a closure radius $r_{\mathrm{close}}$, we constrain the baryon distribution and infer a best-fit value for $r_{\mathrm{close}}$ to be $9.2^{+9.9}_{-4.9}\,r_{\rm vir}$ ($1σ$), corresponding to a total CGM mass of $M_{b,\mathrm{halo}} = 18.6^{+7.9}_{-8.4} \times 10^{10}\,M_\odot$. Our results suggest that M31 may harbor a substantial fraction of its cosmic baryon budget in diffuse, ionized gas. This work demonstrates the potential of FRBs as a powerful tool for studying the CGM of nearby galaxies, with future larger samples expected to provide tighter constraints on the baryon content of galactic halos.

Pulse-profile modeling of rotation-powered millisecond pulsars targeted by NICER has enabled mass--radius constraints of several neutron star sources, with implications for the dense-matter equation of state. For the bright isolated pulsar PSR J0030+0451, the inferred mass--radius was previously found to depend strongly on the assumed hot spot model. These hot-spot models yielded different mass--radius constraints, with the statistically preferred model exhibiting some mild tension with results inferred for PSR J0437$-$4715, PSR~J0614$-$3329, and GW170817. We present an updated pulse-profile analysis of PSR J0030+0451 using new NICER observations obtained between 2017 July to 2023 January, increasing the number of X-ray counts by about 50% compared to previous analyses. We jointly analyze the NICER data with archival XMM-Newton observations to better constrain the source spectrum and background. The new analysis significantly reduces the discrepancy between the hot spot models. The inferred mass and radius are $M = 1.43^{+0.20}_{-0.17}\,M_\odot$ and $R_{\rm eq} = 12.68^{+1.31}_{-1.04}$ km (68% credible intervals), reducing the tension with the results from other sources. In addition, the inferred hot spot configurations suggest the presence of intra-spot temperature gradients.

We present an investigation of the full width at half maximum (FWHM, or γ) of absorption features of Type Ia supernova (SNe Ia). We found that, the average value of FWHM can be well predicted with the rest wavelength (λ). The velocity also plays an important role, as objects with a higher velocity tend to have a larger FWHM. Temperature may be the third factor, as we found that, at the same velocity (but different phases), a normal-velocity (NV) object tends to have a larger FWHM than high-velocity (HV) object. Also, 1991T/1999aa-like objects that are believed to have relatively high temperatures show the largest FWHMs if compared at the same velocity. Generally speaking, FWHM evolves very slowly with time and shows no correlation with Δm15, but 1991T/1999aa-like objects are characterized by relatively fast decreasing FWHM. On the other hand, we found that, objects with relatively small FWHMs shows a tighter correlation between absorption depth (A) and Δm15, possibly a sign of higher degree of homogeneity. We also found that A/γ of Si II λ5972 has a strong correlation with Δm15, and more importantly, a relatively slow time evolution, making it a useful luminosity estimator even in the absence of phase information.

X-ray timing of active galactic nuclei (AGN) provides a unique probe of gas accretion onto supermassive black holes (SMBHs). Quasi-periodic oscillations (QPOs), which trace gas dynamics in the strongly curved spacetime around SMBHs, are rare in AGN. These signals often are analogs of high-frequency QPOs occasionally seen in some black-hole X-ray binaries, and their scarcity in AGN can partly be attributed to the low frequencies expected for typical SMBH masses. Intriguingly, robust X-ray QPO detections in SMBH systems have so far been reported only in narrow-line Seyfert 1 galaxies (NLS1s) and tidal disruption events (TDEs). Here we report the discovery of a QPO candidate during the 2018 outburst of the changing-look AGN (CL-AGN) NGC 1566. Numerical simulations indicate that the disk epicyclic oscillations responsible for high-frequency QPOs are damped by magnetohydrodynamic turbulence unless the accretion flow is misaligned and/or eccentric. In TDEs, the stellar debris stream is naturally misaligned with the SMBH spin, while NLS1s may host misaligned disks due to their youth. Motivated by the QPO candidate in NGC 1566, we propose that CL-AGN accretion is also misaligned -- potentially fueled by captured, free-falling broad-line region clouds. This model naturally explains why CL-AGN transition timescales are much shorter than the standard disk viscous timescale. This picture can be tested by searching for QPOs or quasi-periodic eruptions in other CL-AGN.

