arXiv Daily Digest - 2026-06-29

Ab initio modeling has established Wigner's SU(4) and Elliott's SU(3) as dominant symmetries of the nuclear force in light and intermediate-mass nuclei. We ask whether they also govern nuclear binding across the entire chart. Our aim is not high-precision prediction but physical insight, through interpretable, symmetry-based models. From the SU(3) and SU(4) Casimir operators we construct three neural-network (NN) mass models: Feature-Informed NN (FINN) for point predictions, Gaussian-Informed NN (GINN) adding uncertainty quantification, and Wigner-Informed NN (WINN) -- a mass formula using the Casimirs as an operator basis. All are trained on AME2016 and validated on nuclei new to AME2020. The SU(4) operators alone cut the root-mean-square error (RMSE) by nearly half on train and test data, and by about a fifth on extrapolation, relative to the liquid-drop baseline -- showing that Wigner's symmetry carries predictive information beyond bulk properties. Despite its compact form, WINN reaches the lowest validation RMSE, 0.430 MeV -- competitive with state-of-the-art mass models -- which we read less as a benchmark than as evidence that its symmetry basis captures important physics. WINN further reveals i) an enhancement of the quadratic SU(4) Casimir near the neutron dripline, signaling restoration of Wigner's symmetry, and ii) an unexpected gain of the quartic operator in the superheavy region. We thereby elevate emergent symmetries from the hidden order within individual nuclei to a governing principle of the whole nuclear chart.

PAC-Bayesian bounds provide finite-sample guarantees for data-dependent randomized predictors, but applying them to learning-based control is difficult because the natural objective is a quadratic trajectory cost. Such losses are unbounded, non-Lipschitz , and lead to response-dependent Chernoff terms. We employ System Level Synthesis parameterization, which exposes the closed-loop trajectory map of a linear system directly and makes the quadratic control loss amenable to explicit certification. Moreover, we provide a set of PAC-Bayes-Chernoff certificates for posterior distributions over feasible closed-loop responses. For Gaussian disturbance trajectories with arbitrary covariance, we derive an exact one-sided Gaussian transform and a tractable quadratic upper bound expressed through closed-loop sensitivity quantities. We also derive a posterior-localized surrogate for settings where pointwise closed-loop response certificates are unavailable or have support related admissibility issues. Although PAC-Bayes certifies a non-degenerate posterior, the convex quadratic form of the SLS loss transfers the certificate to the posterior mean response. We present a deterministic mean response deployment result that is particularly suitable for control while retaining the stochastic posterior in the bound. Additionally, we provide a data-driven bound for this deployment, transitioning away from an oracle bound. Minimizing this bound naturally results in a learning algorithm for control selection from data. Numerical experiments on a double integrator show that the algorithm acts as a sensitivity-aware finite-sample regularizer, improving held-out cost and reducing closed-loop sensitivity in the low-data regime

We present HORIZON, a self-evolving agent framework that treats hardware design as repository-level code evolution. A Markdown harness is compiled into a project pack containing domain knowledge, an executable evaluator, an acceptance predicate, and a git/runtime policy; a hands-free agent loop then evolves an isolated git worktree, using repository operations for state management, tracing, and replay. This extends prior works of repository-scale self-evolution from EDA software systems, to hardware-design artifacts themselves. We evaluate our approach on ChipBench, RTLLM, Verilog-Eval, and nine CVDP categories, achieving 100\% benchmark completion across all suites with a fully hands-free agentic loop. However, we do not claim that agentic AI for hardware design is solved: these benchmarks are controlled proxies for a much broader engineering problem in chip design. Section~\ref{sec:discuss} examines the limitations of the current study and highlights open research challenges.

Artificial intelligence is driving a revolution in scientific discovery, accelerating everything from hypothesis generation to mathematical theorem proving. However, this rapid acceleration is creating a systemic challenge: traditional human peer review cannot scale to match the influx of AI-assisted science. Ultimately, to resolve this tension, we must also deploy AI to accelerate the verification and review process itself. To frame the discussion around this transition, we propose a taxonomy consisting of four progressive levels of AI-human collaboration in scientific evaluation, and discuss various trade-offs involved with each. As a step toward this future, we introduce the Paper Assistant Tool (PAT), an agentic AI framework built for deep scientific review and verification. PAT ingests full scientific manuscripts and produces a comprehensive evaluation, checking theoretical results, validating experiments, suggesting improvements, and identifying potential flaws. By utilizing inference scaling techniques, PAT is able to identify deeper issues than a single model call alone, achieving a 34% improvement over zero-shot recall on mathematical errors in the SPOT benchmark. Pilot deployments of PAT as a pre-submission tool for authors at two major Computer Science conferences -- STOC and ICML -- demonstrate its ability to identify critical errors and suggest substantive improvements to research papers. By catching errors early, PAT eases the cognitive burden placed on referees, while preserving their control over the outcomes of the review process.

Accurate network traffic prediction is a critical element for efficient resource allocation in dynamic urban cellular networks. However, prediction remains challenging because network demand is influenced by complex mobility patterns, congestion dynamics, and heterogeneous user behavior. This paper introduces the Parameter-Efficient Hybrid Transformer (PEHT), a network traffic prediction framework that integrates urban mobility and congestion information into a Transformer-based architecture. PEHT separates primary network communication features from secondary urban mobility features and incorporates Low-Rank Adaptation (LoRA) into the Transformer encoder to reduce the number of trainable parameters while maintaining high predictive accuracy. A multimodal fusion strategy then injects external mobility and congestion features into the decoder to improve traffic forecasting. Experiments on the Telecom Italia Milan dataset and multiple synthetic congestion scenarios show that PEHT outperforms state-of-the-art baselines in terms of RMSE, MAE, and $R^2$. The implementation is available in the GitHub repository.

Vision-language models must reconcile visual evidence with memorized world knowledge when the two conflict. How they resolve this conflict shapes the reliability of multimodal systems, yet prior work characterizes it behaviorally without a component-level causal account. We combine activation patching across three granularities (residual stream, attention heads, and MLP sublayers) with model-component ablation studies and mechanistic analysis. Across three VLM families, we find that visual grounding emerges by default, whereas prior grounding depends on a small set of causally necessary attention heads (2.5-4.8%) concentrated in the second half of the network. These heads enable answers from stored world knowledge (e.g., "red" for a strawberry) despite conflicting visual input. Ablating them flips predictions from knowledge-grounded to visually grounded answers in 68-96% of cases under prior-knowledge prompts, but changes only 0.8-7.5% of visually grounded predictions, establishing an asymmetric causal structure. The identified heads decompose into routing heads, which modulate information flow, and writing heads, which directly project answer tokens into the residual stream. This structure is consistent across model families and scales, revealing a sparse causal circuit underlying perception-knowledge conflict in VLMs.

The transition from static chat bots to autonomous agents--equipped with persistent memory, tool-use protocols, and multi-agent collaboration--has fundamentally expanded the AI threat landscape. Current defense mechanisms, such as perimeter security and training-time alignment, remain external to the agent's active reasoning loop. Consequently, they fall short: a fully aligned agent remains highly vulnerable to runtime hijacking via memory poisoning, tool-chain manipulation, or multi-agent protocol attacks. To address this critical gap, we introduce the Agent-Native Immune System (ANIS), the first biologically inspired, endogenous defense architecture embedded directly within the agent's cognitive loop. Our framework presents four primary contributions. First, we design a six-layer Immune Tower (L0-L5), distinctly incorporating Barrier Immunity (L1) as a non-cognitive, physical-and-logical isolation layer. Second, we establish a unified taxonomy of Agent Viruses and Agent Vaccines, formalizing the critical distinction between superficial non-parametric defenses and robust parametric vaccines. Third, we conceptualize the Harness Triad--Meta, Self, and Auto--a self-monitoring, meta-cognitive automation backbone that drives Continual Immune Learning (CIL), enabling vaccines to dynamically adapt to novel threats. Finally, we establish a rigorous theoretical demarcation between model alignment and agent immunity: while alignment provides a static "constitutional" value foundation during training, ANIS serves as the dynamic "law enforcement" mechanism during runtime. We conclude by framing open challenges for the field, including immune protocol standardization, novel evaluation metrics such as the Autoimmunity Rate (false-positive intervention rate), and the co-evolutionary dynamics between pathogens and vaccines within collective intelligence ecosystems.

Test-time adaptation (TTA) has emerged as a promising paradigm for mitigating distribution shifts in deep models. However, existing TTA approaches for anomaly segmentation remain limited by their reliance on pixel-level heuristics, such as confidence thresholding or entropy minimisation, which fail to preserve structural consistency under noise and texture variation. Moreover, they typically treat anomaly maps as flat intensity fields, ignoring the higher-order spatial relationships that characterise complex defect geometries. We introduce TopoTTA (Topological Test-Time Adaptation), a novel framework that integrates persistent homology, a tool from topological data analysis, into the TTA pipeline to enforce geometric and structural coherence during adaptation. By applying multi-level cubical complex filtration to anomaly score maps, TopoTTA derives robust topological pseudo-labels that guide a lightweight test-time classifier, enhancing segmentation quality without retraining the backbone model. The approach avoids reliance on method-specific raw-score thresholding for mask binarisation, preserves connectivity, and generalises across both 2D and 3D modalities. Extensive experiments across six standard benchmarks (MVTec AD, VisA, Real-IAD, MVTec 3D-AD, AnomalyShapeNet, and MVTec LOCO) demonstrate an average 15% F1 improvement over state-of-the-art unsupervised anomaly detection and segmentation methods, with the largest gains on anomalies exhibiting complex geometric or structural variations. These findings suggest that integrating topological reasoning into test-time adaptation provides a principled route to structure-aware generalisation, bridging the gap between geometric learning and robust adaptation.

Early detection of oral cancer markedly improves clinical outcomes, yet specialized diagnostic tools remain scarce in low-resource settings. Smartphone-based screening is a scalable alternative but needs lightweight models that run within edge-hardware constraints. Hybrid classical-quantum architectures are emerging candidates for parameter-efficient learning, yet most rely on qubit hardware that needs cryogenic operation, unsuitable for edge deployment. Continuous-variable (CV) photonic quantum computing, which operates at room temperature, offers a complementary route. We investigate a hybrid classical-CV quantum classifier for oral cancer detection from smartphone images. The pipeline combines a MobileNetV1 feature extractor, principal component analysis to 16 dimensions, and a parameterized CV-QNN of displacement, interferometric, and Kerr gates on a photonic backend. We propose a simplified $Φ\circ D \circ U_1$ CV-QNN architecture that cuts trainable parameters 40-45% relative to the standard CV-QNN layer of Killoran et al. (2019a), and identify dimensionality-reduction and encoding-restriction strategies that mitigate barren plateaus, raising loss-gradient variance by roughly 58 orders of magnitude. Whether the simplified layer beats the full layer is width-dependent: the full layer holds a small but significant edge at two qumodes, whereas the simplified layer is significantly better at four qumodes using 44% fewer parameters. The strongest model, a four-qumode simplified CV-QNN with only 18 parameters, attains the highest validation AUC of all models, exceeds a 55-parameter classical baseline using 67% fewer parameters, and reaches 100% calibrated test accuracy across all seeds. These results support CV photonic quantum machine learning for parameter-efficient, room-temperature medical image classification and motivate progress toward edge quantum AI.

Recently, Large Language Model (LLM)-based Text-to-Speech (TTS) models have achieved remarkable naturalness. However, the standard Supervised Fine-Tuning paradigm often converges to statistically averaged prosody, limiting emotional expressiveness. While preference-driven optimization offers a promising alternative, existing approaches suffer from two structural mismatches: information conflict, where content and emotion in a shared latent space produce conflicting gradients, leading to reward hacking and semantic degradation; and scale gap, where sparse sentence-level rewards struggle to guide dense frame-level generation. To overcome these challenges, we propose HPRO, a hierarchical progressive reward optimization framework. Within HPRO, we introduce the HD-Emo codec as a novel differentiable reward model to resolve the information conflict. It extracts speech into distinct content and style preference tokens, structurally isolating emotional optimization from semantic content. Building upon this structured preference space, HPRO bridges the scale gap by progressively aligning frame-, word- and sentence-level objectives. Experiments demonstrate that HPRO significantly enhances emotional expressiveness, while effectively preserving linguistic intelligibility. The code and audio samples are publicly available at https://xxh333.github.io/hpro-demo/.

Understanding how performance scales jointly with model size and data is a central problem in modern machine learning. Existing theoretical works on scaling laws typically describe generalization as a function of data or compute, often in fixed-feature or infinite-width regimes and for online SGD. Here, we instead study how generalization scales with the number of trainable parameters and the number of samples in a feature-learning model. We analyze $\ell_2$-regularized empirical test error minimization in a quadratic two-layer network in a finite-sample setting with structured data. This setting allows for an explicit characterization of the generalization error as a function of the number of samples, model width, and regularization. Our results reveal a phase diagram with distinct scaling regimes as the number of parameters varies. In particular, the generalization error follows data-dependent power laws controlled by the spectral structure of the target. We further characterize the transitions between regimes, including the onset of interpolation, and their impact on generalization.

Autonomous coding agents now open and merge pull requests in shared repositories at scale, and the field evaluates them the way it has always evaluated components, one agent at a time, on isolated benchmark tasks. Yet agents that each pass their own tests still leave repositories that accumulate problems no single contribution accounts for. We ask whether this problem belongs to the individual agent or to the repository where it accumulates. We study integration friction, the cost of integrating a contribution into a codebase that other contributors are concurrently changing. Across more than 930,000 agent-authored pull requests, we measure how much of the variation in friction stays with the repository after the contribution, its author, its size, and its agent are accounted for. About half does, and it survives full controls. In the same repositories, agent-authored contributions concentrate this repository-level friction roughly twice as much as human ones (intraclass correlation 0.30 versus 0.16), a gap that holds after controlling for codebase size, age, task shape, process maturity, and merge path. The risk is a property of the ecosystem, not the agent. AI-native software is therefore better measured and governed at the ecosystem level than one agent at a time.

Search-based test generation tools such as EvoSuite produce compilable and high-coverage unit tests at scale, but their suites are often hard to read and maintain. LLMs can generate more natural tests, yet direct generation remains brittle, with compilation rates of only 51-78% in our study. We introduce TestHumanizer, a hybrid SBST+LLM approach that uses LLMs as controlled refactoring layers over compilable SBST suites to improve naming, structure, and developer-oriented clarity while preserving behavior and compilation validity. We evaluate TestHumanizer on 350 classes from Defects4J and SF110. EvoSuite generates 15 suites per class, and each suite is refactored under three context configurations using gpt-4o and mistral-large-2407, yielding 31,500 refactorings. TestHumanizer reaches 88-98% compilation rates, close to EvoSuite's 100% baseline and clearly above direct LLM generation. Structural coverage is largely preserved, typically within 1-2 percentage points, and 86-95% of refactorings satisfy a composite faithful-refactoring threshold. Refactored suites also improve predicted readability, reduce control-flow and cognitive complexity, and mitigate structural smells. The summary-based setting offers the most robust trade-off, while long code-centric prompts are more prone to hallucination-induced failures. A developer study on 30 classes and 444 test methods confirms significant gains in perceived readability and willingness to adopt, with Wilcoxon p less than 0.01 and substantial inter-rater agreement. Overall, LLMs are most effective not as standalone generators but as validation-gated refinement layers over robust SBST outputs.

Causal representation learning for time series has developed strong identifiability results in discrete-time latent causal models, but identifiability in continuous-time latent stochastic differential equation (SDE) models remains largely open. We address this gap using environment-induced shifts in diffusion covariance. We study additive-noise latent SDEs observed through an unknown nonlinear diffeomorphism, with shared drift but environment-specific diffusion covariance. We show that two diagonal diffusion regimes with pairwise distinct coordinate-wise variance ratios identify the latent coordinates up to permutation and scaling, without any sparsity assumption on the drift. We first prove this result for linear Ornstein--Uhlenbeck systems and then extend it to general additive-noise latent SDEs. Under mild smoothness, the instantaneous drift-Jacobian causal graph is identifiable up to the same permutation. We propose a two-stage estimator for latent disentanglement and optional graph recovery; experiments on synthetic systems confirm the predicted identifiability boundary, and an application to Hardanger Bridge monitoring data illustrates the approach on real sensor trajectories.

Flow Matching (FM) has achieved remarkable generative performance, yet it suffers from exposure bias due to discrepancies between training and inference. Existing mitigation strategies typically rely on static constraints or external heuristics. In this work, we propose that exposure bias itself inherently contains dynamic signals that can guide its own rectification. To leverage this, we introduce DEFAR (DirEctional-Frequency Adaptive Rectification). This framework simulates the single-step inference process during training to identify exposure bias. It utilizes directional and frequency-adaptive feedback signals from the bias itself to enhance the model's bias tolerance. It consists of two key components: (1) Anti-Drift Rectification (ADR). ADR treats inference-time drift as a signal to learn the direction to steer deviated states back toward the target. ADR endows the model with intrinsic active self-rectification capabilities; (2) Frequency Compensation (FC). Empirically, we observe that accumulated bias often stems from a lack of low-frequency components in high-noise stages, and exposure bias carries the missing frequency. FC leverages the bias itself as a self-feedback weighting factor to reinforce the missing frequency components. Experiments on CIFAR-10, CelebA-64, and ImageNet-256/512 show that DEFAR outperforms prior baselines and further demonstrates favorable scalability, compatibility, and inference robustness.

Temporal link prediction is usually evaluated by predictive performance on unseen edges, but in probabilistic temporal graphs this criterion can conflate model error with irreducible uncertainty. We study this issue by characterising an inherent estimation--prediction tradeoff in binary logistic models where regimes that maximise Fisher information and improve parameter recoverability are also those with the highest entropy, making individual predictions intrinsically harder even under perfect parameter recovery. We propose a probabilistic causal framework for generating temporal graphs with transient edges and known ground-truth causal structure, allowing temporal link prediction to be evaluated jointly with causal parameter recovery. For the proposed binary logistic parametrisation, we derive the Cramér--Rao bound and validate the tradeoff between parameter estimation error and irreducible predictive loss. Our results show that predictive accuracy alone may not reflect whether a model has learned the underlying causal mechanism, motivating benchmarks that distinguish reducible model error from intrinsic process uncertainty.

Physics-informed neural networks (PINNs) have emerged as a powerful tool for solving nonlinear partial differential equations (PDEs), including battery electrochemical models. They typically en-force conservation laws within the loss function to ensure physically consistent solutions. Tradi-tional numerical methods such as finite difference, finite volume, and finite element techniques, re-ly on discretization and can be computationally expensive for nonlinear systems. To address this challenge, PINNs offer improved scalability, particularly for reduced-order models like the single particle model with electrolyte (SPMe). The SPMe describes lithium-ion battery dynamics through coupled diffusion, transport, reaction kinetics, and voltage equations. Despite these advantages, training SPMe-based PINNs from scratch for different battery chemistries or operating conditions is demanding and often leads to slow convergence. To overcome this limitation, this work introduces a transfer learning framework for SPMe-PINNs. The model is first pretrained to learn general elec-trochemical dynamics and then adapted to a target battery by transferring weights, freezing se-lected layers, and fine tuning the remaining parameters, including estimating key electrochemical variables. Validation using PyBaMM demonstrates accurate voltage prediction, indicating that the proposed approach preserves electrochemical consistency while reducing training time and ena-bling efficient generalization across batteries.

We propose a framework for reward allocation in fully delegated AI cooperatives where humans are represented by agents that contribute data and participate in model updates under heterogeneous value constraints. The key idea is to credit only those updates that remain admissible after screening them against each principal's value profile. We formulate value-conditioned gradient filtering, online marginal contribution signals, and cumulative revenue settlement within a traversal learning (TL) substrate. TL is especially attractive here because it performs decentralized backpropagation without the quality loss associated with aggregation-centric distributed learning and, we argue, offers a finer attribution substrate than FedAvg-style federated learning by preserving explicit traversal and gradient paths. The framework is positioned against data valuation, federated contribution estimation, personalized federated learning, and pluralistic alignment.

Extracting dynamic 4D object interactions from massive, in-the-wild monocular videos offers a highly efficient data collection pathway for scaling Embodied AI and training VLAs. However, existing monocular 4D reconstruction methods primarily focus on isolated objects, often failing under the severe occlusions and complex dynamics inherent in multi-object interactions. To bridge this gap, we propose HAT-4D, the first agentic framework designed to reconstruct the 3D geometry, temporal dynamics, and physical interactions of multiple objects from a single video. By integrating VLMs with a multi-level human-in-the-loop feedback mechanism, HAT-4D efficiently resolves depth ambiguities and interaction-induced occlusions during 3D generation and 4D propagation, yielding physically plausible assets without relying on expensive multicamera rigs. As a scalable data engine, HAT-4D facilitates the creation of MVOIK-4D, an open-world benchmark for monocular 4D interaction reconstruction, accompanied by a novel multi-dimensional evaluation protocol focused on physical plausibility and temporal consistency. Extensive experiments demonstrate that HAT-4D achieves SOTA performance on most evaluation metrics, while maintaining competitive semantic alignment. Ablation studies show that introducing a small amount of human feedback improves interaction reconstruction. Moreover, the data produced by HAT-4D effectively improves baseline performance when used for fine-tuning. Our data and code are available at https://lijiaxin0111.github.io/HAT4D/

Algorithmic developments in Strategic Classification have been mostly limited to linear classifiers in settings where the best response has a closed-form solution or can be easily approximated. While some work has explored the role of non-linear classifiers in strategic settings, progress in this direction is impeded by the computational intractability of the strategic behaviour. Addressing this, we present a novel method for approximating the best response by exploiting Lagrangian duality. By reformulating the strategic response as a constrained optimisation problem, we can construct a Lagrangian that is amenable to first order optimisation methods. This approach reproduces closed-form strategic behaviour in linear settings and can be straight-forwardly applied to non-linear settings. We show how the Implicit Function Theorem can be used in conjunction with our proposed response formulation during classifier learning to compute the total gradient of the loss. This connects the classifier parameters directly to the consequent strategic behaviour, yielding a novel training algorithm that can exploit this relationship. Experimental evaluation shows that the resulting models achieve improved strategic accuracy on common machine learning datasets.

While interpretable models such as concept bottleneck models (CBMs) and program synthesis methods enable verification of model decisions, their evaluation is typically limited to simple tasks, leaving complex reasoning on real-world images largely unexplored. We introduce COCOLogic-V2, an object-centric dataset for visual inductive reasoning on real-world images covering a broad subset of first-order logic. By categorizing samples into positive variants, near-boundary (NB), and far-from-boundary (FB) negatives, COCOLogic-V2 enables fine-grained diagnosis of model accountability. Our evaluations show that models tend to separate positive and FB samples well but fail on NB samples, while perceptual noise and large rule-induced search spaces pose additional challenges in few-shot settings. Together, these results highlight that visual inductive reasoning remains an open challenge and COCOLogic-V2 provides a concrete foundation for advancing methods in this direction.

This study analyzes Sri Lankan migration and remittances over 32 years (1994-2025). Using a 384-month harmonized dataset, we apply exploratory data analysis, stationarity corrected time-series modeling (ADF, Johansen, VAR/VECM), and supervised learning. Results reveal remittance inflows are primarily driven by external macroeconomic variables, specifically exchange rate dynamics and global oil prices, rather than domestic indicators. Impulse response analysis confirms the asymmetric impact of currency depreciation and oil price shocks. Predictively, multivariate machine learning models outperform traditional univariate approaches; Ridge Regression achieves a 73.8% accuracy improvement over SARIMA (Annualized RMSE: USD 494.8 Mn). The optimized framework projects 2026 remittances at USD 9,001 million under stable conditions. These findings highlight the structural dependence of remittances on global economies, emphasizing the need for robust exchange rate policies, skilled migration, and formal financial channels to enhance long-term economic resilience.

Multi-agent systems (MAS) built on large language models (LLMs) provide a promising framework for solving complex tasks through role specialization and structured interaction. However, their performance is often limited by miscoordination and, more fundamentally, the lack of fine-grained credit assignment across agents. Existing approaches typically rely on coarse-grained feedback, making it difficult to identify which agents or interaction steps are responsible for errors. We propose Gradient-Based Connections (GBC), an approach for fine-grained attribution and optimization of multi-agent systems. GBC models a MAS as a computational graph and introduces gradient-based connection weights to quantify the influence of each agent's output on downstream agents at the token level. By constructing an attribution graph and propagating task-specific loss signals backward, our method enables precise identification of error sources and targeted prompt optimization. We further develop AgentChord, an efficient implementation that leverages prefix-based gradient computation. Experiments on MultiWOZ and τ-bench show that GBC improves multi-agent performance and outperforms strong single-agent and multi-agent baselines, and higher attribution quality is associated with greater optimization effectiveness. Code is available at: https://github.com/yxc-cyber/AgentChord.

Predicting human item difficulty is central to educational assessment, where reliable estimates support fairness and effective test construction. Existing methods often depend on costly human calibration or item-level textual representations, providing limited evidence about the cognitive processes that make items difficult. We argue that difficulty should be viewed not only as a property of item text, but also as an observable consequence of the problem-solving burden an item induces. Large Reasoning Models (LRMs) offer scalable process evidence through reasoning traces, but such evidence must be structured to support interpretable modeling. To this end, we introduce Epi2Diff (Episode to Difficulty), a framework that maps LRM reasoning traces into cognitively grounded episode sequences. These episodes group trace segments into functional problem-solving states, enabling difficulty to be modeled through reasoning scale, effort allocation, and state transitions. Epi2Diff extracts compact episode-dynamic features and combines them with semantic item representations for human difficulty prediction. Experiments on four real-world human difficulty datasets show that Epi2Diff consistently outperforms strong baselines, including fine-tuned small language models, LLM in-context learning, and supervised LLM adaptation. On SAT-derived classification benchmarks, Epi2Diff achieves an 8.1% average relative gain over supervised LLM fine-tuning baselines. Further analyses show that harder items induce more effortful, iterative, and implementation-centered episode dynamics, rather than merely longer responses. These results demonstrate that cognitive episodes in LRM reasoning traces provide a predictive and interpretable process representation for human item difficulty, offering a new lens for educational measurement with reasoning models.

Embodied agents operating in decentralized and partially observable environments have attracted growing attention in recent years. However, existing large language model (LLM)-based agents often exhibit behaviors that are misaligned with their partners or inconsistent with the environment state, leading to inefficient cooperation and poor task success. To address this challenge, we propose a novel framework, Learning Laws of Cooperation (LLawCo), that enables embodied agents to autonomously align with both their partners and task objectives. Our framework allows agents to reflect on past failures to extract misaligned behavioral patterns, which are used to derive high-level behavioral laws, such as "Talk when necessary" and "Wait for partner." These laws are explicitly incorporated into the agents' chains of thought via supervised fine-tuning, aligning their reasoning with task requirements and the behavior of other agents. To evaluate our approach, we introduce PARTNR-Dialog, a large-scale multi-agent communicative and cooperative planning benchmark built on the PARTNR environment. Experiments on existing tasks and our new benchmark demonstrate significant improvements in cooperative efficiency and task success rates. Across four backbone LLMs, our method achieves average success rate improvements of 4.5% on the PARTNR-Dialog benchmark and 6.8% on the TDW-MAT benchmark over state-of-the-art open-source communicative agent frameworks. See the LLawCo project page for details: https://www.merl.com/research/highlights/LLawCo

Identifying robust associations between cardiac imaging phenotypes and clinical diseases is fundamental to population-scale cardiovascular research and reliable risk stratification. However, current phenome-wide association studies rely on pre-defined, single-variable phenotypes or expert-crafted features, which limits their ability to capture clinically meaningful non-linear effects and cross-phenotype interactions. To address this, we propose CPAgents, an iterative phenotype-Composition framework for cardiovascular Phenome-wide association study (PheWAS) that automatically constructs and validates interpretable composite phenotypes (e.g., polynomial, ratio, and interaction forms) from base imaging features. Specifically, our system coordinates three agents: (i) an Analyst that identifies statistical pathologies and nominates candidate transformations; (ii) a Proposer that generates constrained, medically and statistically motivated expressions under numerical safety rules; and (iii) a Verifier that evaluates candidates using multi-stage criteria and produces transparent evidence trails for accepted phenotypes. Evaluated on a population-scale cardiac imaging cohort, the discovered composite phenotypes markedly improve disease discrimination: across 72 classifier-disease-metric combinations, our variants achieve the top rank in 56 cases versus 18 for baselines, with gains observed across all nine clinical disease categories. Our framework yields compact, clinically interpretable phenotype formulas with transparent evidence trails, enabling scalable discovery of stronger phenotype-disease associations beyond expert-driven feature selection.

Reinforcement learning with verifiable rewards (RLVR) has significantly improved the reasoning capability of large language models, reaching expert or even superhuman performance in domains such as competition math. However, whether weaker agents and humans can actually harness this capability is far less certain, with RLVR documented to drift reasoning toward idiosyncratic patterns such as poor readability and language mixing. Tandem training is a recently introduced paradigm that targets this compatibility problem: a trained, stronger senior co-generates each rollout with a frozen, weaker junior, and the two are rewarded as a team, so the senior is pushed to reason in ways the junior can follow. Yet this paradigm has so far been demonstrated only in proof-of-concept settings, leaving open whether it scales to the long chains of thought of the modern RLVR pipeline. In this work, we propose Tandem Reinforcement Learning (TRL), which carries the tandem training paradigm into RLVR. In TRL, the senior and a frozen junior alternate stochastically to co-generate the reasoning, the resulting generation is rewarded, and the standard GRPO loss is applied to the senior. Training Qwen3-4B-Instruct on competition math, we find that TRL matches vanilla GRPO on solo reasoning capability while three properties emerge together from the same rollout structure: stronger handoff robustness with the junior, reduced distributional drift from the junior, and a chain-of-thought more legible to the junior. Our results demonstrate a promising route for RLVR with practical payoffs in multi-model communication and human compatibility.

Echocardiography is the most widely used non-invasive cardiac imaging modality, providing essential information for cardiovascular diagnosis. Interpreting an echocardiogram requires synthesizing complementary evidence across multiple heart views to identify abnormalities and produce structured clinical reports. While recent efforts focus on improving classification performance, most models lack explicit diagnostic reasoning and spatially grounded anatomical evidence, limiting clinician trust. We present EchoSonar-R, a multi-view reasoning-enabled vision-language model that jointly performs multi-label disease classification and report generation from echocardiography studies. EchoSonar-R combines a spatiotemporal video encoder with a structure-aware cardiac detector that provides spatially grounded anatomical cues to improve interpretability and clinician trust during cross-view reasoning. EchoSonar-R is trained in two stages: supervised fine-tuning (SFT) on reasoning-annotated targets, followed by Group Relative Policy Optimization (GRPO) with task-specific rewards that jointly align classification and report generation within a unified reinforcement-learning framework. Across a private multi-view dataset and two public benchmarks, EchoSonar-R improves macro balanced accuracy by 17.1% on the private set and 6.1% on MIMICEchoQA over the strongest baseline, achieves a GREEN clinical faithfulness score of 0.800, and produces interpretable reasoning traces grounded in multi-view visual evidence.

Physics-informed neural networks (PINNs) have recently emerged as a promising framework for addressing the Calderón inverse problem from limited boundary data. In this work, we revisit neural Calderón inversion by introducing multiscale boundary excitations based on randomized wavelet functions and investigating the role of Fourier-feature encoding (FFE) for representing sharp conductivity variations. We propose a physics-informed reconstruction framework that represents the unknown conductivity and the associated family of electric potentials with separate neural networks conditioned on the applied boundary excitations. The governing elliptic PDE is enforced through physics-informed residuals, while finite Dirichlet-to-Neumann (DtN) data are incorporated through boundary losses. Using synthetic data from a finite-difference forward solver, we evaluate the method on conductivity fields with inclusions, sharp interfaces, smooth profiles, and heterogeneous media. Results show that the framework recovers dominant conductivity structures from finite boundary measurements with relative errors between $3\%-12\%$ approximately. We show that FFE improves the reconstruction of localized sharp features, particularly for inclusions and interfaces, but are not universally optimal, with raw-coordinate networks performing competitively for smoother fields. These results highlight coordinate representations and boundary excitation design as key factors in neural Calderón inversion.

Jailbreak attacks bypass LLM safety alignment, yet their mechanisms remain poorly understood. We provide evidence that attacks do not comprehensively eliminate safety features, but instead selectively suppress specific attention heads. We identify two functionally differentiated types: Adversarially Compromised Heads (ACHs) concentrated in early layers, which are suppressed under attacks, and Safety-Aligned Heads (SAHs) in mid-layers, which maintain robust activations even when attacks succeed. Ablation studies support the causal role of ACHs and the contribution of SAHs to robust activations: suppressing a small number of ACHs is sufficient to induce jailbreak-like behavior on normally refused inputs, while removing SAHs substantially weakens mid-layer safety activations. Token-level attribution further shows that ACH suppression is driven specifically by attack-template tokens, providing a mechanistic account of why attacks can bypass refusal decisions through ACH suppression while leaving internal safety signals sustained by SAHs -- a phenomenon we term Robust Harmful Features. To validate the practical significance of this robustness, we show that simply reading these persistent activations -- without any training -- yields competitive aggregate detection performance with strong adversarial robustness.

We propose KL-Coupled Policy Regularization (KCPR), a policy coordination framework for Reward-Punishment Reinforcement Learning (RPRL). Based on KCPR, we derive KL-Coupled Soft Optimality (KCSO) and develop its deep realization, klDMP. Unlike existing RPRL approaches that optimize reward-seeking and punishment-related policies largely independently, KCPR enables direct interactions between companion policies by treating each as a dynamically learned prior for the other. KCSO yields coupled soft-optimal policies and KL-regularized Bellman operators, allowing reward and punishment information to jointly influence value propagation. To improve learning stability, we introduce a companion-prior softening mechanism and evaluate separate replay-buffer designs for balancing reward- and punishment-related experience. Experiments in grid-world and Gazebo robotic navigation tasks demonstrate that klDMP improves safety and learning stability while maintaining competitive task performance compared with DQN, SQL and softDMP. These results suggest that policy-level coordination provides an effective mechanism for integrating multiple behavioral objectives and may serve as a useful design principle for reinforcement learning systems with interacting motivational processes.

In-context segmentation (ICS) requires a model to segment target regions in a query image using only a few reference images and their corresponding masks, without updating any parameters. Despite recent progress, prior ICS studies have largely overlooked a critical aspect: system robustness, ie, whether the model can produce stable segmentation results for the same query under different references. In this work, we revisit ICS from the robustness perspective and introduce a novel paradigm, Concept-Guided In-Context Segmentation (CG-ICS), which performs segmentation by extracting high-level semantic concepts from references rather than relying solely on low-level visual matching. Specifically, CG-ICS introduces a concept reasoning module that uses an MLLM to propose candidates and a SAM3-driven scoring function with tree-search refinement to select reliable textual concepts, together with a parallel visual exemplar route that provides query-side spatial grounding via a simple context construction. Both the textual concept and the visual exemplar are then used to activate the segmentation capability of a frozen SAM3 backbone. Extensive experiments on standard ICS benchmarks demonstrate that CG-ICS not only achieves state-of-the-art accuracy but also substantially improves robustness, yielding a more reliable ICS system with significantly reduced variance across diverse reference choices.

Wearable devices produce large, high dimensional training logs for everyday runners, and interpretation rather than data collection is now the limiting step. This paper evaluates five dimensionality reduction models, three autoencoder variants, PCA, and a Variational Autoencoder, on their ability to compress nine sensor runner profiles into a single scalar performance indicator, the latent score. Because the setting is fully unsupervised, model quality is assessed along two complementary axes: reconstruction error (Mean Squared Error) and latent score interpretability, measured via Spearman and Kendall rank correlations, Mutual Information, and Permutation Importance. These are combined into a composite selection criterion that prevents selecting models on reconstruction accuracy alone. Feature rankings from the four metrics are aggregated via a modified Borda count, and their stability is confirmed by bootstrap validation. A two feature linear baseline is included to anchor the comparison. Deep autoencoder achieved the lowest reconstruction error and the highest composite score. Once the PCA hidden layers were widened, the deeper variants became closely competitive with Deep AE on the composite criterion, indicating that the limiting factor was hidden layer capacity rather than the one dimensional bottleneck. Running pace, aerobic decoupling, and average heart rate emerged as the dominant latent score drivers across all models and resampling runs, consistent with established physiology.

Test-Time Adaptation (TTA) methods commonly update the affine parameters of normalization layers to adapt deployed models under distribution shifts. However, per-channel affine parameters perform axis-aligned scaling and shifting, making them geometrically incapable of correcting cross-channel structural changes induced by distribution shift. To address this limitation, we propose MixTTA, a lightweight plug-in module that equips normalization layers with a low-rank cross-channel transformation, enabling inter-channel mixing at each layer. To ensure that the low-rank branch captures only cross-channel interactions, we also propose Decoupling Projection that enforces strict separation from the diagonal affine path, along with Spectral Projection that prevents rank-1 collapse under non-stationary test streams. MixTTA can be seamlessly integrated into any existing normalization-based TTA method. Experiments in both standard and wild TTA settings show consistent improvements over strong baselines while mitigating adaptation failure under challenging conditions. The source code is publicly available at https://github.com/delta6189/MixTTA.

Graph-based fraud detection is essential for safeguarding large-scale transaction systems, where undetected anomalies may lead to substantial financial losses and security risks. Real-world fraud graphs pose two coupled challenges: sparse and imbalanced supervision, where verified fraudulent labels are scarce and heavily skewed toward benign accounts, and representation dilution, where spatial message passing may oversmooth camouflaged anomalies while spectral filters may suppress fraud-relevant mid- and high-frequency irregularities. To address these challenges, we propose ADC-GNN, short for Attention-guided Diffusion-Contrastive Graph Neural Network, a unified framework that combines diffusion-guided feature augmentation, contrastive representation learning, and multi-hop spectral attention for few-shot graph fraud detection. The diffusion component is formulated as a feature-space denoising augmentation mechanism rather than a full topology-generative graph diffusion model: it constructs noise-perturbed node-feature views under a cosine schedule and uses contrastive learning to stabilize node representations across perturbations. The spectral attention module further adaptively emphasizes fraud-relevant hop-level and relation-level cues. We evaluate ADC-GNN primarily on three public benchmarks and additionally report a proprietary real-world telecom transaction dataset with approximately 60,000 records as a private case study. Under the 1% training setting, ADC-GNN achieves consistent improvements over original graph fraud baselines and four protocol-consistent recent graph anomaly/fraud baselines on the public benchmarks. Additional analyses on split stability, training ratios, oversampling alternatives, module-level ablations, diffusion schedules, and runtime and memory-consumption comparisons further characterize the effective operating regime of ADC-GNN.

Modern JVM software increasingly integrates multiple programming languages, such as Java, Kotlin, Groovy, and Scala, within a single application. Supporting such interoperability requires JVM compilers to perform cross-language compilation while reconciling subtle semantic differences across language boundaries. Errors in this process can lead to critical miscompilations, yet existing compiler testing techniques focus exclusively on isolated, singlelanguage compilation. To address this gap, we present CrossLangFuzzer, the first differential testing framework for cross-language JVM compilation. CrossLangFuzzer leverages the Kotlin compiler's unified intermediate representation (IR) to synthesize cross-language test programs. It further applies seven mutation operators to diversify generated test programs and improve bug-finding capability. Evaluated on the latest versions of five major JVM compilers, CrossLangFuzzer uncovered 32 confirmed bugs, including 15 in Kotlin, 4 in Groovy, 7 in Scala 3, 2 in Scala 2, and 4 in Java. CrossLangFuzzer is open-source at https://github.com/XYZboom/CrossLangFuzzer

Video generation models have emerged as a promising paradigm for embodied world simulation. However, both general-domain video generators and robot-specific data fine-tuned models can still produce physically implausible manipulations, including discontinuous motion trajectories and inconsistent robot-object interactions, which limits their reliability as world simulators. Through extensive experiments, we find that such physical instability mainly arises from two factors: deformation of moving objects and implausible spatio-temporal correlations among interacting entities, particularly during contact. Building on this observation, we propose PhysisForcing, a scalable training framework that strengthens physical consistency by focusing supervision on physics-informative regions through joint optimization of pixel-level and semantic-level features. The framework consists of a pixel-level trajectory alignment loss, which supervises DiT features using reference point trajectories, and a semantic-level relational alignment loss, which aligns DiT features with inter-region relations extracted from a frozen video understanding encoder. Extensive experiments on R-Bench, PAI-Bench, and EZS-Bench show that PhysisForcing consistently improves embodied video generation over strong baselines, improving the Wan2.2-I2V-A14B and Cosmos3-Nano base models on R-Bench by 22.3\% and 9.2\% (7.1\% and 3.7\% over vanilla finetuning), with the Cosmos3-Nano variant attaining the best overall score. Beyond generation, as a world model under the WorldArena action-planner protocol it raises the closed-loop success rate from 16.0\% to 24.0\% and further improves downstream policy success, indicating that physically aligned video models yield stronger representations for robotic manipulation.

The AI community has framed the relationship between large language models (LLMs) and world models as a dichotomy: LLMs predict tokens; world models simulate reality. Yann LeCun argues in 2022 that reaching general intelligence requires abandoning autoregressive token prediction in favour of latent-space architectures. This framing is unnecessarily binary. Two claims will be defended. First, LLMs are a degenerate special case of world models: the state space is the set of all token sequences, the only action is appending one token, and world models are therefore a strict generalisation of LLMs, not a replacement. Second, there is a natural continuous spectrum from NTP to JEPA, with multi-token prediction, future-summary prediction, and next-latent prediction as intermediate stations already populated by current research. Moving along this spectrum relaxes the LLM constraints one by one. It also progressively surrenders the two practical advantages that make LLMs trainable at scale: internet-scale self-supervised data, and a transformer architecture co-designed for discrete token prediction. Both are examined as open research questions: the data question (the cliff from self-supervised text to instrumented action-labelled environments) and the architecture question (whether the transformer generalises to continuous-state prediction, or whether a new primitive is needed).

This article addresses the combinatorial complexity inherent in modern high-tech system design by presenting automation-in-design (AiD) as a transformative paradigm. We propose computational design synthesis (CDS), a framework utilising deep learning and generative AI to automate the creation of novel systems. Two case studies (e-drive system design and spatial dimensioning problem) serve as proof-points for this approach. The AI-driven methods used in the case studies represent a fundamental shift in engineering, advancing from simulation-based optimisation towards autonomous design with minimal human supervision.

Developers may reference vulnerabilities in pull request discussions through both explicit identifiers, such as CVEs or GHSAs, and implicit security-related language (e.g., "unauthorized access" or "SQL injection"). Prior work has primarily focused on explicit identifiers, potentially overlooking vulnerability discussions that lack formal references. Bots and coding agents are becoming more common in pull requests, raising new questions about how different accounts communicate about vulnerabilities. In this registered report, we describe our planned study of vulnerability communication in pull requests by humans, bots, and coding agents. Building on the AIDev-pop dataset, we analyze explicit vulnerability references and implicit security-related signals across pull request titles, descriptions, review comments, commit messages, and timeline discussions. We further investigate whether these references are associated with vulnerabilities introduced or fixed in the modified code and how they relate to pull request review activity and outcomes. This study contributes a large-scale empirical investigation of vulnerability communication practices in modern software development.

Last-iterate convergence and generalization guarantees in first-order convex learning hinge on the monotonicity of the update operator. While linear averaging preserves the monotonicity of gradient updates, this property is often violated when gradients are aggregated non-affinely, as in modern pipelines enforcing constraints like adaptivity, privacy, robustness or fairness. Whether it is possible to design non-affine aggregation rules that maintain monotonicity has remained an open question. We answer this question negatively: we prove that the monotonicity of aggregated gradients is preserved if and only if the aggregation rule is positively affine. Consequently, non-affine aggregation prevents steady convergence and substantially degrade algorithmic stability. We quantify these drawbacks and propose a path forward by identifying sufficient conditions under which monotonicity can be restored. Our results provide a unified theoretical framework explaining the disparate failure modes observed in modern learning systems.

Neural operators provide deep neural networks for learning mappings between function spaces. Among them, the Fourier Neural Operator (FNO) is particularly effective: its spectral convolution relies on low-dimensional Fourier-domain representations and can handle inputs at different resolutions. This design aligns well with settings where the Fourier basis diagonalizes the underlying operator, such as linear, constant-coefficient PDEs on periodic domains, in which Fourier modes evolve independently. However, nonlinear PDEs may benefit from an additional inductive bias, as they exhibit structured interactions between modes, governed by polynomial nonlinearities. To capture this inductive bias, we introduce the Higher-Order Spectral Convolution, a spectral mixer that extends FNO from diagonal modulation to explicit n-linear mode mixing, aligned with the dynamics of nonlinear PDEs. Our experiments on standard benchmarks show that the proposed Higher-Order FNO (HO-FNO) retains the efficiency of FNO-based architectures and consistently improves over other spectral neural operators. HO-FNO also performs on par with or better than state-of-the-art transformers and state-space models on several datasets, with stronger gains in highly nonlinear regimes, such as the Poisson equation with polynomial forcing, where a single HO-FNO layer outperforms FNO models with up to 16 layers. We open-source our code for reproducibility at: https://github.com/AlexColagrande/HO-FNO.

Software and scientific knowledge co-evolve, yet they are catalogued in separate corpora that rarely speak to one another. We bridge them at global scale by linking World of Code (a near-complete mirror of public version-control history) to Semantic Scholar and OpenAlex through a typed cross-corpus graph of 69.8M edges over eight relation types (paper-to-software mentions, software-to-paper citations, software dependencies, authorship, affiliation, and identity bridges). Anchoring on 18,247 curated science repositories, we ask two reciprocal questions: what is the impact of science on software, and of software on science? To test whether this Science-Software Supply Chain (S3C) view is feasible, we run basic investigations rather than claim a definitive measurement. The two directions appear to illuminate different, complementary strata: the literature's reach into software is dominated by a reproducibility and packaging layer (nf-core, Nextflow, Bioconda) and sequence-analysis tools, whereas software's reach back into science is proxied by a largely invisible machine-learning and data-science infrastructure tier (PyTorch, seaborn, NLTK). The direct paper-names-software channel is too sparse to rank: a human-curated gold benchmark links none of its 65 in-scope cases. Dependency reuse stands in as a proxy and is at most weakly coupled to citation count and to stars (Spearman rho=0.36). Our most cautionary finding is about measurement itself: the reuse-citation coupling flips sign and confidence across two reasonable ways of pairing a repository with a citation count, through papers that name it (n=137, rho=0.05, CI straddling zero) versus DOIs a repository declares for itself (n=1,067, rho=0.13, CI [0.07,0.19]). With linkage this sparse, the sign of a headline correlation depends on which gap one tolerates, so we report both and refrain from a strong decoupling claim.

We introduce a physics-constrained neural network (PCNN) for the rapid prediction of rigorous coupled-wave analysis (RCWA) outputs in the form of Jones matrices. Starting from energy conservation in lossless layered periodic structures, we use the fact that RCWA outputs lie on a Stiefel manifold. This energy constraint is enforced as a hard condition by projecting onto the manifold using differentiable symmetric orthogonalization. The resulting surrogate enforces energy conservation by construction while preserving differentiability for gradient-based inverse design. The performance and generality of the proposed approach are demonstrated through the inverse design of a diffractive waveguide combiner for augmented reality glasses.

Low-Rank Adaptation (LoRA) has become the standard tool for parameter-efficient fine-tuning of large pretrained models. When applied sequentially across tasks in Continual Learning (CL), the standard assumption is that each new task requires a dedicated low-rank adapter. In this work, we challenge this assumption empirically and structurally. We show that task-specific LoRA adapters in CL exhibit significant low-rank redundancy: the subspaces spanned by adapters trained on different tasks substantially overlap, and in many cases earlier adapters can faithfully represent later tasks. Building on this observation, we propose LiteLoRA, a plug-and-play gating mechanism that learns at train time whether to recruit a new adapter or reuse existing low-rank representations. Our method reduces the number of active adapters by 20-70% while matching or exceeding state-of-the-art performance on standard CL benchmarks, revealing that structural redundancy is pervasive and that selective learning is sufficient to achieve stability without sacrificing plasticity.

Frontier large language model training consumes massive accelerator fleets and long wall-clock computation, making stability failures costly when they occur. After a numerical or a hyperparameter fault has already destabilized the training dynamics, it may continue for thousands of steps while loss and gradient norms still appear normal. We study mechanism-driven detection of training instability by deriving internal monitors from the functional role of each critical module and from the earliest computational sites where failures are expected to produce measurable signatures. For low-precision flash attention, we monitor the spectral entropy of a QK bilinear decomposition, whose first-order term becomes abnormal before the loss fully collapses. For MoE routers, we derive indicators from their role in expert selection. Our fault-injection experiments on low-precision attention, large learning-rate, and combined faults show that these signals provide distinct signatures for different failures, triggering thousands of steps before loss divergence.

Degradation-aware prompts, conditions, and latent priors are increasingly used in image restoration, yet they are usually judged by a single endpoint: whether the restored image obtains higher PSNR. This is a weak test of semantics. A condition can help by adding capacity, acting as a global correction bias, or exploiting dataset shortcuts, without becoming an interpretable degradation prior. We propose BiDeMem, a bidirectional degradation memory for explainable image restoration. A query built from restoration features and input statistics retrieves a compact top-k subset of memory slots. The same selected slot identity supports the restoration path at inference time and a training-only forward-degradation explanation path. The study centers on verifiability in a controlled multi-degradation NAFNet setting. New controls separate the gain from a correction head alone, a dense query prior, and a static global prior: these variants are 0.2588, 0.2586, and 0.2839 dB below BiRank, respectively. Strong residual supervision and a wider degradation head also remain below the full bidirectional memory model. Intervention probes show that BiRank preserves restoration quality while increasing wrong-prior and native-prior sensitivity, framing degradation memory as both a restoration module and a falsifiable explanation mechanism.

Traditional meta-heuristics often rely on fixed population sizes, manually chosen search scales, and externally attached parameter-control modules. This paper presents the \textit{Metabolic Multi-Agent Optimizer} (MMAO), a cross-domain optimization framework in which adaptation is derived endogenously from a private-public metabolic resource loop. Each agent carries internal energy, a continuous role state, motion or structural memory, and local search history, while the population shares a communal resource pool. Fitness improvements are converted into normalized metabolic gains through a robust progress scale and a recent success statistic; the same closed loop then regulates sensing intensity, search amplitude, role drift, branching, pruning, respawning, and elite reinvestment. In the continuous setting, MMAO uses energy-regulated symmetric zero-order probing and role-interpolated motion. In the discrete setting, the same control law is instantiated through structural sensing, local route improvement, guided perturbation, and energy-weighted edge reuse. The paper combines an implementation-faithful formulation with a reproducible experimental study on a CEC2017 subset (10D/30D, 20 seeds) and five TSPLIB instances (100 discrete runs in total). The current evidence supports MMAO primarily as a parameter-light, self-calibrating optimization framework whose main validated originality lies in metabolically endogenous resource allocation across heterogeneous search behaviors, rather than as a universally superior optimizer.

We develop a quantitative theory of the Random Language Model (RLM), an ensemble of stochastic context-free grammars, in a scaling limit where the number of hidden symbols $N \to \infty$ while the grammar temperature $\tildeε_d \to 0$ at fixed $x = {\tildeε}_d \log N$. In this limit, the model admits a controlled description based on a large-deviation principle over rule-usage patterns. A semi-annealed approximation maps the problem to a class of Random Energy Models with nontrivial combinatorics. We show that the RLM exhibits a condensation transition at a critical value $x_c=1/8$, below which rule usage concentrates and language statistics acquire a nontrivial dependence on corpus length. A second characteristic scale at $x=1/2$ marks the onset of entropy reduction from its maximal value. Across these regimes, we derive explicit scaling laws for the number of distinct rules, entropy, and related observables, identifying distinct scaling, saturation, and critical regimes controlled by the interplay of grammar size, corpus length, and temperature. The theory resolves previous ambiguities regarding the existence of a thermodynamic transition and explains the slow approach to the large-$N$ limit as a consequence of the dependence on $\log N$. It further provides a unified framework in which universal statistical properties of language emerge from typical realizations of generative grammars, with implications for both natural language statistics and the behavior of large language models.

Vision-based assessment can provide convenient and cost-effective evaluation in Traditional Chinese Medicine (TCM) rehabilitation training, where action quality assessment (AQA) from computer vision offers a promising solution. Existing automatic AQA frameworks for physical therapy typically rely on skeletal data captured from a single viewpoint, which is inefficient for TCM techniques such as acupuncture or Tuina that involve dense hand self-occlusion and complex hand-object interactions. To address these challenges, we propose CME-AQA, a cross-view, multimodal vision-based assessment framework that integrates visual-pose fusion to enhance understanding of environmental context and leverages both first-person and third-person videos during training to improve inference robustness. We collected two dual-view datasets, TCM-AQA61-A (Acupuncture) and TCM-AQA61-T (Tuina), each containing synchronized first-person and third-person recordings of 61 subjects with expert annotations. Experimental results show that our approach achieves superior or comparable mean performance against competitive baselines, achieving over 10% relative improvement in weighted F1 over the best competing method on key rating tasks such as Needle Depth and Quick Needle Insertion, while also reducing mean absolute error in quantitative measures such as insertion time and manipulation frequency. Testing on a CPR dataset further demonstrates comparable performance on several posture-based criteria, suggesting applicability to related structured simulated clinical skill assessments where participant motion is central to evaluation. Overall, CME-AQA enhances assessment accuracy for structured TCM rehabilitation training and facilitates more convenient and effective training-oriented skill evaluation.

Post-hoc controllability of fair machine learning models, the ability to control the trade-off between fairness and accuracy after training, is valuable for practical deployment. Existing post-processing methods provide such post-hoc controllability but often suffer from significant accuracy degradation, whereas in-processing methods achieve efficient trade-offs but require computationally expensive retraining for each change in trade-off ratio. To achieve both post-hoc controllability and efficient trade-offs, we propose a novel fair classification algorithm that learns effective feature representations to improve the trade-off efficiency of post-processing fair classifiers, by a gradient-based optimization approach. Experimental results on real-world datasets demonstrate that our method achieves trade-off efficiency comparable to, or even surpassing, in-processing methods, without requiring any retraining.

Real-world object removal is challenging due to two key difficulties: the target object's non-local effects, such as shadows and reflections, which are difficult to model, and the fact that user-provided masks are often inaccurate or incomplete. With billions of parameters and tens of denoising steps, diffusion-based models achieve strong removal performance at the expense of substantial computational cost, limiting their use in interactive applications and on edge devices. To address these challenges, we present OSOR (One-Step Object Removal), which simultaneously achieves efficient, effect-aware, and mask-robust object removal. Concretely, OSOR introduces: (1) an occupancy-guided discriminator for precise boundary supervision, enabling stable single-step diffusion training; (2) an alpha head that leverages knowledge from pretrained diffusion models to predict appropriate removal regions with minimal overhead, thereby handling imperfect masks; and (3) a semantic-anchored verification pipeline (SAVP) that filters noisy instruction-based triplets to produce effect-aware supervision at scale. Using SAVP, we curate CORNE, which contains 280K verified removal pairs, and further annotate AnimeEraseBench and TextEraseBench to evaluate performance on more complex removal tasks. Experiments show that OSOR surpasses strong multi-step diffusion baselines in perceptual quality while achieving $4\times$ to $30\times$ faster inference.

Micro-expression recognition is challenging due to subtle and short-lived facial muscle movements. Existing methods rely heavily on apex-onset frames, overlook fine-grained inter-frame dynamics, and separately model spatial and temporal information, limiting generalization across datasets. To address these challenges, we propose STAG, a dynamic ROI-AU-coupled spatial-temporal network that jointly models motion flow and adaptive facial connectivity. The framework extracts optical flow from discriminative frames using magnitude-based selection and temporal attention. A dual-branch architecture combines an enhanced graph attention network for structured spatial reasoning with a transformer encoder for temporal modeling. A bidirectional cross-attention module enables mutual refinement of spatial and temporal features, while AU-guided dynamic connectivity adapts facial region interactions according to muscle activation patterns. The transformer captures subtle temporal dynamics beyond apex-based approaches, improving semantic consistency and interpretability for explainable micro-expression recognition. The fused representation is optimized using focal loss and evaluated on CASME II, 4DME, DFME, NaME, SAMM, and SMIC-HS. Extensive experiments demonstrate improved robustness, generalization, interpretability, and computational efficiency, confirming the effectiveness of adaptive relational reasoning, AU-guided dynamic connectivity, and deep spatial-temporal feature fusion for accurate cross-dataset micro-expression recognition.

Knowledge graph question answering (KGQA) aims to answer natural-language questions by reasoning over structured facts. Existing multi-hop KGQA methods mainly rely on topic-centered expansion, which faces two key challenges: the search space rapidly grows with noisy mixed-type paths, and retrieved paths may fail to satisfy the semantic constraints of complex questions. To address these challenges, we propose OPI, an ontology-guided evidence path inference framework for multi-hop KGQA. OPI introduces a relation-centric ontology graph to capture the head-tail type constraints of relations, providing a compact interface for answer-side constraints. Based on this ontology graph, OPI first introduces a bidirectional retrieval mechanism by mapping the predicted answer type to compatible final-hop relations and combining topic-side prefix expansion with answer-side final-hop matching, thereby suppressing noisy mixed-type expansion. OPI further adopts an iterative refinement strategy to reassess retrieved paths and candidate answers under the question context, filtering type-compatible but question-irrelevant evidence for more reliable answer prediction. Experiments on WebQSP, CWQ, and MetaQA show that OPI substantially reduces the search space, improves Hit@1/F1 by 4.6/5.0 points on WebQSP and 8.9/3.3 points on CWQ over the strongest prior results, and achieves near-saturated Hit@1 on MetaQA with the retrieval module alone.

JD.com, one of the world's largest e-commerce platforms, serves over 700 million active users and millions of merchants, with a catalog of tens of billions of SKUs. At this scale, high-quality, structured item knowledge underpins a better consumer experience, lower management costs, and higher operational efficiency-yet producing and serving it poses three industrial-scale challenges: fast-emerging concepts, high-quality knowledge production for massive SKUs, and diverse downstream requirements. To address these challenges, we present the JD Oxygen AI Item Center (Oxygen AIIC), an industrial-scale platform built on LLMs/VLMs for item-knowledge production and service. Oxygen AIIC is built around four core pillars: (i) ontology engineering driven by efficient human-AI collaboration, which supports the dynamic evolution and agile expansion of an ontology with millions of entries; (ii) a "Semantic Search then Discrimination"(S2D) knowledge identification architecture that, combined with throughput improvement strategies, enables scalable, extensible, and high-throughput AI Item Library production for tens of billions of SKUs; (iii) self-evolving item-understanding LLMs/VLMs that improve in a stable and controllable manner, enabling knowledge production with 94.2% precision and 82.8% recall; and (iv) a unified item tunnel that serves as the data and service hub. Oxygen AIIC now covers tens of thousands of JD categories and processes hundreds of millions of item updates per day on Huawei Ascend NPUs. It has accumulated hundreds of billions of item-knowledge assets. Deployed across core business scenarios-including search, recommendation, operations, category planning-Oxygen AIIC has delivered measurable gains at scale. Search-traffic coverage reaches 80.4%, item-information quality issues drop by 37%, the automated fill rate of core attributes during item listing exceeds 80%.

Understanding model predictions is essential for physical applications, where outputs often inform safety-critical decisions, such as structural load assessment, weather warnings, and clinical diagnosis. Shapley values satisfy many desirable properties as an attribution method, but their computational cost during inference hinders their practical use. Current amortized explainers, such as FastSHAP, are limited to homogeneous inputs, which is problematic for physical applications where data often comes from irregular grids and geometries. We introduce OperatorSHAP, a grid-agnostic attribution method and training procedure that allows us to train FastSHAP-like explainers for neural operators. We establish a theoretical framework for attributions in function space, connecting to Aumann-Shapley values. We further show that OperatorSHAP's explanations are consistent with state-of-the-art discrete Shapley values across resolutions and transfer across grid sizes without retraining.

Data fusion, also known as truth discovery, is a data integration problem that aims to determine the correct value or set of values for each attribute of an object when presented with potentially conflicting values from multiple sources. Data fusion tasks belong to two main categories: single-truth scenarios, where each attribute has only one correct value, and multi-truth scenarios, where multiple values can be valid simultaneously. This paper investigates the use of Large Language Models (LLMs) in data fusion tasks for tabular data. Various prompting strategies, encompassing both single-truth and multi-truth scenarios, are investigated empirically. Domain-dependent, domain-independent, zero-shot and one-shot prompts are evaluated on three different benchmark datasets. Experimental results demonstrate that LLM-based approaches outperform traditional unsupervised truth discovery methods, such as DART and LTM, across all datasets. The codebase of this study has been made publicly available on GitHub.

Context and Problem. Software development produces interrelated artifacts like software architecture documentation (SAD), software architecture models (SAMs), and source code, whose relationships are essential for maintenance and consistency checking. However, automatically recovering links between these artifacts (traceability link recovery (TLR)) remains difficult to deploy in practice. Method and Aim. We present an accessible tool landscape for ARDoCo's TLR approaches: the ARDoCo REST API exposes four TLR pipelines (SAD-SAM, SAM-Code, SAD-Code, and SAD-SAM-Code) via HTTP endpoints with asynchronous execution and caching; TraceView is a browser-based frontend with a guided wizard and interactive multi-panel exploration of recovered links and inconsistencies; and TraceViz, which is a VS Code extension that overlays trace links directly onto documentation in the IDE. Results and Conclusion. All three components are publicly deployed and usable. A preliminary study for TraceViz's in-IDE visualization confirmed that it improves developer comprehension during software understanding tasks. The tool landscape makes state-of-the-art TLR accessible to architects, developers, and tool integrators. Video. We provide a screencast of our ARDoCo Tool Landscape and how it is used here: https://youtu.be/IOTEPZQ3tVs

Large language models (LLMs) have increasingly moved from standalone text generation systems to agents that invoke external tools, access environments, and execute multi-step tasks. However, conventional function-calling benchmarks mainly evaluate task completion and API correctness, while privacy evaluation benchmarks typically focus on final responses or privacy judgments. Neither perspective captures purpose-bound information flow across an executed multi-tool trajectory. Motivated by this limitation in current agent evaluation, ToolPrivacyBench audits whether task-private atoms are routed only to authorized tools and downstream sinks, thereby evaluating both task completion and privacy over-disclosure during tool use. The benchmark contains 2,150 cases, including 1,150 fully synthetic privacy-sensitive business workflows and 1,000 cases adapted from existing multi-tool and function-calling benchmarks. Each case is represented by a policy knowledge base. After an agent executes against mock business backends, the evaluator compares recorded tool arguments and backend audit logs with this policy knowledge base. The evaluation covers nine widely used agents to characterize purpose-bound privacy over-disclosure. The results show that successful tool execution does not imply appropriate privacy disclosure: an agent may complete a task while transmitting unnecessary private information through intermediate tool calls. ToolPrivacyBench therefore formalizes a need-to-know disclosure boundary, under which each tool should receive only the information necessary for its stated purpose, and uses trajectory-level auditing to identify privacy over-disclosure in multi-tool workflows.