We investigate the impact of a Coronal Mass Ejection (CME) on the transport and acceleration of relativistic protons in the solar wind using a coupled 3D Magnetohydrodynamics (MHD) simulation and a test-particle approach. The CME is driven by a spheromak injected into a Parker solar wind at a heliocentric distance of 0.139 AU. The trajectories of 5 GeV protons, injected towards the CME from 3 AU, are integrated in the guiding-centre approximation limit and scattered in velocity space with a mean free path $λ_{\|}$. Our results show that the CME can increment the protons energy by several GeV. The acceleration occurs during the time particles stream along the portion of a magnetic field line subject to compression downstream of the quasi-perpendicular portion of the CME-driven shock. In our configuration, the maximum energy gain, which is of the order of a few percent per shock crossing, occurs when the shock approaches 0.3 AU. Large energy gains require multiple passes through the acceleration region, which is made possible by the combined action of the mirror force and pitch angle scattering. The efficiency of the acceleration on time scales of the order of hours scales as $λ_{\|}^{-3/2}$. Energy spectra harden for decreasing parallel mean free path $λ_{\|}$.

We present a comprehensive search for non-thermal high-energy $γ$-ray emission from the nearby merging galaxy cluster Abell~119 ($z=0.044$) using archival observations spanning over seven decades in photon energy. Our analysis combines hard X-ray data from INTEGRAL/ISGRI (30--100~keV), MeV $γ$-ray observations from COMPTEL (0.75--30~MeV), and GeV--TeV data from the DArk Matter Particle Explorer (DAMPE; 3~GeV--1~TeV). No statistically significant emission is detected at the cluster position in any energy band. In the hard X-ray regime, we derive a $3σ$ upper limit of $F_{30-100\,\mathrm{keV}} \lesssim 8.5 \times 10^{-11}$~erg~cm$^{-2}$~s$^{-1}$ from ISGRI mosaic imaging. Reanalysis of archival COMPTEL data yields 95\% confidence-level upper limits ranging from $\sim 9 \times 10^{-11}$ to $\sim 8 \times 10^{-10}$~erg~cm$^{-2}$~s$^{-1}$ across 0.75--30~MeV. In the GeV--TeV range, DAMPE constrains the differential energy flux to $\sim 10^{-12}$--$10^{-10}$~erg~cm$^{-2}$~s$^{-1}$ (95\% confidence level). These results provide independent multi-band constraints on the reported GeV excess from recent Fermi-LAT studies. While our DAMPE limits do not exclude the flux levels claimed in those analyses, the absence of confirmation across keV--MeV--GeV bands indicate that any non-thermal emission from Abell~119 remains tentative.

Using the Spitzer and WISE images, we discovered 42 mid-IR luminous dusty supernovae with local integral-field spectroscopy data. The observed mid-IR emission indicates the presence of newly formed dust, or pre-existing dust heated by the radiation from the supernovae or circumstellar medium interactions. We carried out a systematic analysis of the supernova host environments and their dust properties, for understanding the dust-veiled exploding stars, and whether such an intense dust production process is associated with their local environments. We find that dusty supernovae prefer the locations with higher EW(Hα), lower metallicity, and heavier host extinctions compared to typical SN types, and they show the same increasing sequence in the values of EW(Hα) and oxygen abundance from hydrogen-rich, type IIn and hydrogen-poor dusty supernovae. These differences in environmental properties of different dusty SN types indicate the diversity of their progenitors. We also found that one marginal correlation is a negative correlation between the SN dust mass and star formation rate. This means that SNe would be more mid-IR luminous and more dust-rich at the region with lower star formation rate. However, the SN dust mass show no correlation with the metallicity and the host extinction, which were thought to be key factors affecting the mass-loss history of progenitors and the CSM environment of SNe. Therefore, the dust formation process in SNe might be insensitive to metallicity and the dust condition of their host environments.