We present SBridge, a precise approach for identifying functions in binaries that are similar to the given source code functions. Identifying reused code in binaries is critical for security, particularly for detecting propagated vulnerabilities. Although binary-to-binary comparison is feasible, leveraging source code as the reference is more practical because source code is easier to collect and analyze directly without compilation. However, significant gaps between source and binary representations, including function inlining, create challenges in cross-domain function detection. Existing approaches primarily rely on string literals or structural similarities between entire functions, failing to capture detailed code behavior and generating many false alarms. SBridge addresses these limitations through a key innovation: control block-based function matching, which encapsulates essential functional features by segmenting functions into meaningful units such as conditionals and loops. Leveraging control blocks as a cross-domain representation, SBridge enables precise measurement of function similarity between source and binary code, effectively overcoming challenges posed by function inlining and stripped binaries. For evaluation, we collected 3,904 real-world C/C++ binaries from BinKit. In experiments identifying binary functions identical to input source functions, despite approximately 40% of binary functions being inlined, SBridge achieved 75.13% recall@1 and 80.98% recall@5, outperforming existing approaches, which achieved up to 43.31% recall@1 and 50.2% recall@

Language models (LMs) represent tokens using embedding matrices that scale linearly with the vocabulary size. To constrain the parameter footprint, prior work proposes hashing many tokens into a single vector within encoder-only models. While this offers parameter efficiency, many-to-one collisions prevent its use in causal LMs. In this paper, we propose MultiHashFormer, a new framework that allows hash-based autoregression. Each token is represented as a unique hash signature, a short sequence of discrete hash IDs, generated by multiple independent hash functions. A Hash Encoder compresses this signature into a single latent vector for processing by a Transformer decoder. Then, a Hash Decoder generates the hash signature of the next token, which is then mapped back to text. We evaluate our approach at the 100M, 1B and 3B parameter scales, demonstrating that MultiHashFormer consistently outperforms standard Transformer LMs across multiple benchmarks. Furthermore, we show that our model handles multilingual vocabulary expansion with a constant parameter footprint without any modifications.

LLM-as-a-Judge and self-evaluation pipelines implicitly assume that evaluation is easier than generation. We test this in a controlled in-context QA setting where a context passage is the sole information source and each model judges the answer it generated, removing the parametric-knowledge confound of open-domain comparisons. Across four benchmarks (SQuAD 2.0, DROP, HotpotQA, MuSiQue) and two models, evaluation is not uniformly easier: generation accuracy exceeds self-evaluation on three of four, with multi-hop MuSiQue the exception. Attention analysis reveals why: evaluation attends to context 3--5x less than generation does and barely reads the candidate answer. LoRA fine-tuning confirms the asymmetry is not a training artifact: generation fine-tuning induces over-acceptance and evaluation fine-tuning degrades generation. These findings challenge core assumptions in self-evaluation pipelines.

Insurance fraud remains costly and operationally difficult, particularly in call-centre workflows where many customer interactions begin at FNOL. While recent fraud detection methods mainly rely on structured data, text, or images, repeated speaker identity across calls remains underused as an investigative signal. This paper presents DG^VoiC, a voice clustering framework for customer verification and cross-profile speaker linking on anonymised real call-centre audio. The approach combines sensitive information-aligned anonymisation, speech-focused preprocessing, sliding-window speaker embedding extraction, and cosine similarity based clustering to identify repeated speakers under real telephony conditions. The method was evaluated on 121 recordings, with a curated reference subset of 56 samples in 22 human-agreed speaker clusters. used for validation. The best configuration achieved 96% AMI, 95% ARI, 98% completeness, 100% homogeneity, and 99% V-measure. These results show that speaker clustering can provide a strong additional signal for fraud investigation by helping analysts verify speaker consistency and surface repeated voices across customers.

In recent times, Large Language Models (LLMs) are increasingly being used for legal case judgement summarization. Most prior works have tried traditional extractive and abstractive summarization of case judgements. However, hybrid or extractive-abstractive techniques have not been explored much. In this work, we propose a novel tree-of-thoughts inspired extractive-abstractive summarization approach for legal judgement summarization. We conduct experiments using two popular LLMs, DeepSeek and LLama, and compare among extractive, abstractive and extractive-abstractive summarization. Our experiments show that the proposed extractive-abstractive prompt provides better summaries compared to other types of LLM prompts.

Demand for high-resolution satellite imagery has increased interest in super-resolution (SR) to bridge the spatial resolution gap between freely available missions such as Sentinel-2 and commercial systems like PlanetScope. Because no sensor provides true paired low- and high-resolution observations, SR models are usually trained on synthetically degraded data, creating a domain gap on real cross-sensor imagery. In this work, we provide the first systematic study of how this synthetic-to-real mismatch affects the performance of modern diffusion-based SR models. Using a large, geometrically and temporally aligned dataset of Sentinel-2 and PlanetScope imagery, we evaluate five state-of-the-art diffusion architectures under controlled experimental settings. We also introduce LPIPS-Sat, a domain-adapted perceptual metric based on Sentinel-2 self-supervised features. Our results show two persistent challenges: synthetically trained models degrade sharply on real pairs, while models trained on real cross-sensor data exhibit optimisation difficulties and struggle to adapt to the physical and radiometric diversity. These findings highlight a key limitation of current SR and motivate methods that disentangle super-resolution from domain adaptation.

The machine learning(ML) component of an ML-enabled system evolves through retraining, fine-tuning, and optimization, so previously valid test results may no longer hold. A single evolution step can worsen performance on some test cases while improving others, making regression test prioritization inherently directional. We present Gradient-based Behavior Vector-Parameter Delta(GBV-PD), the first approach to operationalize the behavior vector space for evolution-aware regression test prioritization. GBV-PD represents each test case as a gradient-based vector(GBV), a low-dimensional projection of its loss gradient under the original model. It then projects the observed parameter update of the evolved model onto the same PCA basis and uses the resulting alignment to estimate whether each test case's loss is likely to increase or decrease, without running the evolved model on test cases during prioritization. In an empirical study across classification and regression tasks, GBV-PD consistently outperformed non-directional baselines and remained competitive with a full-gradient reference, while offering better time and storage profiles for repeated updates via reusable GBV caching. These results show that behavior-space ideas can be operationalized into a practical and efficient mechanism for repeated-update regression testing of evolving ML-enabled systems.

Neural video codecs have surpassed classical codecs in coding efficiency but remain impractical for deployment due to cross-platform incompatibility and high computational cost. Existing quantization-based solutions fail to produce deterministic results across diverse hardware platforms, leading to catastrophic decoding failures. We introduce MLVC, a hardware-robust neural video codec designed for practical cross-platform inference. The key idea is to explicitly transmit scale parameters through the hyperprior, which guarantees entropy coding consistency across devices without requiring bit-exact arithmetic. While this increases bitrate overhead, we recover most of the coding efficiency through architectural improvements (gated memory, ReGLU activation), a long-term reference recovery mechanism, and domain-specific perceptual training. On the VCD video conferencing benchmark, MLVC achieves >70% BD-rate (MOS) improvement over hardware HEVC, the strongest deployable baseline, while reaching subjective quality competitive with DCVC-RT, which cannot operate across diverse platforms. Both the encoder and decoder run at 100 FPS on average on commodity NPUs from Apple, Intel, and Qualcomm. MLVC is the first neural video codec to combine competitive compression performance, real-time speed, and cross-platform robustness across diverse consumer devices, making it suitable for widespread deployment. Code will be released.

Lifted inference exploits indistinguishabilities in probabilistic graphical models by using a representative for indistinguishable objects, thereby speeding up query answering while maintaining exact answers. In this article, we show how lifting can be applied to efficiently compute causal effects in relational domains. More specifically, we introduce parametric causal factor graphs (PCFGs) to incorporate causal knowledge in lifted models and give a formal semantics of interventions therein. We further present the Lifted Causal Inference (LCI) algorithm to compute causal effects on a lifted level, thereby drastically speeding up causal inference compared to propositional inference, e.g., in causal Bayesian networks. In addition, we present partially directed parametric causal factor graphs (PD-PCFGs) as a generalisation of PCFGs to handle partial causal knowledge and extend LCI to perform lifted causal inference in a PD-PCFG, thereby extending the applicability of lifted causal inference to a broader range of models requiring less prior knowledge about causal relationships.

Headline type-correctness (TC\%) of LLM autoformalization has climbed from $\sim$53\% to $\sim$76\% in two years, yet this scalar conceals which errors each method resolves. We propose a signal-coverage matrix that crosses the Lean elaborator (pass/fail) with a semantic-equivalence judgment (equivalent/not), sorting every output into one of four cells: true success (TS), type-only (TO), semantic-only (SO), or both fail (BF). On ProofNet\# and MiniF2F-test with DeepSeek V4-Pro across Vanilla, Lean-Retry, Sample-Filter, and Stratified Autoformalization (SAF): (1) the +34 to +36 TS gain across the three elab-feedback methods is $\sim$64\% type-stratum recovery, with SO flat on net (87.5\% of original semantic errors rescued, 8 newly created). (2) The TO-to-TS rate is 23/61 for each method (Wilson 95\% CI [26.6\%, 50.3\%]), and this stratum-level recovery rate predicts $Δ$TS on held-out methods to within 2/186 and renders $Δ$TC linear in the Vanilla elab-fail rate across six (model, dataset) cells ($R^2=0.96$). (3) The two judges disagree by 26 to 37 pp on elab-feedback outputs (vs. 7 pp on Vanilla), with 30 to 56\% of symbolic-judge false negatives traceable to elaborator-forced rewrites. The persistent residual reduces to two gold-formalization errors. TC\% gains should be credited by which cell moved, not by the scalar alone.

We propose an agentic Large Language Model (LLM) framework for active Fault-Tolerant Control (FTC) that transforms fault detection outputs into constraint-aware recovery actions grounded in plant-specific knowledge. The approach couples (i) a multi-agent workflow that decomposes operator duties into monitoring, planning, action synthesis, simulation, validation, and reprompting; (ii) a Digital Process Plant Twin (DPPT) that exposes plant data, models, and a simulation service for pre-execution testing; and (iii) a Graph Retrieval-Augmented Generation (Graph RAG) layer built on the CPSMod ontology, which organizes plant knowledge (structure, function, hybrid dynamics, control context, and fault semantics) into a graph that supports relation-aware, multi-hop retrieval for the agents. Corrective actions are generated as minimal-risk state-machine recovery paths and corresponding discrete commands or continuous setpoint adaptations, then validated deterministically against interlocks, envelopes, and dynamic feasibility before any actuation. If no acceptable plan is found within a bounded time window, control is handed to a safety fallback. The framework is evaluated in simulation on two representative benchmarks: a discrete batch Mixing Module and a Continuous Stirred-Tank Reactor (CSTR) under closed-loop PID regulation. Results with lightweight LLMs (GPT-4o-mini and GPT-4.1-mini) show that semantically grounded agents can derive valid recovery decisions within latency budgets compatible with the respective process dynamics, demonstrating a practical pathway from detection to validated corrective action across both discrete and continuous FTC tasks.

Insurance fraud imposes substantial financial losses and operational inefficiencies, raising premiums and impacting trust among legitimate policyholders. Early detection at FNOL remains a persistent challenge. Existing approaches rely largely on private, text-only datasets, limiting progress on multimodal methods that integrate linguistic, behavioural, and speaker-based indicators. We introduce a synthetic multimodal framework that replicates FNOL conditions. It generates agent-customer dialogue transcripts and two-speaker audios, performs ASR and diarisation. Downstream modules combine NER, regex-based feature extraction, LLM-RAG retrieval, and speaker embeddings in a rule-based risk score to flag narrative reuse, structural inconsistencies, and cross-case voice repetition while balancing sensitivity and false positives. Dataset validation and component-level evaluations show stability and transfer potential, offering a reproducible baseline beyond text-only fraud detection.

Benchmarks of machine learning models often include many datasets, making evaluation expensive. For efficiency, it is preferable to perform evaluations on small, representative datasets instead. The selection of such subsets typically relies on heuristics and is rarely analyzed for the robustness of the resulting model rankings. We introduce a framework to perform the task of selecting datasets subsets with an evaluation of how different selection strategies preserve the global model rankings. Our framework includes bootstrap aggregation, which provides valid confidence intervals, allowing a principled comparison of selection strategies. We consider clustering, design criteria (A/D-optimality), random baselines, and greedy farthest-first (FAFI). For the latter, we derive upper bounds on selection quality in terms of ranking errors as a function of the number of selected datasets. Empirically, in time series classification (TSC, 112 datasets) and in a supplementary natural language processing benchmark derived from MTEB (57 tasks), several selection strategies improve rank preservation compared with random subsets, including simple FAFI. In contrast, in recommender systems (30 datasets), the improvement of strategies over random selection is small and typically statistically insignificant. For TSC, our best-performing strategy achieves a Spearman correlation of 0.95 with the full benchmark model rankings using only five selected datasets. Additional experiments indicate that the effectiveness of selection approaches depends on both the quality of dataset representations and the scale of the benchmarking regime.

Multimodal feature fusion can effectively capture complex patterns in real-world data by integrating complementary information from different modalities. However, in many applications, such as boiler combustion monitoring, equipment failure, inconsistent sensor sampling frequencies, and network delays often cause missing modalities and temporal asynchrony. These issues lead to incomplete and disorderly multimodal data. To address them, previous studies have proposed several data fusion methods that align cluster centers before fusion. However, these methods have two key limitations. First, they cannot guarantee accurate sample-level alignment of data pairs. Second, they do not address significant discrepancies in data sizes across different classes, which may affect subsequent fusion performance. To address these problems, we propose a dual-learning based penalized multi-align clustering model, named DLPMAC. The dual-learning mechanism enables the model to learn prior knowledge from each modality, including semantic and structural information. This helps preserve semantic consistency and structural similarity across modalities at both local and global levels. In addition, the penalized multi-align module performs multi-to-multi data alignment through a penalty mechanism. It allows one sample to form data pairs with different samples from other modalities, thereby improving data-pair alignment accuracy. The penalty mechanism also prevents data aggregation, avoiding the case where excessive samples are linked to a single sample. Experimental results demonstrate the effectiveness of DLPMAC in addressing data alignment and fusion challenges from both sampling and clustering perspectives.

We introduce a new, contextual, multilingual dataset called ToxiREX: Toxic REasoning in ConteXt. The dataset consists of threads of Reddit comments and structured characterizations of what the comments imply, following a systematic toxic reasoning schema developed in a previous paper. Using the schema allows us to capture and explain implicit and context-dependent toxicity, while supporting mappings to existing toxicity taxonomies. The dataset includes comments in six languages (English, Arabic, Turkish, Spanish, German, and Dutch), collected from posts connected to specific major events (e.g. the 2023 Turkey earthquakes; the Russian invasion of Ukraine). We describe the context-preserving preprocessing of the threads. We create a training set of 125 thousand comments which is annotated by a commercially available LLM, and a test set of just under three thousand comments that is annotated by native speakers. We show that apparent disagreements in the test set annotations often reflect defensible alternative interpretations rather than noise. Finally, we provide baseline results by prompting and fine-tuning language models. To produce these results, we develop evaluation strategies for our hierarchical, schema-based predictions. While models perform better than random, there remains a lot of room for improvement, showing the task to be challenging. ToxiREX is the first dataset to simultaneously incorporate multiple languages, conversational context, and implicit toxicity, while using the toxic reasoning schema for rich, structured annotations. Dataset available at: https://github.com/cltl/toxirex

A stash is a storage medium such as Dynamic Random Access Memory (DRAM), Solid State Disk (SSD), Hard Disk Drive (HDD), or Non-Volatile Memory (NVM). This paper presents a disaggregated transactional key-value (KV) store, DiStash, that governs KVs cross pools of stash types. It enables an application to use a single transaction to read and write different copies of one or more key-value pair across the different pools of stashes. It simplifies the application logic by (a) preventing undesirable race conditions that may cause copies of data across different stash pools to reflect different values and/or (b) failures that may result in loss of key-value pairs. A configuration of DiStash may use a pool of stashes as either ephemeral or durable storage. The application dictates whether the content of its participating stashes are inclusive (replicated) or exclusive (tiered). We implement a DiStash by extending FoundationDB. We quantify the tradeoffs with its design decisions using microbenchmarks and eBay's production workload. We open source our implementation at https://github.com/ebay-USC/DiStash.

Pixel-space continuous-token autoregressive (AR) generation directly models images as sequences of raw pixel patches, avoiding discrete tokenization or a separately pretrained tokenizer. However, it faces coupled challenges: high-dimensional patch generation causes large single-step errors, and teacher-forced training creates a train--inference gap that makes these errors accumulate across AR steps. Existing fixes such as $x$-prediction and input noise injection only partially mitigate these issues. Exact rollout training better matches inference-time conditions, but is impractical due to prohibitively slow sequential sampling. We propose \emph{Parallel Rollout Approximation} (PRA), a scalable framework that addresses both challenges jointly. PRA generates low-dimensional intermediate states instead of high-dimensional pixel patches, then maps them back to pixel-space tokens with a pixel decoder, preserving a pixel-in, pixel-out AR interface. It also constructs inference-like pixel inputs through the same intermediate-state-to-pixel path used at inference, independently across positions, approximating the pixel-feedback interface encountered during inference-time rollout while retaining parallel teacher-forced training. On class-conditional ImageNet-1K generation at $256\times256$ resolution, PRA-S with 135M parameters achieves an FID of 2.58, surpassing the previous billion-scale pixel-space AR result of 3.60. Scaling to PRA-L with 511M parameters further improves FID to 1.94, establishing a new state of the art among pixel-space AR models. Beyond generation, PRA achieves higher ImageNet classification probing accuracy than other AR and diffusion baselines, suggesting its potential for unified pixel-space image generation and understanding.

Dense embeddings underpin semantic search and RAG, yet a leaked vector store hands much of the underlying text back to whoever holds it. The attacks that make this possible (few-shot alignment, zero-shot inversion, unsupervised cross-space translation) share one weakness: the protected store is a single global geometry that can be aligned to a known one. A secret global rotation, the usual lightweight defence, is no exception: orthogonal Procrustes recovers it once the attacker has about the subspace dimension in known pairs. We introduce Shard, a retrieval-preserving embedding transform that removes this weak axis. The centred embedding is split into a short public prefix (for stage-1 retrieval) and a private residual sharded into C cells under separate secret keys; the residual is reranked under CKKS, where the keys cancel and leave the inner product exact. A single parameter C runs the design from the global-linear baseline it replaces (C=1) to per-document micro-keys (C=N). Because the rerank is full-dimensional, Shard returns the raw-space nDCG@10 that half-SVD truncation gives up; and because the residual is keyed cell-locally, mapping it back to a common frame under a diffuse known-plaintext leak costs roughly C times more anchors (median 200 to 102,400 at C=256), for a few encrypted queries. The short public prefix leaks far less neighbour structure, and a micro-key limit drives the residual graph to zero with an unlinkable, renewable template. The barrier holds against learned, non-linear and unsupervised aligners, and where a matched-utility noise defence de-anonymises almost every probe, Shard de-anonymises none. We are plain about the limits: within a cell the keys cancel, a targeted attacker needs only about d_priv anchors, and an overlapping reference corpus still leaks through the prefix. Shard is an attack-aware geometric defence, not a cryptographic guarantee.

Knowledge-based Visual Question Answering (KB-VQA) requires models to combine image understanding with external knowledge. Most prior methods use a fixed retrieve-then-generate pipeline with a pre-selected retriever and a static top-k setting, which is not adaptive during reasoning. We propose ProMSA, a progressive multimodal search agent for KB-VQA. Given an image-question pair, the agent iteratively chooses image search, text search, or stop, under explicit tool-call budgets and with deduplication to avoid redundant retrieval. For training, we first use rejection-sampling SFT to learn valid tool-use formats, then optimize the agent with TN-GSPO, a sequence-level RL objective that normalizes updates by both generation length and tool-interaction depth. Experiments on E-VQA and InfoSeek show consistent gains over strong RAG and agent baselines, and improved retrieval and end-to-end accuracy. The code is available at https://github.com/DingWu1021/Promsa.

Speech-based cognitive impairment detection offers a noninvasive, accessible alternative to costly biomarker assays, yet transformer-based models remain clinically uninterpretable. We propose a multi-stage explainability framework that translates black-box transformer predictions into clinically grounded narratives by integrating SHapley Additive exPlanations (SHAP)-based token attribution, theory-informed linguistic features, and a four-stage LLM reasoning pipeline using LLaMA-3.1-70B-Instruct. Built on the SpeechCARE-Adaptive Gating Network multimodal screening model (F1 = 72.11% on the NIA PREPARE benchmark), the framework maps model outputs to four cognitive-linguistic dimensions, including lexical richness, syntactic complexity, and semantic coherence. Physician evaluation on 70 stratified English samples demonstrated strong alignment with patient-level cognitive profiles, and a System Usability Scale score of 82/100 indicated high potential for clinical workflow integration.

Real-world knowledge graphs are often incomplete, lacking many valid facts. Knowledge Graph Completion (KGC) aims to predict missing links using known triples, thereby enhancing graph coverage. A key challenge is modeling diverse relational patterns such as symmetry, antisymmetry, inversion, composition and semantic hierarchy. Existing models such as RotatE can capture symmetric, antisymmetric, inverse, and commutative composition patterns, yet struggle with non-commutative composition. Rotate3D addresses this by introducing non-commutativity via three-dimensional rotations, but still fails to capture the semantic hierarchies prevalent in knowledge graphs. Moreover, both models cannot effectively model one-to-many relations. To overcome these limitations, we propose RelBall, which extends Rotate3D with two innovations. First, our model introduces modulus transformation to model hierarchies, driving abstract concepts toward smaller moduli and concrete instances toward larger ones. Second, it introduces a tail-centric relation ball to model one-to-one, one-to-many, many-to-one, and many-to-many relations. RelBall offers the following advantages: (1) coverage of all relational patterns, including the ones mentioned above; (2) an interpretable hierarchical representation where the modulus directly reflect semantic levels; (3) support for one-to-one, one-to-many, many-to-one, and many-to-many relations. Experiments on multiple datasets demonstrate RelBall's competitive link prediction performance against various baselines.

Software engineering is an intellectually demanding, creative discipline that juggles a web of interdependent tasks to design, build, and assure the quality of increasingly complex systems. As our expectations from software soar - with demands spanning AI-driven products, pervasively distributed and cloud-native architectures, and deeply embedded cyber-physical environments - its complexity steadily increases. In response, a new wave of co-engineering methods and tools, fueled by deep learning, has emerged to augment the process, enhancing automation and decision support. Yet, these advances remain far from delivering the kind of intelligent support that modern software development demands. We call for a new paradigm of human-machine cooperation: one where machines don't just automate routine tasks or predict from learned patterns, but actively amplify engineers' reasoning through the lens of causation. As software becomes smarter, a smarter support is needed.

Factual Error Detection (FED), which is the task of identifying factually incorrect spans in a given text, has long been recognized as an important research problem. However, with the rapid rise of large language models (LLMs), research attention has shifted toward factual errors specific to LLM-generated text (hallucinations) and their detection. As a result, the detection of factual errors in human-written text has been relatively neglected. To address this gap, we first distill a taxonomy of human-induced factual errors by analyzing corrections of newspaper articles, a representative source of text that is guaranteed to be human-written and contains few grammatical errors. Our analysis revealed that there are characteristic categories such as kanji misconversions and numeral classifier errors, which are not focused in existing hallucination benchmarks. Based on the taxonomy, we then evaluate the FED capability of vanilla LLMs on synthesized realistic test cases and real corrections. Experimental results demonstrated that even high-performance LLMs such as GPT-5.4 achieved only word-level F1 score of 52% on the synthetic evaluation data, highlighting the task difficulty. Furthermore, a detailed analysis by detection difficulty revealed the current state of FED.

AI agents are promising tools that can act as flexible behavioral nudges to enhance human cooperation in addressing large-scale societal problems. However, evidence on whether AI agents can effectively boost cooperation remains mixed. We recruited 1,283 participants to play iterated Collective Risk Games in small groups, testing whether AI assistants could nudge participants toward cooperation. By using persuasive framing personalized to each player's Social Value Orientation profile, the AI interventions significantly increased contributions and group success rates. These cooperative effects were short-lived, however, fading after the first few rounds. Strikingly, when the AI treatments were reconfigured to promote selfish behavior through exculpatory framing, the negative effects on contributions and group success were larger and substantially more persistent, particularly for personalized interventions. This asymmetry between prosocial and antisocial persuasion highlights the dual-use risks of AI systems designed to influence group behavior in collective action settings.

Counterfactual explanations (CFs) help understand machine learning models by identifying minimal input changes that would lead to alternative model outcomes. Recent work demonstrates their utility for reconstructing black-box models, enabling third-party auditing of opaque decision systems for fairness and accountability. Still, CF-based reconstruction may suffer from decision boundary shifts, overfitting, and restrictive assumptions requiring online query access to target platforms. We propose REconstruction via Counterfactual-Aware waSserstein opTimization (RECAST) under limited data and restricted access, a behavioral surrogate model based on Wasserstein barycentric prototypes. Our approach addresses decision boundary shifts by incorporating CFs as informative, though less representative, samples for both classes, maintaining high surrogate fidelity in low-sample regimes without requiring online access during reconstruction. To enhance fairness auditing, our method enables systematic group fairness diagnostics. Experiments on real-world datasets and various setups show that RECAST effectively achieves high fidelity and query efficiency, as well as stable results even when the access is limited and noisy.

Recent breakthroughs in synaptic-resolution network connectomics have revealed that brain circuits feature fine-scale structural connectivity, such as pairs of correlated synaptic couplings known as second-order motifs. Large-scale recordings of neuronal activity in networks containing nonlinear neurons reveal macroscopic heterogeneous population dynamics throughout the brain. These findings rekindle the inquiry into this intriguing question: Can microscale synaptic structures contribute to macroscopic heterogeneous dynamics and computations in ways that canonical brain circuit models cannot? To answer this question, we create random RNNs with various cell types, nonlinear non-negative neural responses, and arbitrary marginal and second-order correlated synaptic statistics. We derive mean-field low-rank equations for P-population networks in which the pre- and postsynaptic neuronal population identities determine the synaptic and motif strengths. Our framework requires 2P latent dynamic variables with P variables describing mean population activity and P variables capturing within-population variability. Theoretical and simulational results demonstrate that chain motifs induce correlations in synaptic variability, enabling microscopic fluctuations to be integrated and influence mesoscopic mean population dynamics. We apply this framework to reverse engineer network connectivity that recapitulates the heterogeneous activity across the population in the mouse primary visual cortex. By bridging the gap between synaptic organization and nonlinear heterogeneous population dynamics, our results offer a principled approach and testable predictions regarding the relationship between fine-scale connectivity, heterogeneous dynamics, and functional computations.

Sparse autoencoders (SAEs) provide useful decompositions of Transformer residual streams, but their learned features are usually named post hoc rather than directly connected to the Transformer's token vocabulary. We introduce Vocabulary-Aligned Sparse Autoencoder (VASAE), a method that trains SAE features under vocabulary-aligned anchoring and assigns each feature an intrinsic token name: the token string whose embedding is nearest to that feature. Without reducing reconstruction quality compared with a standard SAE, VASAE produces dictionaries with vocabulary-aligned features. Using a 0.8 cutoff on the nearest-token alignment score, dictionaries trained on GPT-2-small post-residual streams align about 90% of features in layers 0--10. In Llama-3.1-8B, representative shallow and middle-layer dictionaries contain strongly aligned features, including 92.8% in the shallow layer, while the representative final-layer dictionary shows limited alignment. After subtracting the sentence-level mean sparse code, case studies show that many remaining intrinsic token names are relevant to nearby input tokens. These results suggest that vocabulary-aligned anchoring can connect learned features to intrinsic token names during training, complementing post hoc interpretation of learned dictionaries.

Protein language models are standard priors for biological sequence generation, but steering them toward explicit distributional design targets remains largely unexplored. We study a constrained protein generation problem in which sequences must match a desired amino-acid (AA) composition profile while preserving plausible sequence statistics and diversity. The motivating application is synthetic feed protein design, where the AA composition of dietary proteins directly determines their nutritional value. We propose a two-stage pipeline in which domain-adaptive fine-tuning (FT) on an in-domain protein dataset is followed by iterative reward-weighted FT via reinforcement learning (RL) anchored against the FT model as a frozen reference. We evaluate the pipeline on two AA compositions and find that FT brings the average composition close to the target, while the subsequent RL enforces specific sequence constraints that FT alone cannot satisfy. We additionally evaluate the design choices of the proposed composition reward term against two baselines and an ablated variant, isolate the contribution of each training stage, and verify that AA composition alignment is achieved without degrading sequence quality.

The widespread collection of fine-grained location data by commercial data brokers creates a re-identification risk that is not widely recognised by the public. While prior research has established that mobility traces are highly unique and that individuals can, in principle, be identified from a handful of spatio-temporal points, such attacks have historically required significant manual effort from skilled analysts, limiting their practical scale. In this feasibility study, we demonstrate in a real world setting that agentic AI fundamentally changes this threat model. We present an end-to-end pipeline in which large language model agents autonomously search the open web, cross-reference public records and social media, and resolve raw coordinate sequences to candidate identities - without human intervention. We evaluate the pipeline on a spatio-temporal dataset containing simulated location points anchored at and around true home and work addresses, focusing on a high-risk disclosure scenario. Our results demonstrate that, from spatio-temporal data and public sources alone, our agentic AI successfully re-identified 18 of the 25 re-identifiable individuals (72%) and 18 of 43 cases overall (41.9%). We discuss implications for Statistical Disclosure Control (SDC) practice and outline the near-future escalation that data custodians and regulators must anticipate. De facto anonymity - an implicit foundation of SDC practice - is shifting. Agentic AI strengthens the case that re-identification is reasonably likely by any means under the GDPR Recital-26 standard, at costs of minutes-and-dollars per target.

Performance numbers reported for hardware are accepted on trust: the reader cannot recompute them, the apparatus is gone, and the silicon itself can be silently wrong, with fleet studies reporting on the order of one core in a thousand returning incorrect arithmetic with no error raised. We make a reported hardware measurement a tamper-evident, independently checkable record. Every quantity in the text, a table, or a figure is bound, by its content hash, to the observation and the verification behind it; the whole is a hash-linked, append-only structure (a transparency log for measurement) that a verifier audits offline without trusting its producer. Matrix products are verified by a probabilistic identity (Freivalds) at O(k n^2) cost under a tolerance we derive from floating-point error analysis and calibrate to the device's own measured residual floor, so a wrong product is rejected with probability 1 - 2^(-k); quantities with no such identity carry an algebraic checksum and a measured reproducibility class. We then treat the check itself as a security object: a probe seed committed for offline reproducibility is an attack surface, and a probe-aware adversary can hide a corruption in the probe's null space, fooling even a quorum of bit-identical witnesses, while a Fiat-Shamir challenge derived from the claimed output closes this. Driving the device from an unprivileged tenant's reach, with a di/dt power virus and a thermal soak, neither moves the calibrated tolerance nor produces a silent error, placing the physical-fault threat at the rare defective part or the privileged attacker and marking the boundary at which the record must compose with a hardware root of trust. We demonstrate the construction across Blackwell and Hopper GPUs and report a residual-floor and reproducibility map by precision, size, and device.

Geometry Problem Solving have increasingly adopt the neuro-symbolic paradigm, combining neural intuition with symbolic rigor. However, current frameworks suffer from severe bottlenecks in two core stages: autoformalization, which treats multimodal translation as a static task decoupled from downstream solver compatibility, and theorem prediction, where solvers frequently hit a deductive impasse due to fixed rule libraries. To address these, we propose SD-GPS, a solver-driven framework that treats the symbolic solver as an execution oracle throughout both formalization and deduction. First, Solver-Driven Autoformalization unifies supervised formal-language adaptation and solvability-guided reinforcement learning into a single module built on QwenVL3-2B, making executability the central training signal. Second, Verified Theorem Proposing introduces an impasse-aware agent that proposes local auxiliary lemmas from current proof states, ensuring soundness by filtering all proposals through symbolic verification. Empirical evaluations on Geometry3K and PGPS9K demonstrate that SD-GPS consistently outperforms existing MLLM, neural, and neuro-symbolic methods across standard completion, multiple-choice, and cross-modal reference regimes, proving that closing the loop between multimodal perception and symbolic execution significantly improves geometric reasoning, offering profound insights into how neural agents can be grounded by formal systems to achieve verifiable problem-solving capabilities.

We present Home3D 1.0, a modular image-to-3D generation system that produces high-quality 3D assets from a single reference image, targeting interior design and e-commerce applications. Given a photograph of a furniture or decor item, the system outputs a mesh with physically-based rendering (PBR) materials, and the mesh can be decomposed into material-specific components. The pipeline is organized into four tightly coupled modules: Geometry reconstructs a watertight mesh through latent SDF modelling with a geometry VAE and a coarse-to-fine flow-matching DiT; Texture predicts multiview albedo observations, reprojects them onto the mesh, and completes unseen surface regions with a 3D texture field; Material uses MatWeaver to obtain component masks through video-based segmentation and UV-space voting, then retrieves and bakes PBR maps from a curated material library through hierarchical multi-modal matching; and Parts generates material-editable semantic part meshes with a PartVAE and PartDiT, decoding multi-head part-specific SDF fields in one pass. Each module is evaluated independently with dedicated metrics, highlighting both the current system capability and the remaining gaps toward broader deployment.

Current multimodal reflection mechanisms for long video understanding predominantly rely on closed-loop self-reflection within internal parameters. Lacking objective external evidence, models are frequently trapped in blind confidence and often fail to correct errors. Furthermore, applying reinforcement learning to multi-stage reflection pipelines introduces severe policy coupling, which is exacerbated by a critical scarcity of dedicated training data. To address these limitations, this work proposes Reflect-R1, the first Evidence-Driven self-correction framework for long video understanding. The framework constructs a three-stage pipeline consisting of intuition, verification, and arbitration. By dynamically retrieving objective visual evidence to verify initial intuitions and autonomously executing multiple temporal searches to resolve conflicts, it completely breaks the hallucination loop. To overcome policy coupling, we design a stage-decoupled reinforcement learning algorithm named SD-GRPO that independently computes advantage functions across different reasoning stages. Concurrently, we construct a dataset of 120K samples to bridge the training data gap. Extensive experiments on benchmarks such as VideoMME and LongVideoBench demonstrate that Reflect-R1 achieves state-of-the-art performance. Our method significantly improves the genuine rectification rate and enables authentic self-correction strictly grounded in objective evidence.

Traditionally, the architecture of high-performance computing (HPC) systems is tailored for speed, while highly secure computer systems must sacrifice speed for security. However, a wide range of scientific domains, such as the life sciences, call for a combination of performance and security to allow processing sensitive data at scale. Here, we present RAMSES (Research Accelerator for Modeling and Simulation with Enhanced Security), an HPC system designed from the ground up to deliver high performance within a robust security framework. RAMSES integrates hardware-based memory encryption of AMD processors with state-of-the-art file encryption from IBM Storage Scale and the Thales CipherTrust manager, establishing an HPC platform that ensures continuous encryption throughout the data life cycle - at rest, in transit, and in use - in compliance with major data protection standards (European General Data Protection Regulation, ISO/IEC 27001 certification, and Federal Information Processing Standards). In addition, we implemented advanced operating system hardening, a multi-layered security architecture, and mandatory multi-factor authentication to adapt the HPC environment to increased security demands. Benchmark results from the biomedical sector demonstrate that the performance impact of the secure environment is limited and that integration of the conflicting requirements speed and security can be achieved while preserving a coherent, flexible, and user-friendly system.

As a prominent symptom of Parkinson's disease (PD), turning impairment is evaluated through parameters such as turning angle, duration, and particularly, the number of steps required to complete a turn, which directly reflects motor dysfunction. Accurate step counting is challenging due to variability in real-world turning movements and atypical shuffling patterns in parkinsonian gait. Existing methods are predominantly wearable-based, requiring users to wear and manage dedicated devices, which can be inconvenient for continuous daily use. To address this, we propose a passive, video-based framework that estimates step count in a coarse-to-fine manner using diverse motion representations. Specifically, an initial step count is estimated from foot movement signals derived from 3D human mesh recovery, providing high-level motion structures. To incorporate fine-grained motion details, a motion encoder learns complementary gait dynamics from mesh and optical flow to refine the initial estimate. In this process, coarse foot movement signals query the pixel-level motion cues via cross attention to capture subtle parkinsonian gait dynamics. To handle varying video lengths, we partition each video into clips and integrate clip-wise motion embeddings via multiple instance learning (MIL) for step count residual prediction. Extensive experiments show our method consistently outperforms existing step counting methods on real-world PD turning datasets.

Contextual bandits with graph-structured arms arise in recommendation, citation retrieval, and social advertising, where arms connected on a graph tend to share reward signal. Standard dimensionality reduction ignores this structure, inflating exploration cost by a factor of $d/k$. We propose GraphDR-LinUCB, which projects arm features onto the graph's low-frequency spectral subspace and runs linear UCB in the resulting $k$-dimensional space. We prove the first $\wtO(k\sqrt{T})$ regret bound for spectral-projection-based contextual bandits, reducing dimension dependence from $d$ to $k$; a perturbation argument extends this to noisy graphs, with an explicit penalty for reward-smoothness mismatch and graph-estimation error. Our central theoretical finding is that the high-frequency reward component need not incur a worst-case linear-in-$T$ penalty: its actual cost depends on its realized impact along the played path, not on its total energy. A simple spectral comparison between subspaces ($Γ_k$) predicts which reducer wins on a given dataset, correctly calling five of six real-dataset outcomes without any fitted threshold. Across a synthetic benchmark and six real datasets (MovieLens, Amazon, LastFM, ogbn-arxiv, MIND), GraphDR-LinUCB reduces cumulative regret by $15\times$ over full-dimensional LinUCB and outperforms competing graph-aware methods on five of six; the single failure is precisely where the graph's spectral subspace is misaligned with the reward.

Theory-of-mind evaluations of large language models typically use dyadic social-deduction games, where every observable cue points to a single hidden side, so a model with strong language priors can score well without ever simulating opponents' incentives. We extend the Werewolf game with a Jester, a third faction whose utility on peer suspicion is inverted because it wins by being voted out, so optimal play requires reasoning across three opposing utility functions. Across 60 games on GPT-4.1, DeepSeek-V3.1, and Llama-3.3-70B with Jester self-learning on and off, the Jester wins 60-70% of games while Werewolves never exceed 20%, and GPT-4.1 wolves vote the Jester out on day 1 in 60-70% of games, a strictly self-defeating action. Self-learning helps DeepSeek and Llama but hurts GPT-4.1, with the cost landing on Villagers rather than Werewolves. Only DeepSeek learns the subtle strategy of looking suspicious without looking intentionally suspicious, and it gains the most from the loop. Triadic incentive structure exposes a layer of multi-agent reasoning that dyadic deduction games leave invisible.

Long-term time series forecasting finds extensive applications in domains such as power demand, traffic flow, meteorological observation, and renewable energy dispatch. Forecasting dynamically varying long-term time series poses inherent challenges, including statistical nonstationarity, local high-frequency disturbances, and coupled cross-period dependencies, which make it difficult for lightweight models to balance parameter efficiency and forecasting performance. To address this issue, this study presents TA-SparseMG, a lightweight cross-period forecasting model built on SparseTSF's sparse cross-period modeling framework. It incorporates three key modules: a trend-aware reversible instance normalization module, a scale-adaptive gated denoising module, and a multiscale gated-attention MLP forecasting module. The trend-aware normalization module captures input-window statistics and calibrates forecast-window distributions, effectively mitigating distribution shift. The scale-adaptive gated denoising module performs feature smoothing and residual suppression before period rearrangement, thereby reducing interference from high-frequency perturbations. The multiscale gated attention prediction module strengthens the prediction head's adaptive representational capacity via conditional gating and feature modulation. Extensive experiments across multiple LTSF benchmarks demonstrate that the proposed TA-SparseMG consistently achieves superior, stable performance. Ablation studies confirm that each module independently improves distribution adaptation, input robustness, and cross-period feature mapping capability.

Recent phone-class mobile SoCs expose practical NPU execution paths for on-device vision-language model (VLM) inference, but developers still lack phase-level guidance for mapping VLM pipelines across heterogeneous backends. We present a hardware-in-the-loop characterization of VLM inference on the Qualcomm SM8750 (Snapdragon 8 Elite), covering phase throughput, cache-state effects, 100-run thermal stability, energy, heterogeneous CPU/NPU pipeline configurations, and visual-token-budget sensitivity. Using FastVLM-0.5B as an end-to-end case study, together with encoder-only measurements across four architecture families, we show that phase matters: NPU execution is highly phase-dependent, delivering 1.64x speedup for prefill but only 1.18x for decode, while vision encoders achieve 20-45x speedups over CPU. These gains translate into 10.47 degrees C lower steady-state temperature and 2.52x lower energy, avoiding thermal throttling in always-on settings. Finally, we show that a four-step graph rewrite enables previously unsupported encoders, such as Phi-3.5-V, to reach the QNN path with up to 22x speedup, providing a practical porting recipe for mobile VLM deployment.

Differentiable partial differential equation (PDE) solvers underpin solver-in-the-loop ML training, gradient-based optimal control, and inverse problems, yet the practical cost of obtaining correct, usable gradients from a given solver on a given problem is largely undocumented. Integration effort, computational cost, gradient accuracy, and numerical conditioning vary widely across solvers and are discoverable only by trial and error. We introduce Mosaic, an extensible benchmarking framework for differentiable PDE solvers that standardizes access to solver gradients. Each solver is packaged as a containerized component (Tesseract) exposing a uniform gradient API regardless of language or automatic differentiation (AD) strategy, enabling researchers to evaluate, compare, and build on non-trivial physical solvers. Our evaluation of 14 solvers across fluid dynamics, structural mechanics, and heat transfer demonstrates that the benchmark surfaces practically relevant differences: order-of-magnitude variation in computational cost and Jacobian conditioning, alongside structural incompatibilities that eliminate solvers from realistic tasks entirely. Despite this variation, all solvers that produce gradients converge to similar optima, indicating that the practical barriers are memory limits, numerical stability, and setup compatibility rather than gradient accuracy alone. Mosaic is open-source and available at https://github.com/pasteurlabs/mosaic.

Wearable devices and Internet of Things (IoT) sensors require on-sensor processing of biosignals and environmental data, including computationally demanding operations such as nonlinear activation functions for neural network inference, sensor calibration curves to map raw readings to physical units, and signal preprocessing functions like logarithmic compression and power operations for feature extraction. These functions exhibit significant complexity, often involving transcendental operations and multivariate dependencies that are costly to implement digitally. Analog function approximation provides a power-efficient alternative by performing these computations in the analog domain, thereby reducing the energy overhead associated with analog-to-digital conversion and subsequent digital processing. Flexible Electronics (FE) present a particularly attractive platform for wearable applications due to mechanical flexibility and low-cost fabrication, but impose strict constraints on circuit density and power consumption, making efficient analog implementations critical but challenging. This work introduces Analog Kolmogorov-Arnold Networks (AKANs), developed via hardware-software co-optimization, to approximate these complex multivariate functions accurately under hardware imperfections. Our method incorporates circuit-level error modeling during training and applies pruning at both software and hardware levels to reduce area and power. Validation across multiple benchmarks demonstrates that our proposed pruning methodology not only reduces hardware cost but can also improve approximation accuracy by regularizing spline parameters. Results show up to 55% area and 50% power savings, with average reductions of nearly 30% across datasets, highlighting AKANs as a robust and generalizable framework for low-power analog function approximation in FE.

Recent advances in Human Activity Recognition (HAR) from wearable sensors have shown that multi-modal deep learning models consistently outperform their uni-modal counterparts. Modalities can include IMUs, RGB cameras, audio signals, and others. One important aspect of multi-modal deep learning is the sensor fusion approach we apply. Over recent years, multiple fusion paradigms have been proposed for multi-modal HAR. However, to the best of our knowledge, no head-to-head comparison of these paradigms exists on a common multi-modal HAR benchmark dataset. To address this research gap, we systematically compare seven state-of-the-art sensor fusion methods on the recently released HARMES dataset, which comprises 61 hours of fully labeled IMU, audio, and ambient humidity data. The chosen dataset focuses on 15 household and personal hygiene activities of daily living (ADLs). By applying the seven different fusion techniques to a state-of-the-art multi-modal model architecture, we show that Gated Multi-modal Fusion achieves the highest macro F1-score (0.82), surpassing the concatenation-based late fusion HARMES paper baseline of 0.76 by +6pp under leave-one-participant-out evaluation. All code used in our experiments is made publicly available on GitHub.

Mixed-precision computation has been introduced in deep neural networks (DNNs) as an effective approach to reduce latency, energy consumption, and memory footprint. However, efficiently mapping mixed-precision networks onto multi-precision spatial architectures poses several challenges. These include determining the appropriate precision for each layer, balancing layer-wise accuracy sensitivity to quantization against architectural heterogeneity and system-level constraints, and accurately estimating the system-level cost of heterogeneous precision assignments. This work presents SEADA, an efficient methodology designed to address these challenges. SEADA comprises: (i) a configurable system-level analytical cost model of a multi-precision spatial accelerator architecture; (ii) a fast mapping tool that identifies near-optimal mappings of DNN workloads onto the target integer accelerator; (iii) analytical models for floating-point layers to estimate the overall benefits of mixed-precision execution; and (iv) a per-layer precision selection methodology based on bit-level entropy, enabling efficient assignment across multiple numerical precisions. SEADA's efficiency provides designers with a robust framework for the design-space exploration of multi-precision architectures.

Temporal variation poses a unique challenge for named entity recognition (NER) in historical texts, where entities drift in surface form and salience across time. While language models (LMs) have made progress in various NLP tasks, their ability to reason about temporality, especially in diachronic contexts, remains limited or at least, questionable. In this paper, we systematically study how temporal metadata can be structurally embedded into NER models using a range of lightweight fusion strategies. We experiment with both absolute and relative temporal representations, injected into Transformer-based architectures via early or late fusion mechanisms such as cross-attention, adapters, and concatenation. Our evaluations on French and German historical datasets reveal that late fusion strategies yield more robust and temporally generalisable performance, particularly in early and noisy periods.

Spatial intelligence is essential for low-altitude unmanned aerial vehicle (UAV) perception, collaboration, and navigation. However, existing UAV benchmarks often emphasize image-level recognition, single-view understanding, or narrow answer formats, leaving 3D spatial inference, multi-view collaboration, scene dynamics, and diverse task formulations insufficiently evaluated. To address these gaps, we introduce SpatialUAV, a real low-altitude UAV benchmark comprising 4,331 curated instances across 14 fine-grained task types, covering semantic discrimination, spatial relation, aerial--aerial collaboration, aerial--ground collaboration, and motion understanding. SpatialUAV organizes all samples into a unified visual-input--question--answer schema, while supporting seven input configurations and nine answer formats, including option labels, region identifiers, geometric values, cross-view correspondences, and free-form motion descriptions. To ensure reliable and grounded evaluation, our data construction pipeline integrates detector-assisted regions, depth supervision, metadata-derived rules, extensive manual annotation, blind filtering, and multi-turn human validation, together with task-specific metrics for heterogeneous outputs. Evaluating representative vision-language models across three categories, we show that current models remain far from human-level performance, with pronounced bottlenecks in cross-view association, structured grounding, geometric reasoning, and temporal viewpoint understanding. These results offer empirical guidance for advancing low-altitude UAV spatial intelligence. Code and data are available at https://github.com/Hyu-Zhang/SpatialUAV.

Vision-Language-Action (VLA) models have demonstrated strong capabilities in robotic manipulation, but their performance degrades significantly in long-horizon tasks due to cumulative error propagation. This limitation largely arises from static feature fusion mechanisms that rely on fixed weights to combine visual, language, and action representations, preventing the model from adapting to different phases of task execution. To address this limitation, we propose S$^2$-VLA, a framework that introduces a State-Space Guided Adaptive Attention (SSGAA) mechanism. SSGAA maintains a belief state that tracks task progression and generates dynamic gating weights to adaptively fuse information from three complementary sources visual features for spatial perception, task intents for high-level task planning, and temporal action sequences for execution consistency. This adaptive fusion allows the model to shift its focus throughout task execution, aligning with the evolving requirements of different task stages. Despite its compact 2B parameter size, S$^2$-VLA consistently outperforms larger 7B-scale models and achieves state-of-the-art performance on long-horizon manipulation benchmarks, including LIBERO and SimplerEnv. highlighting the importance of adaptive feature fusion for long-horizon robotic manipulation.

Mixture-of-Experts (MoE) language models scale model ability with sparsely activated experts, making this architecture a standard recipe for modern large models. However, sparse activation does not remove the deployment burden of storing and serving all experts, and the available deployment budget can vary substantially across devices, users, and workloads. Existing MoE compression methods are still largely fixed-budget, typically optimizing one compressed endpoint at each chosen target budget. We study a different setting: converting a large pretrained MoE LLM into a nested family of deployable subnetworks across budgets. Our method first ranks expert FFN channels by their importance, then lets each expert learn a discrete action to prune its channels. By gradually increasing cost pressure, a single action-training run exports a series of action masks from high to low budgets, each of which identifies a reliable smaller subnetwork nested in the ranked base model. Moreover, we use a single recovery fine-tune at a mid pruning budget (40%) to recover degraded model quality and transfer the recovered model to other unseen budgets. Overall, our framework surpasses recent MoE compression baselines. Specifically, on Qwen2-57B-A14B, our method retains ~99.8% of base performance while pruning 50% of routed expert parameters even without fine-tuning. For deployment, our pruned subnetworks deliver real memory reduction and throughput gains, and further support realtime online budget switching with kernel-level co-design.

Vision transformers have become a dominant architecture for visual recognition. However, standard models do not explicitly encode the planar symmetries that arise in many vision domains. We introduce a family of vision transformers equivariant to arbitrary discrete subgroups of $\mathrm{O}(2)$, providing a unified framework that generalizes prior flipping- and $D_4$-equivariant transformer architectures. Our construction yields equivariant analogues of the core transformer components, together with expressivity guarantees for the resulting layers. In particular, we show that whenever $H \le G$, the class of $G$-equivariant ViTs embeds naturally into the class of $H$-equivariant ViTs. We also prove that, in the single-head setting, the corresponding equivariant self-attention layer realizes every $G$-equivariant self-attention map representable by ordinary self-attention. We further construct a $D_6$-equivariant model based on hexagonal patches, making the architecture compatible with six-fold rotational symmetries. We evaluate the resulting models on the PatternNet aerial image dataset in artificially data-scarce regimes across subgroups of $D_4$ and $D_6$. Our experiments compare two equivariant attention mechanisms and analyze how the choice of homogeneous-space configurations used in the nonlinearities affects performance. Preliminary results under matched parameter budgets indicate that equivariance can improve recognition accuracy, motivating further study of how discrete symmetry groups shape transformer-based visual recognition models.

Demand forecasting at the bottom of a retail hierarchy requires predicting tens of thousands of correlated long-horizon series across products, stores, and regions. Modern systems must scale across massive catalogs, capture shared demand dynamics, and remain interpretable enough to be trusted. Classical statistical methods need a separate model per series and are hard to manage at scale; deep autoregressive models struggle as the joint state grows to tens of thousands of dimensions; and recent graph-based forecasters, while capturing cross-entity dependencies, often produce opaque long-horizon forecasts. We propose GNBAN (Graph Neural Basis Attention Network), an end-to-end architecture combining heterogeneous graph representation learning with an interpretable basis-decomposition head. Retail data are represented directly as a heterogeneous graph derived from the relational schema, so a single model serves the entire catalog. Rather than predicting the horizon directly, GNBAN decomposes each forecast into trend, seasonal, and generic components. Its key innovation is a per-basis attention mechanism: each basis function keeps its own learnable query and retrieves information independently from the entity's historical neighborhood, letting different bases specialize to distinct temporal patterns while preserving interpretability. On two large-scale benchmarks, M5 Walmart and Favorita Grocery Sales, evaluated under matched protocols, GNBAN improves volume-weighted WRMSSE by roughly 4-5% over a matched graph baseline. Qualitative analysis shows the learned decomposition exposes trend, seasonal, and residual demand drivers without post-hoc explanation methods. These results demonstrate that scalable relational forecasting and interpretable forecast decomposition can be achieved together in a unified graph-based framework.

Deep learning models for surface electromyography (sEMG) can benefit substantially from subject-specific (re-)calibration, since no sufficiently large and diverse datasets are available to train fully generic decoders. However, for user acceptance, the number of repetitions that can realistically be collected during calibration is severely limited, which increases the risk of overfitting and, in extreme cases, can even degrade performance compared to the uncalibrated model. Classical overfitting indicators such as validation performance and regularization with early stopping are difficult to apply in this low-sample regime, as they require additional held-out data that is rarely available in practical calibration scenarios. In this work, we investigate a recently proposed class of memorization indicators based solely on the activation statistics of rectified linear units (ReLU) in deep neural networks, which can be computed directly from training data without any extra validation set. We conduct a transferlearning experiment on a benchmark sEMG dataset, where a convolutional neural network is first pre-trained on multiple subjects and subsequently fine-tuned on individual users using only a small number of repetitions. During calibration, we monitor both decoding performance and the activation behaviour of the last hidden layer. Our results provide first evidence that decreases in test accuracy during fine-tuning are ac companied by characteristic changes in activation rates, indicating that activation-based memorization indicators are a promising tool for early spotting of unsuccessful learning in low-sample sEMG calibration settings.

The widespread adoption of Artificial Intelligence (AI) has led to increasing concerns about energy consumption, yet there is a lack of standardized methodologies to accurately estimate AI inference energy consumption, particularly across various tasks and architectures. In this study, we propose a task independent, layer-wise energy estimation model for AI architectures. Our model is evaluated on a large dataset of more than 100,000 layers for 295 neural network architectures across 3 widely-used tasks and 3 distinct hardware platforms. Our approach achieves a median error of 19.6%, outperforming state-of-the-art methods. We further show that layer-wise decomposition generalize to new tasks without complete retraining, by leveraging shared layers across architectures. It offer tools, insights and a precise methodology to empower stakeholders in designing energy-efficient AI systems.

Statistical adversarial detection (SAD) treats detection as a two-sample test. Given a reference set of clean examples (CEs) and a batch of queries, potentially containing an unknown mixture of CEs and adversarial examples (AEs), SAD decides whether the query distribution drifts away from the CE distribution while controlling the false-alarm rate. Existing SAD-based methods mainly use maximum mean discrepancy (MMD) to measure the distributional discrepancy. However, MMD's distributional properties limit its ability to capture characteristic uncertainty patterns of AEs that are crucial for detection: AEs typically exhibit abnormal feature spread (i.e., global uncertainty) and instability under perturbations (i.e., local uncertainty). To close the gap, we propose Uncertainty-aware Statistical Adversarial Detection (USAD), which explicitly captures these uncertainty patterns with two new statistics: (1) Variance Discrepancy (VD), which measures the difference in feature spread between AEs and CEs to capture global uncertainty differences. (2) Perturbation-based Covariance Discrepancy (PCD), which compares feature covariance under Gaussian perturbations to capture local uncertainty differences. By aggregating VD and PCD, USAD achieves superior detection performances over baseline methods against various adversarial attacks, highlighting the importance of considering characteristic behaviors of AEs for effective SAD. Our code is available at: https://anonymous.4open.science/r/USAD.

This paper addresses the lack of explicit memory mechanisms in current object detection models and proposes Hippocampus-DETR, a novel detection framework based on biological hippocampal memory modeling. This framework integrates a hippocampal memory network module, HipNet, into the DETR architecture and systematically simulates the anatomical structure and functional organization of hippocampal subregions, including the entorhinal cortex, dentate gyrus, CA3, CA1, and subiculum. Through this design, Hippocampus-DETR realizes pattern separation, pattern completion, importance filtering, and information integration of visual encoding features. During training, different memory submodules are optimized using a layer-wise training strategy, ultimately forming a memory system with memory retrieval and completion capabilities. Experimental results demonstrate that Hippocampus-DETR achieves higher detection accuracy than current mainstream models. More importantly, models equipped with this framework also exhibit excellent generalization ability and data efficiency in tasks such as few-shot image classification, multimodal feature construction, and image restoration. Subsequent experiments further validate the functional necessity and internal interpretability of each memory submodule. This study not only provides a novel object detection framework, but also offers a feasible technical pathway for integrating neurocognitive mechanisms with deep learning models, highlighting its significant value in improving model learning efficiency and task robustness. The project is available at https://github.com/2186cloud/hipnet.

Multimodal large language models (MLLMs) are increasingly deployed as embodied planners in egocentric environments, where task success requires not only achieving instructed goals but also acting in socially appropriate ways. While explicit goals may render certain actions optimal, implicit social norms often impose hidden constraints. Existing evaluations typically focus on explicit goal achievement or direct norm knowledge, seldom assessing whether planners can infer and apply these hidden constraints within action sequences. We introduce NormAct, a benchmark for embodied social-norm interactions that evaluates plans on Goal Achievement, Norm Compliance, and overall Task Success. NormAct uniquely embeds hidden norms within ordinary tasks, testing whether models can realize them without explicit instruction. Experiments with state-of-the-art MLLMs (GPT-5.4, Claude Opus 4.7, Gemini 3 Pro) reveal a significant gap: models achieve explicit goals in 67.3\% of cases, but comply with hidden norms in only 26.4\%. Cue-condition experiments indicate that this gap stems not from a lack of general social knowledge, but from challenges in activating and grounding relevant norms in context. To address this, we propose NormPerceptor, a context-conditioned cue generator that infers scene-relevant norms prior to planning, increasing Task Success from 24.2\% to 46.7\%. Our results underscore the importance of enabling embodied agents to proactively detect hidden norms, ground them in visual evidence, and integrate them as action-planning constraints. Our benchmark is publicly available at https://huggingface.co/datasets/Caleb196x/NormAct.

Peptides are a promising therapeutic modality that combine the chemical tunability of small molecules with the target specificity of macromolecular therapeutics. However, designing antigen-specific binding peptides while avoiding toxicity remains a major challenge for therapeutic peptide discovery. Here, we present Pepti-drift, a toxicity-aware latent refinement framework that generates peptide candidates through a single antigen-conditioned drift step. In a peptide embedding space, Pepti-drift learns to attract generated peptide latents toward antigen-matched binding peptides while repelling them from toxicity-associated regions. This is challenging because binding-promoting physicochemical features often overlap with toxicity-associated features in peptide representation space. To address this, we introduce a warm-up strategy to stabilize this competing objective by first learning binding-oriented attraction and then increasing toxicity repulsion.

Traffic matrices (TMs) capture network-wide origin-destination demand and are central to traffic engineering, yet accurate whole-matrix forecasting remains challenging when prediction must be performed under the memory, update, and training-budget constraints of online network control. This paper investigates whether compact quantum-inspired recurrent models can provide effective TM forecasts without relying on dedicated graph, transformer, or diffusion modules. We adapt gated quantum-inspired Kolmogorov-Arnold network fast-weight programmers (QKAN-FWPs) to direct multi-step Abilene TM forecasting, where each model predicts the next 20 five-minute frames of a 144-channel origin-destination (OD) matrix from a two-hour history. We benchmark three QKAN placement variants against a matched-size long short-term memory (LSTM) network, a larger LSTM, and a classical gated fast-weight programmer under a shared fixed-budget training protocol. Among the evaluated recurrent models, G-QKANFWP achieves the best pooled root-mean-square error (RMSE), while using only 22.4% of the larger LSTM. It also outperforms both the matched-size LSTM and the classical G-FWP baseline, indicating that the gain is not due to gated fast-weight framework alone. Convergence and channel-wise analyses further show that the quantum-inspired variants obtain lower validation-loss area under the learning curve (AULC) than matched-size recurrent baselines, while G-QKANFWP and GQKAN-FWP achieve substantially more OD-channel wins. These results identify a classical slow programmer with a quantum-inspired fast programmer as a promising accuracy-efficiency design for resource-conscious network traffic-matrix forecasting.

Time-triggered communication protocols rely on trusted components known as guardians to enforce adherence to predetermined network schedules. Network-agnostic guardians offer an efficient and scalable distributed solution with reduced implementation cost and complexity compared to network-aware alternatives. However, this efficiency is based on the guardian's dependence on the controlled node for clock synchronization, which introduces a vulnerability: a malicious node can exploit this dependency to launch timing attacks against its guardian and eventually interfere with messages from other nodes on the network. In this paper, we establish a theoretical lower bound on the attainable clock synchronization precision between a node and its network-agnostic guardian. Building on this result, we introduce a timing attack that leverages the unavoidably imperfect clock synchrony to cause controlled and undetected de-synchronization of the guardian. The attack enables a malicious node to cause collisions with targeted critical network messages. We evaluate the effectiveness of the attack using a FlexRay field bus network model implemented in the OMNeT++ simulation framework. Our results show that the attack is able to remain undetected with 100% success and disrupts the transmission of the critical messages of the target node by causing collisions with them with 100% success.

We present MMD-Reg, a novel correspondence-free approach to point-cloud registration that is differentiable and has linear computational complexity in the number of points. We model registration as a nonlinear least-squares problem based on the Maximum Mean Discrepancy, approximated using random Fourier features. The resulting objective can be solved efficiently with standard methods such as Levenberg-Marquardt, and the solution is differentiable via the implicit function theorem. This allows MMD-Reg to be used as a differentiable optimization layer within end-to-end trainable models, supporting registration under challenging conditions such as poor initial alignment and partial overlap. We demonstrate this Neural MMD-Reg formulation by integrating the layer with a set transformer, training the resulting model in supervised and unsupervised settings, and comparing its performance against recent learning-based methods. We also evaluate standalone MMD-Reg, comparing its accuracy and scalability against widely used non-learning-based registration methods.

Dynamic Time Warping (DTW) is a cornerstone for time series classification, but its reliance on Euclidean distances fails to capture latent cross-channel correlations in complex multivariate data. We propose a hybrid Quantum Dynamic Time Warping (qDTW) architecture, replacing the classical distance metric with the parameterized geometry of a quantum Hilbert space. Through structural ablation on benchmarks up to $C=8$ spatial dimensions, we establish fundamental topological rules for quantum sequence alignment. We introduce a Unified Pre-Embedding Adjoint Ansatz that decouples trainable entanglement from classical data, eliminating the severe phase-scrambling and information bottlenecks inherent to traditional measurements. We demonstrate this decoupled architecture allows untrained quantum kernels to act as highly expressive baselines, while parameterized training effectively untangles deeply overlapping hyper-dimensional data. Furthermore, we identify a strict spatial-temporal expressivity tradeoff: temporal depth (data re-uploading) is necessary for dimensionally restricted univariate circuits, but applying it to wide multi-qubit registers triggers chaotic frequency-spectrum explosions and representation collapse. By navigating these topological hazards, our multivariate quantum architecture outperforms classical baselines, setting a new standard for integrating parameterized quantum circuits with dynamic programming

Training small language-model agents for long-horizon interactive tasks requires both fast imitation and reward-driven improvement. On-policy distillation (OPD) provides dense teacher guidance and typically improves rapidly in the early stage, but its gains saturate once the student approaches the teacher, limiting the final performance ceiling. Reinforcement learning (RL) directly optimizes environment rewards and encourages exploratory improvement toward a higher reward-defined ceiling, but sparse and delayed feedback makes early-stage learning much less efficient than OPD. In this paper, we propose ATOD (Annealed Turn-aware On-policy Distillation), a hybrid online distillation algorithm that explicitly exploits this complementarity. (1) ATOD uses an annealed OPD-RL schedule: OPD dominates early training to approach teacher-level behavior, while RL is gradually strengthened to drive reward-based exploration. (2) ATOD introduces Turn-level Disagreement-Uncertainty Reweighting (T-DUR), which softly amplifies high-utility turns and improves dense supervision in long trajectories. Experiments on ALFWorld, WebShop, and Search-QA show that ATOD consistently outperforms competing post-training baselines: across the three student sizes, ATOD improves average success rate by 3.03 points over OPD and 23.62 points over GRPO, while surpassing the corresponding teacher models by 2.16 points.

A minute lexical variation can reverse the procedural meaning of an instruction even when the rest of the sentence remains unchanged. In automotive maintenance instructions, this pattern often appears when an action phrase turns an instruction into its procedural counterpart. The entities, modifiers, and surrounding context remain largely invariant, while the action phrase determines the procedural relation. We define this task as Complementary Action Modeling (CAM). Given a maintenance instruction, the goal is to identify or generate its procedural counterpart by modifying the action phrase while preserving the remaining sentence context. This task focuses on three aspects: distinguishing complementarity from surface similarity, controlling generation at the action-phrase level, and evaluating relational correctness using retrieval, overlap-based, and human evaluation. Using a German automotive maintenance dataset, we examine these questions through candidate matching and controlled Seq2Seq generation. The results show that complementary maintenance instructions are best modeled as procedural associations grounded in subtle lexical cues. They should therefore not be treated as ordinary cases of sentence similarity or synonym-based paraphrasing.

World models for language agents come in two useful forms. An agent-based world model calls an LLM API and reasons flexibly in language, but its errors appear as hallucinated state changes that are hard to score with ordinary regression losses. A parameterized world model is a trained transition predictor; its errors are easier to measure with quantities such as NodeMSE, delta accuracy, and validity accuracy, but it is usually weaker as a standalone planner. We compare these two families on four graph-structured planning benchmarks and introduce operational hallucination metrics for the agent-based case. The comparison motivates \textbf{Grounded Iterative Language Planning} (GILP), which trains only a small parameterized backbone and combines it with API-based agent reasoning. The backbone supplies valid actions, predicted state deltas, risk, and value; the LLM drafts an action and imagined delta; and a consistency gate asks for revision when the two disagree. On real GPT-4o-mini calls, GILP reduces hallucinated-state rate from 0.176 to 0.035. In calibrated simulator ablations, it raises success from 0.668 to 0.838 while adding only ~22% extra LLM calls.

Hierarchical predictive coding provides an interpretable framework for perception as error-driven inference in multi-layer generative models, while sparse coding imposes parsimonious latent representations through explicit sparsity constraints. Their combination yields hierarchical sparse predictive coding models with appealing computational and neuroscientific properties, but practical use is often limited by the cost of iterative latent inference. In such models, each input may require many recurrent refinement steps before a useful sparse representation is obtained, and this burden becomes more severe as the hierarchy deepens. We study this bottleneck by holding the hierarchical sparse energy fixed and varying the inference procedure. The comparison includes four schemes: classical iterative inference based on ISTA, an accelerated MFISTA reference, structurally informed amortized inference using a LISTA-style bottom-up encoder adapted to the hierarchical model, and a hybrid method in which this fast amortized initialization is followed by a small number of corrective energy-based refinement steps. Under this shared objective, we measure reconstruction quality, sparsity, latency, and stability on static image benchmarks. The results show that a shallow LISTA-style initializer plus short corrective recurrence improves over pure amortization while remaining much faster than long iterative inference.

Reliable monitoring of hydroelectric generators requires descriptors that capture both electrical loading and electromagnetic field behavior. This work investigates operating-regime identification in the Porjus U9 10-MW Kaplan hydrogenerator using synchronized measurements from ten stator-mounted Hall probes and six rotor-current channels. Seven steady guide-vane-opening settings are considered, and each 300s record is divided into 1s windows. The resulting windows are represented by spatial Fourier descriptors of the circumferential air-gap field, probe-wise temporal flux indicators, and channel-wise RMS rotor-current features. Correlation analysis and principal component analysis are used to examine how the feature groups vary with the operating point, and Random Forest, radial-basis-function support vector classification, and multilayer perceptron models are evaluated for supervised identification of the guide-vane-opening state. The analysis shows that RMS rotor-current features mainly track the loading axis, while the magnetic-flux features reveal complementary information associated with spatial imbalance, waveform distortion, and weak low-frequency modulation. Spatial descriptors alone provide limited separability, yielding test accuracies below 27%, whereas rotor-current features alone reach about 84-85%. Combining flux and current information gives the most discriminative representation; the SVC-RBF model achieves 99.5% test accuracy and macro-F1 score. The results indicate that distributed air-gap magnetic sensing, when fused with rotor-current measurements, can support accurate and interpretable data-driven monitoring of Kaplan hydrogenerator operating regimes.

Knowledge Distillation (KD) enables training smaller student models under the guidance of larger teacher models, and the widely adopted TRL library implements it. Yet, TRL treats both models symmetrically, missing opportunities to exploit their pronounced asymmetry in memory footprint, and communication requirements. This paper presents an HPC-aware methodology for KD that decouples teacher and student partitioning efficiently. Our approach achieves up to 67% higher samples-per-second than TRL by avoiding unnecessary teacher-model data structures and selecting the best split strategy. We combine vertical and horizontal partitioning of models, deriving an analytical expression that identifies the existence of inflection points between splitting regimes. These results showed that exploiting teacher--student asymmetry through topology-aware parallelism notably accelerated GKD training on production HPC clusters at our company

Long-context language models suffer from position bias, where information in middle positions is underutilized. Attention Sorting addresses this by iteratively reordering documents based on attention patterns, but its multiple sort-and-generate cycles increase deployment cost. We hypothesize that position bias is the primary bottleneck and propose Debiased One-Pass Attention Sorting, which estimates a per-prompt position-bias curve from the low-attention majority of documents and uses it to correct raw attention scores (via subtraction or division) to enable single-pass sorting. Our experiments on two models refute this hypothesis in the tested setting: on LLaMA-2-7B-32K-Instruct, debiasing produces identical results to uncalibrated single-pass sorting (94.83\% containment accuracy), while on YaRN-Llama-2-7b-64k, debiasing improves accuracy by 8.67 percentage points but remains 14.84pp behind iterative sorting, closing only 37\% of the gap. These results suggest that position-bias correction is insufficient to match iterative sorting, and that repeated reordering provides additional benefits beyond bias correction.

Hybrid attention models that mix full and sliding-window attention across layers offer a promising approach to efficient long-context inference, but the critical question of \emph{which layers} should retain full attention remains unsolved. Existing methods use either fixed periodic patterns or attention-based heuristics that may not capture what matters for downstream accuracy. We propose NLL-guided layer selection, a training-free method that directly measures each layer's importance by computing the negative log-likelihood degradation on answer tokens when that layer uses sliding-window instead of full attention. On LongMemEval with Qwen3-4B, our method achieves 64.6\% accuracy using only 1/4 full-attention layers, matching the 1/2-FA periodic baseline (65.0\%) while halving the computational budget. NLL-guided selection outperforms the SWAA-reported periodic 1/4-FA baseline by 10.4 percentage points and a matched LightTransfer-style baseline by 26.4 percentage points. De-confounding analysis shows the signal is consistent with long-range attention needs rather than generic layer sensitivity. The method requires only $\sim$15 minutes of one-time calibration, advancing the efficiency-accuracy Pareto frontier for long-context LLM deployment.

Retrieval-augmented generation (RAG) enhances LLMs by incorporating external knowledge to support response generation. However, conflicts between retrieved context and parametric knowledge have emerged as a critical challenge in RAG systems. To mitigate such conflicts, numerous studies have attempted to identify and edit knowledge-related internal neurons, aiming to improve the ability of LLMs to rely on contextual evidence during generation. However, these neuron-level approaches may introduce unintended cascading effects that compromise the general capabilities of LLMs, as the modified neurons are often entangled with broader model behaviors and functionalities. In this paper, we introduce SHIFT, a novel framework that reformulates neuron-level modification as learnable gate modulation, allowing LLMs to adaptively regulate internal activations for knowledge conflict resolution. Technically, our SHIFT equips LLMs with a lightweight gate module and optimizes fewer than 0.01% trainable parameters while keeping the backbone model frozen. During generation, the gate module adjusts the model's internal representations to adaptively leverage contextual and parametric knowledge. Extensive experiments on six datasets validate the effectiveness of our SHIFT in comparison with various competing baselines. All datasets and code are available at https://github.com/OpenBMB/SHIFT.

Training-free compression methods for large language models (LLMs) often use calibration data to guide compression decisions. ROCKET, a recent method combining sparse-dictionary factorization with multi-choice knapsack problem (MCKP) allocation, derives its per-layer factorization from an output reconstruction objective but uses weight-space Frobenius error as the MCKP allocation cost. We investigate whether aligning the allocation cost with the output-space objective improves compressed model fidelity. On Qwen3-8B at 50\% compression, our ROCKET-ActCost achieves +0.8 percentage points higher average accuracy across 8 zero-shot benchmarks (53.1\% vs 52.3\%), but increases WikiText perplexity by 16\% (61.46 vs 52.98). This accuracy-perplexity tradeoff reveals that different allocation objectives favor different downstream metrics. The high correlation ($>$0.99) between weight-space and output-space errors limits allocation divergence, explaining the modest effect size. On Llama-3.2-1B at 20\% compression, the two methods produce near-identical results (53.3\% vs 53.5\% accuracy, 14.45 vs 14.66 PPL), suggesting that the effect of the cost function is minor at lower compression ratios.

The existence of adversarial attacks is often attributed to the presence of non-robust features in neural networks. While prior defenses reduce their impact via pruning, masking, or feature recalibration, we instead propose to jointly learn to amplify and attenuate these signals through a simple activation scaling mechanism. To this end, we introduce Activation Amplification and Attenuation (A3), a lightweight plug-in module that enhances adversarial robustness with minimal modifications of the activations. A3 dynamically rescales the activations using a learnable mask and a scaling factor derived from the original activation magnitudes. The influence of adversarial perturbations can be amplified or attenuated using the same learnable parameters by simply flipping the sign of the scaling operation. The amplified signals serve as negative references to construct novel contrastive and ranking loss functions. Experimental analysis shows that learning to degrade the predictions in amplification mode simultaneously improves adversarial robustness in attenuation mode. Moreover, A3 relies on only a small number of learnable parameters, with most of its behavior being determined by the scaling mechanism rather than additional network capacity. Extensive experiments demonstrate that integrating A3 into different backbones, datasets, and training methods consistently improves adversarial robustness while introducing negligible computational and memory overhead compared to existing plug-in modules. Code is available at: https://github.com/tgoncalv/A3.

Continuous attractor neural networks (CANNs) are the canonical computational framework for how the brain encodes continuous variables such as spatial position, head direction, and movement direction, and explain the activity of hippocampal place cells, entorhinal grid cells, and head-direction cells. CANN research, however, is fragmented: most results rest on lab-specific implementations, general-purpose simulators lack CANN-specific abstractions, and the path from spike trains to attractor geometry in real recordings lacks a standardized toolkit. Here, we present a comprehensive open-source toolkit that unifies the full CANN research workflow. It combines three tightly integrated components: 1) canns, a Python library on BrainPy/JAX that provides standardized 1D/2D CANNs, spike-frequency-adaptation variants, grid cell networks, hierarchical path-integration models, and brain-inspired attractor architectures, together with curated datasets, task generators, an analyzer module and trainer modules for biologically plausible plasticity; 2) canns-lib, a Rust acceleration backend delivering hundreds-of-times speedups for spatial-navigation workloads and modest gains for Ripser-based persistent homology; 3) ASA (Attractor Structure Analyzer), a PySide6 pipeline applying persistent homology and cohomology to experimental neural recordings to detect ring-like and toroidal attractor signatures in real data. The toolkit ships with full-detail reproducible pipelines that recover recent CANN results including SFA-driven anticipative tracking, theta sweeps in head-direction/place/grid systems, and hierarchical path integration.

World models are often used for planning by rolling learned dynamics forward. Many planning environments, however, are not vectors or images; they are graphs of agents, tools, skills, routes, and dependencies. In these settings, a local prediction error may stay local or spread through the graph, and the failure mode changes again when edges are predicted rather than fixed. This paper studies long-horizon rollout error in Graph World Models (GWMs). We formulate a unified fixed-edge and dynamic-edge GWM framework with action nodes for node-, edge-, and graph-level decisions. We develop graph-valued rollout bounds that separate topology-induced amplification from model-induced amplification, and we introduce a joint node-edge operator for dynamic-edge rollouts. Guided by the analysis, we propose Error-Aware GWM, which combines spectral regularization, rollout consistency, and critical-node weighting. Across synthetic topologies and heterogeneous agent-graph testbeds, rollout error and planning regret grow with horizon, dynamic-edge training is needed when structure evolves, and Error-Aware GWM prevents long-horizon divergence while preserving prediction accuracy. Real-world graph benchmarks clarify the scope of GWMs: they are most useful for dynamic graph rollout and agent planning, while specialized graph models remain strong on static or sparse prediction tasks.

Reinforcement learning (RL) post-training improves the reward alignment of flow-based generators, but often degrades perceptual quality in ways that are not captured by the reward proxy. We identify a simple structural signature of this drift: across three post-training methods (NFT, AWM, DPO), RL fine-tuning inflates the per-step velocity norm $\|v_θ\|$ by $5\%$ to $15\%$ relative to the reference. A form of norm inflation has been studied in classifier-free guidance (CFG), where rescaling the velocity back to a reference norm at inference time can mitigate the resulting artifacts. However, this inference-time correction does not transfer cleanly to RL: rescaling $v_θ$ to match $\|v_{\text{ref}}\|$ at inference time neither improves reward nor fixes the quality degradation, because the inflation is co-adapted into the model weights. Furthermore, an adjoint sensitivity analysis shows that velocity magnitude rescaling carries no coherent first-order reward signal at the batch level, indicating that suppressing norm inflation is unlikely to remove a consistently reward-carrying component. Since inference-time renormalization fails while norm suppression carries no reward cost, training-time intervention is the appropriate strategy. Together, these findings motivate \methodname, a hinge penalty that activates only when $\|v_θ\|$ exceeds $\|v_{\text{ref}}\|$ and composes additively with any velocity-local base loss. Across two base models, three post-training methods, and two reward proxies, \methodname consistently improves MLLM-judged image quality and forensic realism while preserving reward, with gains that amplify under few-step inference and are not explained by early stopping.

Optimizing functionals over the space of probability measures is now ubiquitous in machine learning. A widely used approach is to perform the optimization directly over the Wasserstein space, but many objective functionals of practical interest are non-convex along Wasserstein geodesics, making the analysis of standard first-order methods challenging. In this work, we study a class of objectives over the Wasserstein space that admit a difference-of-convex (DC) decomposition and we lift the classical convex-concave procedure (CCCP) to this setting. Under smoothness and strong convexity assumptions on the convex components of the decomposition, we prove almost stationarity along the iterates of the resulting algorithm. Our main focus is on the Maximum Mean Discrepancy (MMD) and the Energy Distance (ED) functionals, for which we develop explicit Wasserstein DC decompositions, and establish local convergence of the scheme under mild assumptions. Empirically, we show that well-chosen DC decompositions yield faster and more stable convergence than Wasserstein gradient descent on these MMD objectives.

Offline reinforcement learning enables policy learning from fixed datasets without additional environment interaction, making it appealing for safety-critical applications where online exploration is costly or unsafe. Diffusion-based decision-making methods have recently achieved strong performance in offline RL by modeling rich, multimodal trajectory distributions. However, existing diffusion planners are typically risk-neutral and therefore may overlook rare but catastrophic outcomes that are crucial in real-world deployment. In this work, we propose RS-Diffuser, a risk-sensitive offline diffusion planning framework that combines diffusion-based trajectory generation with distributional value critics. RS-Diffuser learns a diffusion planner over future state trajectories, a separate inverse dynamics model for action decoding, and a Monte Carlo distributional critic that estimates the full return distribution of candidate plans through quantile regression. At sampling time, we incorporate a risk-sensitive guidance signal into the denoising process, using gradients computed from tail-aware objectives such as Conditional Value at Risk to steer generation toward desired risk profiles. As a result, a single trained model can flexibly produce risk-averse, risk-neutral, or risk-seeking behaviors by changing only the inference-time risk parameter. Extensive experiments on risk-sensitive D4RL and risky robot navigation benchmarks demonstrate that RS-Diffuser achieves state-of-the-art performance, improving both overall return and worst-case robustness while reducing safety violations.

Constrained single-objective numerical optimization requires a careful balance among feasibility, objective convergence, and computational efficiency under a fixed function-evaluation budget. This paper proposes DE-2LS, a late-stage, locally search-enhanced variant of differential evolution built on the RDEx framework. The proposed method preserves the original RDEx components, including mutation and crossover operators, success-history adaptation, archive mechanism, population-size reduction, and $ε$-based constraint handling. A lightweight coordinate-pattern local search is added as a guarded polishing component around the current best solution. It is activated only in the late stage of the run, uses a small evaluation budget, and accepts candidates through a feasibility-aware comparison rule. Ablation results show that the finalized DE-2LS configuration achieves the best U-score among all tested variants, confirming that controlled late-stage refinement is more effective than aggressive or premature local search. In the direct comparison with RDEx, DE-2LS achieves a 5.58\% gain in U-score. In the four-algorithm comparison, DE-2LS obtains the highest overall U-score of 80968 and the best total rank of 48 among RDEx, CL-SRDE, and UDE-III. These results indicate that DE-2LS improves the exploitation capability of the RDEx-based search framework while preserving its speed advantage under the combined speed-accuracy scoring criterion. The source code of DE-2LS is available at https://github.com/ChauhanDikshit?tab=repositories.

Unconstrained single-objective numerical optimization requires a careful balance among global exploration, late-stage exploitation, and function-evaluation efficiency. This paper presents DE-2LS, a late-stage, local-search-enhanced differential evolution framework built on RDEx for unconstrained single-objective optimization with variable bounds. The proposed method preserves the original RDEx evolutionary search engine and introduces two conservative refinements: a smoothed exploitation-biased branch-rate update in the late search stage and a guarded coordinate-pattern local-search that serves as a budget-aware refinement mechanism. Since the considered setting is unconstrained apart from variable bounds, all selection and local-search acceptance decisions are based solely on objective values. To determine the final algorithm configuration, we conduct a staged ablation study by testing multiple settings of the EB-rate smoothing mechanism, the initial EB-rate, the standard-branch Gaussian sampling scale, the selection-pressure parameters, and the local-search coefficients. The final configuration is selected using a U-score-based evaluation that jointly reflects solution quality and convergence speed. Experimental results show that DE-2LS consistently improves the original RDEx in direct head-to-head comparison. In particular, DE-2LS increases the U-score from $33602.0$ to $37448.0$, corresponding to an improvement of $11.45\%$. Moreover, compared with several competitive and IEEE CEC-winning algorithms, DE-2LS achieves the best overall U-score of $178966.5$, outperforming the others by $34.43\%$. These results show that a carefully designed late-stage local-search strategy can improve both convergence speed and the final objective quality of the algorithm. The source code of DE-2LS is available at https://github.com/ChauhanDikshit?tab=repositories.

Training binary neural networks (BNNs) from scratch is dominated by the straight-through estimator (STE), whose forward/backward mismatch produces severe accuracy degradation as networks deepen. We study an orthogonal axis: when and where binarization is enforced during training. We introduce StoMPP (Stochastic Masked Partial Progressive Binarization), which gradually replaces clipped weights and activations with their hard binary counterparts layer by layer from input to output, using stochastic partial masks with soft refresh. StoMPP delivers two complementary benefits. As a standalone training rule, it provides a fully STE-free procedure that improves over vanilla STE with gains that grow with depth (ResNet-50 BNN: +18.0/+13.5/+3.8 on CIFAR-10/100/ImageNet), and the pattern holds across ResNet-18/34/50, MobileNetV2, and BERT fine-tuning. Composed with surrogate gradients by applying STE only to frozen entries, it reaches +27.1/+19.8/+17.7 over vanilla STE on the same setting. Underlying both regimes is a single mechanistic finding: progression order is decisive. Forward layerwise progression prevents depth collapse, reverse progression collapses to near-chance, and binary-weight networks (without binary activations) are insensitive to order. We trace this asymmetry to activation-induced gradient blockades: a committed binary activation severs gradient flow upstream, and ordering controls when these blockades form. To isolate the progression's contribution from any benefit conferred by STE, we conduct all ablations in the STE-free regime; the resulting characterization (schedule, refresh, ordering, dynamics) thus reflects the progression itself rather than its interaction with surrogate gradients.

Large language models (LLMs) have attracted widespread attention from academia and industry, yet their deployment raises critical security concerns regarding robustness and reliability. Planning, a core component of intelligent behavior, remains challenging for LLMs, which often produce infeasible or incorrect solutions in long-horizon decision-making tasks due to inherent complexity. In this paper, we propose a symbolic feedback-driven iterative self-refinement framework to enhance the robustness and reliability of LLMs in long-horizon planning. Specifically, a natural language prompting mechanism is introduced to map logical symbols into natural language descriptions, enabling LLMs to better capture task constraints and semantics. We further design a symbolic verifier that identifies errors and converts them into corrective instructions interpretable by the LLM, thereby guiding self-refinement. In addition, we leverage a plan recognizer to infer goal reachability, facilitating more effective guidance toward desired goals. Empirical results demonstrate that the proposed framework consistently improves both feasibility and correctness in long-horizon planning tasks. This highlights its effectiveness in enhancing the reliability of LLM-based planning and potential to enable more trustworthy AI systems.

Vision-Language-Action (VLA) models enable instruction-driven robotic manipulation, but they inherit oversized language backbones from pretrained VLMs whose capacity far exceeds what is needed for short robotic instructions. This raises a basic question: how much of a VLA model is actually necessary for closed-loop control? In this work, we study architectural redundancy in VLA models by using transformer block removal as a controlled intervention. We introduce \textbf{Drop-Then-Recovery (DTR)}, an analysis protocol that removes selected blocks from a pretrained VLA model and then fine-tunes the resulting model to measure whether the removed capacity was necessary for downstream control. To make this intervention reliable, we propose \textbf{GateProbe}, a one-shot virtual-gate sensitivity metric that ranks blocks by their contribution to the downstream action loss. Across multiple VLA architectures, manipulation benchmarks and even real-robot industrial scenarios, we find a strong asymmetry in post-removal recoverability: \ul{\textit{language backbones are highly redundant for standard robotic manipulation tasks, whereas vision and action pathways are substantially less tolerant to removal}}. On LIBERO, removing half of the LLM blocks even improves OpenVLA-OFT from 95.0% to 98.3% under the same downstream fine-tuning budget, and retaining only two language blocks still recovers baseline-level performance. These results suggest that current VLA benchmarks may exert limited pressure on deep language grounding and compositional instruction understanding, and that future VLA architectures should allocate capacity more deliberately across language, vision, and action components. The code is available at https://github.com/s1ghhh/VLADrop.

Single-cell perturbation models can reduce costly wet-lab screening by predicting how cells respond transcriptionally to interventions. While recent generative models improve population-level prediction, individual generated cells are not explicitly checked for biological consistency. We introduce PerturbCellRL, a reinforcement learning (RL) framework that post-trains a pretrained single-cell transcriptomic generator using a suite of cell-level verifiers as rewards. These verifiers define four rewards: Pearson top-k similarity, RMSE top-k proximity, DE Spearman, and Pathway activity. The Pathway activity verifier rewards cells whose pathway responses match known perturbation biology. We evaluate PerturbCellRL on multiple genetic and chemical perturbation benchmarks. Across these benchmarks, PerturbCellRL improves over the pretrained flow-matching generator on reward-aligned evaluation metrics and a held-out evaluation metric. Moreover, PerturbCellRL remains competitive with state-of-the-art methods on population-level metrics. Together, these results frame trustworthy single-cell prediction as verifier-guided generative alignment, moving beyond matching expression distributions toward predictions whose single-cell perturbation effects are explicitly checked for biological consistency.

This report investigates the extension of pretrained General-Purpose Audio Tagging (GP-AT) models toward spatially grounded Sound Event Localization and Detection (SELD). The proposed AT2SELD framework couples a pretrained AT backbone with compact First-Order Ambisonics (FOA) spatial processing, track-wise SED and Cartesian DOA estimation, permutation aware supervision, and calibration. It characterizes how semantic audio priors support localization-aware scene analysis under data, computation, and deployment constraints. The framework is developed through informed multi-stage Neural Architecture Search (NAS). Stage 1 shows that spectral FOA descriptors, based on magnitude, phase, and Intensity Vectors (IVs), provide the most reliable interface for semantic-to-spatial transfer. Stage 2 identifies early residual spatial encoding as the main capacity-sensitive component, while late track-wise abstraction and recurrent smoothing act mainly as refinement stages. Stage 3 shows that late cross-stitch coupling improves semantic-spatial interaction, whereas early fusion is costlier and less effective. Diagnostic evaluation analyzes the selected architecture under class balancing, focal loss, activity-conditioned DOA supervision, threshold calibration, and transfer across STARSS23, TAU2019, TAU-NIGENS2020, and TAU-NIGENS2021. Focal loss improves the activity point, active-only DOA supervision mitigates inactive target dominance, and validation-selected thresholds recover calibration without replacing spatial learning. Cross-dataset and oracle-activity analyses indicate strong fixed source localization on TAU2019, transferable representations from TAU NIGENS2021, and meaningful but uncertain behavior on STARSS23. Overall, GP-AT priors appear promising for SELD design when embedded in spatial-aware architectures and optimized through integrated calibration and deployment oriented strategies.

Multi-vehicle collaborative perception (MvCP) is considered as a key technology to facilitate automated driving (AD), where real-time MvCP under limited resources is significant for reliable AD. In this paper, we formulate a lightweight acceleration scheme for intermediate-fusion (IF) MvCP, which can adapt to both situations of limited computation and communication resources. We provide a relaxed definition conditional additivity and analyze the conditional additivity for various DNN linear layers. On this basis, we focus on the IF-MvCP based on additive feature fusion, and derive the MvCP precision consistency of the forward and backward feature fusion position (FP) adjustments among linear layers. Through experiments, we further validate the precision consistency of the FP adjustment method. Moreover, we propose an FP adjustment among linear layers (FALL) scheme for MvCP acceleration without precision loss theoretically. Simulation results show that the proposed FALL can reduce MvCP latency by up to 74.8% under limited communication resources and by up to 30.3% under limited computation resources.

Transformer models rely on attention mechanism to capture long-range dependencies but suffer from quadratic complexity, limiting their scalability to long sequences. Kernel-based linear attention reduces this complexity but typically relies on fixed or weakly learnable kernels, restricting expressiveness and performance. In this work, we propose Flexformer, a flexible linear Transformer that learns attention kernels in a fully data-driven manner. Flexformer builds on random Fourier feature-based linear attention and treats spectral frequencies as trainable parameters, enabling the model to learn a broad family of attention kernels. We develop both stationary and nonstationary variants, with the latter offering strictly greater expressiveness. Extensive experiments on language modeling and sequence classification demonstrate that Flexformer consistently outperforms baselines. Moreover, Flexformer can be effectively distilled from pretrained Transformers to recover softmax attention and exhibits strong kernel transferability across domains, achieving both high efficiency and competitive performance on long-sequence tasks.

While pre-trained models have achieved remarkable success in code search, their multilingual capabilities remain a major hurdle, plagued by data imbalance, cross-lingual semantic interference, and the loss of critical information from existing unified representations like Abstract Syntax Trees (ASTs) or Intermediate Representations (IRs). Furthermore, conventional contrastive learning strategies often rely on simplistic hard negative sampling while overlooking the potential of mining hard positives to learn code's intrinsic semantic invariance. To address these challenges, we introduce UNICS, a framework for multilingual code search built on a two-stage training strategy. In the first stage, UNICS is pre-trained on a novel dataset we constructed, which uses pseudo-code as a unified representation to learn a cross-lingual, algorithm-level logic that preserves full semantic fidelity. The second stage employs a multi-task transfer learning strategy that adapts this general knowledge to specific languages by decomposing code into semantic slices (e.g., API calls, function bodies) and incorporating tasks for hard positive mining and cross-lingual dynamic hard negative sampling. Experimental results demonstrate that UNICS achieves state-of-the-art performance across multiple multilingual and cross-lingual benchmarks, showcasing superior generalization and performance balance, especially in zero-shot transfer tasks to low-resource languages.

Large Language Models (LLMs) inference is typically deployed under a static resource assumption, where models execute a fixed computational graph regardless of the runtime environment. However, real-world cloud infrastructure is inherently dynamic, characterized by fluctuating availability (e.g., spot instance preemption) and tiered Quality-of-Service requirements. In such volatile settings, static models are inflexible: they either crash under resource constraints or waste compute on redundant operations. To bridge this gap, we propose Learning to Allocate (L2A), an end-to-end framework for resource-adaptive inference. Unlike prior methods that condition only on input difficulty, we formulate inference as a constrained allocation problem conditioned on both the input and the runtime resource budget itself. We introduce lightweight, budget-conditioned and input-aware gating networks integrated into the LLM. These gates are trained via a unified objective that jointly optimizes task performance, logical consistency, and resource costs along three axes matching how real-world dynamics manifest: layer skipping for memory and depth pressure, head pruning for throughput contention, and reasoning-token reduction for latency tightening. This lets the model learn a budget-aware policy beyond input difficulty alone: it adaptively configures its computational footprint with respect to real-time resource dynamics, maximizing reasoning depth when resources permit while enforcing strict frugality when budgets tighten. A single L2A model traces the entire compute-accuracy Pareto frontier on Llama-3-8B and Qwen-3-4B: at up to 34% realized layer sparsity, it stays within 0.6% of the dense baseline on GSM8K, with the same gap holding zero-shot on out-of-distribution tasks, while every static or heuristic baseline requires a separately tuned model and still drops by 5-10% at comparable inference time.

Enterprise Knowledge Graphs (KGs) are increasingly used for internal search, analytics, and question answering, but building natural-language interfaces for private enterprise graphs remains costly. We present KG2Cypher, a data-centric pipeline for building enterprise text-to-Cypher systems from existing KGs. KG2Cypher first constructs an executable Cypher query from observed graph facts and then uses LLMs to generate its associated natural-language question. The resulting Text-Cypher pairs are validated with an LLM judge and human validation, and are converted into candidate-aware SFT data. The trained generator is served with class-conditioned schema prompting, entity retrieval, and LoRA-based inference. We evaluate KG2Cypher in Korean enterprise settings, where short search-style queries and schema paraphrases make language grounding difficult. LoRA SFT improves execution-result F1 from 0.806 to 0.950 on broadcast-program queries and from 0.70 to 0.92 on company queries. In an 11-class setting, KG2Cypher achieves 95.2% exact match, 99.9% execution rate, and 0.964 execution-result F1.

Process reward models (PRMs) enhance the reasoning capabilities of large language models (LLMs) by providing fine-grained feedback, yet training PRMs typically requires expensive stepwise annotations. Outcome-supervised PRMs offer a scalable alternative by learning from final-answer correctness alone, but this introduces a fundamental *credit assignment* challenge, i.e., attributing outcomes to responsible reasoning steps. Existing approaches rely on either uniform or causal assignment, both of which fail to anchor credit in step correctness and thus hinder process error identification. In this work, we propose Outcome-Supervised Process Reward Modeling via **L**earnable **C**redit **A**ssignment (**LCA**), an outcome-supervised PRM framework that jointly learns credit assignment and reward modeling under the principle of *Weakest Link Assignment: a reasoning chain is as strong as its weakest link*. To address mutual dependence between credit assignment and reward modeling, we formalize outcome-supervised PRM as a Multiple Instance Learning (MIL) problem and introduce Softmax-Weighted-Sum (SWS) pooling, an MIL pooling technique tailored for strong dependence and redundancy among reasoning states. We prove Bayes consistency of our algorithm under mild assumptions. Extensive experiments demonstrate that **LCA** consistently outperforms state-of-the-art outcome-supervised PRMs across multiple tasks and backbones. Code is available at https://anonymous.4open.science/r/LCA.

Programming adaptive behaviors at the cellular level is a long-standing goal that raises the question of how probabilistic computation can be implemented in biochemical systems. Chemical reaction networks (CRNs) provide such a substrate and have been shown to realize probabilistic models, including hidden Markov models and factor graphs, with dynamics reproducing Bayesian inference and belief propagation. However, encoding these algorithms typically requires prohibitively large reaction networks, and classical CRN reduction techniques do not directly apply. By recovering the factor graph structure encoded in Napp--Adams-compiled CRNs, we transport recent factor-graph reduction results to their chemical implementations, obtaining significantly smaller CRNs while preserving the belief-propagation fixed points on surviving variables.

The rapid spread of fake news poses increasing threats to information ecosystems, especially as AI-generated misinformation under Generative Engine Optimization (GEO) poisoning allows adversarially crafted content to be systematically surfaced by retrieval systems, contaminating LLM reasoning. In this paper, we propose Tree of Evidence (ToE), a hierarchical evidence reasoning framework for automated fact-checking that models each claim as a dynamically expanding argument tree. ToE integrates a reinforcement learning-driven multi-source retrieval agent, an evidence evaluation agent, and an argument tree aggregation algorithm to iteratively decompose, retrieve, and verify claims through an explainable evidence chain. We further provide a theoretical analysis of the retrieval process, deriving a formal error bound that guarantees the learned policy converges to a neighborhood of the information-theoretically optimal policy. Experiments across multiple datasets and backbone LLMs demonstrate that ToE achieves improvements ranging from 4 to 24 percentage points over competitive baselines, with particularly pronounced gains on adversarially poisoned inputs.

Bash scripts are the cornerstone of system administration and DevOps automation, where code quality directly impacts system stability and security. In automated Bash script generation using Large Language Models (LLMs), two interconnected failures emerge: unauditable "black box" reasoning and critical robustness vulnerabilities in generated code. To address both, we propose BashCoder-R1, a novel framework for robust and explainable Bash script generation. Our pipeline combines: (1) Continual Pre-training (CPT) to specialize the model on Bash paradigms; (2) Long Chain-of-Thought Supervised Fine-Tuning (L-CoT SFT) on expert-validated reasoning-and-code samples to emulate proactive risk-aware thinking; and (3) Robustness-Aware Group Relative Policy Optimization (R-GRPO), a reinforcement learning phase optimizing a weighted reward for syntax correctness, robustness (via shellcheck), and format correctness. We evaluate on BashBench, a new benchmark of 952 real-world tasks (773 single-line, 179 multi-line). BashCoder-R1 achieves SyntaxPass (100.00%/94.97%), RobustWarnRate (4.01%/16.47%), RobustPass (95.99%/79.33%), FuncRate (93.01%/93.85%), and FullRate (90.04%/73.18%) for single-line/multi-line tasks, outperforming the strongest baseline DeepSeek-V3.2 (Reasoning) by 37.82% and 20.18% in FullRate. Human evaluation on Functionality, Robustness, and Clarity further confirms BashCoder-R1 achieves the highest quality ratings.

Discrete diffusion language models (dLLMs) recover masked tokens in parallel, offering significant speedups over autoregressive (AR) generation. However, such promising frameworks face a fundamental architectural design dilemma: \ding{182} Adopting bidirectional attention achieves strong generation quality by allowing each position to access the full context, but is inherently incompatible with KV caching, limiting inference throughput in batch-serving scenarios; \ding{183} Conversely, causal attention enables efficient cached inference but loses all right-side context, substantially degrading generation quality. This paper introduces Bifocal dLLMs, a new paradigm that resolves this dilemma through \emph{asymmetric bidirectional context}. Analogous to bifocal lenses, we instantiate the paradigm as \textbf{R2LM} (Right-to-Left Mamba), which combines two complementary mechanisms: $a$) standard causal attention providing precise left-context with full KV cache compatibility, while $b$) a lightweight reverse Mamba SSM sidecar supplying compressed right-side context without breaking cacheability. Comprehensive experiments on continued pretraining of Qwen3-1.7B with 60B tokens demonstrate that R2LM achieves $2.4\times$ to $12.9\times$ higher throughput than bidirectional dLLMs and $1.9\times$ to $2.9\times$ speedup over AR baselines in batch serving through parallel decoding with KV caching, while exceeding the causal baseline on most benchmarks and surpassing the bidirectional dLLM on average.

Despite their strong general capabilities, large language models (LLMs) often remain unreliable when outputs must be numerically precise. A key reason is the training objective: standard cross-entropy treats numeric tokens as unstructured categories and ignores the metric structure of their values. We address this mismatch with Smooth Maximum Mean Discrepancy (SMMD), which builds on the classic MMD by incorporating value-distance kernels over numeric tokens and graph-based smoothness. With this kernel defined over a numeric sub-vocabulary, SMMD aligns the predicted numeric distribution to the target via kernel matching and smooths the prediction-target residual over the induced kernel graph to encourage local consistency. We evaluate SMMD on four numeric-target tasks: mathematical reasoning, arithmetic calculation, clock-time recognition, and chart question answering, across multiple open-weight LLM and VLM backbones. SMMD consistently improves accuracy over both cross-entropy and recent numeric-target losses; analyses show complementary effects between MMD and smoothness and underscore the importance of distance-based kernel design. Code is available at https://github.com/Zuozhuo/smmd-loss.

Compositional generalization, the ability to solve complex problems by combining solutions to simpler sub-problems, is a fundamental capability of both natural and artificial intelligence, and a key mechanism underlying chain-of-thought reasoning. However, the theoretical underpinnings of compositional generalization remain poorly understood: when and why does decomposing a problem into parts yield more efficient learning than solving it directly? We study this question through the canonical problem of learning to simulate semiautomata (predicting the outcome of $T$ steps of sequential computation), a model that captures state tracking, regular language recognition, and modular arithmetic. We show that an autocurriculum-based approach building on Part I of this series, recursively decomposing longer sequences into shorter sub-problems, learning to solve them, and composing the solutions, achieves dramatically better statistical complexity than direct methods. (i) For a setting inspired by supervised fine-tuning (SFT) where the learner receives interactive feedback on intermediate states of the computation, curriculum facilitates learning from only $2^{\mathcal{O}(\sqrt{\log T})}$ tokens of supervision; i.e., subpolynomial in the sequence length $T$, overcoming the $Ω(T)$ token barrier required by direct simulation. (ii) For a setting inspired by reinforcement learning with verifiable rewards (RLVR), where the learner improves a pre-trained reference model using an outcome verifier, we show that curriculum reduces the requirement on the reference model from coverage at the full sequence length $T$ to coverage at a shorter block length $B \ll T$, an exponentially weaker condition.

Prosodic emphasis varies across languages, emotions, and speaking styles, yet existing emphasis detection models are largely trained and evaluated on monolingual neutral read speech. We introduce MMEE (Multilingual Multi-Emotion Emphasis), a corpus of 10,000 professionally recorded expressive utterances (14.13 hours) across 7 languages and 34 emotion/style categories, with three-level perceptual labels (10 annotations per sample). We benchmark two state-of-the-art architectures under monolingual, cross-lingual, multilingual, cross-emotion, cross-dataset, and data-scale settings. Monolingual models show limited zero-shot transfer, degrading across typologically distant languages, while multilingual training substantially improves robustness. Models transfer robustly between high- and low-arousal emotions; bidirectional transfer between synthetic and perceptual benchmarks suggests shared prosodic structure; and performance stays robust even at smaller training scales.

We show that for tall matrix parameters, like projection matrices in the MLP layers, the Muon update can have row norms that are arbitrarily non-uniform. This can lead to a self-reinforcing feedback loop whereby neurons receive persistently small updates and eventually do not contribute meaningfully to network outputs. This problem is effectively mitigated by an additional row normalization step, but current methods do this in a way that moves the Muon update geometry away from the polar factor of the momentum matrix, which we find is undesirable. We propose Aurora, an optimizer that enforces row-uniformity of matrix parameter updates while respecting Muon's polar factor geometry. Aurora outperforms Muon in our pre-training experiments and, when combined with existing methods, achieves state-of-the-art performance among spectral optimizers on the optimizer track of the modded-nanoGPT speedrun. Additionally, we find that Aurora's empirical gains over Muon scale with the MLP expansion factor, suggesting that Aurora may allow for effective training of very wide MLP layers.

We introduce a neural network-based framework for learning time series estimators through a process we term decision-theoretic pretraining. Analysts specify a generative world, a distribution over data-generating processes, and a target decision objective. A neural network trained on stratified simulations from this world approximates the corresponding optimal decision rule, yielding a neural estimator that provides forecasts, parameter estimates, predictive intervals, or model-selection for zero-shot inference on previously unseen time series. The joint specification of the generative world and objective enables the estimators to directly approximate process-level, finite-sample properties: near-optimal risk, bias control, minimax performance, and uniform calibration. Our experiments demonstrate that these neural estimators can outperform traditional baselines such as maximum likelihood estimation and model selection via AICc, for the same model structural model classes. Furthermore, even when trained purely on simulations of structural models, they achieve competitive or state-of-the-art forecasting accuracy on major real-world benchmarks, compared with statistical, neural or large pre-trained models. We illustrate the framework by addressing two longstanding challenges: finite-sample bias and miscalibration in AR(p) models, and the forecast combination puzzle. These applications highlight the approach's main advantage: its ability to approximate solutions to analytically intractable or computationally prohibitive time series problems, including complex structural equations or optimality criteria. Ultimately, by enabling explicit control over decision-theoretic trade-offs, the framework equips analysts with highly efficient estimation tools tailored to their specific analytical needs.

Recent work has shown that fine-tuning large language models (LLMs) for social warmth degrades factual reliability and increases sycophancy. We investigate a related but distinct failure mode: warmth fine-tuning also weakens adversarial safety, making models more susceptible to jailbreaks and harmful output generation. We examine whether this reflects an inherent consequence of empathetic adaptation or an artifact of data construction. To address this, we introduce a persona-driven rewriting pipeline that conditions user turns on low agreeableness and pairs this with warm, de-escalating assistant responses. Across three experiments on four models, our approach reduces jailbreak susceptibility and harmful output rates relative to generic warmth fine-tuning baselines, while preserving conversational warmth. Representational probing provides suggestive evidence that this conditioning reduces the geometric alignment between warmth and compliance directions in latent space. These results show that safer empathetic fine-tuning is achievable through data design alone, without safety labels, harm detectors, or changes to the training objective.

Large Language Models (LLMs) still struggle with the ``lost-in-the-middle'' problem, where critical information located in the middle of long-context inputs is often underrepresented or lost. While existing methods attempt to address this by combining multi-scale rotary position embeddings (RoPE), they typically suffer from high latency or rely on suboptimal hand-crafted scaling strategies. To overcome these limitations, we introduce a layer-specific positional embedding scaling~(LPES) method that assigns distinct scaling factors to each layer. LPES achieves a more balanced attention distribution without fine-tuning model parameters or increasing inference delay. A specially designed genetic algorithm is employed to efficiently select the optimal scaling factors for each layer by incorporating Bézier curves to significantly reduce the search space. Extensive experiments demonstrate that LPES effectively mitigates positional attention bias and delivers consistent improvements across multiple long-context benchmarks, yielding up to an $11.2$\% accuracy gain on the key-value retrieval dataset.

While voice control is rapidly becoming a ubiquitous vector of human-AI communication, the risks facing these systems remain poorly understood. This is, in part, a product of the difficulties in scaling strictly digital adversarial workflows to the physical world. These scale barriers have led the community to abstract away key acoustic factors relating to detectability and the influence of geometry on acoustics. These methodological and metrological shortcomings undermine our understanding of risk. We illuminate these issues through real-world testing, conceptual discussions, and a novel, high-throughput reality simulation framework. By testing over 8 million adversarial evaluations, we demonstrate that acoustic awareness yields relative Word Error Rate increases of up to 94.5\% under Whisper and wav2vec. We employ this framework to explore a formalize and operationalize a Dual-Form Signal to Noise Ratio to decouple source stealth from victim attack efficacy, resolving a crucial limitation in current works. This lays the groundwork for repeatable, verifiable research that embraces, rather than abstracts, the acoustic environment.

Historical encrypted manuscripts present a challenging problem at the intersection of cryptology, linguistics, paleography, and computer vision. Current automatic decipherment approaches usually rely on a two-stage pipeline: transcription of cipher symbols from manuscript images, followed by decryption into plaintext. However, this design is sensitive to transcription errors, which propagate to the final output. We present Direct Image Decryption, an end-to-end approach that directly maps encrypted manuscript images to plaintext, bypassing the intermediate transcription stage. Using the Copiale cipher as a case study, we build a synthetic data generation pipeline to create large-scale cipher-like training data and compare the traditional pipeline with the proposed joint architecture. Results show that joint image-to-plaintext modeling is a promising alternative to traditional transcription-based pipelines.

Automatic Speech Recognition systems are notoriously both sensitive to adversarial and benign perturbations. While this has been repeatedly demonstrated using reference datasets, detecting such behaviors in deployed systems is incredibly challenging, due to the absence of oracle knowledge of the true transcription. We demonstrate that employing a certification-inspired mechanism can significantly decrease WER, increase recall, and decrease the Spearman correlation between confidence and WER. We achieve this through a dual-gate diagnostic pipeline: a Two-Sided Atomic Audit that accumulates statistical wealth to certify both token existence and adversarial exclusion, and a Rank-Based Tournament that selects the winning sequence. Our evaluations across four diverse architectures demonstrate up to a 55% relative reduction in Word Error Rate, while also providing granular word- and sentence-level certifications to enhance acoustic security.

In this paper, we are the first to examine the correlations between class frequency and the multi-scale noise schedule within diffusion models. For score-based generative models, low-density regions often lead to inaccurately estimated scores, thereby compromising the generation quality. Although the multi-scale noise schedule can alleviate this issue during the diffusion process, low-frequency classes still face the challenge of large low-density regions, resulting in more inaccurate estimated scores than high-frequency classes. Furthermore, high-frequency classes tend to dominate the score space, causing a convergence of most data points towards generating samples from these classes. Consequently, samples generated within low-frequency classes exhibit suboptimal quality and limited diversity. To address this challenge, we propose the \textit{Class-frequency Guided (CFRG)} noise schedule, leveraging the insight that low-frequency classes should be endowed with larger-scale noises. To illustrate the effectiveness of our method, we conduct experiments on various tasks, including image generation, image classification, and text-to-image generation, using imbalanced datasets, \textit{i.e.}, CIFAR-100-LT, and ImageNet-LT. By employing the CFRG noise schedule, we achieve substantial improvements over baselines, manifesting the crucial role of frequency statistics in noise schedule design.

Selecting near-optimal hybrid MPI+OpenMP configurations for molecular dynamics workloads on modern HPC clusters has traditionally required exhaustive empirical benchmarking, consuming allocation budget proportional to the number of configurations evaluated. This work investigates whether a cold-start Random Forest surrogate, trained once on a small, structured benchmark dataset, can reliably predict execution performance and recommend high-performing configurations without further cluster runs. The training dataset comprises 54 LAMMPS+SPICA runs of the antimicrobial peptide Tritrpticin on a hydrated DOPC bilayer (4 354 coarse-grained beads), spanning 18 hybrid configurations on 1-8 AMD EPYC 7662 nodes of the Lovelace cluster at CENAPAD-SP, with three independent replications each. Nine topology and resource features feed five regressors that predict loop time and four internal LAMMPS timing fractions (Pair, Kspace, Comm, Modify). In-sample mean absolute error is 0.49 s on loop time (4.0 % relative). Feature importance localizes predictive signal in topology variables (OpenMP threads and MPI/OpenMP ratio dominate; raw node and core counts contribute under 3 %). Leave-one-dimension-out generalization reveals that accuracy is governed by hardware regime membership: within a common regime (single-node, multi-node, or shared threading tier) the surrogate ranks configurations correctly, and degrades when targets cross architectural boundaries. The result is an interpretable map of where the surrogate's recommendations can be trusted, useful for scoping further benchmark campaigns at a fraction of their nominal cost.

Randomized Smoothing (RS) provides rigorous robustness guarantees for neural networks without architectural constraints, yet its adoption is limited by extreme computational costs. Standard RS requires tens of thousands of model evaluations per input and forces practitioners to commit to fixed sample sizes a priori. In this work, we present a novel meta-learning framework for anytime-valid certified robustness that adaptively deploys computational resources. By using a lightweight meta-learner to predict image-specific priors for a sequential E-process, we achieve a 20-fold reduction in sample complexity compared to traditional methods while maintaining rigorous statistical guarantees. Beyond raw efficiency, we demonstrate how anytime-validity enables adaptively allocating compute based upon application-specific risk thresholds, a form of resource triage impossible under classic certification frameworks. That this is achievable while also providing similar certification performance demonstrates that our approach provides a pathway for real-time, safety-critical certification deployments.

In financial forecasting, predictive performance depends not only on which model is trained, but also on how the trained model is deployed. We study this issue in multi-horizon volatility forecasting. Our starting point is that a trained multi-output (MIMO) forecaster does not define a single deployable predictor: by changing the inference-time rollout rule, the same trained model induces a family of forecasts with different accuracy and cost profiles. Across 20 stock-volatility series, three forecast horizons, and architectures ranging from linear models to PatchTST, we find that non-default rollout rules often improve over standard MIMO deployment. However, the best fixed rule varies substantially across architectures and horizons, making any single static replacement unreliable. We therefore evaluate validation-based deployment policies over the induced rule family. Under the primary MSE objective, validation-selected singletons provide a low-cost improvement over default MIMO, while small rule subsets recover much of the benefit of larger ensembles at substantially lower inference cost. We also find that policy rankings are metric-sensitive: MSE-selected policies do not transfer uniformly to QLIKE, a finance-standard volatility loss. These results show that inference-time deployment is a meaningful source of adaptiveness in financial forecasting, and that trained volatility forecasters should be evaluated not only by their architecture, but also by their deployment policy.

Large language models (LLMs) are increasingly used to generate, classify, and annotate data whose outputs feed downstream hypothesis tests. However, LLM-based research is easy to p-hack: a researcher can tune the prompts, decoding parameters, or output format until a desired result is reached. We propose a protocol to mitigate p-hacking in LLM-based research: preregistering the experiment and eligible models, and then running it on the first eligible LLM that is released after the preregistration. The researcher finalizes the procedure on current models, preregisters the analysis plan together with a set of eligible future models, and runs the confirmatory analysis on the first eligible model released afterward. Because this model does not exist at commitment time, it cannot be hacked against; furthermore, configurations that hack one model frequently do not transfer to the next. We evaluate the protocol on two tasks whose true values are known. Across 20 models from four providers and 11 LLM-analysis configurations, the protocol would have blocked successful transfer of the p-hack in 73.9% and 72.7% of cases in the two tasks. Additional analyses reveal that mitigation remains substantial under several stress tests. Finally, putting money where our mouth is, we followed our own protocol and preregistered our experiment. The preregistered experiment confirmed the protocol's effectiveness: out of the 7 configurations that hacked the prior model, the hacking failed to carry over in 6 configurations on the first eligible model released afterward.

Pervasive data contamination -- stemming from measurement errors, outliers, or adversarial corruption -- has motivated the development of robust statistical methods. In this context, we propose a two-stage Adversarial Contamination-resistant Iterative Hard Thresholding (AC-IHT) algorithm for high-dimensional regression with contamination. Our nonconvex algorithm achieves minimax near-optimal (up to logarithmic terms) estimation by iteratively updating the coefficient vector and the contamination vector with different thresholding scales. We further demonstrate that our AC-IHT estimator is signal-adaptive: under proper signal conditions, it adaptively attains a sharper estimation rate and more accurate support recovery. Moreover, it enjoys the strong oracle property, laying a theoretical foundation for asymptotic inference. Numerical experiments confirm its superior finite-sample performance. Finally, we discuss theoretical extensions of the proposed procedure to generalized linear models and to heavy-tailed noise settings.

Edge devices increasingly invoke large language models (LLMs) through API services for context aware edge intelligence, while edge generated data may be collected to improve LLMs and may introduce sensitive, copyrighted, harmful, or outdated information into model behavior. Machine unlearning offers a practical way to remove the influence of undesired data without retraining LLMs. However, existing methods still face two gaps. The first is API only black box access, where target model parameters and internal logits are unavailable. The second is how to preserve retained utility when unlearning target data and retained data share highly similar prompt structures or semantic patterns. To address these challenges, we propose Controlled Behavioral Divergence (CBD), an API only black box unlearning framework. CBD uses two auxiliary models to create controlled behavioral divergence between retained inputs and unlearning target inputs, converts this divergence into an unlearning relevance score, and routes unlearning related prompts away from the target LLM. To improve discrimination accuracy under high similarity between target and retained data, CBD constructs a gradient statistics based discriminative basis by estimating empirical Fisher matrices and solving a regularized generalized eigenvalue problem, guiding the unlearning signal toward target specific information rather than shared prompt structures. Compared with eleven white box and gray box unlearning baselines, CBD achieves a better unlearning utility trade off and its performance varies little across settings. On ToFU forget10, CBD approaches the retrained reference on the forget set while raising model utility to 74.90, about 15% above the second best baseline. On WMDP, it lowers hazardous knowledge accuracy to 25.68, near random guessing, while preserving MMLU accuracy of 52.67. Code is at https://github.com/DGL-codes/CBD.

World models in partially observed environments rely on latent representations that summarize interaction history, but in many modern LLM-based architectures predictive performance fails to reflect representation quality due to history bypass, rendering the latent state unidentifiable. Strict latent state mediation, requiring predictions to depend only on the latent state and action, is a classical principle that resolves this, but enforcing it in text-based settings is an open challenge: textual latent states are discrete and non-differentiable, precluding variational training, and expressive LLM decoders readily ignore the bottleneck. We show how to make strict mediation work in the text domain. We formalize why it is necessary, showing that strict mediation makes representation quality empirically testable while history-leaky architectures break this connection. We then introduce textual latent states, which are discrete, interpretable, and variable-length, and factorized GRPO (fGRPO), a tree-structured reinforcement learning method that enforces strict mediation during training. Experiments on TextWorld and ScienceWorld show preserved one-step prediction accuracy alongside up to 57\% gains in representation quality and 98\% improvements in rollout performance, increasing with task complexity and horizon.

Probe-based uncertainty estimation (UE) has emerged as a prominent approach to detect hallucinations in Large Language Models (LLMs) by learning uncertainty from internal model signals. Yet, recent methods vary simultaneously across feature design, training data construction, and evaluation setting, obscuring what actually drives performance. To address this issue, we propose a factorised study of probe-based UE under matched conditions. Our results show that raw hidden states and attention features are difficult to outperform in-domain. However, under distribution shift, structured and compressed features are more robust, suggesting that in-domain performance alone is insufficient to measure progress. Furthermore, prompting and label construction significantly affect probe behaviour. Building on these best-practice findings, we train benchmark-based pretrained probes that transfer reasonably well to open-ended factual generation, providing a stable off-the-shelf baseline. Our work encourages more deployment-oriented evaluation of probe-based uncertainty estimators. The code repository is available at https://github.com/ponhvoan/ProbeUE.

Inspired by advances in natural language processing and computer vision, "time-series foundation models" (TSFMs) have recently been introduced with the promise of strong generalization across diverse time-series tasks, including forecasting, classification, and anomaly detection, as well as across domains such as healthcare, climate science, and manufacturing. However, their utility for gas-sensing data remains largely unexplored. To address this gap, this paper systematically evaluates recent TSFMs on electronic nose (E-Nose) data. In particular, we investigate whether embeddings produced by representative TSFMs, including Chronos-2 and MOMENT, provide effective representations for gas identification and concentration prediction. Specifically, we show that fine-tuning is necessary to achieve satisfactory performance on E-Nose data, and fusing TSFM embeddings with representations learned by specialized predictive models can further improve the performance, suggesting both the potential and limitations of current TSFMs for gas-sensing applications.

Search agents powered by large language models (LLMs) are increasingly used to solve complex information-seeking tasks, requiring multi-step retrieval and reasoning to fulfill user goals. However, existing benchmarks often assume that user queries are complete and explicit, overlooking the fact that real-world search requests are frequently vague, underspecified, or even factually incorrect. In deep search scenarios, such ambiguity can propagate along multi-step reasoning chains and lead agents toward incorrect search trajectories. To address this gap, we introduce DiscoBench, a benchmark for clarification-aware deep search, designed to evaluate whether search agents can proactively identify ambiguity, ask effective clarification questions, and recover correct reasoning paths through user interaction. DiscoBench contains 211 samples and 463 ambiguity instances across 11 real-world domains, covering four ambiguity types. We further design a user simulator for multi-turn interaction and evaluate model performance from four perspectives: task utility, ambiguity detection, interaction strategy, and cost efficiency. Experiments on representative LLMs show that ambiguity detection and effective clarification are distinct capabilities, and that repeatedly searching instead of asking for clarification often performs worse than direct guessing, highlighting a critical gap between retrieval ability and interactive problem-solving in current search agents.

Artificial intelligence is transforming biodiversity monitoring by enabling automated analysis of ecological imagery collected from camera traps, drones, satellites, underwater platforms, and other sensing systems. These tools can expand the scale and speed of conservation assessments, yet many computer vision models remain difficult to inspect, making it challenging to determine whether predictions are based on ecologically meaningful signals or on spurious correlations, sampling biases, and other artifacts that may undermine conservation decisions. We argue that explainable artificial intelligence (XAI) should become a standard component of ecological model validation because conservation practitioners increasingly depend on understanding not only whether a model is accurate, but why it is accurate. We provide practical guidance for applying XAI to three common ecological computer vision tasks: image classification, object detection, and image segmentation. To illustrate how XAI can support ecological model auditing, refinement, and deployment, we present two case studies using aerial imagery: harbor seal detection and cetacean anatomical segmentation. These examples demonstrate how explanation methods can identify biologically meaningful cues, reveal false positives driven by background and shape confounds, uncover edge and occlusion effects, and guide data collection, augmentation, and retraining strategies. More broadly, they show how explainability can help assess whether model reasoning aligns with ecological understanding. We conclude by identifying key challenges and opportunities. By making model behavior more transparent and scientifically interrogable, XAI can help ensure that AI-supported ecological evidence is more reliable, understandable, and actionable for biodiversity conservation.

Recently, Large Language models (LLMs) find application in several fields. This extends to hardware definition and synthesis. However, most works at the intersection of LLMs and hardware generation focus on text-based tasks, creating a gap for multi-modal LLMs for RTL design. In this work, we introduce MultModLM, a benchmark for evaluating LLMs on the task of generating hardware schematics from RTL (Register Transfer Level) descriptions. The dataset consists of 99 diverse RTL modules spanning arithmetic, control, and state-based designs. To address the challenges of non-unique schematic representations, we propose a multi-stage evaluation framework combining rubric-based scoring, self-evaluation, cross-model assessment, blind evaluation, and human validation to enable exhaustive evaluation. Through experiments on state-of-the-art LLMs, we observe that while models can generate visually interpretable schematics, their functional correctness remains constrained. Furthermore, we find that LLM-based evaluators exhibit near-zero agreement with human raters, revealing, as a key finding, that LLM-as-a-judge paradigms are unreliable in structurally precise domains. These findings suggest that reliable evaluation of multi-modal hardware outputs remains an open challenge, motivating the need for more robust and domain-aware evaluation methodologies, as well as tools for structural evaluation, so as to enable formal equivalence checkers.

Vue.js is a popular framework for building modern web applications. As Vue.js functionality and tooling support grow, ensuring its reliability (through automated testing) is becoming increasingly important. Although model-based testing has been successfully used to automate graphical user interface (GUI) testing on other platforms, its application to Vue.js remains challenging: Transition candidates, which are spread across router configurations and single-file components (SFCs), must be concretized and normalized into an executable page transition graph (PTG) for testing. To address this, we propose the LLMVue framework, which uses a large language model (LLM) to generate a PTG from Vue.js source code. LLMVue infers component hierarchies and route transitions, merging them into a unified PTG across multiple SFCs. We evaluated LLMVue on a collection of ten open-source Vue.js projects from GitHub, using GPT-4o as the LLM backbone. The constructed graphs demonstrate high precision and recall, with low graph edit distance. LLMVue -guided testing also significantly improves the coverage and exploration efficiency, compared to a random exploration baseline (with the same time constraints). To the best of our knowledge, this is the first use of LLMs for model-based GUI testing of Vue.js applications using source-level PTG extraction.

Obesity is a global health crisis associated with metabolic disorders such as type 2 diabetes and cardiovascular disease. This study employed single-cell RNA sequencing to reconstruct the developmental trajectory of human adipocytes from adipose tissue samples. Our analysis identified 15 transcriptionally distinct cell clusters, including 7 transitional states, revealing the dynamic process of adipocyte differentiation. We detected 16 functionally active signaling pathways mediating cellular communication between adipocytes and their progenitors. Among these, insulin-like growth factor (IGF) and fibroblast growth factor (FGF) pathways emerged as the most prominent networks, showing consistent activity across differentiation stages (p<0.05). The study revealed depot-specific differences, with visceral adipocytes undergoing additional extracellular matrix remodeling absent in subcutaneous differentiation. Spatial analysis further showed that IGF signaling was particularly active in perivascular niches, while FGF activity dominated in mature adipocyte zones. These results provide the first comprehensive map of human adipocyte development, highlighting IGF and FGF pathways as potential therapeutic targets. The identified signaling networks offer new insights for developing interventions to promote healthy adipose expansion or inhibit pathological fat accumulation. This work advances our fundamental understanding of adipose tissue biology while providing clinically relevant data for metabolic disorder treatments.

We find that explicit reasoning does not necessarily translate into better multimodal emotion recognition (MER) accuracy, even though it makes predictions more interpretable. Specifically, for reasoning-based MLLMs, fast thinking by triggering direct answers often outperforms slow thinking after deliberative reasoning. Our empirical analyses show that fast thinking improves recall with broader and more confident predictions, whereas slow thinking favors precision through conservative filtering of incorrect categories. Building on these insights, we propose MER-R1, a reinforcement learning framework that turns slow-fast complementarity into explicit optimization. Dual-objective disentanglement separates recall and precision into two optimization signals, allowing them to be jointly optimized rather than traded off against each other. Slow-fast confidence calibration further aligns the final slow-thinking answer with fast-thinking intuition, strengthening correct emotions while suppressing incorrect ones. In this way, MER-R1 unifies the recall-oriented intuition of fast thinking with the precision-oriented selectivity of slow thinking. We further provide theoretical justification for this synergy, showing that it mitigates variance-induced interference during optimization. Extensive experiments on MER-UniBench and MME-Emotion show that MER-R1 achieves state-of-the-art performance and makes reasoning genuinely benefit emotion recognition.

In recent years, with the emergence of Temporal Knowledge Graphs (TKGs), research on learning entity and relation representations in TKGs has attracted increasing attention, giving rise to a large number of TKG embedding methods. TeRo is a simple and efficient temporal knowledge graph embedding approach. However, TeRo does not do well in modeling the mapping properties of various relations, such as one-to-many, many-to-one, and many-to-many. Meanwhile, it also has limitations in the expression of temporal information. To address these issues, we propose a novel TKG embedding method named TeRoR. This method divides the temporal evolution of entity embeddings, and conducts independent rotation transformations on head and tail entities in the complex vector space to strengthen temporal information modeling capacity. In terms of relational characteristics, we train a radius to constrain the rotated and translated head entities within a circular region centered on the tail entity, which effectively captures the diverse mapping properties of relations. Experimental results demonstrate that TeRoR achieves competitive performance against state-of-the-art models on four distinct TKG datasets.

LLM-agent simulation faces a joint grounding and scaling problem: agents should act in environments that reflect real urban constraints, yet direct online LLM calls for city-scale populations are computationally prohibitive. We present GenWorld, an empirically grounded urban simulation infrastructure that combines a building-level synthetic city, a structured agent-environment interface, and offline compilation of LLM-derived decision signals into lookup policies for scalable rollout. In a reference instantiation for Higashihiroshima, Japan, GenWorld grounds 196,608 synthetic residents in census and geospatial data, validates demographic consistency against census tabulations, and uses YJMob100K mobile-phone data as a commuting-distance diagnostic. We demonstrate the infrastructure through three reproducible cases: a full-city weekday rollout, a weekday-weekend behavioral contrast, and a warning-response perturbation with auditable replanning traces. These cases support GenWorld as a reproducible platform for grounded and scalable LLM-agent studies, while calibrated forecasting for traffic, evacuation, or policy outcomes remains future work.