The "knee" of cosmic ray spectra may reflect the maximum energy accelerated by galactic cosmic ray sources or the limit of the galaxy's ability to bind cosmic rays. Measurements of individual energy spectra are a crucial tool to understand the origin of the knee. Energy reconstruction and composition identification are foundations of the individual energy spectra measurements. One of the main scientific goals of Large High Altitude Air Shower Observatory (LHAASO) is measuring the cosmic ray energy spectra and composition from ~10 TeV to ~EeV. In this work, a novel method for reconstructing energy and logarithm mass (lnA) based on a superposition model is introduced. Energy and lnA are reconstructed using two universal, composition- and energy-independent calibration lines. For zenith angle below 40 degree, the energy and lnA biases are within +-5% and +-0.3, respectively, across all compositions. The method uses particle densities-measured by LHAASO's electromagnetic and muon detectors at a fixed distance from the shower axis-rather than integrated particle counts in annular bands. The density-based approach improves resolution for both energy and lnA, especially for heavy nuclei. The resulting energy resolution ranges from below 5% to ~15% above 1 PeV, the best mass resolution for iron achieved is below 25% above 10 PeV. The hadronic model dependencies of energy and lnA are also reported. These dependencies scale with lg(E/A) and are nearly independent of primary composition.

Interstellar objects provide a unique view into the formation of other star systems. Here we present spectroscopic observations of the recently discovered interstellar object 3I/ATLAS between a heliocentric distance of $3.7$ to $1.8$~au on either side of its travels through perihelion. We obtained several observations with the Keck-I/LRIS, Keck-II/NIRES, Gemini/GMOS, and UH88/SNIFS spectrographs, covering a wavelength range of $0.3 - 2.5~\mathrm{μm}$. We report the continued emission of both Ni and CN, along with post-perihelion detections of Fe and a weak detection of $\mathrm{C_3}$. We determine the spectral slope across optical and NIR wavelengths and find a positive spectral slope in the optical, with values ranging from $\sim 21 - 27\%$ in the blue regions ($0.4 - 0.55~\mathrm{μm}$) to $\sim 6 - 10\%$ in the red ($0.65 - 0.9~\mathrm{μm}$) regions. In contrast, the NIR showed a negative spectral slope of $\sim -0.9 \%$ between $0.9 - 1.5~\mathrm{μm}$ and $\sim -2.3\%$ between $1.9 - 2.5~\mathrm{μm}$. 3I/ATLAS shows a clear turnover in its spectral shape at $\sim 1.1~\mathrm{μm}$, corresponding to scattered light from the dusty coma. Finally, in the NIR, we do not find an increase in the depth of the water features identified in an earlier NIR observation of 3I/ATLAS. Our observations of 3I/ATLAS in the NIR show a similar shape to the NIR spectrum of 2I/Borisov as it approached perihelion.

Monte Carlo simulations were conducted using the Particle and Heavy Ion Transport code System (PHITS) to investigate the role of secondary cosmic rays in the generation of long-duration bursts from thunderclouds and to clarify the conditions of the electric field region responsible for particle acceleration. The simulations utilized realistic secondary cosmic-ray spectra, including gamma rays, electrons, positrons, and muons, as input. The simulation results indicate that gamma rays provide the dominant supply of seed electrons for long-duration bursts, regardless of the geometry or strength of the electric field region. They also reveal the structure and strength of the electric field region required to produce gamma rays exceeding several tens of MeV, which have so far been detected only by high-altitude observations. Furthermore, the fluxes of long-duration bursts estimated from the simulation results were compared with observational data to constrain the properties of the electric field region. In particular, the comparison with measurements at Yangbajing, located at an altitude of 4.3~km, helps narrow down the possible range of electric field strengths and configurations.