Small Language Models (SLMs) are increasingly being considered for deployment on edge devices such as laptops, enabling private, low-latency, and locally personalized applications. However, personalization requires models to adapt over time to evolving user- or task-specific data, placing them in a continual learning setting. This creates the risk of catastrophic forgetting, where learning new information degrades performance on previously learned tasks or broader model capabilities. Recent benchmarks such as TRACE have shown that continual fine-tuning can significantly degrade the general abilities of aligned large language models. In this work, we present a study for sequential LoRA personalization of SLMs. We save model checkpoints after each adaptation stage and evaluate them on current tasks, previously seen tasks, and a fixed reference set. This checkpoint-level protocol enables us to monitor task performance, forgetting, and reference set drift over time. We show that lightweight reference set distributional diagnostics can reveal model-specific instability patterns during sequential LoRA personalization of SLMs, including cases where task-level metrics alone hide harmful adaptation. We hope this can highlight new research avenues for monitoring stability of SLMs in a continual learning setting.

As large language models are increasingly deployed in real-world systems, safety failures can still lead to harmful outputs and dangerous misuse. We argue that the essence of safety is adversarial: many failures arise not from natural inputs alone, but from strategic attempts to evade model policies and safeguards. However, existing general-purpose model development largely overlook this adversarial nature, and often remain insufficient for realistic safety scenarios involving planning, tool use, and multi-step reasoning, causing measured safety performance to overestimate real deployment robustness. To address this gap, we present Yuvion LLM, a large language model built for adversarially robust content safety and broader AI safety. Yuvion LLM treats adversarial robustness and agentic capability as first-class objectives. Its pipeline combines adversarially aware data construction, knowledge-enhanced continued pretraining, and policy-grounded multi-task safety post-training, including risk-aware supervised fine-tuning and reinforcement learning-based policy optimization, together with safety-aware agentic reinforcement learning for tool use and multi-step reasoning in complex safety scenarios. We further introduce the Yuvion LLM RiskEval (YLRE), a collection of 93 benchmarks across four evaluation categories, covering diverse open and internal evaluations with a focus on safety, adversarial robustness, and real-world capability requirements. Across these evaluations, Yuvion LLM demonstrates clear advantages on safety-focused benchmarks and particularly strong robustness under adversarial conditions, while maintaining solid overall capability. Notably, Yuvion-8B outperforms most state-of-the-art baselines, including substantially larger models such as GPT-5.4 and Qwen3-MAX, on several safety tasks.

Cross-platform deployment of offensive comment detection for Chinese social media suffers performance degradation. The paper proposes a dual-threshold hard mining method to address this. First, the clean-Chinese-base RoBERTa is finetuned on COLD to establish a binary baseline for fair comparison. Second, a three-class fine-labeled test set covering Weibo, Xiaohongshu, Tieba, and Zhihu is constructed, domain distances from the source are quantified using Jaccard and Proxy-A Distance, as well as the degradation bottleneck of the baseline under domain shift is systematically revealed. Herein, a dual threshold hard example mining strategy is proposed. High- and low-confidence error-prone samples are filtered from unlabeled corpora by prediction confidence. The model is secondarily finetuned under implicit contexts with merely a small set of manually labeled hard examples, realizing low-cost cross-platform domain adaptation. Experiments reveal significant performance gains of the optimized model across four platforms.

Discrete audio representations have become increasingly popular for building multimodal text-audio systems and integrating audio capabilities into Large Language Models (LLMs). However, numerous studies report performance degradation on various downstream tasks due to information loss during discretization. To address this, we propose a novel approach combining temporally compressed discrete tokens with dimensionality-reduced continuous residuals. Our framework consists of a hybridized discrete-continuous focal modulation codec and a hybrid Transformer. This architecture performs autoregressive inference in the discrete domain, coupled with non-autoregressive prediction and continuous residual upsampling. Experimental results show that our approach significantly improves the retention of speaker characteristics compared to discrete-only methods, while simultaneously reducing the number of required autoregressive steps.

This paper presents Parallel ARC (P-ARC), a parallel variant of the Adaptive Robot Coordination (ARC) approach to multi-robot motion planning (MRMP). P-ARC proposes a parallel variant for each of the three main stages in ARC: initial individual solutions, conflict detection, and conflict resolution, exploiting the independence created by ARC's decomposition of the MRMP problem. Additionally, we employ an OR-parallel multi-start strategy to both ARC and P-ARC, creating a hybrid parallel strategy OR-P-ARC. We evaluate the impact of the different parallel strategies for ARC using a set of scaling 2D mobile and planar manipulator scenarios with up to 128 robots to control for conflicts and work distribution across the stages of ARC. Additionally, we demonstrate planning time speedups approaching 4X over the sequential version for large Panda multi-manipulator teams in real-world inspired scenarios when deploying 16 CPU cores.

Using robots to estimate the location of the radiation source is an effective way to improve efficiency and safety. Existing methods focus on planning the robot's path to achieve precise estimation, typically approaching the source. However, approaching the source increases the risk of radiation damage to a robot. In addition, a path-planning algorithm designed solely for radiation source localization (RSL) limits the flexibility of missions that deploy robots into radioactive environments. This study presents an automation framework for robotic RSL that leverages a physics-informed machine learning (PIML) model to precisely estimate the source location, regardless of measurement paths, in unknown environments. Physics-inspired model tensors have been designed for PIML to handle attenuated gamma-ray flux signals from unknown obstacles, and multiple models are computed in parallel to improve the robustness and precision of the RSL. The proposed method is evaluated in high-fidelity simulation environments using Monte Carlo particle transport across diverse randomized domains, including spatial scales, radiation source types, obstacle materials and geometries, and robot trajectories. The method is also validated through physical experiments on configurations that are not included in the simulation-based evaluation. The continuous learning technique is applied in real-robot deployment to enhance the practical applicability of the online robotic RSL system. The proposed method advances robot radiation perception from pointwise flux detection to spatial intelligence.

Byzantine-robust federated learning seeks to protect distributed model training from malicious or corrupted clients without requiring access to their private data. FLTrust addresses this challenge by introducing a trusted server-side root dataset that assigns trust scores to client updates for more robust aggregation. In this work, we propose FOGGYTRUST, a hierarchical extension of FLTrust that localizes trust computation to fog nodes, allowing the framework to better handle globally heterogeneous data while preserving robustness within locally homogeneous client groups. We further show that this two-level architecture can simultaneously address distribution mismatch in trust estimation and client drift across groups by combining local trust-based aggregation with heterogeneity-aware global optimizers such as FedAdam and SCAFFOLD. Across benchmark datasets, FOGGYTRUST achieves its strongest gains on more challenging heterogeneous settings, particularly on CIFAR-10 under Krum and Trim attacks, where it achieves an over 50% improvement over FLTrust. We also test FOGGYTRUST in a real-world safari dataset to show the promise of hierarchical trust networks for robust federated learning in socially impactful, safety-critical settings such as distributed wildlife monitoring.

Dyslexic learners increasingly use artificial intelligence (AI) tools to support reading, writing, organisation, and study-related tasks. However, their lived experiences with these tools remain largely underexamined. This paper proposes DysLexLens, a low-resource LLM framework, designed to analyse dyslexic learners experience with AI through online forum discussions. DysLexLens is designed as an end-to-end, evidence-traceable architecture which transforms noisy social media posts into a dictionary-driven corpora, provides knowledge-graph (KG)-based question reasoning, generates verifiable query responses, and enables response evaluation through quantitative and human-grounded assessment. DysLexLens has four key features. First, it employs a dictionary-driven filtering method to construct a more focused Reddit corpus on dyslexia and AI, filtering out noisy and weakly related posts to improve the relevance of data collected from low-resource forum contexts. Second, it integrates LLM-assisted semantic analysis with KG-based query reasoning to uncover meaningful patterns. Third, it has quantitative evaluation metrics (RAGAS and Query Robustness) to measure LLM-generated response performance. Fourth, it provides structured qualitative validation guidelines for assessing response quality, with a specific focus on hallucination and evidence alignment. We demonstrate the effectiveness of DysLexLens using dyslexia-related Reddit forum data and 30 questions. The results show its potential generalisability to other low-resource forum data contexts. DysLexLens, sample data, questions and evaluation results are available at Github to support reproducibility.

Masked Diffusion Models (MDMs) promise fast, parallel language generation, but their reverse transition factorises across token positions -- an approximation that breaks down in the few-step sampling regime where parallel generation ought to provide the greatest efficiency gains. Flow Language Models (FLMs) sidestep this limitation by learning a continuous flow that transports noise toward clean sequences represented in Euclidean space, inducing a flow map that can be distilled for single-step generation. However, this makes complex tasks requiring multi-step reasoning problematic for FLMs, as FLMs are forced to decode every token during generation. To address this, we introduce Masked Language Flow Models (MLFMs), which incorporate masking into FLMs using a continuous stochastic interpolant to bridge partially masked and clean sequences. This design enables conditional generation via continuous flows and allows pretrained MDMs to be converted into MLFMs through a simple, lightweight adaptation. Leveraging this flexibility, we propose a novel sampler that alternates continuous denoising with the discrete unmasking of confident tokens to better support multi-step reasoning. We evaluate our approach on GSM8K and MT-Bench and find, for the first time, that flow-based language models can be scaled to solve downstream reasoning and instruction-following tasks.

Operations Research (OR) provides a rigorous framework for high-stakes decision-making, but effective OR modeling requires substantial domain knowledge, mathematical abstraction, and solver expertise. Recent LLM-based systems automate parts of this pipeline, yet remain limited by low accuracy on complex problems, opaque outputs, and narrow solver support. We propose COOPA (COoperative OPerations Agent), a modular LLM-agent architecture for interpretable and scalable OR decision support. It combines three components: iterative confidence-based modeling, which generates multiple candidate formulations, self-evaluates them across modeling dimensions, and selects one using a max-min confidence criterion; element-level provenance and confidence explanations, which link variables, parameters, constraints, and objectives to quoted source text and provide an audit trail for human verification; and multi-solver routing to specialized optimizer agents for different OR problem classes. Across three OR benchmarks, eight LLM backbones, and four baselines under identical conditions, COOPA achieves the best macro-average accuracy on six of eight backbones and improves over the strongest baseline by up to 6.7 percentage points. A within-system ablation isolates the contribution of iterative confidence-based modeling, while additional analyses and case studies illustrate the value of source traceability and multi-solver dispatch.

Physical or operational constraints often impose communications limitations on autonomous agents. Such limitations complicate monitoring or multiagent coordination. Even when strong communications are absent, some information may still be available. The remainder of the relevant agent state may be reconstructed via estimation. The actions taken by an agent are a potential source of information -- as the agent interacts with the environment, these actions may be observed even in the absence of explicit communication. We investigate using actions to estimate the state of an agent, using reinforcement learning to develop policies which make the estimation problem more tractable. Policy observability is encouraged through the training reward and is analyzed using simulation of the trained agent. In an aircraft tracking problem a policy with enhanced observability is found that has minimal impact on nominal task performance.

We present Qwen-Image-2.0-RL, a post-training pipeline that applies reinforcement learning from human feedback (RLHF) and on-policy distillation (OPD) to improve both the visual quality and instruction-following capability of the Qwen-Image-2.0 diffusion model. To provide reliable reward signals, we construct task-specific composite reward models by fine-tuning vision-language models with a pointwise scoring paradigm and chain-of-thought reasoning. For text-to-image generation, the reward models cover alignment, aesthetics, and portrait fidelity dimensions. For image editing tasks, the reward system addresses instruction-following accuracy and face identity preservation. Building on this reward system, we develop a scalable GRPO-based RL training framework, incorporating a hybrid classifier-free guidance (CFG) strategy to preserve pre-trained knowledge, prompt curation via intra-group reward range filtering, and per-category reward weight calibration. To merge the task-specialized RL policies for T2I and editing, we propose on-policy distillation as the final training stage, which consolidates multiple teachers into a single student model through trajectory-level velocity matching. Extensive evaluation shows that Qwen-Image-2.0-RL achieves 57.84 overall score on Qwen-Image-Bench (+2.61 over the base model), Elo ratings of 1193 in text-to-image arena (+78) and 1349 in image edit arena (+93), demonstrating consistent gains in aesthetic quality, prompt adherence, and editing accuracy.

Recently, test case selection (TCS) techniques have been explored to support the operational evaluation of deep neural networks (DNNs) under limited testing budgets, where labeling cost is a primary concern and uncovering model failures early is a key objective. Although prior studies report promising results, existing empirical evaluations focus almost exclusively on vision-based DNNs and datasets, leaving it unclear whether prior findings generalize to LLM code models. This paper presents a large-scale replication study of TCS techniques in the context of LLM code models. We re-examine established TCS strategies originally proposed for DNNs and complement them with statistical sampling strategies not previously evaluated for TCS. We assess their effectiveness on three code-related classification tasks: clone detection, vulnerability detection, and technical debt prediction. The study spans 17 task-specific fine-tuned model instances, 7 predictive features, and 13 selection strategies, including 12 feature-aware strategies and simple random sampling (SRS) as a feature-agnostic baseline. We evaluate performance along two dimensions: accuracy estimation and early failure discovery. The results indicate that only a subset of findings reported for vision-based DNNs generalize when TCS is applied to LLMs for code. In particular, uncertainty-based features are effective for early failure discovery, while representation-based features are more robust for accuracy estimation. At the same time, performance varies substantially across tasks and models, indicating that TCS effectiveness is context-dependent. Overall, this study provides empirical evidence on the replicability of TCS techniques beyond vision-based deep learning and offers insights into their use for the operational evaluation of LLMs for code.

While many explainable AI (XAI) methods have been proposed, most are not designed for time-series forecasting models and often rely on the implicit assumption that timestamp features are independent. This assumption ignores the fundamental property of temporal dependence and can lead to explanations that violate the sequential and causal structure of the data. We introduce \textsc{KARMA}, a method for explaining time-series predictors by constructing a Markov surrogate model that captures the temporal dependencies learned by the predictor. Our approach revolves around three main aspects: identifying the minimal history length $K$ that is predictively sufficient for the model, estimating the best-fitting $K$-order Markov transition kernel from the discretized history space, and a five-level global explanation hierarchy that can be derived from the Markov transition kernel, which we illustrate using real-world weather data (Beijing PM 2.5). We also certify using complex synthetic data with known true causal edges that KARMA (i) recovers the data causal structure as learned by the model via a controlled experiment and (ii) identifies temporal dependencies better than established attribution methods such as TimeSHAP.

Ultra-fine entity typing (UFET) assigns highly specific types to entity mentions, but current approaches struggle with types in the long tail. We hypothesize that a key limitation is the reliance on sentence-level context, since disambiguating evidence is often spread across multiple sentences. Testing this has been difficult because all existing UFET resources are sentence-level. We present Narrative-UFET, a controlled extension of UFET in which each entity mention is paired with an automatically generated short, coherent narrative. Synthesizing narratives lets us isolate the effect of specific discourse properties. We experiment with two paired variants: one in which the entity's type is held constant across the narrative (Maintain) and one in which it shifts (Change). We show that narrative context yields consistent improvements on long-tail types over sentence-level baselines, with the Change variant providing the stronger signal. A comparison against naturally occurring contexts shows that synthetic narratives yield stronger gains, indicating that controlled discourse construction can surface signals that real text leaves implicit. Substantial room for improvement remains, suggesting open directions in both discourse modeling and narrative construction.

Large Vision-Language Models (LVLMs) exhibit sophisticated reasoning but remain susceptible to object hallucination. Deviating from the prevailing attention intensity assumption, we reveal a deeper dynamic structural misalignment: hallucination is triggered at decision-critical steps where specific attention heads, acting as risky mediators, decouple from visual evidence to lock onto language priors. This establishes a pathological shortcut that bypasses visual grounding. To dismantle this, we propose Fox (Faithfulness and Observational-flow via eXpression-rectification), a training-free inference-time framework. Fox diagnoses structural misalignment using a visual attention entropy probe to localize risky mediators unsupervisedly. We then execute a targeted causal intervention via numerical logit saturation to physically sever the shortcut path. Finally, a conflict-gated cooperative decoding strategy reconciles interventional faithfulness with observational fluency. Extensive experiments demonstrate that Fox achieves SOTA performance, outperforming SID by 29.1% while preserving linguistic richness. Code is available at https://github.com/Cc2021start/Fox.

Web-agent benchmarks overwhelmingly measure depth -- pinning one obscure answer behind a chain of constraints -- while breadth, exhaustively enumerating a closed set and filling each item's attributes, is barely evaluated, especially outside English. Breadth is also hard to build: certifying that a gold set is complete and every cell correct is far costlier than checking a single answer. I introduce \textsc{Ko-WideSearch}, a Korean breadth-search benchmark built by an automated synthesize-and-verify pipeline. Each task names a set-parent entity -- a TV season, a dynasty, a league, an administrative region, an election -- and asks for its full membership plus a per-item attribute table, graded by Item-, Column-, and Row-F1. It spans 228 tables over 190 entities and sixteen categories across three difficulty tiers, set by two structural knobs I dial independently -- table width and a 2-D composite key -- so cross-product membership climbs from 0\% to 100\% across the tiers. A single normalization-aware comparator is shared between gold construction and grading, so stable date and count columns are not over-dropped on formatting alone. Across twenty web agents, the failure is consistent: agents recover the set but not the rows (e.g.\ Item-F1 92.8 against Row-F1 53.7), accuracy falls steadily as the knobs harden, and neither more search nor more spend closes the gap. Broken down by cell, the hard part is finding the right value, not formatting it: open-ended free-text cells fail most, while cells with a standard answer such as a date or a name usually come out right.

We introduce a categorical framework called ODYSSEY for constructing verifiable, local truth-preserving foundation models as compositions of foundries: building-block architectural components that specify a cover of local contexts, local representation families, restriction maps, gluing rules, obstruction policies, update obligations, and human-facing views. A foundry is an organized sheaf of knowledge that carries within it an argumentation component. Concrete foundries are built from generic foundries such as evidence/argument, operational decision, institutional/financial, market meaning, scientific challenge, research-program, assistant-build, and evaluation-harness foundries. Universal Foundry Learning (UFL) formalizes foundry construction as a composition of left and right Kan extensions, with left Kan extension rolling local artifacts into candidate foundries and right Kan extension enforcing the restriction, gluing, obstruction, and argumentation conditions required for promotion. Foundry SQL (FSQL) is a small typed query surface for slicing maintained foundry artifacts that uses TICKET (Topos Integration using Causal Kan Extension Transformers) certification for admitting external or pre-built models into durable ODYSSEY state. ODYSSEY is fully implemented and tested across a wide spectrum of concrete foundries, showing that the same categorical machinery supports domain construction, artifact replay, sheaf diagnostics, grounded Toulmin/local-LLM scrutiny, residual-obstruction ledgers, and optimized TICKET-compatible causal-claim extraction across heterogeneous sources. This paper is to be presented as a 2.5 hour tutorial at ICML 2026. The tutorial home page is at https://bit.ly/4ajS0nA.

3D scene inpainting is essential for reconstructing areas corrupted by occlusions or limited viewpoints. While recent methods leverage Gaussian Splatting (GS) for efficient 3D editing, they often depend on precise multi-view segmentation masks and are inherently constrained to object removal tasks. We propose CoIn, a novel framework that bridges 2D inpainting models and 3DGS through a multi-stage consistency pipeline. Our approach first generates initial inpainted images using a diffusion model, enabling the use of arbitrary-shaped masks and diverse tasks like object insertion. We then introduce Reference Adaptive GS with Feature Attention to reconstruct a coarse 3D scene by adaptively weighing towards a reference view (2D -> 3D). This 3D representation provides geometric guidance to the diffusion process via GS-based Reference Feature Warping, ensuring multi-view consistency (3D -> 2D). Finally, a Texture-Enhancing Discriminator refines the 3D scene to achieve high photometric realism (2D -> 3D). Experiments show that CoIn, effectively leveraging bidirectional information flow, achieves state-of-the-art performance and effectively handles both object removal and object insertion with flexible mask input.

Current humanoid reinforcement-learning policies excel at free-space motions but struggle with contact-rich tasks, as pure kinematic tracking cannot resolve the physical ambiguities of interacting with objects and uneven terrain. To address this, we introduce SceneBot, a unified motion-tracking framework capable of handling freespace locomotion, terrain traversal, and whole-body manipulation. SceneBot conditions a single policy on both reference motions and per-link contact labels, explicitly defining expected environmental interactions. To overcome the lack of annotated interaction data, we propose a hindsight scene reconstruction approach that infers scene-interaction graphs from retargeted human motion. Trained on 7.5 hours of this reconstructed, contact-rich data, SceneBot successfully generalizes to unseen motions and environments. Our results demonstrate that SceneBot is the first general framework to seamlessly unify free-space and contact-rich behaviors executing complex, long-horizon tasks like carrying a box upstairs and establishing contact conditioning as a powerful interface for humanoid control. All code and data will be open-sourced. More demos and information are available at: https://ericcsr.github.io/scenebot/

Reinforcement learning from human feedback (RLHF) in production does not always have a synchronous reward signal. Code-execution verifiers, slow judge ensembles, and queued human review can return several gradient steps after the rollout that produced them, breaking the synchronous-reward assumption underlying standard PPO. We address this gap with Retroactive Advantage Correction (RAC): each pending slow completion is queued, aged through a non-negative kernel, and reinjected as a clipped residual into the next optimiser step's advantage. We prove that under an unbiased clipped importance ratio, the cumulative RAC correction is exactly unbiased when the effective delay kernel reinjects all of its mass, and carries a bias linear in the unreinjected fraction otherwise; at the no-delay identity kernel it reduces to V-trace. On a tabular Markov decision process (MDP) proof-of-concept, RAC reduces the closed-form policy bias by up to 47.9x at the two-slow-channel configuration, beating wait-for-slow at lower wall-clock cost. RAC integrates with PPO and GRPO through a two-line reward-manager patch.

Accurate assessment of tumor proportion score (TPS) in non-small cell lung cancer (NSCLC) is critical for treatment planning and prognosis. Key challenges include the tedious manual work required to annotate each slide, combined with the limited number of experts certified for this task. Multiple instance learning (MIL) has proven to be an effective approach for predicting TPS scores at the slide level; however, existing methods struggle with non-expressive (zero class) images. Our approach involves two models: (1) an embedding-extraction and multiclass-classification network that captures the histopathological features of individual patches, and (2) a MIL model that aggregates these embeddings to predict zero-inflated beta (ZIBeta) parameters representing the overall TPS probability distribution for the entire slide. Using only slide-level TPS scores as labels, we demonstrate how this end-to-end framework can leverage a novel distribution-based architecture to improve prediction accuracy and explainability. ZIBeta modeling significantly outperforms baseline linear and ridge regression while capturing expected accuracy through distribution concentration.

Reward models for Reinforcement Learning from Human Feedback (RLHF) pool preferences across thousands of annotators and fit one global affine calibrator, collapsing raters with systematically different rating-scale offsets and slopes into a single average-rater fit that does not match any individual annotator. PEBS is a per-rater empirical-Bayes shrinkage estimator: it fits per-rater affine calibrators on a held-out slice of each annotator's ratings and applies Morris-James-Stein empirical-Bayes shrinkage toward the population mean, in closed form and without retraining the reward model. On PRISM, PEBS reduces within-user held-out RMSE by 8.58% over the pooled population-slope baseline. The procedure replicates on PluriHarms harm ratings (Qwen-2.5 base, in-family) with a +9.66% RMSE reduction over the same population-slope baseline. PEBS is a closed-form post-hoc estimator for annotator-specific affine calibration in RLHF reward modeling; it leaves the reward base model unchanged and estimates only the rater-level map used at inference time for new ratings.

This paper formulates frame-level freeway risk assessment as a multi-agent scene graph-level binary classification problem, where each video or trajectory frame is labeled risky if any TTC- or PET-based conflict violates a specified severity threshold. We construct a relation-aware graph per frame with vehicles as nodes and two interaction types as edges: same-lane (longitudinal) and adjacent-lane (lateral), augmented with physics-informed edge features aligned to rear-end and lane-change conflict mechanisms. Building on a structured benchmarking suite of non-graph models and graph baselines, we propose HIA-GAT, a dual-stream heterogeneous graph attention network that processes longitudinal and lateral interactions through dedicated attention pathways and fuses them via a conflict-type-aware gating mechanism with event-level gate supervision derived from SSM conflict attribution. Experiments on the NGSIM I-80 and US-101 freeway datasets across nine TTC and PET threshold configurations show that HIA-GAT achieves the best average risk-ranking performance (AUC 0.835 on I-80 and 0.867 on US-101), with the largest gains on PET-only (lane-change) settings where relational structure is essential. Beyond accuracy, the learned gate provides interpretable per-vehicle attribution of dominant conflict type, supporting actionable, real-time freeway safety monitoring. We show that graph structure is critical for modeling lateral conflict risk, while longitudinal risk can often be captured by non-relational aggregation.

Prompt injection is the top security risk for LLM-integrated applications, yet every defense proposed so far has been broken. We prove this is not a coincidence: in shared-embedding architectures that lack enforced control-data separation, perfect prompt-injection prevention is mathematically impossible. We formalize prompted systems as Prompted Action Models whose outputs include control-authoritative actions: refusal decisions, tool authorization, policy routing, and memory writes. We define Semantic-Faithful Control (SFC), the property that such behavior depends only on the meaning of untrusted input, not on how it is encoded. We then prove SFC is unachievable within the shared pipeline, via three results: a provenance-recovery impossibility (shared representations make trusted and untrusted content statistically inseparable, bounded by total variation distance); control-path exposure (untrusted tokens enter control-relevant computation through the same attention value-aggregation that determines outputs); and a finite-coverage invariance gap (finite training cannot certify invariance over infinite semantic-equivalence classes). We ground each quantity in measurements on production tokenizers and models. The result is structural, not a gap in current defenses. It mirrors the code-data confusion in Von Neumann machines that gives rise to buffer overflows, a vulnerability class that took decades of layered defenses (DEP, Write-XOR-Execute, ASLR, stack canaries, and ultimately memory-safe languages) to contain, because no single mechanism sufficed. The implication is the same: prompt injection cannot be eliminated by better in-pipeline classification or alignment alone. It requires architectural separation of instruction and data channels. We identify the root cause and the class of solution it demands.

Generative models have achieved remarkable success in data synthesis, though recent advances driven by increasing model scale have introduced challenges in computational cost and efficiency. Quantum machine learning offers a promising alternative, representing complex data distributions using compact, highly expressive models. Here, we propose QDiffusion-TS, the first quantum generative diffusion model for time series synthesis, and validate it on the IQM quantum processor. The framework extends a classical diffusion architecture by replacing feed-forward components within the denoising transformer with quantum neural networks, yielding a hybrid quantum transformer that reduces the number of trainable parameters in each replaced component by nearly three orders of magnitude. Evaluated on financial time series from Apple and Amazon, the model generates synthetic data that more accurately reproduces the real distributions, reducing Wasserstein distance by approximately 44% relative to its classical counterpart across both datasets. In a downstream forecasting task, augmentation with the generated data improves predictive performance by up to 71% in RMSE over a baseline trained solely on real data. These results show that quantum enhanced architectures can consistently match and frequently surpass classical performance with substantially fewer parameters, establishing a practical framework towards more efficient and scalable data-driven generative modelling.

Fairness measurements in the form of disaggregated evaluations often rely on demographic signals that are legally constrained or culturally sensitive. Race and ethnicity signals are among the more difficult signals to curate and use for this task. This paper presents Privacy-Preserving Probabilistic Race/Ethnicity Estimation (PPRE) as a method for enabling fairness measurements with respect to race/ethnicity for U.S.\ LinkedIn members in a privacy-preserving manner. PPRE applies privacy technologies (specifically: secure two-party computation, differential privacy, and additive homomorphic encryption) on top of two race/ethnicity demographic signal sources (the Bayesian Improved Surname Geocoding estimator and a sparse golden survey set of self-reported demographics) to power a fairness measurement solution with respect to US-based race/ethnicity demographics. We detail its privacy guarantees and demonstrate its application on candidate- and viewer-side fairness measurements. We close with a transferable framework for institutions seeking to implement similar privacy-preserving measurement infrastructure.

Multi-token prediction has been shown to increase data density during training, improve downstream text-generation quality, and serves as the defacto approach for self-speculative decoding. Existing foundation and open source models that use MTP heads commit to a static tree-based attention topology throughout the entire generation sequence, meaning the speculation depth, and thus the compute required during verification, stays constant regardless of the context. This is fundamentally misaligned with the entropy patterns of natural language where low-entropy regions often support reliable multi-step drafting, while high-entropy regions require more conservative speculation. To address this, we propose Entropy-guided Multi-Token Prediction (EntMTP), a training-free scheduler that toggles between tree-based attention topologies from a set of task-specific pareto-optimal trees conditioned on a running estimate of local generation entropy. By matching speculation depth to context predictability, EntMTP maximizes expected accepted-token throughput across the full distribution of generated text without sacrificing generation quality. When evaluated across Humaneval, ShareGPT, GSM8k, and Litbench benchmarks, EntMTP consistently achieves a 1.15x speedup against Hydra and peak speedup of 1.36x against Medusa baselines respectively.

The variations in vocal effort range (e.g. whisper, soft, neutral, loud, shout) alter production and speech acoustics, reducing intelligibility and limiting the robustness of any subsequent speech technology. Classification is challenging since effort lies on a continuum, adjacent categories are easily confused, and labeled data remain scarce. Prior SSL approaches with wav2vec2, HuBERT, and AST improve performance on the AVID corpus but still suffer from boundary errors. In this study, we introduce WavLM for the first time in vocal effort classification and benchmark it against wav2vec2 and HuBERT. To address data scarcity, we conduct a systematic study of augmentation strategies, covering RIR convolution, additive noise, time masking, speed perturbation, band-limiting, MixUp, and CutMix. Augmentation consistently improves WavLM, with gains ranging from +0.6% to +1.8% absolute. We further propose Gaussian-neighbor soft labels, which further reduce near-boundary confusions by modeling the vocal effort continuum. Our best system, WavLM-BASE with gradual unfreezing, augmentation, and Gaussian-neighbor soft labels, achieves 78.2% mean accuracy, establishing a new state-of-the-art on AVID.

Social media popularity prediction aims to forecast the future reach or influence of online content from early-stage observations. Accurate prediction enables key downstream applications, such as advertising optimization and strategic content planning by users, creators, and platforms. Despite substantial progress, existing popularity prediction works often fail to jointly consider multimodal content and temporal social interaction signals. Moreover, the literature remains highly fragmented across datasets, modalities, observation windows, prediction targets, and evaluation protocols. This fragmentation prevents fair comparison and obscures a systematic understanding of how textual, visual, temporal, and interaction-based signals jointly shape popularity dynamics. To address these challenges, we introduce MMG-Pop, a Multi-modal Graph-based Popularity Prediction benchmark, which unifies datasets, modalities, temporal interaction signals, and representative baselines under a standardized evaluation protocol. Furthermore, we propose MMG-PopNet, a unified multi-modal graph-based network that jointly models the aforementioned multi-modal signals and graph-structured social interactions. Extensive experiments on MMG-Pop, comprising four datasets across Bluesky and Reddit platforms, demonstrate the superior performance of MMG-PopNet and yield new insights into cross-platform training generalization, multi-task prediction benefits, multi-modality contributions, and LLM prediction limitation. These findings establish a unified foundation for future research on social dynamics modeling and intervention under heterogeneous modalities and socially-aware agentic ecosystem paradigms.

We introduce the context-ready transformer, a new recurrent neural network architecture built from a D-layer transformer block that pre-contextualizes each token before it enters the block. During left-to-right generation, a correction network combines the previous position's block output -- a cached summary of past context -- with the current token embedding, so the tokenenters the block already contextualized rather than as a raw embedding. At sequential inference, the correction chain makes the architecture a recurrent neural network. For training, we unroll the correction process K times over the full sequence, processing all positions in parallel at each step. A pretrained transformer can also be converted to a context-ready model by adding a zero-initialized correction FFN and fine-tuning. We evaluate across widths, depths, block sizes, and two datasets, with all comparisons against standard transformers, variants, and ablations. A D=5 model beats a 12-layer transformer while generating 1.7x faster on an A100. With K=10, a single-layermodel (D=1) beats a 6-layer transformer with a 2.6x inference speedup, and sequential inference matches parallel K=10 to within 0.01 PPL. The architecture benefits most from wide representations and long contexts. On a pointer-chasing task, D=1 trained with BPTT solves all 10 composition levels, while standard transformers exhibit staircase-like depth dependence.

Speech emotion recognition (SER) often relies on hard consensus labels that collapse annotator disagreement. We study distribution-based supervision for 9-class SER on MSP-Podcast 2.0 using a WavLM-Base multitask model for categorical emotion and dimensional VAD. Hard-label training is compared with targets from primary and merged primary--secondary annotator vote distributions. Distributional objectives improve alignment with human vote distributions, reducing JSD/KLD relative to hard-label training. Analysis shows that hard supervision partly benefits from assigning ambiguous utterances to the residual Other class, whereas distributional supervision redistributes uncertainty across emotion categories. Entropy-stratified evaluation shows that high-ambiguity utterances remain challenging, but distribution-based supervision better captures perceptual uncertainty. These findings support moving beyond hard labels toward targets that reflect listener disagreement.

Large Language Models (LLMs) possess broad conceptual knowledge acquired through large-scale text pretraining, yet their potential to supervise models in other modalities remains underexplored. In this work, we propose LaViD--Language-to-Visual Knowledge Distillation--a simple and effective framework for transferring high-level semantic knowledge from a language-only teacher to a vision-only student model. Instead of relying on paired multimodal data, LaViD elicits conceptual signals from an LLM by prompting it to generate multiple-choice questions (MCQs) that probe semantic distinctions between visual classes. Each class is mapped to a soft label distribution over these MCQs, forming a rich conceptual signature that guides the student through an auxiliary distillation loss. Notably, despite using a language-only teacher without access to image data, LaViD consistently outperforms recent methods like MaKD that distill from vision-language models across multiple fine-grained benchmarks. It also achieves competitive or superior performance compared to state-of-the-art visual distillation methods such as DKD and MLKD, with further gains when combined with logit standardization. On the Waterbirds dataset, LaViD substantially improves worst-group accuracy, demonstrating enhanced robustness to spurious correlations with distillation. Code is available at https://github.com/lliangthomas/lavid.

Accurate prediction of bottom-hole pressure (BHP) and CO2 plume migration is essential for safe geological carbon storage, yet practical simulations often rely on truncated domains where artificial boundaries distort pressure diffusion and CO2 saturation footprints. In this study, we evaluate how boundary-condition fidelity affects BHP and CO2 plume prediction by comparing ten reduced-domain boundary treatments against full-domain reference simulations in homogeneous and heterogeneous reservoirs. We test uniform pore-volume multipliers, transmissibility modifiers, corner-adjusted pore-volume corrections, layered corrections, and gradual modifiers using BHP RMSE, NRMSE, peak pressure deviation, and plume Intersection over Union (IoU) as performance metrics. Our results show that conserving corner pore volume is the most important requirement for truncated-domain modeling. We find that uniform treatments which neglect corner storage generate large pressure errors, with BHP RMSE of 362 to 382 psi in the homogeneous model and 250 to 304 psi in the heterogeneous model, and yield plume IoU values near 0.80 to 0.84, indicating roughly 16 to 20% of the combined plume area is misrepresented. Corner-adjusted scenarios substantially reduce pressure errors and raise plume IoU above 0.94, but we observe that transmissibility correction is not universally beneficial. In homogeneous reservoirs, uniform transmissibility adjustment improves pressure fidelity; in heterogeneous reservoirs, it can over-restrict flow across variable-permeability boundary faces, increasing BHP error and contracting the predicted plume. We find the gradual modifier with transmissibility correction provides the most consistent performance, achieving BHP NRMSE below 3.7% and plume IoU above 0.97 in both reservoir types.

Activation patching is the primary tool in mechanistic interpretability. It attributes causal responsibility for a model behavior to each of its individual components by estimating its natural indirect effect (NIE). Re-deriving the activation patching estimand from causal mediation analysis, we find that the NIE does not solely capture the causal effect through the specific component. It also contains interaction effects (INT) that measure how much the component's causal effect itself depends on the state of other components in the model. A natural response may be to try to eliminate INT by adjusting the estimator or unit of analysis, but each of these potential remedies has predictable failure modes. We demonstrate these failure modes in the GPT-2 IOI circuit; components whose causal importance is conditional on the state of other components are either invisible or artificially inflated, and INT variance explains the previously documented instability of faithfulness scores. We prove that INT scales with the distance between clean and patched component activations, is negligible when the model is locally affine, and decomposes combinatorially into pairwise and higher-order group interactions. Despite its inevitability, INT is not a nuisance to be eliminated, but rather a diagnostic for interpretability studies. Its individual and group-level magnitude and sign signal when causal conclusions are prompt-dependent, and when greedy NIE-based component ranking will miss mechanisms only discoverable through combinatorial search.

[Context] Screen reader users navigating webpages by element list often encounter accessible names such as "Read more" that are valid under the W3C Accessible Name and Description Computation specification but uninterpretable in isolation. The surrounding elements that would make these names meaningful exist in the page but are not linked to the target by any mechanism. No prior work has empirically studied how to select which surrounding elements are contextually relevant to a given target. [Objective] This registered report investigates whether human-perceived contextual associations between webpage elements can be recovered from the accessibility tree using link prediction, and whether the learned associations generalize across websites. [Method] We will construct a dataset of human-annotated contextual associations on 35 websites, stratified across the Tranco top-million list, with three independent annotators per page. Each page is represented as a graph derived from its accessibility tree, augmented with spatial and semantic features from the DOM and CSS. We compare four machine learning models (MLP, GCN, GAT, and SEAL) against two heuristic baselines under leave-one-site-out cross-validation with a pre-registered statistical framework, using Hit@K and MRR. [Results] We have conducted a five-site author-annotated pilot study to establish the pipelines and parameterize the power simulation, with pilot Hit@10 ranging from 0.16 to 0.85 across four learned models and 0.08 to 0.30 across two heuristic baselines. The final results will be reported after the planned experiments and analyses are completed. [Conclusion] The study contributes a human-annotated dataset of contextual associations on webpages, an empirical evaluation of link prediction for context selection on accessibility-tree graphs, and a cross-site generalization analysis.

Large Vision-Language Models (LVLMs) specialized in healthcare are emerging as a promising research direction due to their potential impact in clinical and biomedical applications. However, progress is constrained by the scarcity of high-quality medical multimodal data, concerns about robustness in safety-critical settings, and the narrow and potentially contaminated evaluation benchmarks that limit reliable assessment. To address these issues, the field requires state-of-the-art solutions to be fully open and reproducible systems in which all components can be inspected, evaluated, and improved. This work introduces Aloe-Vision-Data, a large-scale, quality-filtered mixture which integrates both medical and general domains across multimodal and text-only sources, designed for direct use in model fine-tuning. Building on this dataset, we train the Aloe-Vision family of medical LVLMs, openly released with full weights, training recipes and data, in two scales (7B and 72B). Through comprehensive benchmarking, we demonstrate that high quality training mixtures produce balanced LVLMs which yield significant gains over the baseline models without compromising general capabilities, achieving competitive performance with respect to state-of-the-art alternatives. To support reliable evaluation, we introduce CareQA-Vision, a carefully curated vision benchmark derived from MIR and EIR exams, the residency entrance exams for medical and nursing specialists in Spain, offering novel vision questions with low likelihood of contamination. Finally, we show that current LVLMs remain vulnerable to adversarial and misleading inputs, underscoring reliability challenges in clinical contexts.

Research on agent memory has matured rapidly, but almost entirely on the text side: few existing benchmarks ask, in an interactive environment, when an agent genuinely needs to remember what it saw rather than what it could write down. We introduce DMV-Bench (Code: https://github.com/yyyujintang/DMV-Bench), the first interactive benchmark for multimodal-agent visual memory. DMV-Bench is built on a controlled home-furnishing e-commerce catalogue of 1,000 product variants in which a text-leakage contract keeps the discriminative signal of each task in the pixels alone. Across a chain of autonomous shopping sessions, every visited product image carries a unique, pre-rendered incidental cue, and the agent is later asked to recall a particular cued product and navigate to its URL. Inspired by dual-coding theory, we propose DualMem, a memory architecture that maintains a visual and a verbal code in parallel. On DMV-Bench, DualMem outperforms a caption baseline and three recent multimodal agent-memory systems at every chain length J in {5, 10, 15, 50} on both Gemini 2.5 Flash and Qwen2.5-VL-7B, with the lead surviving controls for memory-bank size and encoding-position bias, and an asymmetric dual-coding regime in which vision carries the cue end-to-end while the verbal channel plays a smaller query-grounding role.

Large language model (LLM) multi-agent systems increasingly depend not only on how individual agents reason, but also on how agents are connected. This paper introduces QueenBee Planner, a framework that treats inter-agent communication topology as a retrievable and self-improving design skill. A pool of worker agents, the task adapter, and the scoring function are frozen; only an outer LLM planner learns to generate temporal communication DAGs specifying who sends information to whom, in which round, who merges messages, and who emits the final answer. Execution traces are distilled into evidence-backed design rules with three actions: \emph{Preserve}, \emph{Modify}, and \emph{Avoid}. To prevent self-evolution from turning lucky runs or plausible but false explanations into policy, QueenBee uses held-out acceptance gates, variance-aware credit, motif-level attribution, transfer trust, insight falsification, and structural deduplication. We evaluate the method on Count-Frequency aggregation and Silo-Bench-style distributed coordination tasks. With fixed workers, self-evolved graph generation produces communication structures that improve over fixed topologies and cold generation. In the CF fulltest setting, the best generated graph reduces RMSE from 12.53 for the strongest fixed topology to 7.87 while also reducing messages, model calls, and token cost; Silo-style results show the same direction of improvement over cold and fixed-topology baselines. These results suggest that multi-agent systems can learn reusable architectural design knowledge rather than merely memorizing task answers.

Large language model (LLM) agents have demonstrated strong capability in sequential decision-making, yet they remains fundamentally reactive in long-horizon tasks. Unlike humans who employ "what-if" reasoning to evaluate potential plans before commitment, standard agents lack an internal world model to simulate future outcomes. Therefore, we propose to internalize future-aware planning by training a single autoregressive model to verbalize both a prospective state rollout and a plan-conditioned success estimate-a textual analogue of the Q-value. Crucially, we identify a format-capability gap: simply fine-tuning agents on look-ahead traces during post-training leads to superficial mimicry of foresight without genuine predictive grounding. To bridge this gap, we introduce a three-stage training paradigm: (i) World Model Agentic Mid-Training (WM-AMT) to inject latent predictive capabilities into the policy; (ii) Format-Eliciting SFT (FE-SFT) to structure this injected capability; and (iii) Foresight-Conditioned Reinforcement Learning (FC-RL) to refine the calibration and utility of the generated simulations. Evaluated on search and mathematical reasoning tasks, our approach consistently outperforms other training baselines. Our results demonstrate that effective internal world modeling in LLM agents requires a capability-first training pipeline to achieve grounded and calibrated foresight.

We present a first study of a diffusion-based approach to accelerated sampling of the $N_f = 2$ lattice Schwinger model. Our work is inspired by recent and growing successes in developing such generative models for ensemble generation in LFT to overcome the well-known critical slowing down problem. We train a U(1)-equivariant score-based generative model to sample gauge link configurations from the marginal Schwinger model. By computing model likelihoods, we obtain unbiased estimates for observables that closely match those produced by MCMC simulations. We also demonstrate improvement over HMC as measured qualitatively by a reduction in topological freezing near critical parameters.

Robots trained on real world data tend to be imprecise, slow, and brittle to perturbations. Improving these policies with reinforcement learning (RL) is an appealing alternative, but this process often requires expensive training in the real world. Performing policy improvement in simulation instead provides a far cheaper alternative, but unconstrained RL in simulation can exploit contact and dynamics mismatches, resulting in unsafe behaviors that do not transfer to hardware. Common forms of regularization can furthermore limit improvement by overconstraining to an imperfect behavior prior. In this work, we propose Support-Constrained Off-Domain REinforcement (SCORE), a real-to-sim-to-real framework that constrains RL in simulation to the support of a generative policy pretrained on real data. We instantiate this constraint through flow steering, restricting SCORE to actions the base policy can already produce, which ensures transferable behaviors while maximizing policy improvement. Improving a policy with SCORE requires minimal effort: it learns from sparse rewards, avoids distillation, and leaves the base policy untouched. Across eight real-world dexterous multi-fingered robotic manipulation tasks, SCORE improves average success rate from 37.8% to 89.9%, compared to 59.5% for the best baseline, and reaches success in 36.8% fewer steps than the base policy. Ultimately, through extensive experiments and ablations, we show that simulation can substantially improve real-world manipulation policies when policy optimization is appropriately constrained, introducing a new paradigm for real-to-sim-to-real policy improvement. Videos and code are available at https://weirdlabuw.github.io/score/.

How should we evaluate generation systems that combine autoregressive (AR) and diffusion decoding? We study this question through Speculative Refinement (SpecRef), a training-free hybrid method that warm-starts a masked diffusion language model from an AR draft using entropy-guided selective masking. Evaluating SpecRef across six benchmarks (HumanEval, MBPP, GSM8K, BBH, ARC-Challenge, HellaSwag) with three distinct evaluation protocols (execution-based pass@1, exact-match, log-likelihood scoring), we surface several findings relevant beyond our specific system: (1) code benchmarks conflate structural discovery with logical correctness: providing a syntactic scaffold lifts accuracy from near zero to over 20% without changing the model, indicating that much of the baseline failure is structural; (2) a refinement tension phenomenon where multi-stage correction degrades already-correct tokens, exposing benchmark saturation ceilings invisible to single-model evaluation; (3) log-likelihood and generative evaluation produce different model rankings for the same model pair, suggesting they measure different capabilities; (4) standard Python post-processing silently breaks code evaluation for non-AR generators. These observations apply to any multi-stage or non-autoregressive generation pipeline and point toward more diagnostic evaluation practices.

Large language model (LLM) agents operate over long, multi-session interactions in which facts change: a user moves, a price updates, a plan is revised. Acting correctly requires using the current value of a fact and discarding values that have been superseded. We isolate this ability on real conversational data and show that it is a distinct, unsolved failure. On the knowledge-update subset of LongMemEval, replacing an agent's full context with a bounded, self-maintained memory drops accuracy from 92% to 77% even on a frontier model (gpt-5.4), a gap that is statistically significant (paired McNemar p<0.005) and persists across model scale while full-context accuracy saturates near 92%. The bottleneck is therefore memory maintenance, not comprehension, and is not closed by a stronger model. We then ask whether this is merely an undersized memory, and find it is not: as the conversation grows 24x, accuracy falls further (from 68% to 28%), and granting the agent proportionally more memory yields no detectable recovery (28% to 28%, n=25). The failure scales with the length of the conversation, not the compression ratio. We release Supersede, an open reinforcement-learning environment (on the verifiers / prime-rl stack) that turns this measurement into a training signal: agents are rewarded for answering from the current value and penalized for stale ones. Finally, we close the loop and show the gap is trainable: GRPO fine-tuning a small open model (Qwen2.5-3B) on this environment nearly doubles its held-out supersession accuracy on real, unseen conversations (9.0% to 16.7%, a single run), along a monotonic checkpoint curve indicating the learned policy, not the harness, carries the gain. To our knowledge this is the first trainable environment whose reward targets temporal fact-currency, and the first evidence the supersession gap can be trained down, not only measured.

Molecule generation methods that leverage generative models have been successfully applied to drug discovery. However, they often require extensive pre-training, suffer statistical biases in the training data, and might suffer from limited interpretability of generated chemical structures. In this work, we introduce SpectralMol, an algorithm based on evolutionary computation that processes chemical structures as a compact matrix of Fourier coefficients, projected onto a fixed basis to generate position-wise latent vectors for SELFIES decoding. The NSGA-II algorithm enforces diversity and enable separate objective functions rather than collapsed objectives into a scalar reward. The quality of the algorithm was tested against standardized benchmarks. The results show comparable aggregate benchmark performance with a task-dependent profile: SpectralMol is strongest on several multi-parameter optimization tasks. The same benchmark was used to perform an ablation study to demonstrate the advantages of a structured latent matrix. Finally, method was tested on a realistic ClpP-targeted drug-discovery benchmark, comparing it with the reinforcement-learning-based model under a fixed oracle-call budget. SpectralMol generates more docking hits and more diverse scaffolds while maintaining competitive physicochemical properties. The representation adopted in this work can cleanly separates scaffold-level modifications from localized substructure variations, as the former occur with perturbations of low-frequency Fourier modes and the latter with perturbations of high-frequency Fourier modes. The results support the evidence that frequency-controlled evolutionary dynamics provide an interpretable, efficient, and training-free route to multi-objective molecular design.

The global minimum-variance portfolio (GMVP) is the canonical decision built from an estimated covariance matrix, yet covariance estimators are universally evaluated by matrix-norm loss, which is not the object the decision depends on. We characterise exactly how covariance-estimation error maps into GMVP suboptimality. We prove an exact regret identity and a non-asymptotic bound showing decision regret depends on the estimation error only through its action on the portfolio weights, scaled by portfolio concentration and the conditioning of the true covariance. From this we derive the decision geometry: GMVP regret is invariant to a (p-1)-dimensional projection of the p^2-dimensional error matrix, with invariance to the covariance-scale direction as an exact special case. We then apply the framework to heavy-tailed returns (tail index kappa in (2,4)), establishing the regret convergence rate implied by the centred operator-norm rate, and confirm the theory on a skew-t/t-copula simulation design with pre-registered analysis. The decision-focused advantage is a sharper constant and a concentration discount rather than a faster rate; we report an honest high-conditioning boundary of the rate prediction. The results complement recent decision-focused learning approaches by supplying the exact estimation geometry and consistency theory they lack.

In this study, we use a developmental approach to investigate the statistical learning and mental representation of neural language models (NLM). A series of Generative Transformer models are trained on a synthetic grammar. The model states are saved at multiple stages in the course of training. Through analyzing how the internal representations of these models change in the developmental path, we found that NLMs acquire the most abstract global statistical knowledge at the beginning of learning and later acquire the relatively local statistical dependencies. This learning path contains many over-generalizations from the very beginning and these over-generalizations are gradually constrained in the later stage of learning. Based on this observation, we propose a new framework to explain the statistical learning and language cognition of NLMs.

We consider the cubic nonlinear Schrödinger (NLS) equation on two-dimensional flat tori with varying aspect ratios. In this formulation, the choice of aspect ratio governs the Fourier resonance structure, so rational and irrational geometries can exhibit different high-frequency cascade behaviors. We present a geometry-conditioned Fourier neural operator (FNO) for the cubic defocusing NLS equation, where the input consists of the real and imaginary parts of the solution together with the aspect-ratio parameter \(ω^2\). The model is trained to approximate the one-step solution operator and is evaluated on unseen trajectories generated from random-phase initial data using Fourier pseudospectral method. Our numerical experiments show that the learned operator captures the main solution dynamics on both tori and reproduces the distinct Sobolev norm behavior of the two geometries, with stronger \(H^2\)-growth on the rational torus and more constrained behavior on the irrational torus, consistent with the findings of \cite{hrabski2021energy}. We perform ablation studies to examine the roles of retained Fourier modes, activation functions, Fourier-layer depth, and explicit geometry conditioning. The results indicate that including $ω^2$ improves long-time predictive accuracy, especially for the rational geometry, and supports the use of geometry-aware neural operators for learning spectral-transfer phenomena in nonlinear dispersive partial differential equations.

Efficient deployment of large language models (LLMs) in production forces a trade-off between accuracy and cost. Operators often default to a single model that is either expensive for easy queries or insufficient for hard ones. To address this challenge, we propose a two-stage cascaded solution. Stage 1 clusters incoming queries and assigns each cluster to its most cost-effective model. The cost budget for this routing process is set by an interpretable hyperparameter, tuned offline. Stage 2 adds a quality estimation (QE) cascade; when an output from Stage 1 is judged low-quality, the query is escalated to a stronger model. This ensures only hard or low-confidence cases reach the expensive models. On the test datasets, the cascaded system retains 97-99% of the strongest model's accuracy while reducing Time Per Output Token (TPOT). It requires only task-correctness labels and adapts to changes in the model pool without manual reconfiguration.

We address the problem of inferring a directed network from nodal measurements generated by linear diffusion dynamics on the sought graph. Observations are modeled as the outputs of a graph convolutional filter, i.e., a polynomial (with unknown coefficients) of a local diffusion graph-shift operator encoding the latent graph topology, excited with an ensemble of independent graph signals with arbitrarily-correlated nodal components. Unlike prior efforts that considered undirected graphs and white signal excitations, here the graph-shift operator and the observations' covariance matrix are not simultaneously diagonalizable. In this challenging context, we first rely on measurements of the output signals along with prior statistical information on the inputs to identify the diffusion filter. Such system identification problem involves solving a system of quadratic matrix equations, which we show is identifiable under spectral-diversity assumptions on the input covariances. For algorithmic purposes we recast it as a smooth quadratic minimization subject to Stiefel manifold constraints. Subsequent identification of the network topology given the graph filter estimate boils down to finding a sparse and structurally admissible shift that commutes with the given filter, thus, forcing the latter to be a polynomial in the sought graph-shift operator. A joint graph filter and topology identification algorithm is also proposed, which alternates between the aforementioned steps in a mutually reinforcing fashion to offer improved sample complexity. Numerical tests corroborate the effectiveness of the proposed algorithms in recovering synthetic digraphs and real-data case studies, and illustrate their potential utility on urban mobility analyses as well as portfolio optimization.

Multi-head attention conventionally partitions the hidden dimension equally across all heads at every layer, enforcing an identical representational subspace dimension (dh = dmodel/h) throughout the models depth. In this work, we identify this uniform allocation as a fundamental structural bottleneck: due to their restricted dimensional space, early-layer heads are unable to faithfully capture complex, high-dimensional contextual patterns. To resolve this, we introduce the Prism Transformer, a novel architectural paradigm that replaces the static, uniform head configuration with a progressive head schedule. By monotonically increasing the head count across layers, the Prism Transformer naturally establishes a local-to-global representational hierarchy: early layers leverage fewer, exceptionally wide heads to capture complex, local compositional patterns, while deep layers deploy many, narrow heads to decompose these patterns into specialized linguistic features. Crucially, this structural shift is parameter-neutral, compute-neutral, and introduces zero training or inference overhead, preserving identical weight matrices and FLOP budgets as the standard Transformer. Across three model scales (124M, 354M, and 757M), the Prism Transformer consistently outperforms uniform baselines, achieving consistent reductions in validation loss alongside consistent gains on downstream zero-shot benchmarks (including PIQA, HellaSwag, ARC-Easy, and WinoGrande). Our findings demonstrate that non-uniform subspace allocation unlocks latent capacity within the standard Transformer budget, enabling more effective use of model capacity.

This paper studies the problem of regret minimization in Markovian bandits with \emph{non-observable states} and possibly \emph{constrained} decision epochs. The focus is restricted to a ``pure'' regret benchmark, that compares the performance of the learning algorithm to the best \emph{pure policy} which -- akin to optimal policies of stochastic bandits -- picks the optimal arm from start to finish without ever switching. We introduce a generalization of rested Markovian bandits, \emph{self-degrading Markovian bandits}, for which pure policies are always asymptotically optimal.We show that without prior knowledge on the underlying bandit, the regret of algorithms that switch arms rarely necessarily scales super-logarithmically for every bandit, i.e., as $ω(\log(T))$, where $T$ is the learning horizon. Despite the unreachability of the logarithmic regime, we design UCB-NOM, an optimistic algorithm inspired by UCB, of which the regret is nearly logarithmic. Lastly, we show that given prior knowledge on the Markovian bandit in the form of a bound on the bias functions of its arm, a proper instantiation of UCB-NOM achieves $O(\log(T))$ regret. We further show that this prior knowledge allows for a $O(\sqrt{T \log(T)})$ worst-case regret bound for UCB-NOM. Notably, our regret bounds do not depend on the number of states of the underlying Markov chains. Our findings suggest that the non-observability of states is a mild inconvenience in self-degrading Markovian bandits.

This paper describes team HSA_CORAL's submission to the FinCausal 2026 shared task on extracting cause-effect relations from financial narratives via extractive question answering in English and Spanish. We compare three modeling families: (i) encoder-only token tagging with multilingual BERT, (ii) encoder-decoder generation with multilingual BART, and (iii) decoder-only LLMs (Llama 3.1 and GPT variants) using prompt refinement, few-shot demonstrations, and supervised fine-tuning. Across settings, prompting and few-shot examples yield competitive performance, while supervised fine-tuning provides the largest gains. Our best system, GPT-4.1 Mini fine-tuned on combined English and Spanish training data, achieves a tied highest score on the English subtask (score 4.8140) and ranks third on Spanish (score 4.7753) under the shared task's LLM-as-a-judge metric. Overall, the results highlight the value of task-specific adaptation and multilingual fine-tuning for cross-lingual transfer in financial causality QA.

Personality prompting shapes how large language models communicate, yet whether these behavioral shifts affect objective task outcomes remains under-explored. Prior work shows that agents prompted with low agreeableness produce adversarial language, while those prompted with high agreeableness become cooperative, but the relationship between communication style and task performance has not been systematically examined across multiple domains. In this work, we investigate whether personality composition matters for multi-agent team performance by manipulating personality traits across frontier LLMs on three task domains: structured coding, open-ended research collaboration, and competitive bargaining. We find that personality effects depend critically on task structure. In coding tasks, low agreeableness leads to large communication shifts that have little effect on milestone completion. In open-ended collaboration and bargaining, the same manipulation substantially degrades performance. We discuss implications for multi-agent system design and the limits of personality manipulation.

Foundation models for structural biology have achieved remarkable performance in predicting biomolecular structure and show promise for the design of proteins and small molecules. Yet understanding which internal features drive their outputs remains challenging. Standard sparse autoencoders (SAEs), effective on transformer-style sequence embeddings, do not transfer cleanly to pairformer-like architectures: naively operating on pairwise representations yields a quadratic blow-up of features and obscures concepts distributed jointly across sequence and pair representations. We introduce PairSAE, which summarizes pairwise tensors via an N-mode SVD into token-wise interaction roles, then uses a sparse autoencoder to learn a shared set of token-level features that decode into both sequence and pair representations. Evaluated on Boltz-2 activations for PLINDER protein-ligand complexes, PairSAE yields interpretable features that align with UniProt annotations and predict Boltz-2 affinity values. These results indicate that PairSAE links the latent space of foundation models for structural biology to interpretable structural concepts, clarifying what the model "knows" while avoiding pairformer-induced pitfalls that limit conventional SAEs.

Since its initial release in 2020, Darts has become a widely used open-source Python library for time series analysis. A series of foundation models have recently claimed accuracy improvements in zero-shot forecasting, promising a paradigm shift from training custom models to harnessing pre-trained general-purpose forecasters. Foundation models, however, are often released as isolated packages with fragmented interfaces and limited interoperability with common tooling, making joint evaluation and integration within complete pipelines difficult. In Darts, we developed a unified $\texttt{FoundationModel}$ class collection (Chronos-2, TimesFM 2.5, TiRex, PatchTST-FM) that provides standardized, full-cycle forecasting interfaces with minimal external dependencies for integrating foundation models into the ecosystem. Existing Darts pipelines can now use foundation models with only a name change; new pipelines can use them for zero-shot or fine-tuned forecasting, uncertainty estimation, and backtesting, combined with data processing and evaluation tooling, all within a unified framework.

Modern image generation demands a single model that unifies diverse capabilities, including text-to-image (T2I), local editing, and global editing. However, these capabilities are rarely naturally aligned and often conflict. For instance, editing tends to degrade T2I performance, while global and local editing interfere with each other. Consequently, effectively composing these capabilities has become a central challenge for image generation model training. To tackle this, we introduce DanceOPD, an on-policy generative field distillation framework for flow-matching models that routes each sample to one capability field, queries one low-noise student-induced state, and trains with a simple velocity MSE objective. With each capability source defined as a velocity field over the shared flow state space, the student learns from fields queried on its own rollout states to compose expert capabilities. This formulation also absorbs operator-defined fields such as classifier-free guidance. Comprehensive experiments on T2I, editing, realism-field absorption, and CFG absorption show that our approach improves multi-capability composition, strengthening target capabilities while preserving anchor generation quality. We believe this work establishes a practical route for generative field distillation in flow-matching models.

Reinforcement learning with verifiable rewards (RLVR) for training LLMs typically rely on ground-truth answers to assign rewards, limiting their applicability to tasks where the ground-truth solution is unknown. We introduce a \textbf{R}anking-\textbf{i}nduced \textbf{VER}ifiable framework (RiVER) that trains LLMs on score-based optimization tasks without ground-truth solutions, using deterministic execution feedback as continuous-valued supervision. When applying group-relative RL to such continuous rewards, we identify two key challenges: \emph{scale dominance}, where uncalibrated score magnitudes across test instances distort policy updates, and \emph{frequency dominance}, where repeatedly sampled suboptimal solutions can outweigh rare but stronger candidates. RiVER addresses these challenges with calibrated reward shaping that uses instance-wise comparisons and emphasizes top-ranked solvers while retaining bounded feedback for other valid solutions. We train on 12 AtCoder Heuristic Contest tasks and evaluate on Algorithm Engineering Benchmark (ALE-Bench), LiveCodeBench, and USACO. RiVER advances Qwen3-8B and GLM-Z1-9B-0414 by 8.9\% and 9.4\% in ALE rating rank. More importantly, despite training exclusively on score-based tasks without any ground-truth solutions, RiVER also improves the backbones across exact-solution benchmarks such as LiveCodeBench and USACO by an absolute average improvement of 2.4\% and 3.5\%. By contrast, baselines trained with raw execution scores improve ALE rating but fail to transfer to exact-solution benchmarks. These results suggest that score-based optimization tasks, combined with proper reward calibration, can serve as effective training environments for general coding ability without ground-truth solutions.

Efficient sampling of molecular systems at thermodynamic equilibrium is a hallmark challenge in statistical physics. This challenge has driven the development of Boltzmann Generators (BGs), which allow rapid generation of uncorrelated equilibrium samples by combining a generative model with exact likelihoods and an importance sampling correction. However, modern BGs predominantly rely on normalizing flows (NFs), which either suffer from limited expressivity due to strict invertibility constraints (discrete time) or computationally expensive likelihoods (continuous time). In this paper, we propose Autoregressive Boltzmann Generators (ArBG) -- a novel autoregressive modelling framework -- that overcomes these limitations by departing from the flow-based BG paradigm. ArBG circumvents the topological constraints of flows and enables sequential inference-time interventions, while offering enhanced scalability by leveraging architectures effective in Large Language Models. We empirically demonstrate that ArBG leads to significant improvements over flow-based models across all benchmarks, but particularly in larger peptide systems such as the 10-residue Chignolin. Furthermore, we introduce Robin, a 132 million parameter transferable model trained with the ArBG framework which improves over the previous state-of-the-art, reducing the zero-shot energy error, E-W$_2$, on 8-residue systems by over 60$\%$. The code can be found at the following link: https://github.com/danyalrehman/autobg.