High-J CO emission in high-redshift galaxies has been studied primarily on an individual-source basis, limiting our ability to draw population-level conclusions about molecular-gas excitation. To address this limitation, we present a catalog of CO spectroscopy based on archival data from the Atacama Large Millimeter/submillimeter Array (ALMA) for a sample of galaxies at z > 3, focusing on high-J transitions (Jup = 9-17). Combining ALMA archival data with published measurements, we compile CO spectral line energy distributions (SLEDs) for 38 well-studied systems spanning z ~ 3.1-6.9, including 5 hot dust-obscured galaxies (Hot DOGs), 17 submillimeter-bright galaxies (SMGs), and 16 optically selected quasars. The class-median SLEDs rise steeply to Jup = 9 and remain approximately flat through Jup ~ 11-12. SMGs show relatively stronger low- to mid-J emission relative to CO J=9-8, while Hot DOGs exhibit tentative evidence for higher excitation. Comparison with simple excitation models suggests that X-ray dominated region (XDR) heating or dense, shock-heated gas can account for the extended high-J CO SLEDs. A tentative anti-correlation between the CO(9-8)-to-infrared luminosity ratio and excitation among the dusty galaxy populations suggests that the enhanced excitation in Hot DOGs may be driven by XDR heating from obscured AGN activity rather than by shocks.

It has long been suggested that X-ray transients are produced at periastron of stellar-compact object binaries with eccentric orbits. Recoil of matter evaporated from the star by X-rays from matter transferred at periastron increases the orbital semi-major axis and eccentricity. After periastrons the object would be a transient X-ray source, a Galactic analogue of a tidal disruption event (TDE), but recurrent with a gradually increasing period rather than catastrophic.

We present nanoCMB, a minimal but accurate calculator for the unlensed CMB temperature and polarisation angular power spectra ($C_\ell^{TT}$, $C_\ell^{EE}$, $C_\ell^{TE}$) of flat $Λ$CDM cosmologies. Written in $\sim$1400 lines of readable Python, the code implements the full line-of-sight integration method: RECFAST recombination, coupled Einstein--Boltzmann perturbation equations in synchronous gauge with a tight-coupling approximation, precomputed spherical Bessel function tables, and optimally constructed non-uniform grids in wavenumber and conformal time. Despite its brevity, nanoCMB achieves sub-percent agreement with CAMB across the multipole range $2 \le \ell \le 2500$, running in $\sim$10 seconds on a modern laptop. The entire calculation lives in a single, easily modifiable Python script, designed as a pedagogical bridge between textbook treatments and research-level Boltzmann solvers, with every approximation and numerical choice made explicit. We describe the physics, equations, and computational methods in detail, with code snippets illustrating each stage of the calculation. The code is publicly available at https://github.com/adammoss/nanoCMB.

Nearby radio galaxies are among the most promising candidates for the acceleration of ultra-high-energy cosmic rays (UHECRs). In this work, we develop a physically motivated, source-resolved framework to quantify the contribution of the three nearest FR-I radio galaxies$-$Centaurus A, Virgo A, and Fornax A to the UHECR flux measured by the Pierre Auger Observatory. Acceleration spectra derived from detailed jet-acceleration models are combined with numerical simulations of extragalactic propagation, while the more distant radio-galaxy population is treated as a continuous background. By fitting exclusively the measured UHECR energy spectrum, we determine the relative contribution of each source and constrain the fraction of jet power converted into UHECR luminosity. We find that a small number of nearby radio galaxies can account for the highest-energy UHECR flux with acceleration efficiencies of order $10^{-3}-10^{-2}$, while the background contribution remains subdominant. The resulting scenarios yield mass-composition trends broadly consistent with observations and predict distinct levels of secondary neutrino fluxes. These results demonstrate that physically grounded, source-specific modeling of nearby radio galaxies provides a viable and predictive explanation for the origin of the highest-energy cosmic rays.