Many decoding methods for large language models can be understood as shifting probability mass toward outputs that are more likely under the model, either locally at the token level or globally at the sequence level. Therefore, their success depends on a fundamental question: when does sequence probability, that is, the conditional probability of a continuation given a prompt, actually align with correctness? In this paper, we set out to quantify this relationship across decoding methods, models, and benchmarks at four levels: across decoding methods, across hyperparameters within a method, across prompt-answer pairs within a dataset, and across repeated responses to the same prompt. We find that higher sequence probability is often predictive of correctness across prompt-answer pairs within a fixed dataset. However, this relationship does not generally transfer to decoding decisions: increasing sequence probability by changing hyperparameters or methods does not reliably improve accuracy. Further, sequence probability is not a good indicator of correctness for responses to the same prompt. These findings clarify when decoding can and cannot be expected to improve correctness, and provide practical guidance for decoding, self-consistency, and verifier-free self-improvement.

Neural surrogate models offer fast approximate mappings from PDE parameters to solutions, but they typically treat solving as a purely statistical task: once trained, they struggle to correct their own constraint violations and extrapolate beyond the training distribution. Recent hybrid methods promote physical correctness by targeting the PDE residual via gradient descent or Gauss--Newton steps, but inherit the compute cost and instability of the underlying classical optimizers. We show, theoretically and empirically, that numerically minimizing the PDE residual can be an unreliable proxy for reconstruction accuracy in ill-conditioned systems, explaining why these methods often do not make accurate predictions despite achieving low residuals. We propose error-conditioned Neural Solvers (ENS), built on a different principle: rather than an optimization target, the PDE residual field is passed as a direct input to the network at each iteration, enabling it to read the spatial structure of its own errors and learn an update policy to iteratively correct its predictions. Across four PDE families, ENS attains the highest prediction accuracy in the large majority of settings, with gains reaching $10\times$ on turbulent Kolmogorov flow, while avoiding the expensive compute cost of hybrid methods. ENS's learned correction policy generalizes under distribution shift, including zero-shot parameter changes and cross-equation transfer, where its relative advantage is largest in the ill-conditioned regimes where residual minimization is least reliable. Project website: https://neuralsolver.github.io/.

Agentic artificial intelligence shows great promise for radically improving the pace of innovation in hardware/software co-design research across computer architecture, systems, compilers, and VLSI. Thus far, however, applications of AI in these contexts have generally been demonstrated in isolated settings on small-scale problems, due to the difficulty of designing and deploying complex AI-infused hardware and software development workflows. This paper introduces CHIA, an open-source hardware/software co-design framework for agile and principled research on the application of AI to co-design. CHIA treats the productive construction and scalable deployment of the co-design flow itself as a first-class objective. In CHIA, agentic AI-driven hardware and software design flows are expressed as \textit{CHIA loops}: directed cyclic graphs whose nodes execute various system-on-chip design tools, microarchitectural simulators, software build systems, AI models, evolutionary coding agents, and more. The \textit{CHIA library} provides node implementations for many popular tools, including Chipyard, gem5, ChampSim, FireSim, Hammer (thus several commercial ASIC CAD tools), Vivado, AlphaEvolve, AdaEvolve, and many others. CHIA also provides a broad set of features to conduct principled science around these flows. These include isolation between AI models and hardware tools, profiling mechanisms, fault-tolerant execution, and reliability at scale across hundreds of heterogeneous systems (CPUs, FPGAs, GPUs, etc., across public cloud/on-prem.). To showcase CHIA, we present five CHIA loops as case studies: (1) automatic RTL-to-gem5 simulator alignment, (2) LLM-driven implementation of microarchitectural features in RTL, (3) agentic, IPC-aware critical path optimization, (4) evolutionary architectural discovery, and (5) maintainer-friendly agentic GitHub issue fixing.

LLM coding agents make it tempting to automate empirical research by delegating experiments to them directly, but naive delegation does not scale to large projects: low-rate instruction lapses compound into broken, irreproducible artefacts. To address this problem, we present Glite ARF, an open-source Python framework for running many LLM coding agents in parallel on a research repository without sacrificing reproducibility or auditability. The framework defines a three-role stack: a human researcher chooses which hypotheses to test, coding agents (Claude Code, Codex CLI) implement individual tasks under a fixed structure, and deterministic Python verifier scripts enforce task isolation, immutability of completed work, a corrections overlay, and a materialised project overview. We call this verifier-driven research: the rules of the research process live in code that fails loudly when violated, not in prose that agents are merely asked to follow. Using Glite ARF, we developed our submission to the BEA 2026 vocabulary-difficulty shared task, placing first in the closed track and second in the open track on all three target languages (Spanish, German, Mandarin) and reducing the official baseline RMSE by 29.9% (closed) and 35.9% (open). The campaign comprised 273 tracked tasks (146 experiment runs) across 129 feature sets, run by up to twelve parallel agents orchestrated from a single laptop - with some model training on rented A100s - at approximately \$450 in LLM API spend (\$498 total third-party cost), and structured per-fold provenance let us catch and strip four target-leaking feature sets, correcting an implausible 0.609 RMSE to 0.802. Across three campaigns in three domains, the framework's structural machinery adds only about 1% of wall-clock time. Framework and a public demo project accompany this paper.

Whether political elites organise into rent-seeking coalitions that capture public resources or civic networks that sustain governance is a central question in comparative politics. Yet observing these complex, informal, and adversarial ties at scale has historically required intensive manual coding, while automated text-as-data methods have largely been limited to simple co-occurrence. Recent large language model (LLM) approaches offer a path forward but often rely on proprietary APIs, lack cross-lingual capability, and struggle with scalable entity resolution. We present a modular, fully open-weight pipeline for multilingual joint entity-relation extraction that builds signed, temporal knowledge graphs from massive unstructured news corpora. It combines span-based named-entity recognition (NER) with a three-stage linking cascade mapping mentions to language-independent Wikidata identifiers; a high-throughput, ontology-constrained mixture-of-experts model then uses guided decoding to extract directed, signed relationships grounded in a domain ontology. A full-coverage spot-check against a 3491-relation gold standard shows high textual correctness (68.2% strict to 93.7% lenient). Two large-scale case studies validate the pipeline against the public record. In Austria, it reconstructs a political party's complete lifecycle, dating internal fractures and tracking personnel into successor factions and court convictions. In a Polish corpus, it uncovers the overlapping economic and governance networks of state-enterprise patronage, alongside the structurally balanced, signed conflict network of the polarized Civic Platform (Platforma Obywatelska, PO)--Law and Justice (Prawo i Sprawiedliwość, PiS) duopoly. By bridging raw multilingual text and structured relational data, our framework provides a robust, replicable foundation for cross-national empirical computational social science.

Entity Matching (EM) is a core operation in the data integration pipeline, where records from different sources are compared to determine whether they refer to the same real-world entity. Recent work has incorporated domain information and low-resource learning techniques to better adapt EM systems to realistic settings. While these approaches have demonstrated strong performance, it remains unclear how they behave under varying data constraints and levels of supervision in practice. In this paper, we investigate a state-of-the-art method for low-resource, domain-aware EM--BEACON--and study how its performance is affected by different algorithmic choices and data availability conditions. We conduct a series of targeted experiments to evaluate these variations, providing deeper insight into the role of distribution alignment and the behavior of the BEACON framework.

Digital twins have emerged as a promising paradigm for personalized healthcare, enabling modeling of individual behavior and health trajectories. In cognitive health, early detection of Mild Cognitive Impairment (MCI) remains challenging, where language and conversational patterns serve as non-invasive biomarkers. In this work, we propose a language-based digital twin framework that leverages large language models (LLMs) to mimic the conversational behavior of elderly individuals by incorporating stylometric cues and contextual metadata. To evaluate fidelity and cognitive consistency, we introduce a multi-head conditional variational autoencoder (cVAE) that jointly measures reconstruction quality and predicts cognitive scores. Experiments on the I-CONECT dataset show that the digital twin preserves identity-specific characteristics and achieves reconstruction and MoCA prediction errors comparable to real data, while outperforming baseline GPT-generated responses. These results highlight the potential of language-based digital twins as a scalable and non-invasive approach for personalized and continuous cognitive health monitoring.

Multimodal web agents can assist humans in operating repetitive GUI tasks, where effective task planning is essential for decomposing complex tasks into executable actions. While small open source MLLMs are cost efficient and privacy preserving compared with commercial large models, they suffer from weak planning and limited cross website generalization. To address these limitations, we introduce the planning experience exploration and utilization (PEEU) method, which autonomously explores environments to discover experiences and utilizes hindsight experience to synthesize strictly aligned, high level training data. To quantitatively analyze the generalization behaviors driving this performance, we propose the task decomposition hierarchical analysis framework (TDHAF) to systematically study compositional generalization across three task granularities: low, middle and high levels. Our analysis reveals that mastering low level atomic skills does not guarantee high level planning competence, while high level task training yields stronger OOD generalization. Experiments on real world benchmarks demonstrate PEEU's superior effectiveness: our 7B model achieves 30.6% accuracy, outperforming the much larger Qwen2.5-VL-32B model. These demonstrate constructing hindsight high level tasks and leveraging experiences is crucial for OOD planning abilities of small MLLMs.

Modern generative world models render increasingly realistic action-controllable futures, yet they frequently hallucinate: rollouts remain visually fluent while drifting from the ground-truth dynamics. We hypothesize that hallucination concentrates in low-coverage regions of the state-action space, where lightweight data-centric signals can both detect it and guide mitigation. To test this, we introduce MMBench2, a 427-hour, 210-task dataset for visual world modeling with ground-truth actions, rewards, and live simulators, and train a 350M-parameter world model on it. We identify three distinct hallucination modes: perceptual, action-marginalized, and scene-diverging -- each anchored to a different stage of the pipeline, and develop three signals that accurately predict where the model will fail. To close coverage gaps at training time, we develop a coverage-aware sampling technique; to close them online, our hallucination predictors serve as curiosity rewards for targeted data collection, yielding a data-efficient finetuning recipe that adapts the pretrained world model to entirely unseen environments with as few as 50 real environment trajectories. Overall, our findings reveal that hallucination in world models is inherently a data coverage issue, and that the same signals used to detect it can also be used for mitigation. An interactive web version of our paper is available at https://www.nicklashansen.com/mmbench2

Sparse autoencoders (SAEs) have become a leading tool for interpreting the representations of vision foundation models, decomposing their polysemantic activations into a larger set of sparse, more monosemantic features. The Top-$k$ SAE, a now-standard variant, enforces sparsity architecturally through its activation function, retaining only the $k$ most active latents per input. Because it was designed precisely to avoid the $\ell_1$ penalty used by earlier SAEs and its known drawbacks, it has not been combined with an explicit sparsity regularizer, despite retaining limitations of its own, such as a budget $k$ that is fixed regardless of input complexity and a tendency to overfit to the training value of $k$. We introduce two sparsity regularizers compatible with the Top-$k$ architecture, both acting on the activations before the Top-$k$ selection: an $\ell_1$ penalty on the unselected (off-support) units, and a scale-invariant $\ell_1/\ell_2$-ratio penalty that concentrates the code onto fewer effective units. Both penalties are applied only to the batch-active units, those selected by the Top-$k$ operator at least once within the batch. Across two datasets, three vision foundation models, and a range of $k$, both regularizers consistently improve monosemanticity at no cost to reconstruction quality. The $\ell_1/\ell_2$ penalty further concentrates information into fewer latents, making reconstruction more robust to the inference-time choice of $k$ and improving small-budget linear probing. Our central finding is that hard architectural sparsity and soft sparsity regularization are complementary rather than mutually exclusive.

Neural audio autoencoders have become a core component of compression, feature extraction, and generation. However, while existing systems support variable bitrate, the vast majority of models still operate at a fixed latent frame-rate, allocating equal temporal budget to regions with very different information density, which can result in unnecessarily long sequences. We introduce Elastic Time, a dynamic frame-rate bottleneck that converts fixed-frame-rate autoencoders to dynamic ones. Our method learns a lightweight latent predictor used to decide which frames can be skipped and later reconstructed, enabling efficient greedy boundary selection at inference. Experiments show our method enables deployment-time rate control while improving efficiency-quality tradeoffs relative to baselines. Overall, we provide a flexible mechanism for adjusting temporal resolution in audio autoencoders, potentially facilitating more efficient downstream modeling for generation and long-context tasks.

The push towards expanded ultra-low-power edge computing necessitates hardware capable of operating under extremely strict energy constraints. Traditional Complementary Metal-Oxide-Semiconductor (CMOS) microcontrollers are fundamentally limited in this domain by the von Neumann bottleneck and by the static power leakage inherent to volatile memory. Memristive In-Memory Computing (IMC) offers a promising solution to these inefficiencies by unifying data storage and computation into a single non-volatile component. However, the State of the Art (SoA) predominantly focuses on accelerators designed to be a co-processor for data-intensive computation. This leaves the prospect of standalone, general-purpose IMC microcontroller architectures underexplored. This thesis proposes such an architecture tailored for ultra-low-power edge devices, alongside an instruction set closely derived from the RV32I standard. Using the IMPLY stateful logic paradigm, a complete implementation of the proposed instruction set is provided, and the novel addressing schema required to support computation in the memristive crossbar array is described as well. Then, the energy use and other circuit-level metrics of the proposed architecture are evaluated through simulation and compared against those of traditional microcontrollers. Finally, the functional viability of the design is demonstrated through an application case study, describing how the proposed design could be used in an intelligent environmental sensor node.

Verifying the eligibility of securities as collateral is a key responsibility of the German Central Bank. However, manually verifying these assets against legal and financial criteria within lengthy, semi-structured, and often bilingual prospectuses is a resource-intensive task. While previous efforts utilized traditional Named Entity Recognition (NER) for information extraction, these methods can struggle with OCR noise, linguistic variance, and rigid span-based constraints, and the need for manually annotated training data for each relevant annotation type. In this paper, we present the first case study applying Large Language Models (LLMs) to the eligibility examination process, shifting the paradigm toward a generative Information Extraction pipeline. Our approach decomposes the task into extraction, normalization, and interpretation, allowing for greater flexibility in handling noisy text and interleaved German-English content. We further introduce a value-based evaluation methodology using LLM-as-a-judge, which offers a more semantic assessment than location-based metrics. Our results demonstrate that LLM-based systems achieve high precision (up to 91%) in document-level eligibility, exhibiting a conservative operating profile that minimizes false acceptance.

Gradient equilibrium (GEQ) is a recently introduced online optimization framework that generalizes first-order stationarity from offline optimization and abstracts problems like online conformal prediction. While GEQ has curious similarities with known online learning frameworks, namely regret minimization, prior work has shown that GEQ error and regret are incomparable objectives, leaving open a precise understanding of how GEQ fits into the broader online learning landscape. In this work, we show that GEQ is equivalent to Blackwell approachability in the algorithmic sense. That is, a Blackwell approachability problem can always be solved using queries to a black-box GEQ oracle, with no asymptotic loss in the oracle's error rate, and vice versa. Taken together with known equivalences between approachability, regret minimization, and calibration, these results imply that GEQ is equivalent to these frameworks, as well. Our reductions are efficient and can be used to transfer refined guarantees, such as optimism and strong adaptivity, from regret minimization to GEQ. Along the way, we also identify necessary and sufficient conditions for GEQ, and establish reductions between different notions of GEQ with unconstrained and constrained decision sets.

To avoid moderation and surveillance on social media, some users routinely invent indirect linguistic expressions (ILE) that camouflage sensitive meanings. Such expressions surface as algospeak, euphemisms, and adversarial obfuscation, depending on intent and context, and they involve recurring encoding mechanisms. We propose a comprehensive, mechanism-oriented taxonomy of ILE that abstracts away from communicative goals and instead categorizes the underlying operations through which meaning is encoded and recovered. We evaluate the taxonomy by incorporating it into LLM prompts and comparing it with four existing taxonomies and a no-taxonomy baseline, using 2,000 manually annotated TikTok and Bluesky posts. The proposed taxonomy attains the strongest document- and span-level performance across the three LLMs, achieving an improvement of 4.7% in accuracy and 5.4% in F1 over the best-performing benchmark. The empirical results reveal the importance of a comprehensive, mechanism-oriented taxonomy as a stable scaffold for detecting emerging coded language and a useful input to content moderation. Disclaimer: This paper contains content that may be profane, vulgar, or offensive.

Translation cascades for reasoning translate the query from another language to English, reason in English, and translate the answer back to the original language. This is a competitive approach to multilingual reasoning, but structurally lossy, since each stage discards information later stages may need, including cues for cultural grounding, register, and disambiguation. We examine the benefits of a simple and training-free intervention: a context-aware translation cascade, which additionally provides the original question, the English translated question, and the reasoning trace to the context of the final translation module. We evaluate gains across nine multilingual benchmarks including various task types, three backbone models, and 285 high-, mid-, and low-resource languages, and demonstrate strong gains for open-ended generation across models and resource regimes. We show that the original language question carries most of the beneficial context. Our study emphasizes the need to better design information flow in machine translation cascades for mitigating error propagation, and provides a simple and actionable default strategy: preserve the original user question until the end of the pipeline.

Guided wave-based structural health monitoring (GWSHM) with onboard transducers offers significant potential for the early diagnosis of damage in engineering structures. However, the practical deployment of deep learning models is often hindered by the limited availability of labelled experimental data and the high computational cost of generating large-scale high-fidelity simulation datasets. This study presents a multifidelity transfer learning framework that integrates lightweight physics-based simulations, convolutional autoencoder (CAE)-based deep feature learning, a feed-forward neural network, and limited experimental measurements for accurate damage localisation and sizing in plate-like structures instrumented with piezoelectric transducers. A computationally efficient one-dimensional time-domain spectral element model is employed to generate a large synthetic dataset for pretraining, while transfer learning adapts the model to experimental domains using only a small amount of labelled data. The CAE-based transfer learning framework significantly outperforms its CNN-based counterpart in damage localisation accuracy. The model achieves excellent predictive performance with $R^2$ scores exceeding 0.93 for damage localisation and 0.99 for damage sizing. Its generalisation capability is demonstrated on previously unseen data, showing high prediction accuracy for damage scenarios not represented during pretraining or fine-tuning. The results establish the proposed framework as an accurate, computationally efficient, and practically viable solution for real-world GWSHM applications.

AI healthcare chatbots are increasingly used to support health information seeking and self-management, yet their performance and impact on users remains to be studied. This study examines over 15,000 user reviews from 59 AI healthcare chatbot apps to explore how these systems function in everyday informational and emotional contexts. Topic modeling and interpretive analysis identify three recurring breakdowns: access barriers and service unreliability, user experience and interaction quality, and billing and customer support issues. Privacy and security concerns are associated with the most negative experiences. By framing AI healthcare chatbots as information infrastructures, our findings highlight how failures in access, usability, and trust affect users, offering actionable insights for designers, policymakers, and information professionals aiming to improve digital health systems.

We study the fundamental problem of learning a high-dimensional Gaussian truncated to an unknown halfspace. Lee, Mehrotra and Zampetakis (FOCS'24) recently obtained the first polynomial time algorithm for this problem, but their resulting sample and time complexity bounds are not optimal. Under non-trivial truncation, for any target accuracy $\varepsilon > 0$ and dimension $d$ we give an efficient algorithm that uses $n = \tilde{O}(d^2/\varepsilon^2)$ samples and learns the underlying Gaussian to error $\varepsilon$ in total variation distance. Our algorithm is also fast: its runtime is dominated by the cost of computing the empirical covariance matrix. Both our sample and time complexity are optimal in terms of $d$ and $\varepsilon$ even without truncation: in this regard, we can learn a Gaussian under halfspace truncation for free. The key ingredient behind our result is a novel reinterpretation of the low-degree moments of the truncated Gaussian in terms of a relative truncation parameter. This relative truncation parameter uniquely determines the parameters of the untruncated Gaussian and enables direct parameter recovery. This reinterpretation allows us to circumvent the time intensive projected stochastic gradient descent procedure that is widely used in learning under truncation.

Analog hardware platforms such as coupled oscillators and Analog Ising Machines naturally solve differential equations at a fraction of the energy cost of digital computation, making them attractive for low-power generative modeling, yet a fundamental mismatch exists: modern generative models assume flexible, software-defined dynamics, whereas analog hardware imposes fixed, physics-determined differential equations with limited approximation capacity. This paper introduces Analog Interaction Systems (AIS), a unified framework for hardware-implementable dynamical systems, and empirically characterizes their expressivity gap relative to neural network baselines. Two hardware-compatible mechanisms are proposed to narrow this gap - time-varying piecewise parameters and hidden physical states - and a Wasserstein GAN training procedure is developed to enable training of these models without requiring them to follow a specific trajectory. We characterize how area and power scale with connection density and precision, showing that sparse connectivity and low-bit-width quantized parameters are necessary for practical implementation, and estimate an energy cost of 23uJ per generated image for the chosen architecture, representing a 2-orders-of-magnitude improvement over digital baselines. On MNIST and Fashion-MNIST, our oscillator-based AIS achieves FID scores of 27.6 and 80.8, outperforming the best prior hardware-implementable analog generative models by 3-4x with a 4-bit sparse architecture.

Job-search platforms rely on low-bandwidth query interfaces that often fail to capture the high-dimensional complexity of candidate profiles. We present an end-to-end RLAIF (Reinforcement Learning from AI Feedback) framework to generate \emph{portable} job search queries, terms that abstract away seeker-specific identifiers while preserving generalizable qualifications. This task introduces a highly adversarial reward surface where policy optimization frequently exploits flaws in LLM-as-judge rubrics, resulting in degenerate verbatim-copying behaviors. We conducted comprehensive empirical experiments to isolate the impact of optimization mechanics against structured reward engineering. Our results demonstrate that for critic-free optimizers, performance is overwhelmingly dictated by robust reward shaping, rendering the specific choice of algorithm largely immaterial. While critic-free per-rollout baseline methods (RLOO and REINFORCE++) natively resist reward-hacking, the group-relative advantage normalization in GRPO appears uniquely sensitive to spurious reward signals, making it disproportionately susceptible to exploitation. We show that introducing a deterministic, rule-based reward floor to correct for rewards assigned to verbatim copying mitigates this failure mode, resulting in a substantial $+0.147$ quality improvement on a cross-family evaluation judge. Ultimately, we show that the training-time reward model inflates performance gains by $2.4\times$, confirming that the training success is fundamentally dependent on enforcing reward-shaping disciplines rather than selecting alternative optimizers.

Multi-model LLM systems such as routing, voting, cascades, fusion, and mixture-of-agents are used to beat single-model accuracy. We show that their gain is capped by a quantity the field rarely reports. For any policy whose output is one member model answer, accuracy cannot exceed one minus beta, where beta is the rate at which every model is wrong on the same query. In contrast, the usual diagnostic, average pairwise error correlation rho, cannot identify beta: error laws with identical marginals and pairwise correlations can have different all-wrong rates. A Clopper-Pearson bound on beta gives a finite-sample certificate on the largest gain any router, vote, or cascade could deliver before training a router. Across 67 models from 21 providers, a tetrachoric-calibrated single-factor model still underprices the all-wrong tail: on open-ended mathematics, observed beta is 0.052 versus 0.023 under the full 67-model Gaussian copula, about 2.5 times underpricing, with 90 percent CI 1.7 to 3.4 and k equals 17. The effect recurs on execution-graded code, where beta is 0.079. Re-asking the same GPQA-Diamond questions in free-response rather than multiple-choice form reopens the tail, with beta 0.127 and a five-judge panel with kappa 0.73 to 0.92, locating co-failure in answer format rather than subject. At matched quality, low-rho heterogeneous ensembles beat high-rho Self-MoA, but on checkable tasks in our pool, combining models rarely beats the single best model without a strong query-level routing signal. Gains come from models failing on different questions, not from adding more models.

Large language models (LLMs) are increasingly used to screen and rank job applicants, creating incentives for candidates to strategically manipulate algorithmic hiring systems. We study prompt injection in automated résumé screening, defined as subtle self-promotional text that introduces no new qualifications but is designed to influence LLM evaluations. Using controlled experiments, we show that prompt injection reliably improves applicant rankings when résumé quality is homogeneous and few candidates inject. However, its effectiveness rapidly diminishes as more candidates inject, collapsing when manipulation becomes widespread. When candidate quality is heterogeneous, prompt injection is less effective on average, but can occasionally allow lower-quality candidates to outrank higher-quality ones, raising fairness concerns. Overall, LLM-based screening is most vulnerable when manipulation is rare and candidate quality differences are small. Code and resources are publicly available at: https://github.com/preetb1199/Prompt_Injection_ACL26

Mechanistic epidemiological models are widely used to support infectious disease forecasting and public-health decision making. Bayesian calibration of such models is commonly performed using Markov chain Monte Carlo (MCMC), which can become computationally expensive for high-dimensional nonlinear systems and repeated near-real-time analyses. Here, we investigate simulation-based inference (SBI) using neural posterior estimation as a scalable alternative for Bayesian calibration of a mechanistic SECIR epidemiological model using COVID-19 intensive care unit (ICU) occupancy data from Germany during 2020. We compared SBI and MCMC across multiple epidemic phases using both 31-day inference windows and a substantially more challenging 201-day reconstruction problem involving multiple transmission change points. Posterior agreement was evaluated quantitatively using Wasserstein distances and Kullback-Leibler divergences together with posterior predictive checks. Across the 31-day windows, SBI recovered posterior distributions in strong agreement with MCMC while accurately reproducing observed ICU trajectories. In the 201-day setting, SBI preserved the dominant posterior structure despite increased uncertainty. SBI, by combining CPU and GPU resources, substantially reduced computational runtime compared with MCMC, which was restricted to running on CPUs. Whereas MCMC required approximately 1000 seconds for the 31-day inference problems, SBI achieved comparable posterior and predictive performance in approximately 60-70 seconds on a single GPU. For the 201-day inference problem, SBI required an average of 157 seconds, while the MCMC runs took over 19,000 seconds. Our results demonstrate that SBI provides a rapid and computationally efficient framework for Bayesian calibration of mechanistic epidemiological models, supporting repeated near-real-time inference and rapid outbreak analysis.

Learning governing equations from observed solution data is a fundamental challenge in scientific machine learning, yet the theoretical conditions under which a ground-truth ODE can be uniquely and stably identified from multiple solution observations remain largely undeveloped, and no quantitative analysis of the sample complexity of such learning tasks exists in the literature. To address this gap, we introduce the Hausdorff distance on solution sets as the natural metric for comparing differential equations, since it captures the worst-case separation between two equations over all admissible initial conditions and thus encodes the minimax structure of the identification problem. We establish identifiability bounds for governing ODEs across a wide class of structure equations--ranging from linear ODEs to nonlinear classes with Lipschitz (Hölder)-continuous vector fields--characterizing precisely when two distinct equations can be distinguished from solution data. Using this metric, we derive metric entropy estimates for the relevant ODE classes and analyze sample complexity bounds, quantifying how many solution observations are needed to reliably recover the governing equation.

Time-series forecasting research has been moving steadily toward larger architectures, from specialized transformers to general-purpose foundation models, on the assumption that capacity is what unlocks accuracy. We take the opposite position: most of the gap can be closed at far lower cost by tuning preprocessing rather than scaling models. We use Ridge regression as the testbed, since it has a closed-form solution and interpretable weights, which let the optimal hyperparameters be read off the search directly. We search over context length, local normalization, regularization, and augmentation on eight standard benchmarks and find three patterns. (1) Optimal lookback is strongly series-specific and often non-monotonic in forecast horizon, with fitted power-law exponents ranging from $+0.46$ on ETTm2 to $-0.19$ on Exchange and Traffic, challenging the convention that longer horizons need longer history. (2) Normalizing over a learned trailing fraction of the context, rather than its entirety, is almost universally preferred. (3) Series within the same dataset often disagree on hyperparameters; the optimal degree of cross-series sharing varies from fully shared to fully per-series. The resulting models beat prior linear forecasters on most dataset-horizon entries and exceed Transformer, MLP, and CNN baselines on six of eight benchmarks. The optimized hyperparameters also serve as a diagnostic on the data itself, revealing structures that larger models absorb silently into their learned parameters.

Earth Observation (EO) forecasting aims to predict future Earth surface dynamics from satellite observations under changing meteorological conditions. In this paper, we view this task as a partially observed, weather-driven world modeling problem, in which weather acts as a conditioning signal, while forecasting remains uncertain due to sparse observations and unobserved land-surface states. However, existing methods do not fully capture this setting: deterministic models collapse uncertainty into a single future prediction, while diffusion-based methods typically treat weather variables as undifferentiated conditioning signals, and existing benchmarks focus mainly on reconstruction accuracy rather than whether forecasts respond correctly to changed weather forcing.We introduce EO-WM, a video diffusion transformer for multispectral EO forecasting. EO-WM incorporates a physically informed conditioning framework that represents meteorological forcing through a climatological baseline, weather anomalies, and cumulative physical stress signals. Specifically, it separates baseline and anomaly through distinct conditioning pathways, and accumulates anomalous forcing over time to capture sustained heat and drought stress. To evaluate weather-response behavior beyond standard metrics, we introduce two diagnostic benchmarks: an Extreme Summer Benchmark for severity-aware prediction of vegetation degradation under extreme weather, and a Seasonal Matched-Pair Benchmark for testing response fidelity under changed weather forcing. Experiments show that EO-WM reduces the error in predicted Normalized Difference Vegetation Index (NDVI) decline amplitude by a relative 5.63% and improves directional hit rate by a relative 7.80%, while remaining competitive on standard pixel-level metrics. The benchmarks and model will be made open-source at https://github.com/Luo-Z13/EO-WM.

Large language models (LLMs) are increasingly critical to digital library workflows, yet their ability to process historical language remains poorly understood. Historical difficulty is typically treated as a monolithic barrier, conflating orthographic variation, linguistic distance, and pretraining exposure. In this paper, we propose a diagnostic framework that decomposes this difficulty into four distinct dimensions: tokenization cost, predictive uncertainty (surprisal), semantic robustness, and context sensitivity. We evaluate this framework on three datasets spanning three centuries: (1) a newly curated corpus of 17th-century Italian texts (1610-1689) digitized from original page images; (2) canonical 19th-century Italian "I Promessi Sposi" serving as a high-exposure control; and (3) 18th-century Russian civil print books as a contrastive orthographic stress test. Our results reveal a distinct dissociation between encoding cost and comprehension. While Russian and early modern Italian incur comparable tokenization penalties (25-30% inflation), their predictive difficulty diverges sharply. 17th-century Italian is on average 2.4 times more surprising than its modern equivalent - with academic prose reaching 3.2 times - whereas Russian shows only a modest increase. But predictive uncertainty does not imply representational degradation: embedding similarity remains robust (> 0.85) across all datasets, confirming that models can represent historical meaning even when generation is unstable. Finally, we demonstrate that a minimal temporal context prompt reduces historical surprisal by approximately 60%, offering a simple, model-agnostic mitigation. These findings suggest that while historical text imposes a consistent encoding tax, digital libraries can safely deploy LLMs for semantic retrieval tasks, provided that generative applications are carefully adapted.

The promotion of betting applications on social media platforms has increased significantly in recent years. Many of these advertisements use persuasive techniques that may mislead users, encourage risky behavior, and potentially influence users' mental well-being. However, research on the automated detection of manipulative and deceptive betting advertisements remains limited due to the lack of publicly available annotated datasets. In this work, we introduce a new dataset of betting-related advertisements collected from two widely used social media platforms, Instagram and Reddit. The advertisements were manually annotated for manipulative and deceptive advertising practices. In addition to classification labels, the dataset includes human-provided explanations that describe the reasoning behind each annotation, enabling research into explainable approaches to detecting manipulative advertising. Furthermore, we analyze the strategies commonly used in betting advertisements and examine how these persuasive tactics may impact users' mental health. The proposed framework can also enable practical applications such as browser plugins that warn users about manipulative betting advertisements and automated web crawlers that help regulatory authorities monitor and detect such promotions online.

Reliably quantifying predictive uncertainty is difficult for complex, high-dimensional, or misspecified models. Both fully Bayesian and bootstrap resampling methods provide principled uncertainty estimates but are often too expensive for modern machine-learning models because they require posterior sampling or repeated model refitting. We introduce Ribbon, a scalable approximation to Dirichlet-reweighted bootstrap uncertainty. Ribbon replaces repeated refitting with an influence-function linearization around a single fitted model, preserving the first-order data-reweighting structure of the Bayesian bootstrap while requiring only post-hoc linear algebra. Ribbon approximates the Bayesian-bootstrap or weighted-likelihood-bootstrap refitting target. With a general concentration parameter, Ribbon gives a calibrated Dirichlet-reweighting family whose uncertainty scale can be tuned on validation data. We show that Ribbon is asymptotically equivalent to a flat-prior Laplace approximation under correct likelihood specification and recovers the robust sandwich covariance under misspecification. Across synthetic regression, MNIST classification, and California Housing benchmarks, Ribbon provides competitive predictive performance and improved calibration in several settings while avoiding repeated model retraining.

Recently, a few works have made early attempts to study test-time scaling for embodied tasks. However, two major challenges remain unsolved: (1) reasoning can effectively improve the performance of the policy, but its scaling mechanism has seldom been studied; (2) historical information is essential, as embodied tasks are inherently long-horizon and sequential, making sole reliance on current observations for action scaling inadequate due to the lack of historical context utilization. To address these challenges, we introduce E-TTS, a modular and plug-and-play Embodied Test-Time Scaling framework that unifies reasoning and action scaling for robotic manipulation via history-aware iterative refinement with vision-language verifiers. To support joint reasoning-action scaling, E-TTS performs reasoning-action joint sampling and scoring in a pairwise manner. To better utilize historical information, E-TTS uses a history buffer to store historical context, which is then used by reasoning and action verifiers to evaluate the sampled candidates. Unlike conventional open-loop TTS methods, E-TTS introduces feedback generation into the sampling process to form a closed-loop iterative refinement mechanism, enhancing both inference efficiency and environmental adaptability. Each component functions as an independent and composable module, allowing flexible and adaptive configuration depending on task requirements. To evaluate the advantages of our framework, we conduct experiments across 4 different benchmarks, 6 environments, 3 embodiments, and 4 base vision-language-action models. The experimental results demonstrate that, without requiring additional expert data collection or retraining, E-TTS consistently improves performance, achieving up to a 33.14% increase in simulation and 26.62% in real-world scenarios.

A core principle of object orientation -- that the functionality of a system can be partitioned amongst objects that correspond to individuals in the problem domain -- has influenced how software has been specified, designed and implemented for more than fifty years. Later developments in software engineering sought to build on this principle. But in fact this partitioning is neither natural nor straightforward, and the problems that these later developments sought to mitigate -- the fragmentation and conflation of functionality -- were often, in fact, the inevitable consequences of this founding principle. An easier path to addressing these problems therefore starts by going back, abandoning object orientation, and replacing it with an alternative approach that decouples the individuals of the problem domain from the modules that partition functionality.

This work studies resilient output containment for heterogeneous linear multi-agent systems with actuator cyber-attacks over directed network topologies. The leaders generate bounded locally absolutely continuous trajectories; however, their dynamics, velocity bounds, and motion envelopes are undisclosed to the followers. The cyber-attack model includes state- and input-correlated, as well as bounded exogenous actuator false-data terms. A continuous two-layer adaptive control architecture is proposed. The first layer is a virtual-actuator reconfiguration layer that uses partial state measurements to compensate for actuator attacks in the local tracking-error dynamics. The second layer is a network interface that generates task-space commands via an adaptive interaction protocol. This protocol uses only neighbor-exchanged network-interface states whose dimensions match those of the plant output, and it does not require global graph knowledge for parameter tuning. For directed graphs, under a leader-rooted united spanning-tree condition, a nonsmooth Lyapunov analysis yields asymptotic containment at the command level. The physical outputs then converge to the leader convex hull up to a residual determined by the command-tracking local controllers. Simulation results using a network of quadrotors with damped suspended loads illustrate the performance of attack recovery and containment tracking.

Building persistent embodied agents in unstructured environments demands unified orchestration of heterogeneous tools spanning both cyber (APIs, IoT) and physical (manipulation, navigation) domains, coupled with autonomous recovery from physical failures that inevitably arise over extended operation. Existing systems treat these as separate problems: VLM-based planners lack a unified cyber-physical action space, agent frameworks accumulate unbounded context that degrades temporal coherence, and VLA policies execute open-loop without detecting their own failures. We argue that persistent autonomy requires not a monolithic model but a hierarchical asynchronous architecture with explicit separation of planning, memory, and verification. To this end, we present OmniAct, a framework integrating a multimodal semantic planner for skill routing across unified action spaces, an adaptive hierarchical memory with event-boundary-driven compression for sub-linear context growth, and an asynchronous visual preemption engine that closes the semantic loop during physical execution. Across 40 real-world long-horizon tasks on two robotic platforms coordinating four IoT devices, OmniAct achieves consistent improvements in end-to-end success across all complexity levels, maintains near-flat token consumption over under 100k+ accumulated interaction tokens, and elevates mid-scale open-weight models to proprietary-level performance.

Decentralized Finance (DeFi) applications rely heavily on the order in which transactions are executed, making them susceptible to reordering attacks that enable adversaries to extract Blockchain Extractable Value (BEV). While linear blockchain systems such as Ethereum have inspired extensive research into fair ordering mechanisms, DAG-based consensus protocols have remained largely unprotected despite their growing adoption for scalability and performance. In this paper, we introduce Tilikum, a DAG-based ledger protocol that ensures fair transaction ordering without relying on weak edges. Tilikum achieves ordering linearizability by leveraging median-based timestamp aggregation, or batch order fairness, while maintaining low data redundancy and robust garbage collection. We implemented Tilikum in Rust and evaluated it against representative baselines, namely Narwhal/Tusk, Pompē, Themis and FairDAG. Our results show that Tilikum achieves up to $39\times$ higher throughput than other fair-ordering baselines, while fully blocking state-of-the-art DAG-specific reordering attacks.

In NLP, mental health conditions are often modeled as isolated phenomena, without interpersonal context. We use Reddit posts about long-distance relationships to capture both mental health distress and associated relational triggers. We introduce the Relational Stress and Psychiatry Corpus (RSPC) containing 1,799 Reddit posts annotated by psychiatrists for diagnostic categories, including the most prevalent mood disorders (anxiety and depression), relational stressor triggers, and indications of relationship phase. We benchmark seven fine-tuned transformer models and five large language models across multi-label disorder classification, relational trigger detection, and temporal phase prediction tasks. We find clear task-dependent differences between model families, with Claude-3-Haiku achieving the best disorder classification performance (Macro-F1 = 0.538) and GPT-4o obtaining the strongest relational trigger detection performance (Macro-F1 = 0.519), suggesting distinct model capabilities. We further find strong associations between anxiety disorders and chronic relational uncertainty. Overall, RSPC establishes a benchmark for NLP tasks that consider relational context and supports a shift from individual-centric to context-aware mental health modeling that captures the social and temporal dynamics of distress.

We study a solvable high-dimensional model of generative adversarial network (GAN) training in which a linear generator learns a low-dimensional subspace from data with structured latent covariance. Prior solvable GAN analyses assume unconditional signals with diagonal latent covariance; we extend the multi-feature discriminator setting to class-dependent, correlated, and non-zero-mean latent structure. For the quadratic energy discriminator, all such heterogeneity enters the dynamics through a probability-weighted effective second moment. We prove that the stochastic microscopic training process converges, in the high-dimensional limit, to deterministic ordinary differential equations governed by this effective covariance. In the matched-covariance specialization, the stability analysis yields a mode-wise solvable interval determined by the learning rates and noise level: learning begins when the leading effective eigenvalue crosses the lower threshold, while full recovery requires all relevant effective modes to remain within the interval. This reveals a signal-boosting mechanism: low-rank correlations can lift weak directions above the learnability threshold, whereas overly strong correlations destabilize recovery. Numerical simulations validate the ODE, phase boundary, and boosting mechanism. Experiments on MNIST, FashionMNIST, and CIFAR-10 further show that informed generator covariance improves alignment with the data-driven reference subspace.

Industrial advertising recommender models are continuously improved through architecture evolution. Upgrades such as RankMixer, TokenMixer-Large, and MixFormer show that better structures remain a key source of quality and business gains. Yet developing such upgrades in production is expert-intensive and difficult to scale. Existing automation is insufficient: AutoML mainly tunes hyper-parameters, while effective gains often require cross-module changes under strict constraints; generic LLM coding agents optimize for runnable code, but runnable code does not imply a valid recommender architecture. Candidates may pass local tests while causing silent failures that degrade performance. We present NOVA, a level-aware agent harness for verification-aware architecture evolution. NOVA uses an architecture gradient, an SGD-inspired, non-differentiable update signal that aggregates prior modifications, verification diagnostics, metric feedback, and trajectory memory to guide the next modification. A verification cascade checks structure semantics, local executability, offline effectiveness, and online impact; invalid candidates are blocked early, with failure patterns recorded as forbidden directions. L1--L4 task-level control matches automation to task complexity and risk, routing high-risk tasks to Copilot for human oversight. Deployed in an industrial advertising system, NOVA achieves the highest effective pass rate on L2 ScaleUp and L3 Literature-to-Production tasks (54.5% and 60.0%), reduces silent failures compared with coding-agent baselines, and shortens one literature-to-production cycle by over 13x in human-attended time. In online A/B testing, the selected L3 candidate improves GMV on three pCVR objectives by +1.25%, +1.70%, and +2.02%, while reducing pCVR bias by 58.8%, 66.7%, and 37.3%.

Training-free source selection for LLM families with shared vocabularies arises in scientific string domains such as SMILES, protein, and genomic sequences, where candidate corpora share a tokenizer but differ in prediction targets. This creates an activation-dark regime: representation-similarity metrics can be uninformative without assumptions about label-conditioned error geometry, while classical update-geometry metrics are computationally prohibitive at vocabulary scale. We show that, in a shared-output head setting, representation metrics (e.g., CKA) are non-identifiable for transfer; models can share identical representations yet have orthogonal head updates. The key identity is that head Fisher alignment is exactly a cosine between kernel mean embeddings in the joint activation-error space, exposing activation, error, and coupling factors rather than requiring a materialized Fisher matrix. FisherSketch estimates this cosine directly in a single streaming pass, making K=128,256 head Fisher alignment practical with a 16 KB task signature (m=4096) and a 192 KB per-task streaming state, small enough to store next to a model hash, but encoding transfer-relevant update structure. Beyond source selection, the same signatures and marginals provide a diagnostic instrument for studying whether LLM task similarity is driven by activations, errors, or their coupling; shared-parameter and internal-layer validations, together with Llama-3.1-8B verbalizer-shift experiments, show that FisherSketch remains informative when activation similarity cannot distinguish tasks.

Modern edge computing applications, particularly high-throughput stream processing like Vision Transformers (ViTs), demand massive spatial parallelism and efficient data movement under tight power and area constraints. Coarse-Grained Reconfigurable Architectures (CGRAs) offer a promising paradigm to balance performance, flexibility, and energy efficiency. This paper analyzes the impact of two critical CGRA design choices: processing element heterogeneity and local data reuse support. We evaluate essential computational kernels (Fast Fourier Transform (FFT) and General Matrix Multiply (GEMM)) alongside an end-to-end seizure detection transformer workload across two distinct configurations: a baseline homogeneous architecture and a heterogeneous evolution integrating specialized functional units with an Scratchpad Memory (SPM). Our evaluation demonstrates that the SPM significantly optimizes data movement, reducing memory traffic eightfold compared to a memory-less design. While the heterogeneous architecture achieves superior energy efficiency for data-shuffling tasks, the homogeneous design minimizes area overhead by 4.4x to 8.2x relative to state-of-the-art CGRAs. Furthermore, it sustains a 700 MHz operating frequency, enabling up to a 5x execution speedup over the heterogeneous configuration during matrix computations. Ultimately, this work provides an architectural roadmap for selecting CGRA fabrics based on the arithmetic intensity, performance goals, and resource envelopes of edge-scale workloads.

Language models (LMs) capture large amounts of factual knowledge applicable to a wide range of tasks, motivating the view of their parameters as a knowledge base. An important property of knowledge bases is that different queries for the same fact return consistent results, drawing on a single source of truth. We investigate whether LMs satisfy this property through behavioral and mechanistic analyses. Our results suggest that they encode knowledge in a task-specific manner. Behaviorally, facts acquired on one task frequently fail to co-emerge on others during training. Parameter localization experiments suggest a mechanistic explanation, revealing distinct parameter subsets underlying different tasks for the same fact. Finally, we show that chain-of-thought reasoning draws part of its effectiveness from engaging task-specific parameters beyond those tied to the evaluation task. Our findings suggest that what the model knows and how it is asked are intertwined in parameter space, undermining the "knowledge base" analogy and carrying implications for the reliability and controllability of factual knowledge in LMs.

AI nudification uses generative models to create synthetic non-consensual sexually explicit imagery (SNEACI) of real individuals. Prior work has examined dedicated nudification platforms and model repositories, finding that most targets are female celebrities. However, the anonymous content community, where SNEACI is actively requested, generated, and exchanged, remains unexplored. In this work, we present a large-scale study of AI nudification in the wild, identifying 24,105 SNEACI items. We find a significant shift in target demographics: non-celebrity individuals now account for 55.8\% of targets, compared to only 4.7\% in prior studies, indicating that AI nudification has expanded from targeting public figures to increasingly harming individuals within users' own social circles. Meanwhile, open-source models dominate production, with Stable Diffusion family generating 42.7\% of images and Wan generating 66.5\% of videos, all driven by thousands of shared fine-tuned models and accessible tutorials. Yet the ecosystem runs on a small cohort of active producers, with the most prolific producing 780 items, drives community engagement, shapes target demographics, and disseminates technical knowledge that lowers barriers for new producers. Our work provides an empirical understanding of how AI nudification operates in the wild, revealing the mechanisms that sustain this ecosystem and highlighting the urgent need for interventions in platform governance, technical safeguards, and affected individual protection.

We present a conceptual framework for analyzing dialogue in collaborative problem-solving contexts, with an emphasis on the emerging dynamics of human-AI and multi-agent collaboration. As intelligent systems become active agents capable of autonomous reasoning and strategic cooperation, understanding the dialogic interaction during collaborative problem solving is increasingly important for optimizing and evaluating such partnerships. Our framework addresses key limitations in current analytical approaches through a hierarchical two-layer coding scheme that integrates cognitive and non-cognitive problem solving with metacognitive regulatory mechanisms. We demonstrate its effectiveness and generalizability across nine datasets spanning multiple domains, and provide insights into how humans and agents coordinate their knowledge, skills, and efforts to solve complex problems, showing in particular that metacognitive regulation can be an essential discriminator of deeper collaboration.

Recurrent models must forget in order to remember, yet the state of the art decides what to erase without consulting what is stored -- the gate sees only the arriving token, not the memory it is about to modify. This memory-blind gating is one of three coupled defects in the leading delta-rule architecture (GDN-2): the value-axis erase mask wastes parameters at the scale of the value projection, and -- as we prove -- mathematically prevents the WY-form triangular chunk solver that makes recurrent training competitive with Transformers. We introduce CARVE (Content-Aware Recurrent with Value Efficiency), which resolves all three problems through one principle: erase only on the key axis. This is provably necessary and sufficient for the WY-form solver to remain valid. Within it, CARVE reuses the recurrent output tensor -- already written to GPU memory -- as a free content signal for the erase gate, and replaces the per-value write-gate projection with a single scalar per head. At initialisation CARVE is bit-identical to GDN-2; any quality difference emerges from what the content gate learns. At 1.3B parameters trained on 100B tokens, CARVE achieves WikiText perplexity 15.72 (minus 0.18 vs. GDN-2, a 4.5-sigma effect), leads every recurrent baseline on nine common-sense reasoning benchmarks, and sets state of the art on every RULER retrieval probe -- at 0.4% throughput overhead, 13% lower peak memory, and 19% fewer parameters. Six formal theorems cover memory capacity, Lyapunov stability, gradient flow, expressivity separation, Pareto-optimal chunk size, and hybrid optimality.

Formal semantics has shown that sentence meanings arise by recursively composing lexical meanings, yet much of the literature on semantic universals models either lexicons with fixed signal structures or holistic composition without interpretable lexical parts. We introduce a framework that integrates this fundamental insight of formal semantics in evolutionary modeling, by allowing lexical meanings and a composition function to co-evolve under pressures for conceptual simplicity and communicative accuracy. We apply this framework to the evolution of quantificational meaning. Analyzing the Pareto frontier, we find that the most well-known semantic universal, conservativity, emerges as an efficient system-wide abstraction. The account is sensitive to syntactic structure and helps reconcile tensions between empirical evidence on quantifier learnability and prior evolutionary models. More broadly, the results demonstrate that the picture of sentential meaning developed in formal semantics can be productively combined with evolutionary modeling. The framework offers a template for studying universals that involve global compression within a grammatical category, semantic specialization of syntactic arguments, and the co-evolution of lexical and compositional meaning.

Understanding which genes control which traits in an organism remains one of the central challenges in biology. Despite significant advances in data collection technology, our ability to map genes to traits is still limited. This genome-to-phenome (G2P) challenge spans several problem domains, including plant breeding, and requires methods capable of reasoning over high-dimensional, heterogeneous, and biologically structured data. Current datasets and data repositories, however, are not well-equipped for this task. Current studies do not link gene expression and trait data, and most focus on very specific traits, limiting the breadth of possible correlations. To address this gap, we present the novel Gene-Graph Regression for Arabidopsis Functional Traits (GRAFT) dataset, a curated multi-modal dataset linking gene expression profiles with phenotypic trait measurements in Arabidopsis thaliana, a model organism in plant biology. GRAFT supports tasks such as phenotype prediction and interpretable graph learning. In addition, we benchmark conventional regression and explanatory baselines, including a biologically-informed hypergraph baseline, to validate gene-trait associations. To the best of our knowledge, this is the first dataset to provide multimodal gene information and heterogeneous trait or phenotype data for the same Arabidopsis thaliana specimens. With GRAFT, we aim to foster research to accurately understand the relationship between genotypes and phenotypes using gene information, higher-order gene pairings, and trait data from multiple sources.

3D Scene Graph Generation (3DSGG) represents 3D scenes as structured object-relation-object graphs, providing a compact relational abstraction for spatial understanding. In embodied intelligence settings, the same 3D scene may be observed by agents from viewpoints that differ by yaw rotations. However, current 3DSGG models often fail to produce relation predictions that follow the expected transformation behavior under such viewpoint shifts. This behavior reveals an empirical mismatch related to predicate-level transformation heterogeneity: directional predicates such as left, front, right, and behind should transform with the observation frame, whereas most contact, support, and semantic predicates such as standing on and attached to should remain stable. To reduce this mismatch, we propose Transformation-Aware Decoupling (TAD), a viewpoint-robust 3DSGG framework that decouples relation reasoning according to predicate transformation behavior and is supported by viewpoint-stable object representations. TAD decomposes relation reasoning into two parts: one learns cues that should stay stable across viewpoints, while the other learns directional cues that should change with the observation frame. The two parts are merged for standard multi-label predicate prediction. Transformation-specific descriptors and group-aware auxiliary supervision encourage the two branches to capture complementary relation cues. Extensive experiments on 3DSSG show that TAD achieves state-of-the-art robustness under yaw viewpoint changes without training-time rotation augmentation, while maintaining competitive performance under the standard benchmark. The project page is available at https://tad-predicate.github.io/.

Legal outcome prediction must disentangle objective case facts from adjudicative context. Merit-based rulings rely on factual evidence while technical disposals may hinge on judicial discretion. We propose a Judge-Aware Gated Multi-Task Learning architecture that explicitly models this distinction. We introduce a fine-grained outcome taxonomy to supervise the encoder, enforcing a structural regularization that disentangles distinct semantic pathways. This granular legal curriculum enables our Gated Fusion mechanism to dynamically modulate reliance on judge identity. We evaluate our approach on 13,937 UK Employment Tribunal decisions. We benchmark our design against supervised fine-tuning (SFT) of a Gemma-4 26B-A4B backbone, in which judge identity and the taxonomy are injected as prompt tokens or autoregressive output targets. The two contextual signals compose only weakly when forced through a single autoregressive channel. In contrast, coupling a LoRA-adapted Gemma-4 encoder with our gated architecture defines a new state of the art on this benchmark while requiring an order of magnitude fewer trainable parameters than the generative SFT baselines, with gains concentrated on the most ambiguous and rarest outcome classes. Beyond accuracy, the architecture is interpretable; learned judge embeddings and calibration profiles localize the cases where adjudicative context drives the prediction. These results indicate that, for identity-conditioned classification of legal outcomes, the choice of conditioning interface dominates scale: differentiable structured composition yields more accurate, more parameter-efficient models than prompt-based composition over a substantially larger backbone.

Cloud database systems cannot rely on instance-local disks for write-ahead logging (WAL) durability, forcing WAL onto remote storage. Existing options are unsatisfying: remote block storage like EBS is easy to adopt but adds substantial write latency and cost, while object storage offers excellent durability and low storage cost but is impractical for OLTP due to high latency and per-append cost. Many cloud-native databases, therefore, depend on purpose-built logging backends, which are typically proprietary and tightly coupled to engine-specific replication and recovery protocols, limiting reuse. We present BtrLog, a reusable cloud logging service that combines low-latency durable appends with low-cost archival for the common single-writer architecture. BtrLog replicates log records across a quorum of SSD-backed log nodes in a single network round trip, reducing sensitivity to stragglers in commit latency. To minimize storage cost, log nodes archive records to object storage as large segments, which are written asynchronously and off the latency-critical write path. In our evaluation, BtrLog achieves lower latency than EBS and enables higher end-to-end transaction throughput when integrated into a DBMS.

We study a quantum autoencoder (QAE) for compression-driven anomaly detection in brain MRI data. The approach leverages angle encoding to map image patches into quantum states, followed by a variational encoder-decoder architecture trained to discard information via auxiliary trash qubits. Anomaly scores reflect the degree to which inputs resist compression relative to normal data, with higher scores corresponding to deviations from the learned normal manifold. Evaluated on publicly available brain MRI DICOM datasets, the method achieves a slice-level ROC-AUC of approximately 0.95 and a patch-level ROC-AUC of approximately 0.813, outperforming classical autoencoder and PCA baselines. Analysis of the learned parameters reveals a pronounced encoder-decoder asymmetry, where effective anomaly detection arises from structured information compression within the encoder rather than increased parameter magnitude or decoder expressivity. This results in a controlled compression-reconstruction trade-off with a clear operating regime that supports principled threshold selection. Qualitative evaluation further shows that the QAE produces spatially localized anomaly heatmaps aligned with tumorous regions. The results, supported by promising baseline performances, demonstrate that quantum autoencoders provide an interpretable and controllable mechanism for anomaly detection based on incompressibility with respect to a learned latent representation. This work highlights the potential of quantum autoencoders as a principled tool for studying compression dynamics in quantum machine learning, with promising implications for decision support in medical imaging workflows.

As embodied AI and world models increasingly operate in dynamic 3D environments, visual perception must move beyond passively interpreting given observations toward actively deciding what to observe. We study this problem through camera planning in dynamic 3D story worlds, where the camera must not only generate smooth motion, but also decide what visual evidence should be acquired before it moves. We formulate this capability as Narrative-Grounded World Visual Attention, where the camera acts as an embodied observer that determines what to observe, how to compose the observation, and how to shift attention over time under narrative intent and physical 3D constraints. To realize this capability, we propose Look-Before-Move, a camera planning framework that separates observation specification from motion execution. It first builds a Semantic Observation Contract to convert directorial intent into executable visual constraints, then performs Monte Carlo Viewpoint Search to find narrative-compliant and geometrically feasible viewpoints, and finally applies Semantic Trajectory Grounding to connect selected viewpoints into continuous, collision-aware, and temporally coherent camera motion. We further construct a dynamic 3D Story World Benchmark based on StoryBlender, covering 50 stories, 457 scenes, and 1585 shots with animated characters, semantic scene configurations, and executable 3D environments. Experiments show that our framework improves subject perception, intent consistency, and trajectory quality over representative baselines, demonstrating the importance of organizing visual attention before generating camera motion.

The primary goal of Remote Sensing Image Change Captioning (RSICC) is to automatically generate descriptions of changes between remote sensing images captured at different time points. Existing models still rely on a single autoregressive generation paradigm, which tends to prioritize learning easily generated vocabulary over capturing discriminative differences between images. To address this, we reframe the training paradigm and propose a novel Difference Feature Modeling (DFM) framework. Specifically, we introduce a Text-guided Gated Contrastive Loss (TGCL) to guide the vision encoder to extract critical features from a text-modal perspective. Additionally, we incorporate a pre-trained Change Detection model to transfer stable change detection knowledge. In order to further enhance the representation, we design a Joint Feature Modeling (JFM) module to achieve the fusion of multi-scale difference representations, thereby capturing comprehensive spatiotemporal variations between multi-temporal images. Extensive experiments on multiple datasets demonstrate the effectiveness of our approach.

Synthetic data is increasingly used to enable the development and evaluation of AI systems in domains where access to real-world data is restricted. In healthcare, clinical documentation presents particular challenges due to its sensitivity. This work introduces a synthetic clinical notes pipeline and dataset designed to support the development of clinical AI tools while avoiding the privacy risks associated with real patient data. The dataset is generated using a modular pipeline that combines structured patient generation, semi-structured patient journey simulation, and unstructured clinical note generation using large language models. The pipeline is designed to prioritise internal consistency across longitudinal patient records, while also capturing variation in writing style, note structure, and clinical detail. Additional mechanisms, including LLM-based validation and augmentation steps, are used to improve faithfulness, realism, and diversity of the generated notes. We release a dataset of 70 synthetic patients, each associated with 20-50 clinical notes spanning a full hospital journey. The dataset is provided at multiple levels of validation, enabling users to balance realism and scalability depending on their use case. This dataset supports the development, testing, and evaluation of clinical AI systems, including summarisation tools, coding models, and decision support systems, without reliance on real patient data.

Multi-agent large language model (LLM) systems often rely on verifier and critic agents to suppress hallucinations, but verification is delayed. During this delay, false claims can propagate through the agent network. We model this process as delayed consensus on a graph with grounded corrector nodes. Spectral decomposition by the grounded Laplacian yields a closed-form stability threshold for the verification dose: correction that is too strong or too delayed can turn consensus into oscillation. The most unstable regime occurs when the communication and verification delays coincide; for delay two, the threshold is the inverse golden ratio. The same framework gives a supermodular placement objective and a greedy (1-1/e)-approximation rule for assigning a limited corrector budget to influential nodes. Experiments across five open models confirm the predicted dose-delay oscillations. By contrast, grounded factual answering makes truth an absorbing boundary and eliminates the effect, suggesting that the instability is specific to signed-belief tasks while grounded verification remains stabilizing

Recommendation algorithm iteration is moving from an artisanal, engineer-bound process toward an industrialized research loop, but this transition remains blocked by a structural execution bottleneck: the idea-to-launch cycle still depends on human engineers to generate hypotheses, modify production code, launch A/B experiments, and attribute online results. Innovation therefore scales linearly with headcount rather than compounding with evidence, compute, and accumulated experimental knowledge. We present AgentX, a production-deployed multi-agent system that fundamentally restructures this production function. AgentX operates as a self-evolving development engine: it autonomously generates, implements, evaluates, and learns from recommendation experiments at a scale and pace that no manual workflow can sustain. The system orchestrates four tightly coupled stages in a closed loop. A Brainstorm Agent synthesizes evidence from historical experiments, system architecture, data analysis, and external research into ranked, executable proposals. A Developing Agent translates each proposal into production-ready code through repository-grounded generation and multi-dimensional reliability verification. An Evaluation Agent conducts safe online rollout with guardrail-vetoed A/B judgment, converting both successes and failures into structured knowledge assets. A Harness Evolution layer (SGPO) then distills execution trajectories into semantic-gradient updates that continuously sharpen the agents themselves -- making the system not merely automated, but self-improving.

Software engineering, whether performed by humans or by AI agents, requires reasoning about how software behaves. We call the internal model that supports such reasoning the software world model, and view current code-execution benchmarks as covering one well-studied slice of it -- control flow. In this paper, we take a step toward a broader evaluation by shifting the observable axis to execution resources: alongside test outcome and exception class, we predict peak memory, wall-clock time, and ranked profiler outputs at method and line granularity. We use SWE-bench Verified as the source of data to hold the test close to real-world software engineering tasks. All tested models, frontier ones included, show modest performance and brittle behaviour, suggesting a notable lack of understanding of how software is executed, as opposed to how its source code is written.

Deep learning has shown significant potential in medical image analysis, particularly for disease detection using MRI scans. Accurate and early diagnosis of brain tumors remains challenging due to the complexity of brain structures and reliance on manual interpretation. This work presents an automated deep learning-based approach for brain tumor detection from MRI images using Convolutional Neural Networks and Residual Networks. Transfer learning is applied with two pretrained architectures, ResNet18 and ResNet50, to classify MRI scans into tumor and non-tumor categories. Experiments are conducted on a dataset of 3,929 brain MRI images, evaluating the impact of model depth and fine-tuning strategies. The results show that ResNet18 achieves a higher accuracy of 97% compared to 96% for ResNet50, demonstrating better generalization on limited medical data. The proposed framework enables fast, accurate, and cost-effective brain tumor detection, supporting early diagnosis and clinical decision-making.

The Abstraction and Reasoning Corpus (ARC) contains tasks that require summarizing patterns from limited grid samples and predicting output grids. Recently, many large language model based approaches have attempted to transform it into a text-based reasoning task. However, methods based on open-source models have generally yielded unsatisfactory results, while those relying on closed-source models are too costly. Current efforts mainly focus on data augmentation, constructing ARC-like data for more comprehensive supervised fine-tuning. In this work, we argue that solving ARC-like problems requires not only positive sample supervision but also the ability to improve model reasoning by distinguishing negative samples. To this end, we draw on the idea of preference alignment and propose DiARC, a method that constructs preference pairs to enable the model to distinguish between them. Specifically, we propose three ways to construct negative samples, including output-level visual transformations, DSL-level rule inversion, and task-specific rule editing. The resulting negative samples provide informative near-miss alternatives while keeping the observed demonstrations unchanged. Experimental results across multiple ARC-like benchmarks show that DiARC consistently improves performance over baseline models. The code is released at https://github.com/szu-tera/DiARC.

Large language models (LLMs) can make scientific software easier to use. However, a general model does not automatically know which measurements a particular sensor can support, which algorithms are implemented in the current software, or which conclusions are justified by a computed result. These distinctions are especially important for low-channel electroencephalography (EEG), where sparse spatial coverage and variable signal quality make plausible but unsupported interpretations easy to produce. We present NeuraDock Agent, an open-source architecture that separates a deterministic local EEG engine from a hardware-aware language layer. The numerical engine parses recordings, performs quality control, executes reviewed spectral workflows, and writes machine-readable artifacts. The LLM receives only a compact, allowlisted summary and a versioned context pack. The context describes the seven-channel hardware, reviewed workflows, result fields, implementation boundaries, scientific limits, and reference cases. Raw EEG and dense per-sample arrays remain local We evaluate the system at three levels. First, 12 recordings produced identical structured results over ten numerical repetitions, and a complete Rest/Task run produced identical result, report, and figure hashes over three repetitions. Second, request-capture and failure-injection experiments confirmed the tested data boundary and preservation of local artifacts under HTTP, malformed-output, and connection failures. Third, a boundary-awareness benchmark tested 36 ordinary and adversarial questions under four context ablations and two LLMs, yielding 288 outputs.These results support hardware- and implementation-aware grounding as a practical mechanism for calibrating what an EEG agent accepts, qualifies, or refuses; they do not establish clinical validity or a validated absolute cognitive-load index.