Star-forming galaxies spend most of their lifetimes on the star-forming main sequence, which establishes a tight empirical and statistical relation between stellar mass and star-formation rate. Occasional episodes of rapid star formation can push them temporarily above this sequence, turning them into starbursts. Yet some galaxies display starburst-like traits -- rapid, dense, and compact star formation -- while still remaining within the scatter of the main sequence. These "starbursts in the main sequence" (SBMSs) reveal the complexity and diversity of star formation modes, making them crucial for understanding how galaxies evolve and transition between different regimes. In this paper, we identify SBMSs in the cosmological simulation NewHorizon and follow their evolution across time to uncover their physical origins and the role of this special regime in shaping galaxy evolution. We explain the existence of SBMSs by a comparatively earlier assembly of their stellar mass, driven in particular by more frequent and repeated mergers as the other galaxies, as well as exceptionally productive starburst events triggered by these interactions. As a result, this regime appears preferentially -- though not exclusively -- in the most massive galaxies. The SBMS behavior is not continuous within individual galaxies but instead arises intermittently as a short-lived (~ 30 Myr) evolutionary mode. Nevertheless, such SBMS episodes exist throughout cosmic time across the galaxy population... [abridged]

Rapid TeV flares have been observed from the core of the active galaxy M87. These have been attributed to inverse Compton scattering of disk photons by electrons and positrons accelerated in transient reconnection layers formed in baryon-poor regions of the magnetosphere of the central black hole, M87*. It was previously shown that even a small number of protons accelerated in the same layers can lead to bright GeV proton-synchrotron flares, if protons receive $\gtrsim20\%$ of the dissipated power for reconnecting fields of $\sim$100 G. We aim to investigate the role of Bethe-Heitler pair production in the emission of reconnection-driven flares from M87* in this physical regime. We perform numerical calculations that incorporate inelastic collisions between relativistic protons and photons, as well as photon-photon pair production, and compute the non-thermal radiation from the layer. The numerical calculations are also supported by analytical estimates. We find that disk photons act as targets for Bethe-Heitler pair production. The resulting pairs emit very high-energy synchrotron photons ($\gtrsim$0.1 TeV), which are subsequently attenuated by the disk photon field, leading to further pair production. The synchrotron emission of these secondary pairs produces soft photons, as part of an electromagnetic cascade, enhancing pion production and photon-photon attenuation down to tens-of-GeV energies.

Recently, an intermediate-mass black hole (IMBH) candidate was announced in the Galactic globular cluster Omega Centauri. IMBHs at the lower end of the traditional mass range have also been detected through gravitational-wave transients, though their formation and subsequent growth linking the two mass scales remains a mystery. One way IMBHs may be produced is through the collapse of very massive stars produced by stellar collisions in dense stellar environments. However, IMBHs may be ejected from such environments by either dynamical recoil from binary-single scattering or gravitational-wave recoil following the merger of two black holes. We conduct Newtonian and post-Newtonian binary-single scattering experiments to study dynamical ejection in greater detail. We obtain fits to the probabilities for dynamical ejection, gravitational wave capture, and per-encounter hardening as a function of the binary mass ratio and hardness with respect to its environment. We borrow techniques from survival analysis (commonly used in studies of medicine, epidemiology, engineering, etc.) to develop a model to calculate the probability of IMBH binary ejection vs in-cluster merger. We confirm that the dynamical ejection probability strongly depends on both the mass ratio of the IMBH compared to other BHs in its environment and on the environment's velocity dispersion. We estimate that for a typical Milky Way globular cluster, IMBHs with mass $\lesssim10^3\,\mathrm{M}_\odot$ are unlikely to be retained until the present. Our results also suggest that IMBH mergers with $q\lesssim0.2$ may be detectable at higher redshifts with future gravitational wave instruments such as the Einstein Telescope and Cosmic Explorer.

We investigate the gravitational wave (GW) signals emitted by planetary-mass primordial black holes (PBHs) passing nearby or traversing neutron stars (NSs). While previous studies mainly focused on the detailed waveforms of the signals, we estimate the rate of PBH-NS gravitational-wave events originating from the Galactic center and compute the probability of detecting a signal over 10 years of LIGO-Virgo-KAGRA observations. We examine in detail the case of PBHs bound to NSs, focusing on eccentric orbits that give rise to repeated GW bursts emitted in correlated series, each burst corresponding to a periastron passage. Despite the enhancement from the large number of bursts produced by a single PBH-NS pair, the total number of signals produced in this way remains subdominant to those due to random unbound encounters of PBHs with NSs. We also find that both types of signals have a very small probability $P\lesssim 10^{-8}$ to be detected in a 10 year period.