A central goal of safety research is determining whether a model is misaligned. Prior work has largely focused on detecting concerning behavior. But behavior alone does not establish misalignment: a concerning action can arise from benign causes such as confusion. This motivates model forensics: investigating whether the action was driven by malign intent. In this paper, we propose a baseline protocol for model forensics consisting of two steps, iterated as needed. First, we read the chain of thought (CoT) to generate hypotheses about what drives model behavior. Second, we make edits to the prompt or environment to test these hypotheses. While the CoT is not always faithful, it is a rich source of unsupervised insight that can guide the collection of more rigorous evidence. To evaluate our protocol, we create a suite of six agentic environments where models exhibit concerning behavior, and apply it to each. We establish that Kimi K2 Thinking takes shortcuts due to a genuine disposition towards low-effort actions, by showing this hypothesis successfully predicts its behavior. Through counterfactual experiments, we show DeepSeek R1 deceives out of a desire to be consistent with a previous instance of itself. Our methods nonetheless leave significant room for refinement. For example, when we test whether Kimi K2 Thinking believes it is violating user intent, we find no evidence of such a belief, but without positive controls we cannot confirm our tests would detect it. Overall, we find our simple protocol provides a strong baseline that we hope future work will improve upon. More broadly, our work is a concrete step in developing the growing field of model forensics.

Floods frequently impact regions around the world. Rapid and accurate flood detection is crucial for emergency response and timely mitigation of human and economic loss. The expanding availability of satellite data and advances in artificial intelligence have enhanced monitoring of environmental hazards, but many flood events remain challenging to detect because cloud cover obscures optical satellite imagery. Rambour et al. introduced the SEN12-FLOOD dataset and extracted per-image features using a ResNet-50 convolutional neural network backbone, then fed these features into a gated recurrent unit network to show that temporal information can substantially improve accuracy compared to single-image baselines. More recently, Chamatidis et al. showed that a vision transformer can achieve strong performance with popular convolutional architectures. However, these models typically function as opaque black boxes, making it difficult to interpret their decision boundaries, learned features, and internal reasoning, especially in safety-critical domains like remote sensing. In contrast, topological data analysis (TDA) provides a mathematically grounded framework for capturing global structural features of data. TDA has emerged as a powerful tool for analyzing complex imagery, especially imagery with geometrically interpretable structures, of which floods are a prime candidate. In this work, we systematically evaluate topological descriptors for flood detection using the open-source SEN12-FLOOD dataset. By extracting topological features from each image and incorporating them into neural networks, we demonstrate that topological descriptors carry meaningful flood signals independently and complement existing networks to yield more robust and interpretable flood detection systems.

Risk stratification for pulmonary embolism (PE) is critical for clinical decision-making. Stratification guidelines are based on patient medical records, parameters measured from computed tomography pulmonary angiography (CTPA), and blood tests. However, blood tests are often missing in routine practice. This work studies whether state-of-the-art models can accurately classify risk stratification from only medical records and biomarkers extracted from CTPA images. We benchmark different approaches to combine medical records and cardiac biomarkers with rich pulmonary vascular information; we add vascular biomarkers to tabular models and apply graph neural networks (GNNs) on the vascular tree's intrinsic graph representation. We use a private dataset (n=353) with uniquely complete data for PE risk stratification. Our results show that, among global features, medical records and cardiac biomarkers are the most significant predictors, while vascular biomarkers do not further improve stratification. Even more surprising, even GNNs on vascular graphs fail to outperform strong tabular baseline on global features. We consider hypotheses, on both models and data, that could explain this suboptimal performance. Our investigation suggests that, counter-intuitively, vascular graphs might hold no discriminative information for PE risk stratification. Code is available from https://github.com/creatis-myriad/GENESIS.

To better assist users with completing challenging tasks, AI agents mediate communications, access data, and interact with different APIs. Many employers (and even nation-states) already provide their users with this technology. However, widespread adoption of AI agents creates a new risk to abuse access to user data for another goal: surveilling users. These users might not even have the ability or permission to control the actions and data accesses of the surveilling agents. We introduce and formalize the problem of agentic surveillance: the ability of an AI agent to analyze available information, craft a report, and send it out using available tools. To evaluate surveillance capabilities across different models, we create SurveilBench, a dataset of various reporting scenarios focusing on three domains: corporate, education, and police. We find that some models exhibit emergent (i.e., unprompted) tendencies to help surveillance, but they also report the attempts to surveil users to the government. Finally, we repurpose prompt injections for evading surveillance and develop three evasion techniques that hide from, deceive, or induce over-escalation in surveillance agents. We conclude that agentic surveillance can already be easily implemented and, therefore, call for a comprehensive technical, ethical, and legislative framework to protect users.

Semantic code search and clone detection are essential for software development, maintenance, and reuse. This paper evaluates the effectiveness, efficiency, and scalability of contemporary deep learning models for first-stage recall in large-scale code-to-code search engines. Benchmarking across multiple programming languages and datasets reveals critical limits in the precision and scalability of these models on Terabyte-scale source-code collections. We present LLM-based code normalisation and query-rewriting schemes that yield significant gains in precision for lower-performing models. Our results question the sustainability of resource-constrained deployment and the assumed robustness of current code-specialised LLMs across datasets. We conclude with actionable insights for building scalable, efficient code-retrieval systems.

This paper presents a pre-silicon analysis of XRM-SSD V24/V7.0, a physics-aware predictive firmware scheduling layer for Intel's 3.5D heterogeneous integrated packages (Foveros Direct 3D + PowerVia + EMIB-T + UCIe + HBM5). Using detailed thermal-electrical co-simulation over a 90,000-step LLM inference dataset, we show that proactive workload-density-driven thermal hinting (20-50 ms look-ahead) enables pre-positioning of PowerVia voltage rails. Key results include a thermal-load correlation of R^2 = 0.9911, compensated CPO spectral drift below 0.36 nm (21% of TSMC tolerance budget), and HBM leakage current clamped below 1 MB/hr across all load states. Monte Carlo analysis (N=2,000 trials) confirms robustness under process variation. V7.0 extends the framework to multi-tile architectures with an N x N thermal coupling matrix and two-pole kernel. The approach demonstrates potential for 20-30% released compute and 65-68% EDA guard-band reduction. All metrics are engineering projections from pre-silicon characterization. Silicon validation on Intel 18A platforms is pending. This work highlights firmware-hardware co-optimization as an effective approach to mitigating physical limits in advanced 3.5D packaging.

Business Process Management (BPM) is concerned with the systematic design, execution, monitoring, and improvement of business processes. Formal grammars have emerged as a particularly fruitful formalism for BPM, offering generative, declarative, and analytical capabilities that are uniquely well-suited to process-oriented concerns. This paper presents a systematic literature review of 34 primary studies at the intersection of formal grammars and BPM. We identify seven research streams: (i) process grammars for organizational process design; (ii) process modeling languages evaluated as grammars under the Bunge-Wand-Weber ontological framework; (iii) production-rule grammars for process structural specification and variant management; (iv) attribute grammars for the declarative specification and distributed execution of workflows; (v) graph grammars for the transformation, generation, and semantic analysis of process models; (vi) grammatical inference for process mining and discovery; and (vii) process algebras as grammar-like compositional frameworks for behavioral specification and verification. For each stream, we synthesize contributions, formalisms employed, and limitations. The review reveals that formal grammars have influenced BPM across every lifecycle phase (from organizational design to formal verification and data-driven discovery) yet the seven streams have developed largely in parallel, without cross-stream synthesis. We identify five corpus-grounded open challenges and argue that a deeper, unified exploitation of grammatical theory holds significant promise for advancing the state of the art in BPM.

Context: Generative AI adoption in software engineering education raises opportunities for learning support alongside concerns about superficial learning and academic integrity. Objective: This study investigates how explicit instructional guidance aligned with Bloom's taxonomy supports responsible GenAI use in SE/CS education, exploring student and instructor perceptions of GenAI-supported learning. Method: We designed a Bloom-aligned GenAI framework that articulated appropriate GenAI roles at different cognitive levels. The framework was embedded in course instructions, labs, and assessment across multiple SE/CS courses at Queen's University Belfast and Azerbaijan Technical University. Data were collected via anonymous questionnaires and learning artifacts, analyzed using thematic analysis with Bloom's taxonomy as an analytic lens. Results: Students perceived GenAI as most valuable for higher-order cognitive activities (analysis, evaluation, reflection) and less suitable for foundational learning. Explicit Bloom-level guidance influenced students to use GenAI reflectively, with delayed or intentional non-use when independent thinking was prioritized. Both students and instructors reported pedagogical benefits alongside challenges in cognitive effort and instructional design workload. Conclusion: GenAI's educational value lies in intentional alignment between cognitive learning goals, instructional guidance, and learner self-regulation. Bloom's taxonomy provides a scalable, pedagogy-driven framework for responsible GenAI use in SE/CS education, offering a practical alternative to enforcement-focused responses.

Evaluating LLM agents requires dynamic environments that go beyond static reasoning and zero-sum games. Real-world economic interaction is often open-ended and mixed-motive: agents must negotiate, create positive-sum surplus, compete for scarce assets, and plan under delayed returns. We introduce SidConArena, a new benchmark framework for evaluating LLM agents in open-ended, positive-sum bargaining. SidConArena formalizes a multi-player economy as a finite-horizon partially observable stochastic game with three coupled phases: natural-language negotiation with binding trades, deterministic converter-based production, and sealed-bid auctions for long-term assets. The framework combines structured observations, phase-aware agent dispatching, a neural-symbolic action interface, and asynchronous execution, enabling free-form interaction while preserving rule-grounded evaluation. Across homogeneous and heterogeneous tournaments, stronger frontier models achieve higher economic outcomes, yet agents still misvalue resources, bargain passively, and remain limited in long-horizon investment planning.

As Large Language Models (LLMs) are increasingly deployed in healthcare settings, accurate error detection and correction in generated or existing text becomes critical, as even minor mistakes can pose risks to patient safety. Existing methods for error detection and correction, including automated checks and heuristic-based approaches, do not generalize well across unseen datasets. In this paper, we propose MedGuards as a medical safety guardrail, which is a new framework that treats medical error detection and correction as a multi-agent in-context learning task. Specialized agents separately detect, localize, and correct errors, while a confidence-guided arbitration mechanism resolves disagreements using reasoning traces and confidence scores. This design enhances interpretability, robustness, and adaptability, without requiring additional training of the base LLMs. Additionally, we introduce the Keyword-Prioritized Correction Score (KPCS), a new evaluation metric that considers whether critical keywords within the reference text are generated correctly, providing a more comprehensive assessment than conventional metrics. Experiments across four multilingual medical datasets consisting of clinical notes demonstrate significant improvements by the proposed framework across several metrics and models. Our aim is to enable safer deployment of LLMs in real-world healthcare applications. For reproducibility, we make our code publicly available at https://github.com/congboma/MedGuards.

Traditional evaluations measure a learning algorithm's final performance on an i.i.d. test set, reducing learning to a single aggregate score. This approach obscures a fundamental question: to what extent does learning from a specific example generalize to others? Such per-sample generalization, akin to learning by analogy in human cognition, captures how far the knowledge extracted from one example can transfer, yet remains invisible to standard benchmarks. We introduce the Generalization Spectrum, an evaluation framework designed to expose this hidden dimension. For each training example, we construct a controlled suite of test variants arranged by increasing transfer distance, from exact recall to implementation transfer across languages, context transfer under complete narrative re-framing, category-matched in-domain problems, and an unpaired baseline. By tracking performance across these distances, we reveal not just whether an algorithm learns, but how far that learning extends. We instantiate this framework on competitive programming, using a selection-and-synthesis pipeline seeded with recent problems to mitigate contamination. We first compare three canonical learning paradigms under matched memorization. RL converts memorization into near-transfer more efficiently than SFT-family baselines, while ICL exhibits strong but correspondence-dependent transfer. We then use the Spectrum to diagnose within-family variants. The resulting profiles show that local gains need not expand the generalization radius: abstractions and hints mainly lift local transfer, RFT preserves a stronger far-transfer tail than reference SFT, and self-distillation or hint-assisted RL can reduce far transfer even when local transfer or optimization improves.

We investigate Anderson localization in three-dimensional disordered systems by comparing scalar classical waves with mass and force-constant disorder to electronic tight-binding models with diagonal and off-diagonal disorder. We show that the commonly employed mapping between classical-wave localization and the electronic Anderson model with diagonal disorder is not mathematically justified. Instead, the correct modulus-type formulation reveals that classical-wave systems constitute a distinct constrained disorder class, in which the acoustic sum rule correlates diagonal and off-diagonal matrix elements and prevents any direct correspondence with the standard electronic disorder models. Within a unified eigenvalue framework, we determine localization phase diagrams for all four disorder classes using complementary spectral, eigenvector, and level-statistics diagnostics. We find that classical-wave systems share a key qualitative feature with electronic off-diagonal disorder: localized states occur only near a band edge, while extended states persist in the central part of the spectrum even at strong disorder. At the same time, the acoustic sum rule produces localization topologies that differ fundamentally from both diagonal- and off-diagonal-disorder electronic systems. In particular, for mass disorder we obtain a phase diagram that differs qualitatively from previous results based on the conventional potential-type approach and reveals an extended localized regime near the upper band edge. Our results establish a unified perspective on localization in quantum and classical wave systems and provide new insight into the conditions under which Anderson localization may occur in three-dimensional photonic and acoustic media.

Microbes in marine environments are often confined to thin near-surface layers while being advected by turbulent flows. Because such constrained advection generates an effectively compressible flow, reproduction and transport interact in a nontrivial way. Here, we focus on populations whose growth is governed by an Allee effect and show that sinks and sources, generated by the compressible flow, have dramatic consequences for the survival of such species. We derive analytical expressions for the carrying capacity as a function of the Allee strength in the limit of small and large Damköhler number, which measures the product of the large eddy turnover time and the organism growth rate. Numerical simulations reveal how these two limits connect. In the limit of small Damköhler number, we find a maximal Allee strength, set by the statistics of the compressible flow, that leads to species extinction in fully developed turbulence.

We present an exact variational scheme for the physics of inhomogeneous classical fluids in thermal equilibrium. A joint metadensity minimization principle is proven for the one-body density and the global interparticle distance distribution. The theory bypasses the inhomogeneous two-body density and thus remains computationally simple. A universal excess entropy functional accounts for all many-body correlations in arbitrary pairwise interacting systems. The framework is relevant for neural functional machine learning, for soft matter design, and for predicting structural correlation functions via entropic test-particle and meta-Ornstein-Zernike routes.

We show that Abelian and nonabelian gauge transformations are the analog of $W_\infty$ transformations for higher dimensional quantum Hall effect. The commutator anomaly and the extended algebra of such transformations on the edge modes of a droplet are obtained by purely topological arguments, basically utilizing the two-cocycle in the descent procedure for anomalies and using the fact that there is anomaly cancellation between the bulk and boundary actions. The method relies on the fact that bulk actions are easily constructed in general using the Dolbeault index theorem. The resulting algebras are shown to agree with explicit edge mode calculations for cases where they are available. We also comment on the similarities and differences in the nature of these transformations between two and higher dimensions.

Electrons in layered van der Waals materials possess a layer pseudospin characterizing their wave-function distribution among layers. In twisted structures, this pseudospin forms nontrivial textures, leading to intriguing phenomena such as the layer Hall effect (LHE), where distinct layer Hall currents flow despite the presence of time-reversal symmetry. In chiral bilayers, LHE manifests as an interlayer electric dipole Hall effect with Hall counterflows and a concomitant in-plane magnetic dipole. Multilayers host richer layer-dependent Hall currents, generating interlayer electric multipole Hall effects and in-plane magnetic multipoles. We start from exploring the interlayer electric quadrupole Hall effect in mirror-symmetric twisted trilayers. At small twist angles, interlayer translation efficiently tunes layer Hall current magnitudes. At large angles and low doping, the currents can be well accounted for by adding the contributions from the two individual twisted interfaces. This decomposition allows obtaining layer-resolved Hall currents in large-angle twisted multilayers even without well-defined periodicity.

We determine the role of the fluctuation-induced Casimir force acting between a membrane of cylindrical shape and a bottom electrode in microelectromechanical capacitive switches. For this purpose, the Casimir force is computed taking into account the real properties of both a membrane and a bottom electrode materials with account of surface roughness. The obtained results are compared with those found for the smooth surfaces using the idealization of ideal metal. It is shown that an account of both the real material properties and surface roughness is crucial for obtaining the correct values of the Casimir force. According to our results, at the shortest separations, when the switch membrane is in contact with the transmission line, the magnitudes of the Casimir force may exceed the magnitudes of the electric one depending on the value of the operating voltage. The obtained values of the Casimir force can be used for determining the thickness of the switch membrane, which ensures the necessary magnitude of the restoring elastic force required for a stable cyclic functioning of the micromechanical switch with no pull-in.

Andreev cavities provide a setting in which superconducting proximity spectra are shaped by phase-coherent electron-hole motion along extended trajectories. While such Andreev physics is well established in transport, local spectra in two-dimensional cavities remain largely unexplored in real space. Here we use scanning tunnelling spectroscopy to study confined Cu(111) surface states coupled to superconducting Nb(110). The in-plane magnetic-field scale for the collapse of the resolved low-energy spectrum is controlled by the transverse extent available to Andreev trajectories, while the zero-field excitation energy evolves with the characteristic trajectory length. These trends, together with spatial variations within individual islands and the response to vortex phase textures, are captured by a minimal semiclassical phase-accumulation picture. Our results identify geometry-defined Andreev trajectories as a design principle for phase-coherent superconducting cavities accessible by local spectroscopy.

Moiré superlattices have emerged as a new platform for engineering electronic and optical properties in van der Waals heterostructures, enabling control over correlated and excitonic phenomena. Yet the impact of moiré superlattices on exciton-phonon coupling remains largely unexplored. Here we demonstrate emergent, layer-selective coupling between moiré phonons and moiré excitons in angle-aligned WS2/WSe2 heterobilayers. Using a broadband terahertz phonon transducer, we coherently launch moiré phonons that resonantly perturb the excitonic states. We show that the exciton-phonon coupling is intrinsically modified by the moiré superlattice in a layer-selective manner. A driven oscillator model captures the dynamics, revealing three moiré phonon resonances with distinct coupling to the moiré excitons. First principles calculations show that many moiré phonon modes can arise with distinct strongly hybridized in-plane and out-of-plane vibrations in the moiré unit cells. The calculations further identify the three experimentally observed moiré phonons and their emergent characteristic coupling to the moiré excitons.

Bubble coalescence phenomenon in electrolytic media finds applications in technologies from mineral flotation to electrochemical energy conversion. However, the underlying governing physics still remains unresolved, with longstanding disagreement over the extent to which Marangoni stresses affect the coalescence time by modulating the interfacial mobility. Here, we show that the thin film morphology governs drainage more strongly than the interfacial boundary conditions. We demonstrate experimentally that thin film drainage during bubble coalescence proceeds through three distinct regimes. An initial visco-capillary stage that exhibits a power-law thinning, followed by an exponential decrease in film thickness with time induced by rim stabilisation. The final regime is governed by disjoining pressure and is marked by an exponential relaxation of the film to the equilibrium thickness. We show that, irrespective of the electrolyte type and concentration, film evolution exhibits universal behavior by collapsing onto a single curve when rescaled with the characteristic film thickness and time scale, demonstrating that electrolyte effects act only to renormalize timescales rather than alter the underlying dynamics.

Context-free grammars are minimal models of hierarchical structure in human language, generating structured text from recursive production rules. The Random Language Model (RLM) [De Giuli, PRL 2019], an ensemble of such grammars with random rule weights, exhibits a cross-over from gibberish-like output to structured text as a function of a "temperature", but the location and nature of this transition remained unclear. Here, we show that the RLM exhibits a hierarchy of phase transitions in a double-scaling limit where the grammar temperature $\tildeε_d \to 0$ and the number of hidden symbols $N \to \infty$ at fixed $x = \tildeε_d \log N$. By identifying the relation between RLM and the Random Energy Model, we identify a series of transitions where correlations between symbols emerge, single-symbol marginals become non-uniform, and rule use freezes in a glassy phase. A semi-annealed approximation yields nontrivial scaling laws for rule usage, entropy, and energy, consistent with Heaps' law and context-length scaling observed in large language models.

We investigate extractable work storage in a capacitively coupled double quantum dot (DQD) quantum battery (QB) subjected to experimentally motivated detuning charge noise. The battery is modeled as two interacting charge qubits with an Ising-type capacitive coupling and is charged by resonant microwave modulation of the tunnel coupling channel. Detuning fluctuations are introduced as classical stochastic processes generated from a band-limited 1/f noise spectrum. For each noise realization, the evolution remains unitary, whereas decoherence and loss of contrast emerge after ensemble averaging. We analyze the total ergotropy, its population and coherent contributions, the energy basis populations, a passive ordering violation diagnostic, and the Jensen-Shannon coherence of the noise-averaged state. The results show that resonant tunnel coupling driving selects a dominant E0 <-> E3 population transfer channel in the interacting DQD spectrum. The dominant extractable work is stored in non-passive population distributions, in agreement with recent population ordering interpretations of ergotropy in QBs, while coherence accompanies and supports the resonant transfer as a transient dynamical resource. Detuning noise reduces the energy basis coherence amplitude and also weakens the population transfer pathway responsible for the dominant population ergotropy. This framework provides a noise-aware description of semiconductor QB charging based on extractable work rather than on injected energy alone.

We develop and characterize scanning transmission electron microscopy (STEM) capabilities within a scanning electron microscope (SEM) to investigate the effective lateral coherence of the electron beam (e-beam) in the specimen plane. Using single-crystalline Au flakes and a sample composed of a monolayer of graphene, we obtain high-quality selected-area electron diffraction (SAED) maps and convergent-beam electron diffraction (CBED) patterns, validating the systems ability to probe crystallographic information at an acceleration voltage of 30 keV. Building on these capabilities, we implement a method, which is adapted from techniques traditionally used in transmission electron microscopy, to measure the degree of lateral coherence of the e-beam in the specimen plane of the SEM. By analyzing interference between electrons with two different wave vectors separated by 0.031 per angstrom, we extract a lower limit for the degree of lateral coherence over 5% of the e-beam diameter of approximately 60%. These coherence values are sufficient to enable quantum-coherent electron-light-matter interaction experiments in the SEM.

The classification of mixed-state phases requires criteria beyond two-point correlation functions, such as the decay of the mutual information (MI) and the conditional mutual information (CMI), with the latter encapsulated in the notion of Markov length. Here we show how such static properties of the fixed point of a Lindbladian follow from natural dynamical features of its generators: rapid mixing and frustration-freeness. We focus on systems with long-range interactions, and prove (i) that local Lindbladians satisfying (global) rapid mixing and frustration-freeness have fixed-points whose CMI decays with the shielding distance, and (ii) that (local) rapid mixing together with primitivity and regularity implies global decay of MI. For long-range interactions decaying with a power law with rate $α$, both quantities decay polynomially rather than exponentially, in contrast to the finite- and short-range regimes where exponential decay (a finite Markov length) is expected within a phase. We further show that Gibbs states of long-range, non-commuting Hamiltonians satisfy a local Markov property at any temperatures, extending the recent results (Chen--Rouzé, 2025) for short-range systems to the long-range regime relevant to a variety of experimental platforms. As a numerical example, we study the long-range Ising model both with and without a transverse field. We find regimes in which the polynomial decay of the CMI holds, in accordance with the bounds proven.

Fluid transport in deformable porous media is central to many biophysical and geophysical processes. While extensive studies exist, how porosity governs fluid behaviour in deformable systems during cyclic injection remains elusive. Here, we investigate gas-liquid multiphase flow in a quasi-2D Hele-Shaw cell packed with soft hydrogel particles at different initial porosities. Alternative gas and water injection experiments, combined with high-resolution imaging and continuous pressure monitoring, are used to quantify gas dynamics and pressure evolution. Results show that the gas entry pressure increases as porosity decreases, consistent with a Young-Laplace estimation based on effective pore-throat width. After entry, invasion shifts from cavity-dominated expansion in high porosity packings to localised pore invasion in low porosity packings, with a mixed cavity-fingering regime at intermediate porosity. Pressure fluctuations are linked to pore-scale gas escape and internal gas redistribution. Low porosity packings produce frequent small-amplitude pressure drops, whereas higher porosity packings produce more discrete pressure relaxations. Across cycles, the decreasing mean pressure suggests preferential-pathway reuse and reduced local capillary constraints. Residual gas saturation increases systematically with injection cycles and reaches higher terminal values as porosity decreases. Specific interfacial length increases as available pore space decreases and follows a power-law relationship with gas cluster size, with scaling exponent decreases as porosity decreases and cycling progresses. Together, these results demonstrate that gas trapping in deformable porous media depends on both initial packing structure and cyclically evolving gas-solid interactions. This study provides insights for interpreting porosity-dependent trapping and reinvasion during repeated gas injection.

We investigate a frustrated four-leg spin-$1/2$ ladder using density matrix renormalization group calculations. The uniform system displays three regimes: short-range antiferromagnetic legs, short-range ferromagnetic legs, and an effective spin-2 Heisenberg chain, separated by a crossover and a first-order transition. The spin-2 regime is confirmed through its finite string order parameter, edge-localized excitations, and excellent agreement with a projected $S_r=2$ effective Hamiltonian. Recasting the model as two frustrated two-leg ladders coupled by rung and diagonal interactions, we track how the trivial and Haldane phases of an isolated ladder evolve as interladder couplings are introduced. The resulting phase diagrams reveal crossover and first-order lines whose locations are captured by the spin-2 projection and show how singlet- and triplet-dominated regimes reorganize when two ladders merge into a four-leg structure, clarifying the emergence of effective spin-1 versus spin-2 behavior.

Protein length distributions across the tree of life carry a quantitative signature of organismal complexity. Nonextensive statistical mechanics, through the Tsallis generalized entropy formalism, provides a natural framework for describing complex systems characterized by long-range correlations, scale invariance, and hierarchical organization -- features that classical Boltzmann-Gibbs statistics cannot accommodate. In this work, the complementary cumulative distribution function (CCDF) of protein lengths is analyzed within this framework for the reference proteomes of 22 fully sequenced organisms spanning the domains Archaea, Bacteria, and Eukarya, with deliberate sampling across the animal transition zone from sponges and cnidarians to higher bilaterians. Maximum likelihood (MLE) fitting of truncated discrete q-exponential distributions, with bootstrap 95% confidence intervals (CIs) reveals that the entropic index q resolves into three statistically distinct regimes: superextensive (q < 1) for prokaryotes, unicellular and non-animal multicellular eukaryotes, and basal animals; a boundary regime (CI on spanning unity) for the two cnidarians studied and the basal bilaterian C. teleta; and subextensive (q > 1) for all higher bilaterians, with q increasing monotonically across the four deuterostomes sampled from S. purpuratus (1.033) to H. sapiens (1.147). The q-exponential outperforms the ordinary exponential distribution across all 22 proteomes and becomes progressively more competitive against alternative two-parameter distributions as proteome complexity increases. These results identify the Tsallis entropic index as a continuous, physically interpretable indicator of proteome organizational complexity and extend the applicability of nonextensive statistical mechanics to proteomic systems.

We introduce a finite-element framework for simulating thermal fluctuations in compressible fluids governed by the fluctuating Navier-Stokes equations. The method is designed to preserve the fundamental fluctuation-dissipation balance at the discrete level. This is achieved by defining the stochastic forcing term in the weak formulation, ensuring its covariance is proportional to the discrete viscous dissipation operator. A nodal quadrature rule is employed to eliminate unphysical mesh-scale correlations. The time integration is performed using the Crank-Nicolson scheme to maintain numerical stability and accuracy. The proposed approach is numerically validated in one, two, and three spatial dimensions, demonstrating its capability to correctly capture equilibrium fluctuation statistics across various discretisation parameters.

Optically addressable valley degrees of freedom in transition-metal dichalcogenide heterostructures provide a powerful platform for valleytronic and quantum-optical functionalities. In moiré superlattices, interlayer excitons inherit valley-contrasting optical selection rules while acquiring long lifetimes, electric dipoles, and site-dependent optical responses. However, because conventional measurements typically probe time-integrated valley polarization, the dynamical origin of vanishing polarization has remained elusive. Here, we show that a nearly zero steady-state valley polarization in electrically tunable moiré excitons does not necessarily indicate fast valley relaxation. Helicity-resolved time-resolved photoluminescence reveals a temporal crossing between co- and cross-circularly polarized emission, indicating that helicity-opposite dynamical components coexist and compensate after time integration. A minimal two-channel model, representing A-like and B-like moiré emission channels with opposite optical selection rules and distinct effective decay/depolarization rates, reproduces the observed helicity crossing without invoking a single rapid valley relaxation process. Furthermore, two-dimensional gate-field maps show that the crossing time evolves systematically with electrostatic tuning, demonstrating that the hidden valley dynamics are electrically controllable. These results show that time-integrated circular polarization can give a false-negative indication of valley polarization in multichannel valley emitters.

We report the design and experimental demonstration of an open nanophononic cavity based on a hybrid oxide superlattice composed of SrTiO$_3$ (STO) and YBa$_2$Cu$_3$O$_{7-x}$ (YBCO), combined with a metallic Ni transducer for coherent phonon generation. The STO/YBCO periodic stack acts as an acoustic distributed Bragg reflector supporting confined longitudinal acoustic phonons in the sub-THz regime, while the Ni layer enables efficient ultrafast optical excitation and detection by time-domain Brillouin scattering. Transient reflectivity measurements reveal confined acoustic dynamics and a well-defined cavity resonance, in agreement with transfer-matrix calculations of acoustic reflectivity and mode profiles. These results demonstrate phonon confinement in multifunctional oxide heterostructures and establish complex oxide superlattices as a platform for hybrid nano-acoustic resonators and ultrafast phonon control of correlated electronic phases.

A stochastic risk model is applied to simulating the behavior of a nuclear reactor in a situation where the neutron chain length is described by a distribution with heavy "tails," such as the Pareto distribution. Probabilities of a fluctuation exceeding a critical threshold are obtained, and risk bounds for power-law distributions of jumps are estimated. Functionals of the reactor power maximum, the instant of first reaching the maximum, and the distribution of the overjump magnitude are considered. A relationship between the shape parameter and the physical constants of reactors is obtained, as well as the relationship with the noise spectrum and physical constants. The finite dimensions of a real reactor are taken into account. The autocorrelation function of the truncated Lévy process and its relationship with the frequency filters of the neutron flux monitoring equipment are considered.

Thermodynamics of information identifies information flow as a thermodynamic resource, but whether quantum coherence and collective coupling can enhance it at low entropy-production cost remains unresolved. We address this question for interacting open quantum systems by deriving a thermodynamic uncertainty relation that bounds information flow by entropy production in nonequilibrium steady states. We construct a quantum engine in which N-fold degenerate ground and excited states are collectively coupled to heat baths. Collective jumps enhance heat currents and information flow linearly with N, while entropy production remains independent of N, realizing a high-power, low-dissipation autonomous quantum Maxwell's demon that leverages collectively enhanced information flow to pump heat against a temperature gradient. Beyond steady states, collective interactions amplify the unitary component of quantum information flow, yielding a quadratic enhancement of the free-energy charging power of quantum batteries. Our results reveal scalable advantages of quantum coherence and collective effects in quantum engines.

A key challenge in developing Al/Ge heterostructures for quantum applications is Al-Ge interdiffusion. This process is facilitated by grain boundaries in polycrystalline films, which degrades interface quality and impairs device performance and reliability. Here, we present epitaxial growth of single-crystalline Al(111) on Ge(111) by molecular beam epitaxy, achieving an atomically flat and sharp interface. At the interface, a commensurate 7-Al-lattice/5-Ge-lattice epitaxial relationship is observed, which dramatically reduces the intrinsic lattice mismatch from 28.4% to about 0.1%. Interestingly, this well-ordered interface does not form below a critical thickness of 0.3 nm. Instead, Al initially nucleates as random clusters, which then transform into two-dimensional (2D) islands and, as Al deposition further increases, eventually develop into a continuous film. By optimizing the growth parameters, we have achieved an ultra-flat Al film with a surface root-mean-square roughness of about 0.16 nm and an ultra-thin continuous film with thickness of only 2 nm. These epitaxially grown Al-Ge heterostructures, with their atomically flat surfaces and sharp interfaces, provide a promising platform for studying topological quantum states.

Crossing a continuous phase transition out of equilibrium typically generates topological defects whose density obeys a universal power-law scaling predicted by the Kibble-Zurek mechanism. Recent numerical studies have revealed systematic deviations from this scaling in the presence of weak explicit symmetry breaking, manifested as an additional exponential suppression of defect formation. However, the origin of this correction and a general theoretical framework to describe it have remained elusive. Here, using large-deviation theory, we show that defect formation under weak symmetry breaking is controlled by rare fluctuations that drive local regions into the disfavored symmetry-broken state. This mechanism yields a closed-form expression for the defect density in arbitrary dimensions, valid in the weak-field and weak-noise limits. These theoretical predictions are verified through direct simulations of stochastic Ginzburg-Landau models in one and two spatial dimensions.

Electron spin resonance in scanning tunneling microscopy enabled the study of electronic transitions of magnetic impurities on surfaces at the atomic scale. This ESR-STM technique allows to spectroscopically probe and coherently manipulate spins using an all-electrical method without oscillating external magnetic driving fields. Here, we aim to review recent advancements in ESR-STM. We will discuss possible fundamental mechanisms by which the electric field drives spin resonance based on Heisenberg exchange, Kondo scattering, and Anderson impurity models. We validate theoretical predictions against experimental observations, to understand how electronic correlations, spin exchange, and many-body effects manifest in ESR-STM signals. After reviewing coherent spin control in the STM junction, we discuss potential applications of the ESR-STM method for coherent multi-spin control which enables multiple-qubit operations. Finally, we address recent developments in coupled electron-nuclear spin systems, including hyperfine-resolved ESR spectroscopy, and the driving and polarization of nuclear spins in ESR-STM.

The integer topological charge of a magnetic skyrmion is the standard emblem of topological protection. We ask what happens to that protection when the magnet is made non-Hermitian, with balanced gain and loss or a PT-symmetric anisotropy. A non-Hermitian skyrmion turns out to carry two charges that coincide in the Hermitian limit but part ways under deformation. The charge built from the right state alone is homotopy-protected: the PT flow reduces exactly to a Gilbert-type relaxation on the target sphere, so it cannot change under smooth evolution. The charge built from the biorthogonal left-right pair is complex, loses quantization as soon as the gain/loss is turned on, and breaks down at the exceptional point of the local generator -- a ring on the skyrmion's equator, where the biorthogonal Bloch field itself diverges. Topological protection of a skyrmion is therefore not a single statement once the dynamics is non-Hermitian: it splits at an exceptional point. This is the real-space topological counterpart of the analyticity breakdown a causal response function suffers at an exceptional point, both being manifestations of the same non-Hermitian degeneracy.

The quantum geometric tensor is a fundamental property of quantum states, with broad applications in condensed matter physics, topological phases, and quantum phase transitions. The eigenvalues characterize the scale, anisotropy, and effective rank of random quantum geometry, going beyond scalar quantities such as the trace. Here we study the eigenvalue statistics of the quantum geometric tensor in finite-dimensional parameter-dependent random Hamiltonians. We obtain exact analytical results for the first two nontrivial cases, $N=2$ and $N=3$, with $N=3$ already showing genuine shape fluctuations. We further propose a finite-$N$, arbitrary-$D$ description of QGT eigenvalue statistics and verify it by numerical simulations. Our results provide exact benchmarks and a practical framework for random quantum geometry in finite-dimensional disordered and chaotic systems.

We predict a mechanism for the non-Newtonian behavior of a two-dimensional (2D) electron fluid due to the local Joule heating of electrons. Within this mechanism, the electron shear viscosity is a non-monotonic function of the velocity gradient. This means that 2D electrons form an unusual non-Newtonian fluid, either dilatant or pseudoplastic, depending on the flow regime. We construct and solve the hydrodynamic equations for a hydrodynamic velocity and a corresponding temperature profile for a Poiseuille-like flow geometry. We demonstrate that the considered non-Newtonian non-linearity induced by the local heating is apparently responsible for the nontrivial differential magnetoresistance observed for 2D electrons in ultra-high-quality GaAs quantum wells, so we conclude that a 2D non-Newtonian electron fluid is realized in these systems at high currents.

Hybrid photonic quantum computers, combining stationary matter qubits and flying photonic qubits, offer an intrinsically networked and resource-efficient route to large-scale, error-corrected quantum computation. Their core components are cavity-coupled matter qubits that act as light--matter interfaces, enabling: high-efficiency on-demand single-photon generation, stable near-unity photon indistinguishability and spin--multi-photon entanglement. Semiconductor quantum dots in microcavities are a leading platform for realizing such devices. Yet reaching the performance, reproducibility and spin-coherence thresholds for large-scale error correction remains a major challenge requiring industrial fabrication and control. Here we report thousands of monolithic semiconductor quantum-dot devices fabricated using a III--V pilot production-line process compatible with large-scale deployment. Systematic control of source parameters yields state-of-the-art efficiency and supports a path to optical losses below fault-tolerance thresholds. Using field-quadrature state reconstruction as a stringent joint test of efficiency and indistinguishability, we observe near-unity photon quantum purity stable over tens of minutes and a record single-photon Wigner-function negativity. We further demonstrate seven-partite spin--multi-photon entanglement and spin coherence extendable to microsecond timescales in the low-magnetic-field regime. Finally, photons from distant sources are as indistinguishable as photons emitted successively by a single source. These results establish foundry-compatible III--V quantum dots as a scalable platform for hybrid photonic quantum computing.

The simple electrostatic picture of a trion is that of an excess charge inducing an exciton polarization and binding closer (farther) to the hole (electron) side of it. Trion formation can be forbidden when such spontaneous rearrangement of charges is not allowed by the application of external perturbation, such as electric field. Here we test this hypothesis experimentally using a non-monotonic electric field. We realize this scenario by imprinting the ferroelectric domains at the AA-stacked twisted-hBN (t-hBN) interface onto a monolayer of MoS2 placed over it. The spatially varying in-plane electric field around the domain wall serves the dual purpose of (a) confining and polarizing the 2D exciton in the domain wall, and (b) depleting the free charge carriers from the domain wall. We observe a large quantized exciton splitting confirming strong exciton confinement in the domain wall. Forced by the confining potential, the electron side of the polarized exciton lies closer to the domain with accumulated free electrons, which should ideally prevent any trion formation. Contrary to the laid hypothesis, we observe signatures of quantized charged exciton emission, with an inter-level splitting that mimics the level-splitting of the quantized excitons. This paradox is explained using the many-body picture of exciton polaron, where a conduction band hole attractively binds the polarized exciton and the electron Fermi sea. The results provide a definitive way to unambiguously discern exciton polaron from trion.

Amorphous magnetic films are promising for anomalous-Nernst heat-flux sensing because their low thermal conductivity can enhance the temperature gradient generated by an applied heat flux. However, amorphization often degrades electronic transport and thermoelectric properties, making it challenging to obtain a large anomalous Nernst response in structurally disordered films. Here, we demonstrate nanometer-thick amorphous Fe-Sn films as high-sensitivity anomalous-Nernst heat-flux sensing materials. By systematically controlling composition and thickness, we find that amorphous Fe-Sn nanofilms combine a large anomalous Nernst response with low thermal conductivity, resulting in a heat-flux sensitivity of 0.37 um/A. This value exceeds the sensitivities reported for both amorphous magnetic thin films and representative crystalline topological magnets. X-ray diffraction and Mossbauer spectroscopy show that the optimized films lack long-range crystallinity while retaining local Fe-Sn environments, suggesting that short-range atomic order contributes to the anomalous Nernst response in the amorphous matrix. The sensitivity is also reproduced on flexible polymer substrates, indicating compatibility with mechanically compliant device architectures. These results establish amorphous Fe-Sn nanofilms as a platform for anomalous-Nernst heat-flux sensing and provide a materials design route based on local-structure control and thermal-conductivity reduction.

Self-organization is a defining feature of living systems, with order often maintained through interactions between constituent units rather than centralized feedback. We introduce a tractable mean-field model of self-organized robustness, formulated as meta-optimization over the system's response to perturbations. The resulting interaction structure has an intuitive picture as a dynamically modulated landscape (``seascape'') whose shape is determined self-consistently to accelerate relaxation back to equilibrium. The collective dynamics follows an optimized Wasserstein gradient flow toward an attractor in the space of collective states. When communication is limited, interactions preferentially encode slowly relaxing modes and modes that are frequently perturbed. The model further shows that robust collective states are associated with flatter equilibrium landscapes and predicts a continuum of intermediate ``reservoir states'' in such systems. The model offers a perspective of self-organization as a hierarchical associative memory that operates on the scale of a collective of interacting computational units.

We systematically investigate multiphoton absorption as a polarization-resolved nonlinear optical probe of planar altermagnets (ALMs). We show that the angular harmonic of the altermagnetic spin splitting fixes the lowest optical absorption at which a symmetry-selective response appears: two-photon absorption for $d$-wave order, four-photon absorption for $g$-wave order, and six-photon absorption for $i$-wave order. In each case, there exists a polarization channel locked to the symmetry harmonic of the altermagnetic texture in which the direct $n$-photon contribution to the transition matrix element is absent. This changes the frequency scaling of the absorption rate relative to other polarization channels and provides a direct optical fingerprint of the underlying altermagnetic harmonic. Our results establish a hierarchy of nonlinear spectroscopic signatures that distinguishes $d$-, $g$-, and $i$-wave altermagnetic spin splittings beyond linear response.

I formulate unresolved soft-photon sector of QED as open quantum system. Resolved charged particles and hard photons form the system, unresolved soft photons form the environment, and basic object is the reduced density matrix. A resolved outcome $f$ of multiple hard particles and photons has probability $P_{f}(i) =\sum_n|\langle f,n|S|i,0\rangle|^2=\langle i|F_f|i\rangle$, with Kraus operators $K_n= {}_{\rm soft} \langle n|S|0 \rangle_{\rm soft}$ and effect $F_f=\sum_nK_n^\daggerΠ_fK_n$. The SK formulation places unresolved virtual and real terms in one doubled-contour expansion. At one loop they carry the same on-shell eikonal kernel with opposite signs. This elegantly organizes the QED probability: for the same observable, perturbative order, diagrams, and phase space, the OQS gives the same infrared-finite terms as the full-QED. The soft-photon evolution is a unitary coherent-state displacement driven by the scattering current. The equal-history identity of influence functional exactly normalizes this soft evolution; together with the soft-photon theorem it removes the IR divergent leading-soft factor from inclusive probability. I also derive explicit leading-soft QED realization of scale-parametrized Lindblad evolution on a fixed hard-branch space. Tracing an infinitesimal soft-photon shell produces diagonal jump operators whose entries are fixed by the corresponding eikonal emission amplitudes. The finite-shell map is a completely positive unital Schur channel and, in the sharp scale-invariant leading-soft regime, a dephasing semigroup of a completely-positive-divisible scale flow. The resulting logarithmic visibility slope and monotonic purity loss are off-diagonal predictions of the reduced-state description. The same controlled-displacement dilation gives the Sudakov probability, Poisson soft-photon multiplicities, and the bremsstrahlung number spectrum.

We introduce a counterwork functional generated by a sign-inverting memory-filtered effective protocol. Given an imposed protocol $λ(t)$, the effective protocol $Λ(t)$ is obtained by applying an active protocol-memory kernel to $\dotλ(t)$, rather than by invoking a passive bath response. The counterwork is then the ordinary Hamiltonian work associated with $H(Γ,Λ(t))$, so that Jarzynski's equality applies directly to it. When $Λ(t)$ reverses the endpoints of $λ(t)$, the corresponding free-energy difference satisfies $ΔF_C=-ΔF_W$, and the exponential average of the counterwork is the reciprocal of that of the original work. We derive the kernel normalization realizing this reversed displacement and show, by Jensen's inequality, that the product of the exponentials of the average work and counterwork is bounded by unity, implying $\langle C\rangle+\langle W\rangle\geq0$. Thus negative average counterwork is possible only when compensated by the average work of the original operation. We further discuss the counteroperation under incomplete thermodynamic information, showing that the robust strategy is to enforce endpoint reversal while minimizing dissipated counterwork.

We describe apparatus and experimental procedures for high stability precision measurements of levitated nanoscale particles confined in an ion trap in high vacuum. We discuss methods for particle generation and collection using electrospray emission, for rapid characterization by direct imaging of thermal motion, and for transfer of the particle from the trap where it is collected to a separate analysis trap in order to achieve better vacuum and lower noise. In the analysis trap at high vacuum (pressure $p\simeq10^{-8}$ Torr), we employ thermostatic control of the trapped particle oscillation amplitudes, allowing long-term, precision measurements of oscillation frequencies, from which the charge to mass ratio ($Q/M$) can be deduced. Under these conditions, we achieve $Q/M$ measurement precision approaching $10^{-5}$. This sensitivity will enable, for example, investigations of the surface chemistry of $μ$m-scale levitated materials in ultra-high vacuum environments.

Surface-enhanced IR absorption (SEIRA) and surface-enhanced Raman spectroscopy (SERS) are complementary techniques that allow for ultrasensitive chemical fingerprinting. Non-invasive optical sensing would be significantly improved by a robust implementation of a reusable substrate that combines these techniques. Here, we present an electrochemically-cleanable metamaterial that enables combined real-time SEIRA and SERS in flow. This metamaterial facilitates the study of surface-adsorbed species and diffusion layers, elicits spectral shifts from changes in nanogap refractive index of 1400 nm/RIU, and delivers ultrasensitive analyte detection. Combining SERS and SEIRA clarifies molecular (electro)chemical transformations and tracks changes in selection rules and symmetry breaking at the analyte-electrode interface. This development in enhanced multimodal spectro-electrochemistry is suited for multiple domains, including understanding charge transport mechanisms and interfacial dynamics at electrodes, and is capable of real-time flow monitoring for a wide range of molecular processes.

Spin textures are key for emergent magnetic phenomena such as topological protection and underpin novel spintronic device paradigms based on racetrack memory, logic gates, and neuromorphic computing. Using a coherent diffractive imaging technique called vector ptycho-tomography, in combination with algorithms that are robust to noise, we image the 3D magnetic texture of skyrmion and Hopf topologies with no prior assumptions about the sample. This directly reveals experimentally for the first time an extended 3D skyrmion lattice, including the domain wall shape, topological charge, helicity, and Hopf index. Our findings demonstrate experimentally that dipole stabilized skyrmions in Fe/Gd multilayers exhibit barrel-shaped skyrmion tubes with a twisted helicity, transitioning from N$é$el-type winding at the surfaces to both clockwise and counterclockwise Bloch-type winding in the bulk, that can also be described as fractional hopfions. We image a lattice of 24 skyrmions with topological charge 1, average depth-dependent domain wall width of 23 to 40 nm, depth-dependent twisted helicity from $\pm$155$°$ to $\pm$30$°$, and fractional Hopf index of $\pm$0.3. Over 10 TB of data were analyzed to yield a fully-resolved 3D reconstruction over a >0.4 $μ$m$^3$ volume, with high fidelity down to the Nyquist limit of 8 nm. This method fills a key gap in the current landscape of magnetic imaging by enabling high-resolution, element-specific 3D reconstructions of full-field extended spin textures - offering a new route for exploring the topological complexity of magnetic materials in three dimensions.

We compute the specific absorption rate of a uniaxial single-domain nanomagnet driven by an alternating magnetic field by two methods: i) direct numerical integration of the stochastic (Langevin) Landau--Lifshitz--Gilbert equation (the LLL approach), and ii) linear response theory (LRT) based on the Debye susceptibility with the Néel relaxation time $τ_\mathrm{N}$. We first analytically show that both methods are equivalent for small magnetic field amplitude, and then compute their deviation $Λ\equiv \mathrm{SAR}_{\mathrm{LLL}}/\mathrm{SAR}_{\mathrm{LRT}}-1$ as a function of the magnetic field amplitude for two temperatures chosen on opposite sides of the Debye resonance. One of the main results is that the sign and magnitude of $Λ$ are governed by the dimensionless product $ωτ_\mathrm{N}$, in addition to the linearity parameter $ξ=μ_{s}B_{0}/k_{B}T$ for the easy-axis geometry considered here. Indeed, below resonance ($ωτ_\mathrm{N}<1$), linear response theory overestimates the specific absorption rate. In contrast, above resonance ($ωτ_\mathrm{N}>1$, the regime typical of blocked nanoparticles), linear response theory can underestimate the specific absorption rate by up to $\sim70\%$ at $ξ\sim2$. We expect this work to provide quantitative guidance for the use of linear response theory in magnetic hyperthermia and related nanoscale heat-transport problems, and to serve as a single-particle benchmark for extensions to many-spin and interacting systems.

Superconductor digital electronics and quantum computing with superconducting qubits are promising next-generation computing technologies. When cooled down or operated in the presence of a nonzero background magnetic field $B_r$, superconducting thin films comprising the circuits can trap magnetic vortices that can degrade circuit or qubit performance. In this work, we report a practical solution for eliminating flux trapped during cooldown in ambient magnetic fields, $B_r\leq 60$ $\upmu$T, based on controlled local thermal gradients and moats, etched holes in the superconducting films of the circuit. Thermal gradients created by integrated on-chip resistive heaters move vortices towards the moats, where they become trapped away from circuitry regions and pinning sites. Using magnetic imaging and electrical circuit readout, we demonstrate that this approach is capable of removing magnetic flux trapped during field cooling and magnetic flux nucleated by circuit operation. If used in an environment with basic magnetic shielding, this solution is capable of suppressing all magnetic flux in a large-scale circuit, overcoming one of the long-standing challenges preventing high-performance scalable computing using superconductors.

We study a non-reciprocal XY model on a square lattice, in which spins interact with their nearest neighbors through vision-induced anisotropic interaction. Such anisotropy breaks rotational symmetry and leads to the pinning of the spin orientation along preferred lattice directions. We systematically characterize this phenomenon for different interaction kernels, including modulated, sinusoidal, von Mises, and hard vision-cone couplings, and for two classes of microscopic update rules: Glauber and Langevin dynamics. A central result of this work is the identification and detailed analysis of two distinct contributions that naturally arise in the Langevin formulation, which we refer to as the reactive and the proactive term. We derive the corresponding equations governing both local fluctuations and the global orientation, and use them to characterize the mechanisms responsible for directional pinning. We show that both reactive and proactive contributions can generate global pinning, whereas their role in determining local pinning depends on the specific interaction kernel and may differ qualitatively. Our analysis clarifies the distinction between local and global pinning, explains the emergence of preferred lattice directions in the different models considered, and reconciles apparent discrepancies reported in previous studies. More generally, it provides a microscopic framework for understanding lattice-induced orientational selection in non-reciprocal XY models.

Conventional rheodielectric studies of dipolar suspensions primarily examine flow-induced variations in the principal permittivity components. In contrast, an asymptotic solution of the perturbed Fokker--Planck equation for orientable Brownian dipoles under weak flow predicts the emergence of off-diagonal permittivity components that are linear in the relative flow strength. For planar shear flow, these terms exceed the corresponding higher-order diagonal corrections, leading to a rotation of the principal dielectric axes. This previously unrecognized rheodielectric response suggests new possibilities for flow-controlled dielectric and electro-optical functionalities.

We demonstrate strong coupling between propagating spin wave modes and microwave photons in superconducting resonator-magnetic thin film hybrid circuits. By fabricating the resonator directly on yttrium iron garnet thin films grown on rare-earth-free Y$_3$Sc$_2$Ga$_3$O$_{12}$ substrates, we achieve strong coupling of both Damon-Eshbach and backward-volume spin wave modes to the resonator, with coupling strengths exceeding both the magnon and photon damping rates. Furthermore, we observe nonreciprocal spin wave radiation of the hybrid magnonic mode in the Damon-Eshbach configuration, highlighting the potential for incorporating intrinsic spin-wave nonreciprocity into hybrid magnonic systems. These results open new avenues for integrating spin-wave magnonics with cavity magnonics, and for harnessing spin waves for potential applications in quantum information science.

Loop-erased random walks (LERW), the $O(n)$-model at ${n=-2}$ and Laplacian random walks (LRW) are three realizations of the same random process. While this equivalence holds on any graph, renormalization is possible only via the $O(-2)$-model. To generalize LRWs to $b$-LRWs or to Diffusion Limited Aggregation (DLA), a field theory directly on the Laplacian growth process is necessary. Here we construct an exact lattice action for LRWs and show that its perturbative expansion equals that of LERWs. We then generalize this approach to $b$-LRWs and DLA.

We report the temperature dependence of the substrate-induced magnetic anisotropy in continuous Ni thin films deposited at room temperature using magnetron sputtering on a 128° Y-cut LiNbO3 substrate. Ultrathin films exhibit a pronounced temperature dependence in magnetization, with as-grown films showing isotropic behavior and an unconventional hysteresis branch crossing at relatively low temperatures, whereas no branch crossing is observed in annealed films over the investigated temperature range, indicating suppression of the underlying mechanism. Temperature-dependent magnetization measurements show that post-deposition annealing establishes uniaxial anisotropy with well-defined easy and hard axes, whose strength evolves with temperature due to the nature of residual strain governed by both lattice mismatch and coefficient of thermal expansion (CTE) mismatch between the Ni film and the 128° Y-cut substrate, independent of film thickness for both 5 nm and 100 nm films. The emergence of this anisotropy following annealing indicates enhanced magnetoelastic coupling at the film-substrate interface, as the magnetic response begins to follow the anisotropic strain imposed by lattice mismatch and CTE mismatch.

Chiral-induced spin selectivity (CISS), in which electrons transmitted through nonmagnetic chiral materials exhibit strong spin-dependent transport, has attracted growing interest for spintronic applications. However, a quantitative understanding of CISS remains elusive, partly because most previous studies rely on single-band models. In this work, we theoretically investigate multi-band effects on magnetoresistance (MR)-CISS, which is typically observed in experiments using magnetic conductive atomic force microscopy. To evaluate the spin polarization in MR-CISS, we simulate the nonequilibrium steady-state current using the Gorini-Kossakowski-Sudarshan-Lindblad master equation. We find that spin polarization exceeding 25% can be achieved for realistic inter-band spin-orbit coupling strengths in the presence of on-site Coulomb interactions. These findings highlight the crucial role of inter-band spin-orbit coupling in the mechanism of CISS.

Recent studies have shown that the orientation of crystallographic c axis of Tellurium thin films can be controlled using picosecond long laser pulses. This method provides spatially programmable control of the crystal orientation and is therefore highly attractive for practical applications in functional optical and electronic devices. Previously, it was suggested that laser-induced selective melting and recrystallization can cause the laser-induced reorientation. However, this interpretation remains inconclusive due to limited data. To clarify the mechanism, here we systematically study Te samples under different irradiation conditions. We find that the threshold fluence for inducing optical reorientation depends on the number of laser pulses. The results agrees well with a minimal kinetic model based on the Arrhenius law. Using the model developed, we investigate the condition required to control the optic axis in other two-dimensional materials, such as black phosphorus, WTe2, and SnSe. These findings provide a guide for developing functional electro-optical devices based on anisotropic materials.

In this work we develop an integrable perturbation theory for the defocusing modified Korteweg-de Vries kink solution based on the squared eigenfunction expansion associated with the underlying Zakharov-Shabat scattering problem. We derive the completeness relation for the squared eigenfunctions appropriate to the kink background, establish the adjoint structure needed to handle perturbations of both the continuous and discrete spectral components, and obtain explicit evolution equations for the perturbed kink parameters at leading order. The study of the first order correction shows that perturbations generically produce a radiative shelf in front of the kink. We also apply our results to certain physically relevant perturbations and show that the predictions are consistent with direct numerical simulations.

Strongly dipolar gases, such as dysprosium, erbium and thulium, exhibit dense Feshbach spectra whose level statistics have been associated with quantum chaos arising from couplings among many molecular channels. Here, we combine a precise calibration of the Feshbach spectrum of $^{162}$Dy with spectroscopic measurements of the differential magnetic moments of bound states associated with more than 80 resonances between 0 and 30 G. These magnetic moments provide an eigenstate-sensitive probe of the molecular states underlying the resonance spectrum. We find that the level statistics are not uniform: resonances associated with states near the center of the magnetic-moment distribution display enhanced level repulsion, whereas those near the lower edge remain close to Poisson statistics. Our results reveal hidden structure within the chaotic dysprosium Feshbach spectrum and identify molecular-state composition as a key ingredient in the emergence of quantum chaos in strongly dipolar scattering.

Based on the original Miller-Robert-Sommeria theory, we explicitly compute a statistical equilibrium of two-dimensional turbulent flow on a sphere for a generic initial vorticity field introduced in a previous study. The macroscopic vorticity field corresponding to the obtained statistical equilibrium has a quadrupole structure. The resulting quadrupole structure is topologically consistent with the final state of the long-term time integration of the vorticity equation. However, the statistical equilibrium does not predict the formation of concentrated vortices as seen in the time integration. We also calculate statistical equilibria for the initial vorticity field with a planetary vorticity term, and find a change of statistical equilibria from quadrupole states to zonally symmetric states as the angular velocity of the sphere increases. The quadrupole statistical equilibria show nearly linear relations between the macroscopic vorticity and the macroscopic stream function, implying that higher-order Casimir invariants are virtually ineffective even when all Casimir invariants are considered. The discrepancy between the equilibria and the time integration results emphasizes the importance of mixing barriers, which prevent the relaxation of the evolving vorticity field to the statistical equilibria and allow the point-vortex-like dynamics of coherent vortices to persist.

This paper investigates the stability and large post-critical dynamics of an inextensible spinning fluid-conveying pipe with pinned-roller supports. Replacing the pinned-pinned support of the extensible counterpart with a sliding support removes the axial-stretching restoring mechanism and fundamentally changes the governing equations of motion. Derived here for this configuration, these equations contain a different set of nonlinear terms -- arising from the inextensibility constraint and the bending curvatures rather than the single axial-stretching term -- that drives a post-critical regime with large deflections. The regime is analysed with two complementary methods. The first is a Galerkin discretisation in which the bending curvatures are Taylor-expanded to ninth order, shown to be the lowest order resolving the post-critical amplitude; the standard cubic truncation overestimates the deflection significantly by missing the geometric stiffening from inextensibility. The second is a modified Hencky bar-chain model with a global angular description: a closed, $n$-independent matrix framework with exact trigonometric kinematics, directly implementable in any standard programming environment with matrix routines and adaptable to both extensible and inextensible configurations through a single boundary-condition reduction. The linearised dynamics give an ellipse-like stability boundary in the flow-velocity--rotational-speed plane with semi-axes $U=π$ and $Ω=π^{2}$; three damping regimes are identified, including a high-rotation instability driven by rotating damping. Close agreement between the two methods across linear-stability, bifurcation, and time-history comparisons confirms the ninth-order Galerkin truncation and establishes the modified Hencky bar-chain as a reliable general-purpose discrete framework for spinning fluid-conveying pipes.

People's opinions can change both from their interactions with each other and from their interactions with media sources. Bounded-confidence models (BCMs) of opinion dynamics provide one framework to study such dynamics. In a BCM, the nodes of a network are agents with continuous-valued opinions, and these agents interact with each other via the edges of the network. In this paper, we extend the original Deffuant--Weisbuch (DW) BCM by incorporating influence from two media sources -- one with a positive value and one with a negative value -- to capture the effects of a polarized media landscape. We show both numerically and analytically that our extended DW model exhibits drifting behavior in which a large cluster of opinions shifts toward one of the media agents. We analyze how the drift trajectory and speed depend on the model parameters, and we identify conditions in which drift is promoted or suppressed. Our results provide insight into how competing media sources can influence collective opinion formation in social systems.

Kinetic analyses of experiments often require coarse-grained descriptions, but complex systems rarely conform to the widely used modeling assumptions of Markovianity and thermodynamic equilibrium. Memory is indeed a general and often inevitable consequence of coarse-graining. Markov state models (MSMs) are a popular choice of coarse-grained description, but require microstate assignments -- which are rarely experimentally tunable -- to macrostates that minimize memory. Generalized master equations (GMEs) circumvent this limitation of MSMs by explicitly capturing memory. However, GMEs are difficult to parameterize and usually formally approximate in the experimentally relevant discrete-time setting. Here we introduce a maximum-likelihood-based procedure to parameterize formally exact, physically feasible, discrete-time generalized master equations from experiments and simulations in and out of equilibrium. By adapting algorithms typically used in optimal transport, we construct physical-constraint-satisfying conditional-maximum-likelihood estimators of both exact Nakajima-Zwanzig memory kernels and time-convolutionless GME propagators in discrete time. Applying these estimators to three examples -- experimental recordings of Förster-resonance energy-transfer in an ion channel, experimental nanoparticle tracking of a processive molecular motor, and simulated folding of a benchmark protein domain -- we recover kinetic parameters including relaxation rates, irreversibilities, dwell times, and first-passage times. These results establish discrete-time GMEs as a physically and statistically principled alternative to MSMs for kinetic analyses of experimental and simulated biomolecular systems.

We present a fast, high-order algorithm for the free-space fractional Fokker-Planck equation (FFPE) in arbitrary spatial dimension. Its fundamental solution, corresponding to a Dirac-delta initial condition, is obtained from the explicit Fourier representation by applying a sum-of-Gaussians (SOG) approximation to the nonseparable stretched exponential, using its complete monotonicity as the Laplace transform of a one-sided $α$-stable density. Each Gaussian term is an ordinary heat kernel and therefore factorizes across spatial coordinates. On a tensor-product grid, the separated form can be assembled in $O(MdN)$ work and storage, rather than forming all $O(N^d)$ grid values, where $M$ is the number of Gaussian terms and $N$ is the number of points per dimension. We prove an a~priori error estimate for the pure-fractional fundamental solution and give a parameter-selection procedure for prescribed accuracy over specified ranges of space and time. In numerical experiments the method achieves more than ten digits of relative accuracy, with $M$ growing only logarithmically in the inverse tolerance, and maintains this accuracy in dimensions up to $d=10^{5}$. This exceeds the dimensions reached in comparable radial-quadrature tests, where the integrand becomes increasingly oscillatory as the dimension grows. Because the method represents the fundamental solution as a separated sum of heat kernels, any initial datum given as a finite sum of tensor products can be evolved in closed form using only one-dimensional convolutions. This yields a computable class of high-dimensional solutions that is amenable to error analysis, and tensor neural networks provide one possible way to construct such separated representations for more general data.

Bistable driven-dissipative systems near bifurcations can exhibit noise-activated switching between steady states. Here, we investigate how quantum vacuum fluctuations induce such switching in a biased optical parametric oscillator (OPO), a nonlinear system with intrinsic bistability. We show how microscopic quantum fluctuations driving macroscopic transitions can be controlled with an external bias field that reshapes the OPO steady-state metapotential. We derive analytical expressions for the average switching time and validate them through simulations of the OPO field distribution and inter-state probability flow under bias injection. We further examine how switching depends on bias strength, pump gain, and optical nonlinearity. Our findings clarify how quantum noise can shape macroscopic dynamics and provide a foundation for noise-assisted photonic machine learning and probabilistic quantum gates.

Young's double-slit experiment is conventionally deemed a spatial phenomenon emerging from free-space transport. In this Letter, we invert this perspective to demonstrate that Young's interference can be accessed directly as a pure momentum-space observable. Using a perfect-crystal diffraction to project the field's reciprocal-space profile immediately downstream of the aperture, we resolve the complete hard X-ray double-slit fringe structure without any propagation arm, focusing optics, or imaging detector. This direct capture of the field's invariant momentum marginal establishes a compact, lensless, and propagation-free approach to coherence diagnostics, proving that the fundamental physics of wave interference can be detached from real-space propagation.

Extreme adaptive optics (AO) is necessary for high contrast astronomy at scales of the habitable zone of nearby systems. We seek to evaluate wavefront sensors that approach fundamental limits of wavefront sensing, enabling adaptive optics systems to run faster or on fainter targets. We present the phase-induced amplitude apodisation Zernike wavefront sensor (PIAA-ZWFS): an adaptation of the conventional Zernike wavefront sensor (ZWFS) that leverages lossless apodisation of the pupil to concentrate the starlight in the focal plane. We optimise and evaluate the sensor with a differentiable modelling framework, drawing on concepts from Bayesian experimental design to minimise the variance of a maximum likelihood estimator that uses the system in the high Strehl regime. Our architecture shows state-of-the-art performance in simulation for different apertures, bandwidths, photon fluxes and source sizes, closing the gap to the fundamental limit by a factor 10 (2.5) compared to the conventional ZWFS (optimised ZWFS) in a typical photon-limited case. For extended sources, we show that even an ideal point source sensor rapidly becomes sub-optimal, and our system outperforms it for stellar diameters larger than 0.8λ/D. We verify that these gains do not come at the cost of dynamic range with either linear or non-linear reconstructors. Finally, we present a proof that there must be a trade-off between the information gained about amplitude and phase errors for any wavefront sensor. The PIAA-ZWFS is a viable wavefront sensor operating near the fundamental sensitivity limits.

Plankton fix about 40 gigatons of carbon annually, using photosynthesis to convert $\text{CO}_2$ into $\text{O}_2$ and carbohydrates. These solutes are exchanged with the ocean in a diffusive boundary layer around the organism called the phycosphere. Here, we study how organisms can increase their carbon influx and outflux by actively mixing the surrounding fluid. By developing exact analytical expressions validated by stochastic simulations, we determine the enhanced diffusivity of phycosphere particles as a function of mixing activity, and their resulting fluxes and concentration fields. Hence, we find that plankton can significantly increase their uptake and photosynthetic turnover. Moreover, we find that the phycosphere radius is enlarged both by increased metabolism and by increased diffusive transport further from the organism. These results provide new biophysical insights into marine microbial ecology, with important implications for global carbon capture and climate change.

In this article, we report on the design and the experimental evidence of a Half Maxwell Fish Eye (HMFE), for Silicon Photonics and working at telecommunication wavelength. It is designed by implementing a Graded Photonic Crystal operating in the non-resonant metamaterial regime. The results of 3D Finite-Difference Time-Domain simulations (FDTD) show an excellent broadband focusing capacity. It has been urther fabricated via the Silicon On Insulator (SOI) platform for its compatibility with CMOS technology. Experimentally, its performances are firstly investigated by the means of a fan-shaped set output waveguides. Next, Scanning Near-Field Optical Microscopy (SNOM) characterisation confirms the wavefront curving inside the HMFE lens. Quantitative analysis of the SNOM results demonstrates its excellent focusing performances: the Full Width Half Maximum (FWHM) is $0.466λ_0$ at $λ_0=1550$nm, while the thickness of the lens is $3.18λ_0$.

Observing the decisions and actions of others provides social information that can inform decisions such as whether to follow. We consider a model where agents simultaneously gather stochastic private information, each deciding once sufficiently confident. Observed decisions and indecision provide social information that triggers discrete waves of collective response: a first decision causes others to update and potentially follow, whose decisions in turn provide further social information, generating successive waves. We explore this model across a range of group sizes and report three main findings. First, social information leads to faster and more accurate decisions than individual decision-making, but agent-level accuracy is maximised at a finite optimal group size. This contrasts with the accuracy of the majority choice, which increases monotonically with the number of agents. Second, waves frequently fail to resolve collective indecision, particularly for smaller groups and when the first decision is incorrect, leaving a subgroup of agents unconvinced. Third, these remaining undecided agents are systematically biased and make less accurate subsequent decisions, with this inaccuracy growing with group size.

Evolutionary intermediates connect observed proteins, but the sequence of steps that produced them is rarely recoverable from extant data alone. Here we ask what can, and cannot, be inferred about such intermediates from the endpoints. Using generative sequence landscapes as controlled models of protein-family evolution, we benchmark data-driven reconstruction against ground-truth simulated trajectories. We find that the best point prediction is not necessarily the most faithful evolutionary reconstruction: maximum-likelihood intermediates can be residue-wise accurate yet statistically atypical, whereas conditional sampling better captures the ensemble of plausible histories. Predictability is limited by the topology of the landscape. Constrained, low-mutability regions preserve information about the path, while permissive high-mutability regions open many alternative routes and erase path-specific memory. We also show that sequence divergence alone is an insufficient measure of elapsed evolutionary time; incorporating endpoint mutability provides a more reliable way to place intermediates in the landscape. These results recast intermediate reconstruction as a calibrated probabilistic problem. Rather than seeking a single "true" sequence, data-driven models should identify when endpoints contain evolutionary information, and return realistic ensembles.

Here we combine empirical network analysis with agent-based modelling to understand how different ways of structuring belief systems may affect the polarisation drive, and how the diversity of belief systems in Europe may result in different polarisation trajectories. Using the 2016 European Social Survey, we infer belief networks across 23 European countries via a Bayesian algorithm, revealing that belief systems are predominantly organised around immigration, LGBT rights, and economic interventionism, reflecting the influence of populist discourse across the continent. We further verify a Western-Eastern divide across the national belief networks: in Western European countries, left-right self-identification is a more reliable predictor of broader belief alignment, whereas in Eastern Europe this relationship breaks down. By applying these empirical belief networks into a sociologically grounded agent-based model, we further show that polarisation is amplified by high individual belief rigidity and low susceptibility to social influence, and that cross-country differences in polarisation levels mirror the same geographic divide observed in belief network topology. These findings establish belief networks topologies as a structural driver of political polarisation, with implications for understanding and anticipating polarisation dynamics across diverse European contexts. We find that populations are not polarised when little attention is placed on maintaining internal coherence and polarisation levels are moderate when high attention is placed in both keeping internal coherence and agreement in beliefs with others.

Modern play-by-play data make it possible to test long-standing intuitions about basketball with the same statistical rigour now routinely applied to other professional sports. Using play-by-play data from 7,054 regular-season and 504 playoff NBA games spanning the 2020-2025 seasons, we provide quantitative insights into scoring patterns and the performance of individual players and teams through methods from statistics, network science, and complexity science. Our findings offer an evidence-based perspective on in-season and in-game performance that can inform coaching strategies, player evaluation, and tactical decision-making.

How are psychosocial profiles, mental health, and brain functional connectivity related? Studies have been dedicated to unraveling the associations of social support perception and neural functional connectivity. Additionally, personality traits have been explored by examining brain networks. Research on mental health has been developed using a broad range of methods and different approaches. However, little attention has been devoted to understanding how personality traits and social variables are related, and to what extent these components are reflected in brain functional connectivity and mental health outcomes. In this work, we aim to address these complex relations by using data from the Human Connectome Project, both from surveys and resting-state fMRI. The survey data includes personality traits measures and self-reported social support-related variables, which we will refer to as inner- and outer-self, respectively. It also includes data on mental health outcomes. Using z-score standardized measures, we analyze correlation matrices to evaluate the association between the inner- and outer-self domains. Our results show that the social indicators are more evidently grouped by impact on social experience than by the duality of inner-outer selves. Using a $k$-means clustering algorithm, we separate individuals into two groups according to social profiles. When confronting these results with the mental health outcomes, we show that the more socially desirable cluster exhibited a higher score on positive aspects such as life satisfaction and purpose in life. In the functional brain connectivity, we observe that the cluster with a more socially beneficial profile exhibits lower interconnectivity, especially in the default mode network. The pipeline we present uses a combined analysis of both fMRI and psychosocial variables, which could open the path for more extensive analysis.

Ancestral sequence reconstruction (ASR) is a powerful approach for studying molecular evolution and the emergence of protein function. Yet most ASR methods assume that sites evolve independently, neglecting the epistatic constraints that shape protein structure, stability, and function. This simplification affects both ancestral inference and its evaluation: maximum-a-posteriori reconstructions may over-concentrate probability into a single over-idealized sequence, whereas independent posterior sampling can generate implausible or poorly functional ancestors. Here, we introduce a coevolution-aware ASR framework that combines standard phylogenetic inference with Direct Coupling Analysis (DCA), thereby preserving site-wise ancestral uncertainty while enforcing residue-residue constraints learned from extant protein families. To benchmark the method, we develop a controlled forward-evolution framework based on a DCA evolutionary sampler, allowing reconstructed ancestors to be compared with known ground-truth sequences generated under realistic epistatic constraints. Applied to beta-lactamases and DNA-binding domains, the approach improves reconstruction when ancestral states are epistatically constrained, and yields ensembles of candidate ancestors that are both phylogenetically consistent and statistically compatible with natural protein families. This framework bridges the gap between single-sequence MAP reconstruction and unconstrained posterior sampling, providing a practical route toward ancestral reconstructions that better reflect the coupled nature of protein evolution.

The development of problem-solving competence (PSC) among high school students is foundational for preparing resilient and adaptive citizens. Generative artificial intelligence (GenAI) can support this process, but it may also encourage students to offload part of the cognitive work that is necessary for deep learning. While the individual effects of GenAI use are increasingly studied, its collective consequences for competence development within classroom environments remain underexplored. In this study, we use an agent-based model to simulate the evolution of PSC in a high school physics classroom, where students complete tasks individually, in collaboration with peers, or with the support of GenAI. By comparing classrooms with and without access to GenAI across different peer-network structures, we show that GenAI use can diminish competence development and increase the share of students remaining in lower competence tiers. These results suggest that the educational impact of GenAI should be assessed not only through individual learning outcomes but also through its effects on collective competence dynamics.

Ego networks capture the variety of structural patterns in the social interactions of individuals. Recently it has been shown that ego networks in online settings display universal patterns of tie strength distributions, but it is unclear how constraints such as spatial proximity and bounded social bandwidth affect such generic behaviour in offline settings. Here, we analyse the time evolution of interaction activity in ego networks constructed from offline face-to-face and colocation data, compare them to online communication networks, and explore simple cumulative advantage models that capture the varying preferentiality of individuals for specific social ties. We find that patterns of preferentiality at the population level are similar for online and face-to-face networks, but not for colocation data, suggesting that the latter is a poor proxy of social network structure. We also provide evidence that empirical ego networks exhibit a bandwidth in the way communication events are allocated across connections. A model implementing this notion uncovers evidence of universal scaling between the tie preferentiality and bandwidth of individuals, common to all online and offline systems explored. Our findings strengthen our understanding of the fundamental mechanisms governing human communication and help disentangle the internal and external factors shaping tie evolution across social contexts.

We present $\texttt{SMaSH}$, a $\texttt{Mathematica}$ package to do spinor helicity computations in four spacetime dimensions $\href{https://github.com/aakash-kmr/SMaSH}{\text{(github)}}$. It can handle massive spinor helicity computations with explicit little group indices which is a novel feature. It can also handle massless as well as off-shell spinor helicity variables. It is designed to compute perturbative computations; it comes with predefined three point amplitudes and propagators for any masses and spins (arXiv:1709.04891). It can implement the high energy limit over an expression, check the discrete $\tt{C,P,T}$ transformations, compute contact terms and impose gauge invariance for any scattering process. We have shown the usage of such functions for computing gauge invariant Weinberg minimal amplitudes (arXiv:2506:12431, arXiv:2504:06343). The package can also generate both real and complex numerical kinematics for any $n$-point scattering for arbitrary masses and energy scales by implementing the $\tt{RAMBO}$ algorithm. It is also rich with basic spinor helicity manipulations like Schouten simplification, Clifford algebra manipulation, conversion between spinor helicity and Lorentz vectors, derivative w.r.t. spinors and their scalars, helicity scaling etc.

Unidirectional guided resonances (UGRs) have attracted considerable attention owing to their remarkable ability to radiate exclusively in one direction from single-layer planar photonic lattices without metallic components. Conventionally, UGRs have been understood to require either broken in-plane $C_2$ symmetry or interband coupling between distinct modes. Here, a new mechanism for realizing UGRs is presented, in which out-of-plane radiation near the fourth stop band is mediated by two distinct channels associated with the first and second Fourier harmonics. When the radiation components from the first and second Fourier harmonics cancel each other out in both the upward and downward directions, nonradiative bound states in the continuum emerge. By contrast, UGRs arise when such cancellation occurs only in one direction. By tuning the lattice parameters, the positions of these UGRs can be controlled, allowing them to merge at the $Γ$ point. The two-channel radiation-cancellation model enables UGRs without relying on in-plane symmetry breaking or interband coupling, thereby relaxing lithographic constraints and simplifying device design. It also provides a useful framework for controlling topological singular states in higher-order photonic bands.

Integrated quantum photonics requires bright, high-fidelity photon-pair sources capable of spectral multiplexing, correlation control, and circuit-compatible extraction. Cavity-enhanced spontaneous parametric down-conversion (SPDC) enhances pair generation, but triply resonant operation imposes stringent pump-signal-idler spectral-alignment constraints. Moreover, the trade-off between intrinsic generation and coincidence-to-accidental ratio (CAR) does not capture the usable output flux, which depends on photon extraction. Here, we demonstrate a single-pass-pumped, resonator-enhanced quantum frequency comb (QFC) source based on a periodically poled lithium niobate photonic-crystal Fabry-P$é$rot microresonator. The device yields intrinsic and loaded brightnesses of 69.9 and 1.88 MHz/$μ$W, respectively, and a maximum CAR of 16,000. Frequency-resolved measurements reveal 461 cavity-defined bins spanning 1495-1570 nm, and loaded spectral brightness approaching 4.29$\times$10${}^9$ pairs/(s$\cdot$mW$\cdot$nm). In particular, tuning the single-pass pump deterministically selects correlated frequency-bin pairings within the fixed QFC grid while preserving brightness and pairwise coincidence rates. We further separate intrinsic generation from output photon-pair flux, revealing the loaded-brightness-CAR relation. Together, pump-selectable bin pairing and extraction-aware loading point to tailored SPDC QFCs as chip-integrated nonclassical light resources for multiphoton-interference quantum simulation and programmable frequency-bin quantum information processing.

Nanoscale mechanical oscillators exhibit a plethora of nonlinear phenomena with promising applications for the sensing and clocking of processes down to atomic length scales. Oscillator dynamics are typically probed by electrical or optical means, providing only limited access to the spatial profile of the oscillator motion. Here, we introduce event-based convergent beam electron diffraction for the spatio-temporal mapping of nanoscale mechanical resonators in ultrafast transmission electron microscopy. Employing an optically driven silicon membrane resonator at various driving strengths, we gain access to nonlinear processes with increasing complexity, ranging from a simple Duffing behavior to nonlinear multimode coupling and period-doubling bifurcations. The time-resolved diffraction probing approach supports a spatial resolution down to a few nanometers and a temporal resolution of 5 ns and provides quantitative information on the local membrane bending. Because the diffraction signal responds to local displacement gradients, which become more pronounced as resonators shrink, this approach offers a route toward probing nonlinear nanomechanics at the atomic scale.

Cryogenic energy storage is vital for frontier technologies including deep-space exploration and quantum computing, yet conventional electrochemical energy systems fail below ~230 K due to frozen ion migration. While relaxor-based dielectric capacitors provide high efficiency at room temperature, the intrinsic freezing/growth of polar nanodomains at extended cryogenic regime limits their applications with deteriorated hysteresis losses. Here, we realize superior cryogenic energy-storage performance by designing unit-cell-level disordered dipole-glass state in Pb0.6Sr0.4ZrO3 thin films with composition near antiferroelectric-paraelectric phase boundary. The antiferroelectric-derived dipole-glass introduces enhanced unit-cell-level complexity of dipole interaction that suppresses long-range ferroelectric order. This enables ultralow-hysteresis operation (efficiency > 88%) down to 4 K, delivering record-high energy density (211 J/cm^3) at 9 MV/cm, stability over 10^8 charge/discharge cycles and microsecond-scale charge/discharge capability. This work establishes a dipole-glass paradigm for cryogenic dielectric capacitors, opening a new avenue to highly-efficient energy-storage systems with broad applications in frontier nanoelectronics.

Understanding how individuals repeat social interactions over time is a central problem in the analysis of temporal networks. In social systems, repeated interactions shape processes such as information diffusion, collective coordination, and the emergence of social structure. Existing measures of egonet persistence often conflate genuine behavioural regularities with structural effects such as node degree, making it difficult to distinguish meaningful temporal correlations from random mixing. In this work, we introduce the Neighbourhood Persistency Criterion (NPC), a statistically grounded framework for quantifying egonet persistence across time. NPC combines classical similarity measures with tailored null models controlling for network topology and interaction weights. We apply this framework to high temporal resolution face-to-face interaction networks collected at four Computational Social Science conferences using the SocioPatterns platform. Our results reveal a common behavioural structure across events, characterised by an exploration$\unicode{x2013}$exploitation trade-off in social interactions. While many individuals alternate between both strategies, others exhibit stable interaction patterns throughout the event. Importantly, these behaviours show little systematic association with socio-demographic attributes, suggesting that interaction strategies are shaped primarily by contextual factors rather than stable individual traits. NPC thus provides a flexible and interpretable tool for studying egonet persistence in temporal networks and social systems.

Afshar's double-slit experiment probes wave--particle complementarity by placing a wire grid at the dark fringes of a downstream interference pattern while retaining an imaging basis that appears to preserve which-path information. Here we propose and analyze a time-reversed Young--Afshar configuration in which the corresponding null test is transferred from the downstream field plane to the source-label plane of a time-reversed Young interferometer. In this reciprocal geometry, a point-addressable source illuminates a double slit, while the detector remains fixed. The observed fringe is therefore not a single-shot spatial intensity pattern, but a detector-conditioned response reconstructed by scanning the source coordinate. Consequently, a null in this pattern is not a node of a freely propagating field; it is a source label for which the coherent two-slit transfer amplitude to the selected detector vanishes. A mask placed at such source-plane labels is invisible to that detector when both slits are open, yet becomes visible when either slit is opened alone. We develop the scalar Fresnel model, derive the source-space null condition, introduce a detector-conditioned null-mask loss, and examine how this loss evolves under a tunable which-path marker. The source-space visibility and path distinguishability satisfy the standard duality relation, so no violation of complementarity is implied. The essential new feature is instead a reciprocal, detector-conditioned form of complementarity: Afshar's field-space transparency is replaced by response-function transparency in a reconstructed source basis.

In this paper, we experimentally demonstrate the microscale direction control of thermoosmotic flows using a focused-laser heating. The key is the off-center laser irradiation on an immobilized light-absorbing microparticle, which generates a nonuniform, asymmetric heat source. The resulting thermo-osmotic flows are evaluated using the optically trapped particle tracking velocimetry (ot-PTV), presented in our preceding paper (T. Tsuji, et al., Physical Review Fluids 11, 034901 (2026)). It is shown that the flow characteristics can be modulated by the ionic strength of a sample solution and/or the surface molecular coating of the substrate. In particular, the significance of ionic strength on thermo-osmotic flows are discussed based on the surface potential of the substrate measured by frequency-modulated atomic force microscopy.

We analyze an often used closure model for multi-material hydrodynamics where pressure temperature equilibrium (PTE) is assumed for every state; emphasis is placed on tabular equations of state. This multi-material model is often referred to as the four-equation model. The identification of the admissible set is presented and is proven to be convex, setting the foundation for development of invariant-domain methods for this model. A novel, robust, and efficient method is presented for solving the highly nonlinear system for the equilibrated pressure and temperature with an arbitrary number of materials. Additionally, we provide a detailed analysis of the thermodynamics of the mixture model for general equations of state and prove existence and uniqueness of the pressure-temperature equilibrium solution under some thermodynamic assumptions.

Conventional machine-vision pipelines typically rely on high-quality optics that produce clean, human-interpretable images, and optical design has therefore been driven by image-level criteria such as resolution, aberration correction, and pixel fidelity. However, such optics are often impractical for size-, cost-, or form-factor-constrained applications, where compact meta-optics offer an attractive alternative but operate under strict physical efficiency limits. We propose CODA, a co-design framework that optimizes a continuous-density meta-optic front-end for frozen-model recognition using differentiable image formation and adjoint-gradient updates of Maxwell-based simulations. CODA directly optimizes the cross-entropy loss of a fixed zero-shot CLIP classifier without learned reconstruction, image signal processing, or image-fidelity auxiliary objectives. In a two-dimensional simulated imaging benchmark on ImageNet-100, CODA improves CLIP ViT-L/14 zero-shot accuracy from 53.75 $\pm$ 3.57$\%$ with a focal-concentration baseline to 65.41 $\pm$ 3.99$\%$. The optimized optics further transfer without re-optimization across CLIP, SigLIP, and DINOv2 on ImageNet-100, CIFAR-100, and Food-101. These results demonstrate that, under constrained meta-optic imaging, downstream recognition can be improved by aligning optical design with frozen vision-model objectives rather than conventional image-formation criteria.

Increasing the circulating power in gravitational-wave detectors to the megawatt level is essential for future sensitivity improvement, but this is critically limited by optomechanical parametric instabilities. Current mitigation strategies are projected to be inadequate against instabilities when circulating power reaches a megawatt. Optical feedback offers a novel independent paradigm to mitigate parametric instability. In this Letter, we report the first demonstration of optical feedback control in a full-scale gravitational wave detector. We successfully suppressed an unstable mode at 10.428 kHz, reducing the parametric gain from R = 2 to R < 0.02. This work validates optical feedback control as an effective mitigation scheme for kilometre-scale interferometric gravitational-wave detectors, providing an effective strategy to allow detectors to reach the megawatt level.

We design and implement a Microscale Topological Surgery (MTS) algorithm to detect and enforce topological transitions in two-dimensional tracked interfaces. The method combines classical Lagrangian tracking with an intermittent topological processor that: (i) constructs Eulerian snapshots from which an interface family with microscale-resolved topology is extracted, (ii) infers adjacency topology between dual Lagrangian and Eulerian interface families, and (iii) performs interface surgery to stitch the two families together across microscale defect regions. A novel long-time nonlinear alternating-shear flow is introduced, in which repeated stretching and folding generate rich multiscale interface dynamics with filamentation at microscales. Using the MTS algorithm and a posteriori geometric and material diagnostics, we compute and visualize microscale filament-breakup dynamics. Error analysis and scaling studies demonstrate second-order geometric convergence and optimal computational scaling of the MTS algorithm, with topology-processing costs comparable to those of the underlying Lagrangian evolution. Ensemble simulations generated by pseudo-random perturbations of the flow further reveal coherent droplet size distributions and statistically robust filament-breakup dynamics.

Silicon photonics enables integration of optical components using standard semiconductor processes, greatly improving data communication bandwidth and energy efficiency. However, photonics integrated circuits (PICs) face unique security challenges, such as counterfeit or tampering threats, that conventional electronic security methods do not address. We propose a novel hardware fingerprinting technique that embeds two dimensional photonic crystal patterns into the density control filler regions of a PIC. Each PhC pattern is designed to resonate a specific visible to near infrared wavelengths, producing a distinctive optical signature (based on wavelength, polarization, and incident angle) for each device. Finite difference time domain (FDTD) simulation using ANSYS Lumerical is employed to optimize nanostructure dimensions and spacing so that each device's reflection/absorption spectrum contains unique narrowband peaks. No extra fabrication steps or materials are required beyond standard lithography, keeping costs low. The embedded nanostructures have sub-50nm precision, making forgery extremely difficult. Our method yields a high resolution, scalable fingerprint for silicon photonic chips, enabling cost-effective device authentication and improved supply chain security.

Photonic lanterns (PLs) are waveguides that convert multi-mode input light to single-mode outputs. Wavefront sensing (WFS) and spectroscopy using a PL have been demonstrated, but PL simulations and experiments show significant mismatches. For the WaveDriver project, a proposed Habitable Worlds Observatory pathfinder that uses a PL for WFS as well as for integral field spectroscopy, we manufactured an optical waveguide consisting of an array of seven 19-port PLs in one device. We present laboratory characterization of the individual PLs, consisting of measurements of the principal modes at each PL input using digital off-axis holography. We compare our mode measurements to simulations to assess the variation in the PL manufacturing process. We discuss expected WFS performance in the WaveDriver configuration.

The photonic lantern (PL) is a focal-plane wavefront sensor (WFS) that can be used for second-stage control of extreme adaptive optics (AO) systems. While the number of sensed modes and the dynamic range with respect to each mode have been relatively well characterized, little attention has been paid to the PL's sensitivity, i.e. how measurement noise impacts the accuracy of PL wavefront reconstruction. We compute the PL's sensitivity to photon noise as a function of spatial frequency, and compare it to existing WFSs, using simulations as well as experiments on the muirSEAL testbed. We further assess these metrics in the case where only a subset of PL ports are available for wavefront sensing. In this configuration, the remaining ports are used to spatially and spectrally reconstruct the observed scene using algorithms such as SPADE. Using more ports for wavefront sensing enables greater aberration sensitivity but leaves less spatial information for image reconstruction. This allows us to trade off between fewer samples with smaller aberrations and more samples with larger aberrations. This work sets the stage for AO system design incorporating the PL as a joint WFS and imager.

Digital platforms increasingly shape how people experience and navigate cities, linking virtual information seeking with physical mobility. Despite this interdependence, online and offline activities are often studied separately in urban mobility research. This paper introduces the Feedback Network, a computational framework that captures interactions between cognitive activity in digital environments and behavior in physical space. Using Google Search and Location History data from the same individuals, collected through a data donation framework in Budapest, Hungary, between 2018 and 2022, we examine how online search patterns and offline visitation behavior co-evolve. We combine semantic and spatial analytical approaches. Radius of gyration is adapted to measure variation in geographic mobility and semantic exploration, enabling comparison between physical movement and online cognitive dispersion. A Feedback Network models transitions between search-related and location-related activity clusters and is evaluated using Concentration Entropy, which measures whether behavioral flows are concentrated around routine pathways or distributed across exploratory transitions. The results show that online exploration is more concentrated than offline mobility, suggesting narrower and more repetitive semantic interests, while physical movement remains relatively diverse. Persistent linkages between search and visitation activities related to retail and business services indicate stable cognitive-spatial behavioral loops. The COVID-19 pandemic disrupted spatial routines more strongly than cognitive exploration, widening the gap between digital engagement and realized movement. The findings demonstrate that urban mobility depends on the interaction between informational exposure and spatial encounter and provide a foundation for Hybrid Digital Twins of Society.

Near-zero-index (NZI) media are of great interest for controlling light-matter interactions, but homogeneous NZI materials in the telecom band remain limited and often suffer from significant losses. Photonic crystals provide an attractive alternative due to their tunability and use of low-loss constituent materials. This work presents a predictive framework for designing epsilon-near-zero (ENZ), mu-near-zero (MNZ), and epsilon-and-mu-near-zero (EMNZ) photonic crystals. We demonstrate an equivalence between Dirac cones and EMNZ behavior, showing that a Dirac cone at the Gamma point is both necessary and sufficient to obtain an effective EMNZ response under normal incidence. This result implies that the presence of a Dirac cone fully determines the NZI response of a photonic crystal, independent of the material composition. The prediction is based on mode symmetry universality and effective medium theory. The approach enables switching between ENZ-MNZ and EMNZ regimes through geometric tuning of the unit cell. It is validated on triangular and rectangular lattices. Overall, this work provides a predictive design strategy for NZI photonic crystals, replacing trial-and-error optimization. It may also impact metamaterials in the telecom regime and applications in quantum communication.

Two-dimensional real-space imaging of vibrational polariton transport in planar Fabry--Pérot microcavities is numerically simulated via the mesoscale cavity molecular dynamics approach, which self-consistently propagates $\sim\!2\times10^4$ realistic molecular simulation cells on a two-dimensional grid coupled to the same number of cavity modes. Beyond the well-known polariton ballistic-to-diffusive turnover in the linear response regime, these atomistic simulations reveal a nonlinear freezing mechanism of vibrational polariton transport, i.e., under strong pumping of the upper polariton, the initially ballistically propagating upper polariton completely freezes and localizes energy to molecules at specific locations. This mechanism originates from pump-induced breaking of the in-plane translation symmetry: significant molecular excitations at the pulse hot spot broaden the polariton density of states, thus funneling population to the $k_{\parallel}\rightarrow 0$ band edge with vanishing group velocities.

Selectively controlling the dynamics of molecular enantiomers underlies advances across chemistry, biology, and physics, yet direct imaging of enantiomer-specific motion has so far remained elusive. Here, we image ultrafast enantioselective orientation dynamics in isolated chiral molecules. Unidirectional coherent rotation induced by a femtosecond laser-pulse pair generates equal and opposite out-of-plane orientations of the two enantiomers. Applying this scheme to 2-methyloxirane, we follow the rotational wave packets by time-resolved Coulomb explosion imaging with two orthogonally arranged detectors. The measured angular distributions reveal that the unidirectional rotation is identical for both enantiomers, while the out-of-plane orientations are mirror images that persist through both early-time quasi-classical and quantum dynamics regimes, in quantitative agreement with simulations. We demonstrate that full angular distributions provide richer dynamical information, with some qualitatively different distributions yielding similar orientation factors upon integration. Our approach opens a route to real-time observation and control of chiral dynamics in the gas phase.

We present a study on the time-domain propagation of electromagnetic waves in dielectric materials modeled by a nonlinear Debye medium with random perturbations. Polynomial Chaos Expansions are employed to transform the random nonlinear Debye polarization model into a deterministic framework. We extend the Yee discretization to the resulting coupled system, establish second order accuracy, and verify convergence numerically. We investigate the sensitivity of nonlinear properties to uncertainty, particularly when the amplitude of the input signal is large. Given the challenges in manufacturing where uncertainties can cause optimal parameters to vary and potentially disrupt nonlinear effects, our approach incorporates these uncertainties within the simulation. This can enable the model-based design identification of realizable materials that maintain their desired effects despite variations. The findings from this study contribute to a deeper understanding of wave propagation in complex media, with potential implications for applications in optical communications, material science, and electromagnetic wave control.

We establish a self-consistent analytical model demonstrating that trapped non-wetting liquid phases in narrow capillary channels can be successfully unpinned via non-resonant, second-order acoustic streaming (acoustic wind) coupled with background static drive gradients. Moving away from boundary-guided or resonant mechanisms, our approach exploits the bulk acoustic-wind force density generated by the steady-state momentum flux of attenuated first-order linear wave interactions. By expanding the hydrodynamic equations up to second order, we determine the critical assisted acoustic wave amplitude required to break capillary pinning thresholds and derive an explicit formulation for steady transport velocity under viscous wall constraints. Furthermore, incorporating both boundary-layer wall effects and bulk core thermo-viscous dissipation reveals a natural mathematical optimum condition where the spatial absorption coefficient matches half the inverse distance to the target droplet ($α= 1/2x_0$). This condition is then numerically validated and cross-correlated against legacy industrial frequency baselines, providing a fundamental theoretical framework for minimizing transducer power requirements while maximizing localized mobilization velocities in geological pore networks. Finally, we demonstrate that this optimal operational frequency scales inversely with the transmission distance, providing an analytical framework to optimize downhole acoustic tools according to the spatial damping constraints of the specific formation rather than relying on rigid hardware parameters.

Background: Both between- and within-subject variations in circadian timing matter for health. If lifestyle changes could be used to regulate circadian timing, they would offer accessible and scalable routes to chronotherapy, but this link remains unclear under real-life conditions. Here, we explore how lifestyle 'traits' (such as typical wake time) and 'states' (day-to-day deviations from traits, such as waking up later than typical) explain between- and within-subject variation in acrophase (peak time) of the circadian rhythm of heart rate (CRHR). Methods: We collected free-living wearable data (smartwatch, continuous glucose monitor) from healthy volunteers for up to 4 weeks. The CRHR was derived from activity-adjusted heart rate, and acrophase was defined as time-of-day at daily CRHR peak. Sleep, food, and physical activity 'factors' were calculated and split into traits and states. Using a linear mixed-effects model, we tested how traits and states associate with between- and within-subject acrophase variance. Findings: Data from 105 healthy volunteers (66 female, age = 42.5 $\pm$ 15.7 years) spanning ~2000 days (18.8 $\pm$ 8.30 days each) were analysed. Traits were substantially more influential than states, explaining 42.3% versus 0.9% of total acrophase variance. Accordingly, traits explained 86.5% of between-subject variance, whereas states explained only 1.8% of within-subject variance. Sleep, food and physical activity factors contributed both jointly and uniquely, and lifestyle timing mattered most. Interpretation: Between-subject lifestyle traits explained acrophase better than within-subject lifestyle states. This asymmetry, alongside the considerable overlap between factors, supports sustained, holistic, timing-focused lifestyle adjustments as chronotherapy targets, testable through future interventional studies.

Strategic behaviour in queueing systems has been studied extensively in the behavioural queueing literature, but almost exclusively for systems that admit closed-form expressions for the cost or utility experienced by a strategic user. Evolutionary game theory offers a mature framework for analysing populations whose individual payoffs depend on the composition of the population itself, and would in principle apply to a much wider class of queueing systems; its application has, however, been constrained by the same closed-form requirement. We introduce Discrete Event Population Updates (DEPU), a general algorithmic framework that couples a single long run of a discrete event simulation (DES) directly to an evolutionary population update rule, removing that constraint. We present two implementations: Discrete Event Replicator Dynamics (DERD), which follows an Euler discretisation of the replicator dynamics equation, and Discrete Event Moran Replacement (DEMR), which maintains a finite population updated via Moran-style copying events. Both are applied to a multi-server jockeying model for which no closed-form fitness expressions are available. On the jockeying model considered, DEPU reaches comparable precision tens of times faster than the standard practice of nesting short simulations inside an outer evolutionary loop, and because each operating point then costs only a single simulation run it also makes systematic parameter sweeps tractable. This brings the toolkit of evolutionary dynamics within reach of any system a modeller can build in a discrete event simulator.

In this paper, we study reactive strategies in repeated additive games between two players with finitely many actions. Reactive strategies condition only on the opponent's previous action, making them one of the simplest ways players can respond to past interactions. Additive games include important models of cooperation, such as the donation game and games with a punishment option. We show that, for this class of games and strategies, the conditions for symmetric Nash equilibria reduce to a system of linear equalities and inequalities in the strategy parameters, allowing us to characterise all such equilibria. We establish a one-to-one correspondence between non-empty subsets S of the action set and equilibrium classes, which we call S-supporting equilibria. These are equilibria that use exactly the actions in S when playing against themselves. As a special case, we recover the well-known equalizer strategies as the equilibria supported on the entire action set. To assess which equilibrium classes are most evolutionarily relevant, we complement our analytical characterisation with simulations of social learning dynamics. We find that their prevalence is determined by two factors: how likely they are to be generated and how robust they are against invasion.

Efficient evaluation of sequence data likelihoods and their high-dimensional gradients on phylogenetic trees improves inference under both maximum-likelihood and Bayesian frameworks. Here, we present BEAGLE 4.1, a high-performance library for statistical phylogenetics that incorporates new algorithms to evaluate these gradients on phylogenetic trees. We also provide new hardware implementations for both likelihoods and gradients supporting ARM NEON intrinsics and optimized matrix multiplication units -- called tensor cores -- on NVIDIA graphics processing units (GPUs). We benchmark the performance scaling of the library across a number of patterns and taxa on multi-core CPUs and GPUs, and compare the speedup afforded by NVIDIA and AMD GPUs as well as performance scaling with an increasing number of GPUs. We show that multi-core CPU implementations provide up to a fourfold speedup over single-threaded CPU implementations and up to an tenfold speedup for nucleotide and codon models, respectively, with performance generally improving as the number of taxa and site patterns increases. GPUs outperform multi-threaded CPU implementations for a realistic number of patterns, even for nucleotide models with a small state-space size of 4, while for codon models they provide substantially higher performance gains even for a single pattern or four taxa. Tensor cores on GPUs provide up to 2-fold speedup relative to standard CUDA cores for codon models. Using NEON instructions on ARM CPUs affords up to a $\sim 1.3$-fold speedup over non-SIMD implementation with the speedup going down to 1.1-fold at 8 CPU threads. We provide these new algorithms to evaluate the gradient and efficient hardware implementations for both likelihood and gradient calculations through BEAGLE 4.1, such that they can be readily integrated into phylogenetic software packages.

Stress is an adaptive response coordinated by neural and physiological systems. While acute stress can enhance survival, chronic stress drives structural brain changes, cognitive dysfunction, and increased psychiatric risk. At the cellular level, chronic stress shifts the excitatory-inhibitory (E/I) balance of prefrontal pyramidal neurons toward inhibitory dominance, yet the mechanisms underlying these alterations are still unknown. We here investigate possible mechanisms causing inhibitory dominance using recurrent neuronal networks trained on a working memory task. Chronic stress is modelled as a modulation in synaptic strength or neuronal activity, systematically comparing eight candidate operators against three experimentally motivated signatures of stress-induced prefrontal dysfunction: inhibitory dominance, excitatory hypofunction, and impaired task performance. These signatures are all recovered by a single stress mechanism, stronger inhibitory-to-excitatory synapses. Contrasting naive networks with resilient networks trained under the stress mechanism, we find that resilience training not only preserves task performance under stress, but also confines the network to the same dynamical subspace and energetic regime with and without stress. This resilience comes at a cost: resilient networks generalise less well when the task requires longer memory than seen during training, indicating that resilient networks find a specialised solution tuned to the trained regime. This trade-off between resilience and generalization performance persists across stress magnitude and network size, offering a computational analogue of the shift toward rigid, habit-like behaviour reported in animal following chronic stress.

EEG-based emotion recognition is widely used in affective computing but suffers from poor generalization due to domain shifts caused by inter-subject variability, dataset differences, and recording conditions, especially in cross-dataset settings. Conventional unsupervised domain adaptation methods require source data, which is often unavailable due to privacy constraints. Although source-free UDA addresses this limitation, existing methods still struggle with large domain gaps, noisy pseudo-labels, and unstable adaptation. To address these challenges, we propose an enhanced source-free unsupervised domain adaptation (SF-UDA) framework for cross-dataset EEG emotion recognition. The framework introduces a non-contrastive predictive coding-based self-supervised pretraining strategy to learn robust and transferable EEG representations by modeling temporal dependencies in a reconstruction-based manner. During adaptation, we estimate target-domain structure through class-wise clustering and prediction disagreement, and optimize the model using a dual-stage strategy consisting of Multi-Loss Adaptive Regularization and Localized Consistency Learning, improving stability and neighborhood consistency under noisy pseudo-labels. We also propose a lightweight test-time training mechanism that enables selective online updates for uncertain samples using predictive reconstruction loss and entropy minimization. Experiments on DEAP, SEED, and DREAMER show consistent improvements over state-of-the-art SF-UDA methods, achieving 69.56% and 63.03% accuracy on SEED and DREAMER when trained on DEAP, and 61.38% and 68.90% when trained on SEED.

Ensuring reliable communication for mission-critical vehicles in dynamic environments with limited infrastructure is a significant challenge due to interference and spectrum scarcity. This paper investigates a UAV-assisted vehicular communication framework that leverages multi-antenna beamforming and dynamic bandwidth allocation to provide prioritized and interference-mitigated wireless links. Vehicles are classified according to their service priority, with each class assigned a distinct frequency band to reduce interference. Within each class, optimized beamforming further minimizes transmission overlap and enhances spectral efficiency. The optimization problem is solved using an alternating optimization framework, incorporating two beamforming strategies: one based on successive convex approximation (SCA) and the other derived in closed form. Numerical results indicate that the proposed scheme outperforms baseline approaches that optimize only bandwidth allocation or beamforming in terms of overall system performance. Among the two joint optimization methods, the closed-form solution achieves higher sum rates and generally requires less transmit power, while also exhibiting lower computational complexity compared with the SCA-based approach.

High-resolution passive microwave imaging is important for numerical weather prediction, disaster monitoring, and oceanographic studies, but kilometer-level spatial resolution remains difficult to achieve because of aperture limitations and the high complexity of large interferometric arrays. This paper proposes a beamforming microwave interferometric radiometer (BF-MIR) for high-resolution passive microwave imaging. BF-MIR employs beamforming-capable antennas as interferometric elements in a large sparse array. The enlarged spatial-frequency sampling interval reduces the required number of elements and the cross-correlation burden, while a large aperture-to-sampling-interval ratio factor (ASRF) array design enables narrow-beam spatial filtering to suppress brightness temperature (TB) aliasing caused by spatial-frequency under sampling. In addition, beamforming enables dynamic beam steering across multiple pointing directions, thereby compensating for the limited instantaneous coverage of narrow beams. A beamforming interferometric imaging model is established, and the relationships among spatial resolution, radiometric sensitivity, and effective field of view are analyzed. An image-domain Shift-Accumulate method is further introduced to analyze aliasing, based on which an aliasing suppression strategy is developed. In addition, a three-element proof-of-concept prototype provides preliminary experimental validation of dynamic beam interferometric measurement and dynamic beam observation modes. These results indicate that BF-MIR is a promising architecture for further spaceborne high-resolution passive microwave imaging.

Vehicle license plate recognition is an integral component of intelligent transportation systems. In this work, we present an embedded real-time license plate recognition system customized for developing countries. We address the challenge of handling complex, unstructured traffic scenes with diverse vehicle types while implementing the system on an embedded platform for low-cost deployment. Our method consists of license plate detection on a multi-vehicle image, followed by character recognition on the detected license plates. Both steps use lightweight convolutional neural networks to balance accuracy and efficiency. We also introduce the SL-LPR dataset of Sri Lankan road images, which contains a variety of vehicle types and traffic conditions typically seen in developing countries. On this dataset, the license plate detection and character recognition models achieved 93.6% mAP and 87.88% accuracy, respectively, and were competitive against larger models on several public datasets. To achieve real-time performance in a resource-constrained embedded environment, we applied low-bitwidth quantization using the Brevitas library and implemented FPGA acceleration for the models using the FINN framework. The end-to-end system can operate at 11.5~FPS when implemented on the Xilinx Kria KV260 platform. These results demonstrate that our system is effective for real-time license plate recognition on an embedded device, even in complex traffic scenarios. The SL-LPR dataset is available for research use at: https://github.com/sl-lpr-uom/SL-LPR.git.

Lighting is the dominant energy load in indoor farming, yet most deployed systems still rely on fixed rule-based or schedule-based control. We present LightFARM, a predictive lighting control framework that couples crop illumination with battery-free sensing for more energy-efficient indoor farming. LightFARM combines finite-horizon predictive control with compact models of photosynthesis, thermal dynamics, and sensor energy state. The controller adjusts lighting intensity to balance photosynthetic benefit, electrical power consumption, thermal safety, and sensing-energy feasibility. A key design feature is that the same light-emitting diode (LED) fixtures serve both as the photosynthetic light source for crops and as a controllable energy source for self-powered sensor nodes. We implement LightFARM in a real indoor basil cultivation system and evaluate it through two independent 12-day cultivation trials. Compared with a conventional rule-based baseline, LightFARM reduces lighting energy consumption by approximately 41% and improves energy productivity from 36.1 to 52.9 $\mathrm{g\,kWh^{-1}}$ and from 41.1 to 60.2 $\mathrm{g\,kWh^{-1}}$ ($\approx 46.5\%$ on average). These results suggest that energy-cooperative predictive lighting control is a promising approach to improving indoor farming efficiency under practical resource constraints, while explicitly accounting for the trade-off between energy savings and crop yield.

Reliable, low-latency communication is critical for real-time monitoring and control in modern Smart Grids (SGs). The emergence of 5G networks, with enhanced reliability, significantly lower latency, and native support for massive machine-type communication, offers strong potential to enable advanced grid applications such as state estimation (SE) and fault detection. While existing studies investigate 5G for SG use cases, most rely on simulations or analytical models; experimental validation using real hardware and SG data remains limited. This paper fills this gap by presenting a fully experimental validation of real-time SE over a commercial 5G network using a 5G-based multi-node testbed built with Raspberry Pi (RPi)-based SG nodes and a Typhoon Hardware-in-the-Loop (HIL) real-time simulator. We first characterize 5G communication performance using simulated SG data under varying reporting rates and deployment environments by evaluating Key Performance Indicators (KPIs) such as end-to-end delay, jitter, and frame loss. Experimental results show that the worst-case mean delay observed for the 5G is approximately 6.5x lower than that of our previous LTE cat-M study at the corresponding reporting rate. We then stream real-time voltage, current, and phase-angle measurements-generated by an IEEE 4-node feeder model in Typhoon HIL simulator-to a remote Phasor Data Concentrator (PDC) for SE and fault detection. Results demonstrate that 5G-enabled measurements support accurate SE under both steady-state and dynamic load variations. Furthermore, fault-detection experiments confirm reliable and prompt fault detection, with detection delays as low as 0.80 s.

In this paper, we present an experimental evaluation study of the Alternating Direction Method of Multipliers (ADMM), which is a widely used technique in the distributed optimization of power distribution networks. The focus of this study is on how real 5G communication performance affects ADMM in a fully experimental platform that features commercial 5G connectivity and real-time control. The ADMM-based Distributed Optimal Power Flow (DOPF) problem is solved using the IEEE 123-bus unbalanced distribution feeder subdivided into five areas, each managed by a local controller implemented on a Raspberry Pi. To mitigate the impact of the communication network variability, we propose a delay threshold-based mechanism that yields a 7.75% reduction in convergence time compared to a no-threshold baseline. We also devised a policy to dynamically update the threshold value based on communication and computation conditions, achieving a 26.42% reduction in the convergence time compared with the static optimal threshold. These results demonstrate the potential of adaptive, communication-aware control strategies for real-world Smart Grid (SG) deployments.

This letter presents the design and operation of an electronically reconfigurable pinching antenna (E-pinching antenna) and examines its capability to establish controllable millimeter-wave links that can circumvent blockages. The antenna consists of a low-loss rectangular dielectric waveguide that leaks energy through modular varactor-loaded elements to form tunable radiation points. A copper reflector ensures unidirectional radiation, thereby boosting forward-link efficiency and spatial selectivity. Full-wave simulations demonstrate that the radiated power and transmission coefficient to a receiving patch antenna can be dynamically tuned by adjusting the varactor capacitances. A multi-user scenario is investigated by activating two radiation modules along the waveguide to serve spatially separated receiving antennas isolated by a metallic partition (blockage). Simulation results confirm the capability to establish links to both LoS and NLoS users with minimal propagation losses. The proposed architecture enables a scalable and electronically controllable distributed antenna platform for reconfigurable wireless systems with enhanced blockage mitigation.

Large-aperture wave receivers can contain a large number of candidate sensor locations, antenna ports, or measurement blocks, while hardware and processing constraints allow only a subset to be activated. In this paper, receiver selection is formulated as a finite-SNR modal-recoverability problem, where the selected subset is required to support reliable recovery in every direction of a prescribed modal subspace. However, large trace, log-det, or codebook-distance values alone do not ensure that every prescribed modal direction is resolved. Specifically, the proposed framework consists of three nested recoverability criteria for individual modal degrees, a joint target subspace, and target modes in the presence of active non-target modes. The third criterion, the Schur-complement information floor, provides an exact worst-direction posterior-error interpretation. We further show that stricter recoverability criteria require at least as many active receiver accesses and derive tests that identify reliability targets that remain unattainable even when all available accesses are activated. Next, we specialize the framework to finite spherical-wave sampling and compare greedy receiver-selection rules. Numerical results demonstrate that global log-det is generally more access-efficient at moderate reliability floors, whereas Schur-based selection is more effective at stringent floors. While this paper is motivated by holographic and XL-MIMO receivers, the framework can be applied to general sampled wave systems.

Force-Free Electrodynamics (FFE) describes the evolution of the electromagnetic field in magnetically dominated plasmas, but ceases to be hyperbolic once the magnetic dominance condition $F^{ab}F_{ab}>0$ is lost. We demonstrate that, after the loss of magnetic dominance, FFE may be replaced by a theory of null fields, $F^{ab}F_{ab}=0=F^{ab}\tilde{F}_{ab}$, characterized by the condition that its principal null direction is tangent to a geodesic congruence. In flat spacetime, this theory may be equivalently stated as the condition that the field satisfies $\vec{B}^2-\vec{E}^2=0=\vec{E}\cdot\vec{B}$ and the integral curves of the drift velocity $\vec{E}\times\vec{B}/\vec{B}^{2}$ are straight lines. We develop the general structure and the properties of this theory and test it against 1D PIC simulations using the collision of planar symmetric Alfvén waves. We find that the force-free combined with the null continuation shows remarkable macroscopic agreement with PIC simulations, including the birth and evolution of the null region ($F^{ab}F_{ab}=0$) and the formation of a current sheet. As an independent test, we apply the null continuation to Adhikari's type-changing solution, an exact FFE solution exhibiting finite-time loss of magnetic dominance, and also find macroscopic agreement with 1D PIC simulations.

We present a method for simulating operator dynamics in out-of-equilibrium quantum systems. Due to the rapid growth of complexity in these systems, this is typically speaking an intractable task. However, exceptional progress has been made in recent years to sidestep this barrier, with the introduction of a number of methods that make use of a truncation of the simulation to low-weight (the number of non-trivial terms in a Pauli string basis expansion) observables, which turns out to be a good approximation for many dynamical quantities of interest, e.g., two-point infinite-temperature correlation functions between local operators. In this work, we extend this idea to a leaner truncation protocol, truncating operators based on their diameter - that is, the size of the region on the lattice on which they are non-trivially supported. Using existing analysis for generic circuits we argue that this kind of truncation protocol is physically well-motivated, and show via extensive numerical simulations for a number of systems of interest (here, the kicked Ising model and the Heisenberg XXZ model) that it is effective, and allows us to efficiently and accurately extract local correlation functions and transport properties.

We continue the study of particle production from gravitational inhomogeneities in the early Universe. Focusing on sources active on sub-horizon scales, we derive general expressions relating particle production to the unequal-time two-point function of the stress-energy tensor sourcing scalar, vector and tensor metric perturbations. The resulting particle yield probes the time-like support of this correlator, and in the tensor case the same object controls gravitational wave emission when evaluated on the light-like support. This establishes a phenomenological link between dark matter production and gravitational wave signals, allowing the dark matter mass to be related to the amplitude of the stochastic gravitational wave background. Our results show that, on sub-horizon scales, particle production from inhomogeneous metric backgrounds practically reduces to gravitational scattering. This directly connects the formalism to gravitational freeze-in from the Standard Model thermal bath, while extending it to non-thermal and out of equilibrium sources. We apply the formalism to first order phase transitions and discuss the associated production from scalar and tensor perturbations. The mechanism can efficiently populate gravitationally coupled dark sectors, especially when the perturbations are generated shortly after inflation.

The curvaton provides a simple mechanism for generating strongly non-Gaussian curvature perturbations after inflation, with potentially important consequences on small scales. We study curvaton dynamics beyond the standard quadratic potential and construct the local non-Gaussian map $ζ=F(ζ_{\rm G})$ relating the curvature perturbation to an auxiliary Gaussian field $ζ_{\rm G}$. Curvaton self-interactions make the onset of oscillations field dependent and modify the effective equation of state once the curvaton enters the adiabatic regime. We incorporate these effects using the abbreviated action, which provides a compact way to connect the frozen and oscillatory regimes and exposes sources of non-Gaussianity absent in the purely quadratic case. We apply the formalism to quadratic, monomial, quartic, and cosine potentials, for which we derive the mapping $F(ζ_{\rm G})$ and show that self-interactions can either enhance or suppress the resulting non-Gaussianity depending on the potential and initial conditions. We consider non-perturbative aspects in the strongly non-Gaussian regime, and show how strong non-Gaussianity can suppress the power spectrum. As an application, we provide a bottom-up scenario in which strongly positive curvaton non-Gaussianity allows primordial supermassive black hole seeds at peak amplitudes $\mathcal{A}_{\rm pk}\sim10^{-5}$, which are compatible with the COBE/FIRAS $μ$-distortion bounds. This opens a new primordial scenario for the Little Red Dots observed by the JWST. The axion-like curvaton provides a particularly natural setting for this mechanism.

We study the composition of classically-controlled quantum instruments--the natural quantum analogue of Markov kernels. Classically, Markov kernels compose by integrating one kernel against another. Defining this composition for quantum instruments with continuous outcomes requires an integral of quantum channel-valued functions with respect to a quantum instrument. We construct this integral in the Heisenberg picture using the Okamura-Ozawa normal extension to a von Neumann tensor product. This integral recovers the expected finite formula, preserves normal complete positivity and subunitality, and provides the multiplication for a monad governing the composition of quantum instruments. As an immediate consequence, we identify the category of quantum Markov kernels as the Kleisli category of this monad.

In this article, we study the cosmic evolution of a generalized dilatonic ghost condensate field as a dark energy candidate, formulated from a Lagrangian density with two dominant kinetic terms; one linear and one of arbitrary integer $n>2$ in combination with an exponential potential, which interacts with dark matter through a source term. We analyzed three scenarios: the non-interacting situation $Q=0$ and two different interaction models, $Q\proptoρ_m\dotφ$ and $Q\propto ρ_m H$ to describe the evolution of the present universe. For each interaction $Q$, we perform a detailed phase-space analysis to obtain stability conditions and identify critical points. In all situations, the system reproduces the standard cosmological dynamics and evolves toward late-time dark energy-dominated attractors, with quintessence or phantom features depending on the sign of the coupling parameter $α$ associated with the standard kinetic term. Furthermore, a joint likelihood analysis with Cosmic Chronometers, PantheonPlus, and DESI observations is performed for two values of power $n$ ($n=3$ and $n=5$) to determine marginalized parameter constraints at the confidence levels of 68$\%$ and 95$\%$ for the different $Q-$models. For the interaction term $Q\propto \dotφρ_m$, we find that the direction of the flow of energy depends on the sign of the coupling parameter $α$ associated with the standard kinetic term. However, for the interaction $Q\propto H\,ρ_m$, the direction of the energy flow is independent of the sign of the coupling parameter $α$ and always remains negative, corresponding to an energy transfer from dark matter to dark energy.

Quantum estimation theory provides ultimate precision bounds on parameter estimation, independent of experimental setups. In this article, we apply this theoretical framework to neutrino oscillations, aiming to clarify some subtle issues and reveal the maximum achievable precision of oscillation parameters. First, taking the example of two-flavor oscillations, we clarify how the quantum Fisher information (QFI) depends on the choice of bases when the basis transformation itself involves the parameters in question. Then, for three-flavor oscillations, we compute the QFI matrix for electron and muon neutrino states in the flavor basis and derive analytical expressions and numerical results for both diagonal and off-diagonal elements. The implications of off-diagonal correlations for multiparameter estimation are discussed, and the quantum Cramér-Rao bounds on the precision of oscillation parameters for typical reactor and long-baseline accelerator neutrino experiments are obtained. Our results establish a theoretical benchmark for the ultimate precision achievable in future neutrino oscillation experiments.

In this paper, we introduce and demonstrate a simple geometric algorithm to determine which critical points, both complex as well as real, contribute to the asymptotic evaluation of multiple integrals with exponential integrands of the form $e^{ikf(\boldsymbol{x})}$ over $\mathbb R^d$, for finite $d\ge 1$ and $f$ is analytic. In so doing, the algorithm removes the need to compute the flows of $-\text{Re} (i\nabla f)$ in $\mathbb C^d$ that is required to identify such relevant critical points in Picard-Lefschetz approaches to the derivation of such asymptotic expansions. By contrast, our algorithm relies on the combination of three simple features: the values of $f$ at all the critical points plotted in the complex Borel plane, the concept of adjacency between such points derived from algebraic resurgence/hyperasymptotic approaches and the new result here of a geometric "South-East" rule. The algorithm incorporates functions $f$ that remain bounded or unbounded on $\mathbb R^d$. We illustrate this new approach with both pedagogical and advanced examples, and draw conclusions as to its importance for resolving issues associated with Wick rotations and its implications for path integrals. This is a significant step towards a systematic way of identifying instanton contributions in real-time path integrals.

The Signed Particle Formulation provides a particle-based interpretation of quantum mechanics in phase space, where quantum dynamics are represented through the creation and evolution of signed particles. A central computational challenge in this framework is the evaluation of the Wigner kernel, which generally involves highly oscillatory integrals and can become computationally demanding in time-dependent simulations. This paper proposes an analytical approximation of the Wigner kernel for one-dimensional single body quantum systems by exploiting a series-based representation of the potential function. The resulting expression provides an efficient way to approximate the Wigner kernel and the associated Gamma function, which governs the particle-generation process in the Signed Particle Formulation framework. The proposed approximation is evaluated for several Gaussian-based potential profiles, including single, double, triple, and quadruple Gaussian potentials. Numerical comparisons between the approximated and directly computed Wigner kernels and Gamma functions show that the proposed method captures the main behavior of the exact quantities while significantly reducing the computational cost. These results indicate that the proposed approximation can serve as an efficient computational component for scalable Signed Particle Formulation based quantum simulations.

In this work, we introduce a quantum inverse power iteration (QIPI) algorithm based on the quantum singular value transformation (QSVT) to target arbitrary excited states. Given an energy shift $ω$, QIPI prepares the target excited state by iteratively applying an approximation of the shifted inverse Hamiltonian $(H-ωI)^{-1}$ to a trial state. Prior quantum inverse power approaches typically relied on Fourier decompositions of the inverse Hamiltonian, with numerical quadrature used to reconstruct the transformation, but such methods are highly sensitive to hyperparameter choices and have been observed to be numerically unstable, effectively restricting their use to ground-state preparation. To enable robust excited-state targeting, we investigate two alternative transformation techniques: a Chebyshev decomposition of the inverse (Cheb-inv) and an eigenstate filtering (EF) approach based on QSVT. We find that EF-based QIPI is substantially more robust than Cheb-inv and other decomposition-based approaches due to the symmetry of the applied filtering polynomial, avoiding divergence with respect to the choice of $ω$ and efficiently suppressing off-target eigenstates even in closely spaced spectra. Numerical simulations for molecular Hamiltonians of H$_2$, LiH, and BeH$_2$ show improved convergence and enhanced access to higher excited states relative to other quantum power methods. Assuming standard oracle access to the Hamiltonian, we further provide logical resource estimates in fault-tolerant settings in terms of T gate counts, and conclude that QIPI can achieve high target state amplification with modest polynomial degrees, thereby making it a promising candidate for scalable excited-state preparation in fault-tolerant quantum chemistry applications.

We settle the global nonlinear stability problem for the family of Kerr black holes in the full subextremal range: spacetimes evolving from initial data close to those of a subextremal Kerr black hole as solutions of the Einstein vacuum equation ${\rm Ric}(g)=0$ settle down to a nearby member of the Kerr family at the rate $\mathcal{O}(t_*^{-2-ε_{\mathcal K}})$ in spatially compact regions. For the initial data, we require $\mathcal{O}(r^{-1-ε_0})$-decay for $ε_0>0$ -- more precisely, an arbitrary but finite expansion into terms $r^{-z}(\log r)^k$ where $z>1$, $k\in\mathbb{N}_0$, plus a remainder term with $\mathcal{O}(r^{-3-ε_0})$-decay. Similarly to previous work with Vasy in the Kerr-de Sitter setting, we use a generalized wave map gauge modified using gauge source terms that lie in a suitable finite-dimensional space determined by the expansion of the initial data. Like the final black hole parameters (mass and angular momentum) and the gravitational wave tail, the gauge source terms are treated as unknowns in a nonlinear (Nash-Moser) iteration scheme. We work directly with the tensorial equation and in particular do not rely on reductions to scalar equations (except insofar as a reduction to the Teukolsky equation is used in the proof of linear mode stability). This paper relies on two companion papers by the author. The first one introduces a strong form of constraint damping in the full subextremal range, which we use in our formulation of the gauge-fixed Einstein equation as a black box. The second one provides tame estimates (albeit with weak decay) for forward solutions of a general class of wave-type equations, which we show here to include the linearizations of the gauge-fixed Einstein equation arising in our nonlinear iteration scheme; these estimates are the starting point for our detailed asymptotic analysis.

We show the conservation of the superhorizon curvature perturbations at one-loop level in spatially-flat gauge, including contributions from loop wavenumbers on all scales. In contrast to previous works, we do not assume a hierarchy $k \gg q$ between the wavenumber of the loop integral, $k$, and that of the power spectrum, $q$, and we explicitly include the regime $k \lesssim q$. Taking into account the nonlinear relation between the inflaton fluctuation and the curvature perturbation with the $δN$ formalism, we show that the apparent time dependence of the one-loop curvature power spectrum cancels once all contributions, including boundary terms, are combined consistently.

Long-range forces in the early universe may lead to early structure formation and, perhaps, to primordial black holes. We generalise previous studies of fermions coupled to a light scalar field by considering general scalar-field-dependent couplings in cosmological backgrounds with a constant equation of state. We identify two broad regimes: a scaling regime, in which the scalar field oscillates around a point of vanishing fermion mass, and an asymptotic regime, in which the field evolves toward configurations where the fermions recover their bare mass. We show that the scaling regime arises from an approximate scale invariance in the scalar-fermion action, which becomes an approximate conformal invariance at late times. In the scaling regime, the ratio between the scalar and fermion energy densities is approximately constant. Our work provides a first step toward a general study of the growth of perturbations in this system.

Classical black holes have sharply defined event horizons, but quantum mechanically the horizon acquires a quantum uncertainty, called the `quantum width' by Marolf. We propose a definition of the quantum width by a physical experiment involving the last moment a signal emitted from an ingoing light ray can escape to infinity. Calculations of this observable for spherically symmetric black holes in perturbative quantum gravity reveal that the quantum width depends on the resolution of the probe, and is often much larger than the Planck scale. For example, for Schwarzschild black holes in four dimensions in a particular regime of parameters, a piece of the horizon of size $σ_\perp$ has quantum width roughly $\sqrt{l_P r_s^2/σ_\perp}$.

We compute the classical gravitational Compton amplitude at the fourth post-Minkowskian order, $\mathcal{O}(G^4)$, within the Worldline Quantum Field Theory framework. We derive the associated $N$-matrix element, which provides the gravitational-wave scattering phase shift at the same order. As a nontrivial check, we show that our result agrees with black-hole perturbation theory.

Parameterized Instantaneous Quantum Polynomial (IQP) circuits have proven useful in quantum generative learning models, particularly for binary distributions. However, when applied to non-binary datasets, they exhibit notable limitations: mapping integer values into qubit-compatible binary representations often destroys the original metric structure of the data. In this paper we aim to extend them to a qudits formulation operating on an integer mapping of the data. The IQP quantum circuit is adapted to encode each integer valued pixel into a bit-string of fixed length and quantum gates are transformed to follow the qudit formalism. As a generative machine learning approach, a suitable loss function for the circuit training and the calculation of the covariance matrix among features are developed and validated on the energy deposits from single-particle electron showers in the electromagnetic calorimeter of the CLIC detector. The method proposed in this work can be also extended to other applications that utilize quantum generative machine learning for non-binary data.

We extend Noise-Directed Adaptive Remapping (NDAR), a recently proposed heuristic meta-algorithm that leverages device noise as a computational resource, to optimization problems over discrete (integer) domains. While originally introduced for unconstrained binary optimization, the proposed generalization introduces additional gauge degrees of freedom at the logical level, such that the gauge transformation applied at each iteration is no longer unique, allowing tailoring to particular encodings or quantum hardware. We identify encoding-dependent requirements for NDAR beyond binary domains: feasibility of the noise attractor, existence of compatible gauge transformations that preserve an efficiently implementable circuit family, and a systematic way to select the transform to apply at each step. We analyze these criteria for qudit-native and for binary, one-hot, and domain-wall qubit encodings, using the Max-k-colorable subgraph problem as a running example. We demonstrate that these encodings can exhibit distinct advantages and tradeoffs when integrated within the NDAR framework, particularly in how noise-induced dynamics interact with the solution landscape and choice of encoding. Our results indicate that NDAR-guided noise considerations provide a new criterion for comparing device-level encoding choices for quantum optimization. Finally, we outline directions toward experimental realization in superconducting qudit devices and further algorithmic improvements.