We present new X-ray, optical, and UV observations of the tidal disruption event candidate eRASSt J234402.9-352640, (hereafter J2344). Between 50 and 60 days after peak optical brightness, J2344 exhibited large-amplitude modulations in its 0.2-2 keV emission, when the flux repeatedly dimmed and re-brightened by a factor of ~6, over a ~3-day timescale. These modulations exhibited harder-when-brighter behaviour but were not detected in high-cadence observations obtained 60-70 days and 170-200 days after peak optical brightness, when the system instead exhibited stochastic X-ray variability over timescales of hours. We discuss the different physical mechanisms responsible for such exotic X-ray variability and explore the possibility that the modulations in J2344 were caused by the Lense-Thirring precession of the inner accretion flow around the disrupting black hole.

We conduct simulations of $\sim$kpc-scale cool clouds in the circumgalactic medium (CGM), using initial conditions sampled from a highly resolved cosmological magneto-hydrodynamical zoom-in of a Milky Way-like galaxy. We select ten distinct cold clouds with masses of $m_{\rm{cloud}}$ $\sim$ $10^{4.5-5}$ M$_\odot$, originally resolved at a mass resolution of $m_{\rm{gas}}$ $\sim$ $200$ M$_\odot$. To further resolve small-scale features and physics, we implement a targeted refinement scheme within spherical regions co-moving with each cloud, thereby boosting the local mass resolution by a factor of 1000, reaching $m_{\rm{gas}}$ $\sim$ $0.2$ M$_\odot$ (spatial resolution, $r_{\rm{gas,cloud}}$ $\sim$ $O(\rm{pc})$). The selected clouds have diverse properties, across a broad parameter space, resulting in heterogeneous evolution. For the clouds we study, radiative cooling is the dominant physical process enabling cloud survival, while magnetic fields play a comparatively smaller role. The motion of these clouds is governed not only by drag forces that decelerate them, but also by acceleration from momentum exchange with the complex background velocity field, which can cause them to move faster than ballistic projectiles set by their initial velocities. Our results suggest that the non-trivial details of realistic cosmological initial conditions -- specifically the complex density, temperature, and velocity fields -- may play an important role in subsequent cloud evolution, and that sampling the output of an existing large-scale simulation provides a self-consistent approach to capture these effects without ad hoc assumptions.

Type Ia supernovae (SNe Ia) are common luminous astrophysical transients. SNe Ia serve as distance indicators for measuring the expansion rate of the universe and play important roles in galactic nucleosynthesis. However, ambiguities persist regarding the nature of their stellar progenitors and explosion mechanisms. The recent discovery of \textit{Gaia} hypervelocity white dwarfs (WDs) has provided direct evidence in support of helium-ignited double degenerate SNe Ia. In this study, we investigate the outcomes of helium-ignited double-degenerate WD mergers by performing a set of 3D hydrodynamical simulations with two different codes: \texttt{AREPO} and \texttt{FLASH}. We consider two distinct binary WD systems close to helium ignition, evolving each with both codes while keeping initial conditions fixed. The first binary WD model produces a double detonation of the primary WD and the hypervelocity ejection of the surviving secondary, similar to the canonical dynamically driven double degenerate double detonation (D6) scenario. In the second model, the secondary also undergoes a core detonation, resulting in the complete disruption of both WDs. Notably, despite utilizing distinct numerical solvers, nuclear reaction networks, and mesh strategies, \texttt{AREPO} and \texttt{FLASH} produce broadly consistent outcomes for both sets of initial conditions. While the nucleosynthetic yields differ due to the different nuclear reaction networks employed, the overall agreement between the simulations demonstrates the robustness of the numerical modeling of this scenario. Our results strongly support the viability of both the D6 and quadruple detonation channels for at least some SNe Ia. We explore the prospective observational signatures of this channel, including in the X-rays using \textit{XRISM's} \textit{RESOLVE}.