We construct a new class of exact solutions describing multi-black holes in the Bertotti--Robinson spacetime, using the monodromy-matrix formalism associated with integrable sigma models. Starting from the extremal Reissner--Nordström black hole in the Bertotti--Robinson background, we derive the corresponding coset and monodromy matrices and show that they are governed by nilpotent algebraic structures. This property enables an explicit factorization of the monodromy matrix, allowing for a systematic reconstruction of the underlying gravitational solutions. We extend this construction to multi-center configurations by introducing multiple poles in the monodromy matrix, leading to Majumdar--Papapetrou--type solutions with Bertotti--Robinson asymptotics. Each center is shown to correspond to a regular extremal black hole with an $\mathrm{AdS}_2 \times S^2$ near-horizon geometry, and the asymptotic end likewise approaches a Bertotti--Robinson geometry. We further generalize the framework to stationary configurations in the Bertotti--Robinson spacetime, as well as to a broader class of Israel--Wilson--Perjés-type solutions, by considering more general nilpotent elements. Our results demonstrate that the monodromy-matrix approach provides a powerful and systematic framework for constructing multi-black hole solutions in nontrivial backgrounds, and suggest a promising route toward more general configurations.

Non-linear dRGT massive and bimetric gravities are complicated theories constructed in terms of square roots of matrices. Apart from the technical issues of successfully working with such square roots, there is also a problem of their non-uniqueness. There are claims in the literature that one should better use the principal root. This is a very reasonable conclusion. However, the motivation they give for it is that otherwise there would be non-primary square roots violating the general covariance. In this paper, I would like to show that, if properly understood, the non-primary square roots are also perfectly covariant. At the same time, I recall the relatively old observation that the real problem with such square roots lies in perturbation theory around them. In terms of matrices, it simply does not exist. In terms of the elementary symmetric polynomials used in the Lagrangian density, it is not analytic. Moreover, the non-principal square roots are more prone to getting into the complex domain.

Identifying quantum phases of matter is key to understanding strongly correlated materials, but remains a challenging task for both conventional computers and current quantum processors. Here, we introduce and implement a hybrid quantum-classical neural network for quantum phase recognition by combining a hardware-efficient parameterized quantum circuit and a feedforward neural network. We jointly train both components with superconducting quantum hardware in the optimization loop, to experimentally demonstrate a classifier for the quantum phases of surface code lattices with up to 4x4 sites in a magnetic field. To learn nonlocal features of the topological phase, we train the hybrid neural network to distinguish topological ground states of the surface code from a featureless ensemble of product states. This allows the trained classifier to distinguish topological ground states from randomly chosen product states, even when subjected to any single-qubit Pauli error. The classifier reaches accuracies above 85% in single-shot measurements, and above 99% when averaging over ten measurements. We expect hybrid neural networks such as the one presented here to be a promising approach for characterizing quantum states in scenarios where classical methods exhibit an unfavorable scaling of sample complexity.

With increasing maturity of quantum computers, standard methods for characterizing global properties of their output quantum states via direct measurements and classical post-processing are becoming increasingly impractical due to large measurement costs. Although quantum neural networks could directly process quantum states to identify underlying characteristics with reduced measurement efforts, they often require deep quantum circuits that cannot be implemented on existing devices. To overcome these challenges, we introduce a hybrid quantum-classical neural network that consists of a shallow parameterized quantum circuit, measurements, and a classical neural network. The parameterized quantum circuit performs a nonlocal transformation of the measurement basis, which is jointly trained with the classical neural network to maximize the statistical distance between data obtained by measuring different quantum states. Using supervised learning, we demonstrate that the hybrid neural network distinguishes the topological phase of the surface code from a symmetry-enriched topological phase and random product states. Moreover, this hybrid neural network reduces both inference and training sample complexities of recognizing the topological phase by approximately one order of magnitude compared to a classical neural network trained on randomized Pauli measurements. As this hybrid neural network features a shallow quantum circuit that can be readily implemented on existing quantum computers, it enables the efficient characterization of complex quantum states.

The Stage IV cosmological surveys, such as Euclid, LSST, DESI, and SKA, will deliver observational data of unprecedented volume, calling for efficient and reliable inference tools. This paper presents HIcosmo (High-performance Inference for Cosmology), an open-source JAX-based framework for cosmology inference. In HIcosmo, the forward model, distance integrals, likelihood evaluations, posterior sampling, and Fisher forecasts are all built from JAX primitives, so that gradients and Hessians of the log-likelihood are obtained directly by automatic differentiation, without any finite-difference approximation. The framework implements the $Λ$CDM, $w$CDM, $w_0 w_a$CDM, and interacting dark-energy models, and provides likelihoods for Type Ia supernovae (Pantheon+, DES-SN5YR, Union3), baryon acoustic oscillations (DESI DR1/DR2, SDSS), Planck 2018 distance priors, local $H_0$ measurements, and strong-lensing time delays. Its scope is restricted to background cosmology, with Boltzmann solvers and full perturbation-level likelihoods left to external tools. We validate HIcosmo against the reference implementation of each likelihood and against Cobaya. $χ^2$ values agree to absolute differences of $10^{-6}$-$10^{-2}$, and the marginalized constraints from the two codes differ by less than $0.2σ$ in every analysis tested. Leveraging just-in-time compilation and automatic differentiation, HIcosmo achieves about $8.7\times$ the end-to-end sampling throughput of Cobaya on CPU. As the dataset grows to survey scale, GPU acceleration over CPU reaches up to $20\times$. As applications, we present multi-probe $Λ$CDM joint constraints, dark-energy equation-of-state constraints, and Fisher forecasts for six 21 cm intensity-mapping surveys, including SKA1, MeerKAT, BINGO, Tianlai, and CHIME.

The primary challenge for space-borne gravitational wave (GW) detectors lies in extracting the weak GW signal from instrumental noise that exceeds the signal level by many orders of magnitude. Time-delay interferometry (TDI) addresses this by suppressing the dominant laser phase noise through recombination of time-delayed measurement data. The detector's response to a GW signal is represented in the frequency domain by a response function. Currently, the GW signal response is first expressed in terms of the Doppler frequency shift in a single detection arm, and this formulation is then incorporated into specific TDI combinations to derive the corresponding response function. This paper introduces a generalized formulation for TDI combinations based on multiple geometric links. By extending the representation of the laser Doppler frequency shift to include various geometric configurations, such as round-trip and non-round-trip links, we reformulate 45 second-generation TDI combinations. For several of these, the new formulation significantly streamlines their mathematical expressions and enhances physical clarity. Our results demonstrate that the proposed link-mapping rules not only enable efficient construction of response functions for these TDI combinations but also reduce computational complexity. This approach provides a reliable theoretical and algorithmic foundation for data processing in future space-borne GW missions.

We develop a novel time-evolution algorithm for matrix product states based on the recently introduced hybrid tensor network (hTN) framework. We retain the tensors close to the boundary on the classical computer and offload the highly entangled inner ones to the quantum computer. In our variant, we employ the Basis Update and Galerkin (BUG) integrator to time-evolve the classical tensors, and we develop a coupling scheme between the classical and quantum parts. Our framework admits modular combination with any quantum time-evolution method, such as (classically pre-optimized) Trotterization. The ratio of classical and quantum tensor degrees of freedom can be dynamically adjusted during the time evolution, which can be advantageous when the classical memory requirements become prohibitive. The quantum and classical components can run in parallel during a single time step and are not constrained by synchronization barriers or mid-circuit measurements. We describe the detailed steps and pseudocode for our algorithm specialized for tensor networks originating from the matrix product state Ansatz.

Axion clouds around rotating black holes form gravitational atoms whose tidal transitions can radiate gravitational waves in binaries. For strongly coupled Bohr crossings, transition radiation is governed by the outgoing two-level coherence, not by the transition probability alone. This coherence is suppressed both on the adiabatic branch and in the weak passage limit, but survives for intermediate sweep rates, producing a finite transition waveform and a localized orbital response. In more massive systems, fine and hyperfine transitions produce narrowband gravitational radiation and cumulative departures from vacuum binary waveforms. Coherent tidal crossings offer a gravitational-wave probe of axion-cloud dynamics.

We study the quantum fluctuations of the black hole horizon in three-dimensional Anti-de Sitter (AdS) spacetime. We define a precise protocol to calculate the horizon fluctuations and define a corresponding ``quantum width'' of the horizon. We relate the horizon fluctuations to boundary correlation functions via holography. Working in perturbative quantum gravity, we find that the quantum width is typically of order $(G_\mathrm{N} L_{\mathrm{AdS}}^3 )^{1/4}$, which is parametrically larger than the Planck scale. In detail, the quantum width depends on the scale at which it is measured, diverging logarithmically in the UV. Our results give the most rigorous evidence to date of gauge-invariant fluctuations at scales much larger than the Planck scale within perturbative quantum gravity.

A line-of-sight acceleration (LOSA) of a compact-binary center of mass, imparted for example by a nearby tertiary perturber, imprints a Doppler modulation on the gravitational-wave signal and provides a single-event diagnostic of dynamical formation environments. Waveform-modeling systematics -- missing higher-order modes, spin precession, or orbital eccentricity -- can mimic or mask a non-zero LOSA, making waveform accuracy a leading concern for LOSA inference. We implement LOSA corrections directly in the time domain as a remap of the strain: the treatment applies uniformly to all mode content, spin precession, and orbital eccentricity, and integrates naturally with time-domain inspiral-merger-ringdown models that best capture these effects. We deploy it in the SEOBNRv6EHM (aligned-spin eccentric) and SEOBNRv5PHM (precessing quasi-circular) models, and validate the implementation on simulated signals; we find that SEOBNRv6EHM recovers LOSA correctly on both eccentric and spin-precessing injections, while SEOBNRv5PHM yields a spurious LOSA measurement on eccentric signals. Motivated by the eccentricity hints in the neutron-star--black-hole (NSBH) event GW200105_162426 and the favorable low-mass regime of these sources, we jointly infer LOSA and orbital eccentricity for the five NSBH events in the LIGO-Virgo-KAGRA catalog, reporting the first LOSA constraints on three of them. All five events are consistent with a vanishing LOSA, $Γ\equiv a_\parallel/c = 0$; for GW200105_162426, the joint $(Γ, e)$ posterior nonetheless disfavors both $Γ$ and $e$ being zero simultaneously at 90% credibility, supporting the eccentricity hints reported in previous analyses.

We report a numerical discovery that in a more rapidly accelerating spacetime, the galactic halos on void surfaces develop stronger perpendicular alignments with the directions toward the void centers. We utilize the halo catalogs from the AbacusSummit suite of simulations for $10$ different cosmologies that include one Planck $Λ$CDM, four $w$CDM and five $w_{0}w_{a}$CDM, which share the identical initial conditions except for the dark energy equation of state. For each cosmology, we identify the voids and void-surface galactic halos at $z=0.1$ and determine the probability density functions of the cosines of the angles, $p(\cosθ)$, between the shape axes of void-surface galactic halos and the directions toward the void centers. The numerically obtained $p(\cosθ)$ is fitted to an analytic single-parameter formula derived through an empirical modification of the linear perturbation theory. Eliminating spurious signals caused by the differences in the mass and sphericity distributions of void-surface galactic halos among different cosmologies, we detect a clear net effect of dark energy on the strengths of the perpendicular alignments of void-surface galactic halos, quantified by the single parameter, $d_{t}$. Noting that $d_{t}$ has higher values in the cosmologies where dark energy has more negative pressure and evolves more rapidly, we put forth a bilinear model for the difference in $d_{t}$ between the two cases of $w=-1$ and $w\ne -1$. Demonstrating that this bilinear relation excellently describes the numerical results, we conclude that the perpendicular alignments of void-surface galactic halos should in principle be a powerful independent indicator of the dynamic nature of dark energy.

Verifiable blind quantum computing (VBQC) enables a resource-limited client to securely delegate computations to an untrusted quantum server while maintaining privacy and detecting deviations from the prescribed computation. The noise-robust VBQC protocol of Leichtle et al. achieves this through a round-based structure: the client delegates multiple computation rounds and test rounds, using the test outcomes to detect cheating while tolerating honest hardware noise. The protocol's security proof involves numerous interdependent parameters, making it non-trivial to find a valid parameter set for a given hardware noise level and security target. We formalize this as a constrained optimization problem and develop a practical framework to solve it. The framework yields the protocol parameters that minimize the number of rounds for any given setup. We derive a heuristic formula for the minimal number of rounds to help understand the scaling with noise and security targets and to provide rapid resource estimation. Since the number of rounds depends on noise while the time per round depends on hardware rate, the framework also enables optimization of rate-fidelity trade-offs to minimize end-to-end runtime. We demonstrate both applications through a case study of a trapped-ion server with a measurement-only client, showing how the client's polarization control hardware specifications translate into protocol parameters and runtime estimates, providing concrete guidance for near-term implementations.

Variational excited-state quantum algorithms fail for reasons usually studied in isolation: barren plateaus, symmetry contamination, finite-sampling instability, and hardware cost. Using one small but complete system -- H$_2$O in the STO-3G basis (12 qubits, Jordan--Wigner) -- we assemble these into a single reproducible pipeline, checking every claim against exact diagonalization. The bare qubit Hamiltonian interleaves cation ($N{=}7$) states below the neutral manifold; hardware-efficient and number-conserving ansätze stall at Hartree--Fock, an exact stationary point by Brillouin's theorem, while ADAPT-VQE escapes; variational deflation inherits the contamination and inverts the spectrum, whereas the quantum equation-of-motion (qEOM) subspace method restores the ladder to sub-milli-Hartree accuracy. Particle number is protected \emph{structurally} under shot noise, and a realistic measurement model collapses the thousands of subspace matrix elements to $\sim\!10^5$ commuting groups; a matrix-aware shot allocation then reaches chemical accuracy at $\sim\!3\times10^9$ total shots -- a thousandfold below the naive per-element estimate and reachable in days -- leaving single-circuit gate fidelity, not measurement, as the binding constraint. This work is a teaching and benchmarking reference, not a new method; all code, parameters, and figures are released.

We demonstrate single-electron trapping and detection in a two-dimensionally scalable dual-plane printed-circuit-board Penning trap. We characterize deterministic electron loading, axial damping, axial temperature, and collision-induced magnetron-radius growth at low magnetic fields. These results establish a practical platform for planar Penning traps and identify key next steps toward applications in quantum information science.

Space-based gravitational-wave (GW) laser interferometers, including LISA and Taiji, are designed to observe gravitational waves in the millihertz band and are expected to open up a frequency range that is otherwise inaccessible. The sensitivity and response of these instruments are central to their scientific goals, mission design and parameter estimation capabilities. However, they are commonly modeled as static, equilateral triangular constellations, an approximation that neglects both orbital motion and directional dependence. In this work, we systematically examine the direction-dependent response and sensitivity of dynamic LISA-like detectors over an entire year of heliocentric orbit. Based on an analytical, time-dependent heliocentric orbital model and an adiabatic unequal-arm interferometer configuration, we construct direction-dependent sensitivity curves in the Michelson interferometric channel for dynamic LISA and Taiji. We obtain analytic expressions for the angular-dependent sensitivity and demonstrate the emergence of a quadrant-like pattern in sky maps at low frequencies. We show that, relative to the static approximation, the low-frequency sensitivity varies by roughly $20\%$, which in turn produces about a $70\%$ variation in the directional dependence of the number of detectable GW sources, with even larger discrepancies at higher frequencies. Therefore, for accurate predictions of the total GW source counts and reliable parameter inference for binary systems, it is necessary to employ fully dynamic, direction-dependent sensitivity curves.

We present a phenomenological study of quantum correlations in the decay of $B^0$ mesons into a system of two vector mesons. The decay of the $B^0$ meson into two vector mesons constitutes a bipartite system of two qutrits. The entanglement entropy is used as a measure of quantum correlations in the system of decaying particles. We study the variation of the Rényi entropy with Rényi order ($α$) for the decay channels $B_s^0 \rightarrow φ\, φ$, $B_d^0 \rightarrow J/ψ\, K^{*}(892)^0$, $B_d^0 \rightarrow φ\, K^{*}(892)^0$ and $B_s^0 \rightarrow J/ψ\, φ$ and discuss the significance of entanglement entropy at different Rényi order regimes. The LHCb, ATLAS and Belle collaborations experimental measurements of complex polarization amplitudes and relative phases are used as input for our analysis. A comparison of entanglement entropy for all the $B^0$ meson decay processes, with both vanishing and non-vanishing phases, reveals a strong phase dependence of the entropy. We further present the results of Hartley entropy (Max-Entropy), von Neumann entropy, collision entropy, and min-entropy, each corresponding to different values and limits of the Rényi order. The comparison between the branching fractions of the decay processes and the von Neumann entropy shows a connection between entanglement and decay dynamics, indicating the role of weak and strong interaction in generating quantum entanglement. In addition, we evaluate several other entanglement measures, including linear entropy, I-concurrence, tangle, negativity, logarithmic negativity, Schmidt coefficients, and Schmidt rank for different $B^0$ meson decay processes. Our study demonstrates that entanglement measures provide useful insights into the underlying decay dynamics and may serve as important tools for understanding quantum correlations in high-energy particle physics processes.

Cavity optomechanical systems have emerged as a promising platform for quantum sensing. Quantum mechanics imposes a standard quantum limit (SQL) on the force-sensitivity for the standard homodyne phase quadrature measurement of the cavity's output field. In this paper, we investigate ways to enhance sensitivity beyond SQL for weak-force measurement by adopting a variational homodyne quadrature readout and quantum squeezing. Our study reveals a remarkable improvement in the force sensitivity of a cavity optomechanical sensor at a suitable homodyne angle, compared with standard phase quadrature detection of the cavity output field within a specific frequency band. We show that the force sensitivity can be further improved by intra-cavity squeezing (ICS) or injected external squeezing (IES) of the cavity mode. Both variational homodyne readout and quantum squeezing induce quantum correlation between the amplitude and phase quadratures of the cavity's output field, thereby improving the force sensitivity. The squeezing-enhanced variational homodyne detection scheme can enable high-precision sensing across various hybrid quantum platforms.

Within single-field inflation, primordial black hole and scalar-induced gravitational wave production from enhanced primordial perturbations typically requires a transient non-attractor phase, such as ultra-slow roll. We investigate the physical consistency of modeling such scenarios through analytical Hubble-flow parametrisation. By reconstructing the underlying scalar field potential, we show that even slow transitions in the slow-roll parameters can hide sharp, localised spikes in higher-order derivatives of the potential at the transition from ultra-slow-roll to slow-roll. These are typically not found in analytic potentials. To evaluate the impact of these structures, we implement a UV-filtering procedure based on discrete Fourier transform to systematically suppress high frequency modes in field space in both classes of models. We find that the filter effectively removes sharp features in Hubble-flow-derived potentials. As a consequence, we show that UV-filtered models typically respect Wands duality invariance as the field evolves back from ultra-slow roll to slow roll. Beyond linear perturbation theory, the introduction of spurious UV effects might affect other observables, such as non-Gaussianity and loop contributions. Our results thereby question the robustness of simple analytical Hubble-flow parametrisation for modeling inflationary models with a transient non-attractor phase.

The use of algebraic frameworks based on complex Clifford algebras for the representation and simulation of quantum circuits has been discussed in the literature. Recently, an alternative algebraic approach employing hard-core bosons has been proposed. Hard-core bosons provide a natural representation of multi-qubit systems, in which the tensor-product structure is realized directly and no sign corrections are required, in contrast to realizations based on complex Clifford algebras. Although both approaches are formally equivalent, the hard-core boson formulation exhibits computational advantages. This work reviews and extends the hard-core boson algebra for circuit simulation and presents an efficient implementation. A performance comparison with IBM Qiskit shows substantially improved execution times for simulations. Moreover, a new application is introduced in which the hard-core boson formalism is combined with genetic algorithms for quantum circuit synthesis.

Threshold private set intersection (TPSI) allows parties to reveal their intersection only when its cardinality reaches a prescribed threshold. Existing quantum TPSI protocols typically rely on a third party (TP) to interpret the final results, which deviates from the cardinality-testing paradigm of TPSI. In this paper, we propose a quantum multiparty TPSI protocol with explicit cardinality testing. Our protocol develops a rotation-based quantum construction in which single-photon sequences are sequentially processed through participant-side data rotations, TP--participant masking rotations, and correlated aggregate rotations. This design produces hidden-label measurement vectors: TP can complete the final measurement, but cannot interpret the semantic meaning of the outcomes. Based on these hidden measurements, we further realize the threshold decision through an oblivious linear evaluation (OLE)-based inner product procedure and a lightweight garbled circuit, revealing only \(\mathbf 1[|\bigcap_i X_i|\ge τ]\) before conditional intersection reconstruction. We prove the correctness and security of the proposed protocol, and further validate its feasibility through quantum-circuit simulations implemented on the IBM \textsf{Qiskit} platform.

We obtain an exact analytical vacuum solution of Einstein-Aether theory with a strictly static aether configuration and investigate its maximal extension. The solution depends only on the coupling $c_{14}$ and reduces to the Schwarzschild geometry in the limit $c_{14}=0$. We show that Schwarzschild is an isolated member of this family: for any nonzero $c_{14}$ the spacetime ceases to be a black hole and instead becomes either a naked singularity ($c_{14}<0$) or a wormhole-like geometry ($0<c_{14}<2$). By constructing the complete analytic extension, we demonstrate that the internal infinity of the wormhole corresponds to an extremal Killing horizon. Crossing this horizon leads to a new spacetime region where the causal roles of time and radial coordinates are exchanged, and the spacetime ultimately terminates at a spacelike singularity. The resulting global structure, summarized by the corresponding Carter-Penrose diagrams, reveals a previously unexplored causal completion of the static-aether vacuum spacetime.

Threshold private set intersection (TPSI) allows parties to reveal their intersection only when its cardinality reaches a prescribed threshold. Existing quantum TPSI protocols typically rely on a third party (TP) to interpret the final results, which deviates from the cardinality-testing paradigm of TPSI. In this paper, we propose a quantum multiparty TPSI protocol with explicit cardinality testing. Our protocol develops a rotation-based quantum construction in which single-photon sequences are sequentially processed through participant-side data rotations, TP--participant masking rotations, and correlated aggregate rotations. This design produces hidden-label measurement vectors: TP can complete the final measurement, but cannot interpret the semantic meaning of the outcomes. Based on these hidden measurements, we further realize the threshold decision through an oblivious linear evaluation (OLE)-based inner product procedure and a lightweight garbled circuit, revealing only \(\mathbf 1[|\bigcap_i X_i|\ge τ]\) before conditional intersection reconstruction. We prove the correctness and security of the proposed protocol, and further validate its feasibility through quantum-circuit simulations implemented on the IBM \textsf{Qiskit} platform.

We study primordial-black-hole (PBH) dark matter in the two-dark-dimensions (2DD) framework, a six-dimensional brane-world scenario with two compact extra dimensions and a fundamental gravity scale of order $10\,\mathrm{TeV}$. We calculate the evolution of higher-dimensional PBHs including the recently proposed quantum-gravitational memory-burden effect. For a memory exponent $p=2$, the evaporation rate is strongly suppressed, allowing PBHs with initial masses as small as $\sim10^{-3}\,\mathrm{g}$ to survive until the present epoch. Consequently, PBHs can account for the observed dark matter over a mass range extending from $10^{-3}\,\mathrm{g}$ to $10^{21}\,\mathrm{g}$. We further compute the stochastic gravitational-wave background generated at second order by the primordial curvature perturbations responsible for PBH formation. We show that the conventional four-dimensional formalism for scalar-induced gravitational waves remains applicable throughout the mass range accessible to current and future gravitational wave experiments. The resulting signals can be probed by LISA, DECIGO, and pulsar timing arrays. Using Fisher forecasts, we find that these observations can constrain the PBH mass, dark-matter fraction, and width of the primordial curvature spectrum with high precision. The low fundamental gravity scale of the 2DD framework also permits the production of microscopic black holes at future high-energy colliders. Their decay signatures, together with gravitational-wave measurements, provide complementary tests of higher-dimensional gravity, the memory-burden mechanism, and primordial-black-hole dark matter.

We investigate the dynamics of a charged massive scalar field around Kerr-Bertotti-Robinson Black Holes. By mapping the Klein-Gordon equation into a one-dimensional Schrödinger-like form and employing matched asymptotic expansions, we unveil the magnetic-field-induced quenching of the superradiant instability. We demonstrate that within a specific frequency band, the external magnetic field alters the physical boundary condition at the horizon from a propagating state to a purely exponentially decaying state, strictly locking the dissipation to zero ($\text{Im}(ω) = 0$). Consequently, the system supports two physically distinct classes of stationary bound states. Alongside the classical synchronized scalar clouds (Type-I) existing strictly at the superradiant threshold, we identify a new class of scalar clouds that exponentially decay at the event horizon (Type-II). Supported by precise numerical integrations, we demonstrate the spatial structure of these configurations. Our findings reveal that magnetized astrophysical environments can harbor significantly richer bosonic configurations than standard isolated black holes.

We establish a Ryu-Takayanagi (RT) relation in AdS$_3$/CFT$_2$ using \emph{relative entropy} as the central object, in place of the ultraviolet-divergent von Neumann entanglement entropy. Adapting Hollands' exact result for the chiral relative entropy to a diamond region, we express the boundary relative entropy between the vacuum and a coherent state as a Schwarzian functional, which the Fefferman-Graham dictionary identifies with the asymptotic data of a Bañados geometry; the rigidity of three-dimensional gravity promotes this boundary identification to the bulk. To linear order in the metric perturbation, the relative entropy then equals the variation of the RT geodesic length divided by $4G_N$. The construction rests only on the Bisognano-Wichmann/Borchers theorem and the holographic dictionary, giving a UV-finite, operator-algebraic counterpart to the RT relation.

Traditional quantum control relies on an iterative "simulate-then-optimize" paradigm, where dynamics simulation and control design are decoupled, leading to substantial computational overhead and limited scalability, particularly in noisy environments. Here, we propose an end-to-end quantum control framework based on long short-term memory, in which system dynamics and control strategies are learned jointly in a low dimensional latent manifold. The model directly maps initial states and environmental parameters to both dynamical trajectories and optimized control pulse in a single forward pass. The framework is validated on adiabatic speedup in a two-level system and state transfer in a one-dimensional spin chain under noise, achieving accurate dynamical prediction and control optimization. It improves the fidelity for both tasks and significantly reduces the optimization cost by three orders of magnitude compared with conventional iterative methods, while exhibiting strong generalization to multi-parameter, time-varying noise, as well as to different initial states and driving fields. Our work introduces a data-driven control paradigm based on latent manifold learning, reducing the computational bottleneck of iterative optimization and enabling real-time adaptive control of complex open quantum systems.

We study low-temperature non-equilibrium quantum thermometry with a bosonic probe: a quantum harmonic oscillator strongly coupled to a bosonic bath at temperature $T$ through a Drude--Ohmic spectral density. We treat the probe--bath dynamics both exactly, using the quadratic solution of Boyanovsky and Jasnow, and within a renormalized Gorini--Kossakowski--Lindblad--Sudarshan (GKLS) master equation. From the time-dependent covariance matrix we extract the quantum Fisher information (QFI) for general single-mode Gaussian probe states, including squeezed ones. In the strong-coupling, non-Markovian regime the QFI is non-monotonic in time, displaying bath-memory revivals that make a finite interrogation time $t^*>0$ strictly optimal. By contrast, we prove that the Markovian QFI rises monotonically to its stationary value and develops no interior optimum, so that its optimum is always pinned to the boundary $t^*\to\infty$; this complements existing Markovian precision-rate bounds, which concern $(\mathcal F(t)/t)$ rather than the single-shot QFI $(\mathcal F(t))$. Squeezed initial states yield a large transient advantage that thermalisation eventually erases, establishing squeezing and interrogation time as complementary thermometric resources. At equilibrium, strong coupling replaces the exponential Boltzmann suppression of the low-temperature relative error by a far milder polynomial divergence. As the model maps directly onto circuit quantum electrodynamics, these protocols appear within current experimental reach.

It is well known that the Einstein-scalar system of general relativity can in principle yield non-unique exact spinning naked singularities, which lead to unique Kerr black hole when the scalar field is switched off. It is a challenging task to observationally distinguish these two types of objects. Since accretion process could be a viable diagnostic for this distinction, the purpose of the present work is to explore whether there could be features in the accretion profiles distinguishing the singularity from a Kerr black hole. Here we study the Novikov-Thorne thin accretion to a \textit{new} spinning naked singularity with a scalar charge $σ$ recently reported by Bogush and Gal'tsov (BG). Our study reveals that: (1) The conversion efficiency $ε$ of the BG naked singularity is \textit{independent} of $σ$ and (2) The maxima of emissivity profiles for the BG singularity tend to shift towards the inner disk ISCO boundary $r=r_{ms}$ and peak at a value significantly larger than those of a Kerr black hole with the increase of $a$, $σ$ and relative shrinking of $\sqrt{-g}$. All these effects are \textit{quantitatively} tabulated, which reveal, for instance, that the flux from the naked singularity could be as high as $10^{5}$ times larger than that of a Kerr black hole. Since these distinguishing features are known to be shared also by other models of naked singularity, it is tempting to speculate that such behavior could be hallmark of naked singularities.

Vector polarimetry is an important application frontier for Rydberg-atom-based sensing. While prior research has largely concentrated on developing novel measurement schemes, high-sensitivity vector polarimetry remains an open question. Here we propose a theoretical framework for high-sensitivity detection of radio-frequency (RF) electric field polarization direction, which is particularly suitable for weak-field detection. Under a static magnetic field, the asymmetry in coupling between the Zeeman sublevels of the Rydberg atom and the RF field's polarization components enables the polarization angles to be determined from the atomic absorption index, which is retrieved via homodyne detection by incorporating the Rydberg atom system into an SU(1,1) interferometer. We derive the sensitivity of the polarization angles along with the corresponding standard quantum limit (SQL) and quantum Cramér--Rao bound (QCRB). Our results demonstrate a sensitivity surpassing the SQL across wide angular ranges using either dual coherent states or a coherent state combined with a squeezed vacuum state as input. Significantly, the optimal sensitivity reaches below \SI{e-6}{\degree}, with sensitivities better than \SI{e-3}{\degree} maintained over most of the angular domain. This work establishes a foundation for high-precision vector polarimetry, thereby advancing the development of Rydberg-atom-based quantum sensing and contributing to a deeper understanding of light--matter interactions.

We compute $\langle\hatΦ^{2}\rangle_\text{ren}$ as well as the energy fluxes $\langle \hat{T}_{uu}\rangle_\text{ren}$ and $\langle \hat{T}_{vv}\rangle_\text{ren}$ (where $u$ and $v$ are the standard Eddington-Finkelstein coordinates) associated with a quantum massless real scalar field $\hatΦ$, with a general curvature coupling constant $ξ$, near the inner horizon (IH) of a Kerr black hole, along the axis of rotation. The quantum field is in the Unruh state, corresponding to an evaporating black hole. We renormalize these quantities by the state-subtraction method. We drop the assumption of minimal coupling to the curvature, thereby generalizing the results of arXiv:2203.08502 for the fluxes at the IH. This requires understanding the asymptotic behavior of $\langle\hatΦ^{2}\rangle_\text{ren}$ neat the IH. State subtraction allows us to push the computation of $\langle\hatΦ^{2}\rangle_\text{ren}$ along the axis of rotation in the Kerr interior in arXiv:2409.17464 deeper into the near-IH region, exposing their final asymptotic behavior on approaching the IH. For $\langle\hatΦ^{2}\rangle_\text{ren}$ (a $ξ$-independent quantity in the Kerr case), we find that the approach to its finite asymptotic IH value is given, per $\ell$-mode, by a ringdown phase (namely exponentially damped oscillations), followed by an inverse-power tail, both in the tortoise coordinate $r_{*}$ (which diverges at the IH). Interestingly, in the regime where the ringing dominates, the ringing's complex frequencies are (numerically) found to match twice the well-known classical quasinormal-mode frequencies in Kerr, and the inverse-power tails are found to be $r_{*}^{-2\ell-3}$ (resembling Price's law in the classical black hole exterior, upon replacement $t\to r_*$). [Abridged]

The hoop conjecture, due to Thorne, is a fundamental aspect of black holes in classical general relativity. Recently, generalized classes of regular spherically symmetric static black holes with arbitrary exponents coupled to nonlinear electrodynamics have been constructed in the literature. The conjecture in those spacetimes could be violated if only the asymptotic mass $M_{\infty}$ is used. To avoid such violations, Hod earlier suggested the appropriate mass function and stated the conjecture in terms of what we call the Hod function. The conjecture can then be applied to any given static spacetime to test whether or not it represents black holes. It is shown here that the conjecture is protected in the above constructed class of generalized spacetimes thus supporting them as black holes. However, it is argued that there are factors, including violation of the conjecture, that militate against the proposed \textit{new} class of solutions to be qualifying as black holes. Finally, we exemplify that the Hod mass $M(r\leq R)$ in the conjecture is exactly the \textit{matter} counterpart of the Misner-Sharp \textit{geometrical} quasilocal mass $m(r\leq R)$ of general relativity. Thus any conclusion based on Hod function is strictly a conclusion of general relativity.

Hamiltonian dynamics have been widely implemented on noisy intermediate-scale quantum devices in recent years. In contrast, experimental demonstrations of Markovian quantum dynamics remain limited, because implementing nonunitary evolution on quantum computers is challenging. Quantum collision models provide a natural approach to this problem by coupling the system to ancillas to realize dissipation. However, previous implementations of quantum collision models on quantum computers have typically been restricted to one or two system qubits and fewer than 12 time steps, owing to noise, circuit depth, the overhead of ancilla reset, and limited qubit resources. In this work, we experimentally simulate Markovian quantum processes with local and nonlocal dissipation on both trapped-ion and superconducting quantum computers. By employing hardware-specific ancilla strategies, we realize simulations with up to seven system qubits, corresponding to 13 qubits in total, and 40 time steps. Our results demonstrate that, even for the same physical model, the optimal implementation strategy depends strongly on the hardware characteristics of the quantum computer.

We initiate the classification of supersymmetric degenerate Killing horizons, with closed spatial cross section, away from the near-horizon limit in D=11 supergravity. We prove that all such solutions fall into two distinct classes, depending on lightcone chirality with respect to a Gaussian null coordinate system. For the first class of solutions, the negative lightcone chirality part of the Killing spinor is non-zero on the Killing horizon, and we prove that all such solutions are isometric to supersymmetric near-horizon geometries. In the second class, the negative lightcone chirality part of the Killing spinor vanishes on the Killing horizon. In this case, we prove that the spinorial Lie derivative of the Killing spinor with respect to the Killing vector which generates the Killing horizon vanishes, and that all such solutions with more than 13 supersymmetries are pp-waves.

Quantum light detection and ranging (LiDAR) utilizes quantum entanglement and correlation to improve precision, noise resilience and covertness of target detection. Despite recent advances, the development of a quantum LiDAR system that simultaneously achieves high precision and a large measurement range remains challenging. Here, we demonstrate a quantum amplitude-modulated continuous wave LiDAR with micrometer precision achievable via increased acquisition time and meter-scale measurement range. In our demonstration, the signal photons directly illuminate the target, while the idler photons are non-locally modulated with a high-frequency cosine wave and never interact with the target. By leveraging the non-local modulation and the quantum correlation, the target detection is achieved with a precision of 0.64 $\pm$ 0.06 mm within one second over a measurement range of 2-8 m. As the acquisition time is up to 500 s, the system achieves a precision of 29 $\pm\ 4{\ \mathrm{μm}}$. Furthermore, our system realizes a 50 times precision improvement over the classical single-photon scheme in a background noise 37 dB stronger than the returned probe photons. With these advantages, our method will open venues for the development of high-precision, long-range, and noise-resilient target detection.

A pair of proper cones $(\mathsf{C}_1,\mathsf{C}_2)$ is said to have the Lorentz factorization property (LFP) if every $(\mathsf{C}_1,\mathsf{C}_2)$-positive map factors through a direct sum of Lorentzian cones, i.e., cones over Euclidean balls. Clearly, $(\mathsf{C}_1,\mathsf{C}_2)$ has the LFP if either $\mathsf{C}_1$ or $\mathsf{C}_2$ is a direct sum of Lorentzian cones, and our main goal is to find other examples. We show that such examples cannot be found for pairs $(\mathsf{C}_1,\mathsf{C}_2)$ where $\mathsf{C}_1=\mathsf{C}_2$, or in the case where both $\mathsf{C}_1$ and $\mathsf{C}_2$ are polyhedral. We also focus on the case where $\mathsf{C}_1=\mathsf{C}_\square$ is the square-based cone in $\mathbf{R}^3$. Here, we show that $(\mathsf{C}_\square,\mathsf{C})$ has the LFP whenever $\mathsf{C}$ is a symmetric cone, i.e., a direct sum of Lorentz cones, cones of positive semidefinite matrices over the real numbers, complex numbers or quaternions, and the cone of $3\times 3$ positive semidefinite matrices over the octonions. We leave open the question whether there are more examples, but we show that this list cannot be extended by any strictly convex cone $\mathsf{C}$ or for a cone $\mathsf{C}$ with $\text{dim}(\mathsf{C})\leq 5$. Finally, we discuss an application to a problem in quantum information theory.

Monads are a ubiquitous structure in functional programming used for modelling computational effects. For example, the state monad models the effect of a computation interacting with a memory system. Here we introduce the quantum instrument monad $\mathcal{I}_\mathcal{A}$, which models the effect of a computation interacting with a quantum system with algebra of observables $\mathcal{A}$. It can be thought of as a noncommutative generalization of the state monad. We construct this quantum instrument monad in two versions: a finitary version on the category of sets and a measure-theoretic version on the category of measurable spaces (the latter under the assumption that $\mathcal{A}$ is a type I von Neumann algebra with separable predual). Both versions are strong monads. The construction of the measure-theoretic version is based on a new notion of integral of a quantum-operation-valued function against a state-valued measure.

It is well-established that the Minkowski limit of pure $f(R)=R^2$ gravity breaks down, unlike that of full Starobinsky theory $f(R)=R+αR^2$. We provide a novel interpretation of this phenomenon using the recent thermal analogy between scalar-tensor gravity and Eckart's relativistic dissipative fluids. In this framework, we show that approaching the Minkowski background corresponds to a diverging effective ``gravitational temperature''. This perspective naturally rephrases the strong coupling problem as a thermal singularity, demonstrating that $R^2$ gravity departs infinitely far from General Relativity rather than recovering it.

The dark shadow cast by a black hole, determined by the outermost unstable null circular geodesics (the photon sphere), provides a direct probe of strong-field gravity. In this work, we derive model-independent lower and upper bounds on the shadow radius $r_{\mathrm{sh}}$ for static, spherically symmetric, asymptotically flat black holes in $n$-dimensional ($n\ge 4$) Einstein gravity, supported by an anisotropic matter field. For the lower bound, assuming the matter satisfies the Weak Energy Condition (WEC), we prove $r_{\mathrm{sh}}\geq \bigl(\frac{n-1}{2}\bigr)^{\frac{1}{n-3}}\sqrt{\frac{n-1}{n-3}}\,r_H$, where $r_H$ is the horizon radius. For the upper bound, under the WEC and the Strong Energy Condition (SEC), together with an asymptotic decay condition on the matter fields, we prove $r_{\mathrm{sh}}\leq\sqrt{\frac{n-1}{n-3}}\bigl[(n-1)M\bigr]^{\frac{1}{n-3}}$, where $M$ is the ADM mass. These results reduce to the known four-dimensional bounds and are saturated by the vacuum Schwarzschild-Tangherlini black hole. Our results generalize the four-dimensional shadow bounds to an arbitrary number of dimensions and provide model-independent geometric constraints on the observable shadow of higher-dimensional black hole spacetimes.

Twin-field quantum key distribution (TF QKD) can overcome the fundamental rate loss limit of repeaterless quantum links, but its practical deployment has long been hindered by the requirement of global phase locking between two independent lasers. By revisiting the fundamental principles of optical interference, this work reveals that interference in TF QKD inherently relies only on the instantaneous phase alignment of two independent optical pulses at the moment they temporally overlap, rather than on continuous global phase synchronization. Guided by this insight, we propose and demonstrate a single-sideband-interference TF-QKD protocol that eliminates global phase locking. Each user employs an IQ modulator to generate a weak single sideband as the quantum signal, while the intrinsically phase-correlated optical carrier propagates as a real-time phase reference. Carrier interference at the receiver enables real-time phase extraction and feedback compensation for the sidebands. Unlike prior no phase locking approaches requiring second- or microsecond-level coherence, in principle, our scheme reduces this requirement to nanoseconds. We achieve 98% interference visibility over 100.8 km fibre and secure key rates surpassing the PLOB bound in the high-loss regime, providing a simpler route towards practical long-distance quantum communication networks.

We propose a Lagrangian formulation for a non-conservative gravity model in which the divergence of the energy-momentum tensor in curved spacetime does not vanish. This is accomplished by introducing an arbitrary vector field that couples with the gradient of the Ricci curvature scalar. We first derive the field equations using the Palatini variational approach. Because the connection and the metric tensor are independent in the Palatini framework, the auxiliary vector field dictates whether the manifold geometry is Weyl or Riemannian. By assuming certain physically reasonable conditions on this vector field, the resulting field equations reduce to those of Rastall gravity. Furthermore, slightly different conditions on the vector field furnish unimodular gravity. For comparison, we also employ the standard metric variational approach to obtain the field equations, demonstrating that the same models can be recovered under appropriate conditions. Our key results are the derivation of a covariant Lagrangian formulation for Rastall gravity and a new Lagrangian formulation for unimodular gravity.

Modern quantum physics now enables control of quantum systems at the level of individual trajectories, opening a new frontier that links quantum information theory, quantum many-body physics, and quantum thermodynamics, and uncovers novel non-equilibrium phenomena such as deep thermalization and measurement-induced entanglement. However, a central challenge remains: their characterization relies on measuring nonlinear properties of individual quantum states, a task tantamount to fine-grained cloning of a quantum ensemble. Here, the fundamental laws governing the cloning of quantum ensembles are investigated. First, a general no-cloning theorem for arbitrary ensembles is established from an information-theoretic perspective, even assuming multiple copies of the ensemble's purification. It is then shown that this barrier can be unexpectedly circumvented for physical ensembles generated by finite-time evolutions. Nevertheless, these tasks are proven to remain computationally intractable, even when the full circuit description of state preparation is known. This stands in sharp contrast to the conventional no-cloning theorem, which relies on the state being unknown. Together, these results establish new fundamental principles of quantum mechanics, reveal intrinsic trade-offs among sample complexity, computational complexity, and quantum measurements, and highlight the necessity of problem-specific strategies for probing measurement-induced quantum phenomena.

We prove explicit upper bounds on the propagation velocity of one-dimensional quantum walks with periodic coins of arbitrary period. We treat two complementary settings. First, in a perturbative regime where one transmission parameter is small, we show that the corresponding almost reflecting coin acts as a bottleneck for transport: the velocity is bounded linearly in this parameter, with an explicit leading order estimate. Second, for arbitrary nonzero transmission parameters, we prove a general a priori bound in terms of their harmonic mean, together with a refined version that detects the spatial variation of neighboring coins. Moreover, we prove a general lower bound on the velocity. These bounds apply directly to the corresponding CMV setting.

Fault-tolerant quantum algorithms offer a promising pathway for estimating the ground-state energies of periodic materials that are beyond the practical reach of classical electronic-structure methods. A remaining challenge is finite-size mitigation: quantum algorithms evaluate a finite supercell or finite Brillouin-zone mesh, while materials properties are defined in the thermodynamic limit. In this work we develop a quantum post-processing strategy for the leading two-body finite-size correction. After one-body shell effects are reduced by twist averaging, the dominant residual error is controlled by long-wavelength density fluctuations, which are encoded in the small-momentum static structure factor $S(q)$. We formulate the corresponding operator in a Bloch-orbital basis, construct its block encoding through the density operator, and estimate its ground-state expectation value using an amplified Hadamard test. We also introduce adaptive global and local binary search procedures for identifying the infrared fitting window used to reconstruct the two-body finite size error correction. The resulting cost remains subleading relative to the main ground-state energy estimation routine: the structure-factor correction has leading $\tilde{O}(N_bN_k)^3$ dependence on the Bloch-orbital basis size, avoids the large plane-wave prefactor of full Hamiltonian simulation, and requires only $\tilde{O}(N_bN_k)$ logical qubits. This provides a fault-tolerant alternative to down-sampling, replacing repeated energy calculations on larger cells with targeted measurements of the infrared density correlations that control the finite-size effects.

To answer the question posed in the title, we consider a novel diagnostic, viz., the difference in the times of arrival (TOA) at the observer of two light rays that simultaneously emanate from a source behind a spinning lens and pass by either side of the lens to reach the observer. This is completely different from the usual Shapiro gravitational time delay, where only one onward light ray is reflected back to the observer. The TOA essentially samples the frame dragging caused by the spinning lens, apart from other lens parameters. Assuming a charged \textit{regular} Ayón-Beato and García black hole as the spinning lens partner in some typical astrophysical PSR-BH binaries, which provide the best laboratory for testing the TOA effect, we theoretically study how the prediction depends on the gyromagnetic ratio $\left(Q/M\right)$ and how it compares with those when the role of spinning lens partner is played by the centrally \textit{singular} Kerr-Newman and Kerr black holes. The numerical estimates for two illustrative binary lens systems show $μ$sec level delay at the zeroth order, which should be measurable. However, the TOA predictions under thin-lens approximation are shown to differ only at third or higher orders of smallness indicating that the regular and singular black holes \textit{cannot} be observationally distinguished despite significant qualitative differences existing among them.

Secure multi-party computation (SMC) addresses the problem of jointly computing global functions of private inputs while revealing minimal information about individual data. Two prominent examples of SMC tasks are sealed-bid auction and secure multi-party summation. Existing schemes for quantum auction and quantum summation have largely been developed independently, motivated by distinct applications and employing different computational primitives. In this work, structural symmetries in existing protocols for quantum auction and quantum summation are identified. In particular, it is established that the core auction primitives including revenue estimation, maximum bid identification, and winner determination can be reduced to repeated invocations of a summation oracle acting on suitably defined indicator functions. Conversely, summation protocols can be naturally embedded as auxiliary subroutines within auction frameworks, establishing summation as a unifying primitive underlying a broad class of auction mechanisms. Further, computational, communication and memory costs of these reductions are analyzed and compared with some of the representative existing protocols. The analysis has revealed that the process of implementing summation tasks through currently known auction protocols leads to additional overhead associated with bid-space exploration and winner determination. The proposed framework is protocol-agnostic and applicable across diverse computational models, including gate-based and photonic implementations. Finally, a proof-of-concept experimental realization (numerical validation) of a two-bidder sealed-bid auction using IBM (optical quantum) hardware is demonstrated to establish that the claimed equivalence is not merely formal but experimentally verifiable with the available hardware.

Quasi-periodic eruptions (QPEs) are repeated X-ray bursts originating in galactic nuclei. Of the many proposed models, the favored model is the disk-collision model in which a stellar mass orbiter collides with a disk formed from a tidal disruption event, generating flares twice per orbit. In this model QPEs are tracers of circular extreme mass ratio inspirals (EMRIs) and can be used to infer the EMRI formation rate and estimate their contribution to the stochastic gravitational wave background (SGWB) in the Laser Interferometer Space Antenna (LISA) band. Whether the secondary is a stellar-mass black hole or a main sequence star is still debated and leads to different results for the EMRI rate and SGWB. We obtain fiducial rates -- subject to systematic uncertainties -- of $R_{\rm SE} = 2.88\times10^{-6}$ per galaxy per year for stellar EMRIs and $R_{\rm BHE} = 6.07\times10^{-6}$ per galaxy per year for black hole EMRIs, then estimate their contribution to the SGWB. We find that only black hole EMRIs contribute to the 1 - 10 milliHertz band resolvable by LISA, and depending on the secondary mass and formation radius can contribute from just below the LISA sensitivity curve to roughly two orders of magnitude above it. Stellar EMRIs, being tidally disrupted before reaching the 1 - 10 milliHertz band, only contribute to sub-milliHertz frequencies and remain below the LISA sensitivity curve.

In the context of hyperbolic formulations of Einstein's field equations obtained via gauge fixing, constraint damping is a desirable feature that ensures that violations of the gauge condition and thus of the constraint equations are suppressed in evolution. Besides its utility in numerical relativity, it has played a key role in several (linear and nonlinear) stability proofs of spacetimes as solutions of the Einstein equations. In this paper, we show that an enhanced form of constraint damping can be implemented for the linearization of the Einstein equations around any subextremal Kerr black hole metric. The results proved here are a key ingredient in the author's proof of the nonlinear stability of the subextremal Kerr family.

Squared commutators in the Holstein--Primakoff limit of a spin--cavity system provide a compact way to separate propagation from covariance growth in a finite-temperature reservoir with memory. In the finite-temperature NMQSD construction, the linear quadrature commutator is fixed by the retarded spin--cavity propagator, whereas a quadratic commutator carries the same retarded factor together with a covariance factor. For a zero-mean Gaussian state, \(C_{R_i^2,R_j}(t)=4|κ_{ij}(t)|^2V_{ii}(t)\); the symmetrized expression gives the spin-side and mixed channels. Since \(\bar n\) enters the covariance sector but not the homogeneous retarded kernel, raising \(\bar n\) from 0 to 1 leaves the linear transfer unchanged while increasing the quadratic signal. Varying the bath-memory rate and the counter-rotating coupling within the stable HP region then shows how stored cavity history changes both the transfer weight and its distribution in time. The calculation separates memory-dependent propagation from thermal covariance growth in collective spin--cavity dynamics.

Fisher information plays a central role in statistics and quantum metrology, providing the basis for the celebrated Cramér-Rao bound. In this work, we introduce a new information measure based on higher-order Fisher information and show that it naturally leads to a generalized uncertainty relation for parameter estimation, which can be regarded as an extension of the Cramér-Rao bound. As an application, we analyze the case of quantum phase estimation with a single qubit and compare our theoretical bounds with the well-known established hierarchical bounds. Finally, we experimentally validate the proposed framework using a photonic platform.

Fast quantum control is essential to overcome decoherence in contemporary quantum platforms, yet achieving this in many-body systems remains a major challenge. We show that the time-rescaling (TR) method enables efficient acceleration of closed many-body quantum dynamics, extending its applicability beyond previously studied regimes. Applying TR to the transverse-field Ising model with a longitudinal field, we demonstrate a significant enhancement of quantum annealing performance, maintaining high ground-state fidelity at evolution times where standard adiabatic dynamics breaks down, with only weak dependence on system size. We further demonstrate high-fidelity preparation of Greenberger-Horne-Zeilinger states in many-body systems, where TR extends the accessible system sizes within fixed evolution times. We additionally show that the Mandelstam-Tamm quantum speed limit does not fundamentally limit the acceleration achievable through TR, as the reduction in evolution time is exactly compensated by increased energy fluctuations. These results establish TR as a scalable and experimentally viable approach to fast quantum control in many-body systems.

We investigate the $d$-dimensional Dunkl--Klein--Gordon equation for a scalar particle within an algebraic framework. By employing Schrödinger factorization, we construct the generators of the $\mathfrak{su}(1,1)$ algebra and establish the associated symmetry of the radial sector. The energy spectrum is derived using irreducible unitary representations, and the corresponding Sturmian radial basis is obtained analytically. We analyze the $d$-dimensional Dunkl--Klein--Gordon oscillator and the bound-state sector of the $d$-dimensional Dunkl--Klein--Gordon equation with a Dunkl--Coulomb-like potential. Furthermore, $\mathrm{SU}(1,1)$ coherent states are constructed and their time evolution is analyzed, revealing a characteristic radial oscillation behavior. The results show that the Dunkl deformation introduces parity-dependent modifications in the spatial structure of the system while preserving its underlying algebraic dynamics.

Multicomponent nonlinear Schrödinger equations constitute fundamental models for coherent matter waves in multicomponent Bose--Einstein condensates, spinor quantum fluids, and vector nonlinear optical systems. We develop a vector formulation of the Hirota bilinear formalism for the completely integrable Manakov system that treats the coupled nonlinear Schrödinger equations directly at the vector level rather than through the conventional component-wise decomposition. This framework naturally retains the intrinsic multicomponent representation of the model while providing compact analytical expressions for exact vector soliton solutions. Within this approach, we systematically construct bright, dark, and mixed one-, two-, and three-soliton solutions and show how the underlying vector structure provides a unified description of their nonlinear interactions. In particular, the proposed formalism makes the coupling between the different components explicit while preserving the geometric organization of the vector system throughout the bilinearization procedure. Beyond its analytical simplicity, the framework offers a natural perspective for the study of coherent multicomponent nonlinear excitations and provides a foundation for extending vector Hirota methods to other classes of exact solutions, including rogue waves, periodic waves, and rational solutions.

We theoretically investigate the response of a coherently-driven nonlinear optical cavity to an additional incident single photon. Using a quantum description of the nonlinear dynamics that fully accounts for the quantum fluctuations of the cavity field and for the discrete nature of the incident photon, we characterize the quantum dynamics of single-photon-stimulated jumps from the low-photon-number to the high-photon-number state of the optical bistability loop. We find that the system can exhibit a giant response to this single quantum of excitation, rooted in the phase-transition picture of optical bistability. In addition to shedding light on the role of quantum fluctuations in the non-equilibrium dynamics of a nonlinear optical cavity, our results suggest a strategy for an all-optical single-photon avalanche detector.

We study the geodesic deviation equation in symmetric teleparallel geometry (STG), where the relative acceleration is defined with respect to the STG connection. We analyze the modified Raychaudhuri equation along a geodesic congruence in $f(Q)$ gravity under the Weyl-type nonmetricity ansatz. In contrast to the purely geometrical Raychaudhuri equation obtained in general metric-affine settings, the equation derived here contains matter-source contributions through the trace equation of $f(Q)$ gravity. Different from general relativity, focusing in $f(Q)$ gravity is not automatic, and one must impose an appropriate focusing condition. We collect the model-dependent terms in the modified Raychaudhuri equation into an effective energy-momentum trace $T_{\mathrm{eff}}$, so that the focusing condition can be written as the inequality $T\leq T_{\mathrm{eff}}$, where $T$ is the trace of the matter energy-momentum tensor. We also apply this condition to a flat Friedmann--Lemaître--Robertson--Walker background. The homogeneous and isotropic STG connection admits three branches, each characterized by a single connection function $γ_i$, with $i=1,2,3$. In the symmetric teleparallel equivalent of general relativity, we obtain the corresponding effective traces and derive the resulting constraints on the matter content and the connection functions in each branch.

We develop a general framework to characterize non-vacuum perturbations to relativistic binaries in the gravitational-wave (GW) driven regime, for use in GW parameter estimation studies. The effect of smooth, structured and stochastic perturbations to the binary's motion is reduced to a resonant spectral projection defined on a rolling averaging window, with weights given by Hansen coefficients. This is combined with practical criteria for identifying and evaluating the corresponding dynamical response to perturbations, starting from either analytical models or numerical simulations of binaries in environments. The result is a set of coupled ODEs for the orbital elements that capture epi-cyclic, apsidal and nodal resonances, consistently incorporate feedback from radiation reaction and can be solved efficiently on a coarse time grid. We demonstrate the practical application of the framework in two representative astrophysical scenarios: a compact binary in a variable tidal field and an extreme-mass-ratio inspiral in an accretion disk. We propose the Lagrange-Fourier-Hansen framework as a unified tool for modeling environmental effects in GW templates for eccentric and precessing binary sources, and particularly for bridging the gap between phenomenological prescriptions and realistic models of binaries in environments.

Achieving high-fidelity execution on noisy intermediate-scale quantum (NISQ) hardware requires careful selection of physical qubit layouts, as gate errors, readout errors, and coherence times vary across the device and drift over time. Currently, qubit mapping is performed either through manual inspection of device calibration data or through automated layout pipelines, neither of which provides integrated, interactive layout visualization combined with calibration analytics. In this work, we present the Interactive Quantum Mapper (I-QMapper), a Jupyter-based, open-source tool for noise-aware layout selection, visualization, and analysis on superconducting quantum hardware. I-QMapper offers two operating modes: a general-purpose mode for arbitrary circuits, and a dedicated mode for quantum-chemistry applications, specifically tailored to the Local Unitary Cluster Jastrow (LUCJ) ansatz. Within each mode, a Design panel supports interactive layout construction, while an Error panel provides calibration analytics through four temporal viewing modes (Live, Snapshot, Intraday, and Multi-day range) together with threshold filtering and delta-mode comparison for drift identification. Each layout receives a Layout-Quality Score (LQS) that aggregates the readout and two-qubit gate errors of the layout into a single quality value. Starting from the automatic LUCJ circuit-generation provided by IBM Quantum, we extend it to a multi-programming setting in which multiple circuits are mapped onto a single quantum processing unit (QPU). I-QMapper further supports side-by-side visualization of two quantum backends and layout comparison, and session export for experimental reproducibility. By combining interactive exploration with calibration analytics, I-QMapper aims to support both rapid layout prototyping and informed noise-aware experimental design on NISQ devices.

Following the landmark acquisition of the first image of a black hole, systems comprising a black hole enveloped by a dark matter halo have have attracted considerable attention. This work implements a novel method to analyze the structure of a system composed of a Schwarzschild black hole and a dark matter halo modeled as an anisotropic fluid, while ensuring that the proposed metric constitutes an exact solution of the Einstein field equations. Numerical results show that the connection between the temporal and radial metric functions preserves an inverse relationship, namely $A(r) \approx 1/B(r)$. Furthermore, we analytically demonstrate that the resulting functions satisfy $A(r)B(r) = 1 + \mathcal{O}(10^{-6})$, in excellent agreement with the numerical findings. In addition, the Einstein equations imply that the variation of the product $A(r)B(r)$ is governed by the factor $\varepsilon_h=\frac{8πGρ_0r_0^2}{c^2}$. Finally, we show that the interior Schwarzschild region and the exterior dark matter halo can be joined smoothly via a regular Darmois-Israel junction, without generating any thin-shell curvature steps.

We study how dark matter (DM) impacts the crust and the spectrum of torsional crust oscillations of dark-matter-admixed neutron stars (DANSs). We construct two-fluid equilibrium solutions wherein baryonic and DM interact gravitationally only, adopting a unified nuclear equation of state for the former and a fermionic equation of state with repulsive self-interaction for the latter. At fixed total gravitational mass and DM mass fraction, we find that DM reduces the crust thickness in comparison to pure baryonic-matter neutron stars (NSs). The thinning of the crust is negligible when most of the DM distribution extends beyond the star's baryonic surface. However, the crust thickness can decrease by as much as 12% when the DM distribution is within the star's baryonic surface, i.e., when the star has a "dark core." We support these results by deriving approximate analytical formulas for the crust thickness that agree with our numerical calculations at the sub-percent level in best case scenarios. Next, we derive the equation that describes crustal torsional modes of DANSs in the relativistic Cowling approximation. We find that the oscillation frequencies are in general higher than those of a comparable pure baryonic-matter NS, with the largest frequency shifts happening in the same parameter space where the crust thickness decreases the most. Moreover, we study the degeneracy between DM and baryonic-crustal microphysics effects on these modes. As an example, we study electron screening, which softens the crust's shear modulus, thus decreasing the frequencies. We find that the degeneracy between the competing effects of DM and electron screening can be broken in some regions of the parameter space we explored. Should they be measured, our results suggest that torsional oscillations could be used to infer the existence of a DM core within massive NSs. (Abridged)

The method of types is a fundamental tool in classical information theory, with applications ranging from composite hypothesis testing and universal source coding to the capacity of arbitrarily varying channels. In this work we introduce an empirical operator acting as a quantum analog of the empirical distribution. We show that this empirical operator satisfies combinatorial and large-deviation bounds, which in combination describe a quantum method of types. As an application, we use this quantum method of types to prove a universal achievability result for composite quantum hypothesis testing.

The physical origin of dark energy remains one of the most profound open questions in modern physics. Although cosmological observations tightly constrain the equation of state parameter $w$, this information alone does not reveal the underlying microphysics, as many distinct theoretical models can reproduce the same expansion history. A key discriminator among these models is the sound speed of dark energy perturbations, yet this quantity remains largely unconstrained by current astrophysical observations. In this work, we propose a fundamentally new approach: using collider measurements of beyond-the-Standard-Model (BSM) mediator resonances as a probe of dark energy microphysics. We construct a unified effective-field-theory framework in which a dynamical dark energy scalar is coupled, through symmetry-motivated derivative interactions, to a pseudoscalar mediator in the 2HDM+$a$ model. These interactions naturally induce invisible decays and modify the propagation of the BSM mediator in a dark energy background, leading to measurable distortions of resonance properties at colliders such as the LHC. We show that the decay widths, branching ratios, and kinematic structure of the mediator resonance become sensitive to the propagation properties of dark energy fluctuations, in particular the sound speed. As a result, collider observables provide a direct and complementary handle on dark energy microphysics, with the potential to distinguish between models that are otherwise indistinguishable through cosmology alone. Our results establish a new paradigm in which high-energy collider experiments can probe the physics of cosmic acceleration, revealing a connection between the smallest and largest scales in nature and opening a novel experimental pathway to uncover the fundamental origin of dark energy.

We investigate the thermodynamic role of asymptotic aether alignment for universal horizons in Einstein-aether theory. In the static, spherically symmetric, asymptotically AdS sector with $c_{14}=0$, the known first law for universal horizons contains an additional term whenever the aether is misaligned with the timelike Killing vector at infinity. While this term has recently been interpreted in Hořava--Lifshitz gravity as the contribution of an elliptic charge associated with khronon reparameterizations, no corresponding explanation was available in Einstein-aether theory. We show that, in the same sector, Einstein-aether theory possesses a previously unidentified symmetry of the reduced action, generated by infinitesimal transformations of the form $δu^a=f a^a$, where $a^a$ is the aether acceleration and $f$ obeys an elliptic constraint. We derive the associated current and charge, and show that the aligned limit is naturally interpreted as the ensemble in which this aether-charge contribution vanishes. This provides the Einstein-aether counterpart of the elliptic-charge mechanism in Hořava--Lifshitz gravity and clarifies the thermodynamic significance of asymptotic aether alignment.

This work investigates how torsion and nonmetricity modify tidal accelerations in metric-affine gravity. We derive a projected deviation equation that generalizes the standard geodesic deviation equation to metric-affine geometry, and apply it to the relative acceleration of neighboring autoparallels in the weak-field, nonrelativistic limit. In this regime, the tidal acceleration separates into the usual Newtonian contribution and linear post-Riemannian corrections sourced by torsion and non-metricity. By decomposing torsion and nonmetricity into their irreducible Lorentz components, we identify the corresponding signatures in the tidal tensor and discuss to what extent these contributions can be distinguished. We then show how future direct tidal measurements could be translated into benchmark bounds on post-Riemannian tidal contributions, assuming probe dynamics sensitive to the affine connection. Our results suggest that the tidal acceleration may provide a systematic route toward probing post-Riemannian spacetime features in the future.

Extreme mass-ratio inspirals (EMRIs) are among the prime sources for future space-borne gravitational wave (GW) observatories and provide a useful setting for testing the presence of fundamental fields and possible deviations from general relativity (GR) in both strong and weak gravity regimes. In this work, we study the effect of a massive scalar field on eccentric equatorial EMRI dynamics around Kerr black holes. Considering that the inspiralling stellar-mass object carries a scalar charge and emits scalar radiation together with tensor GWs, we compute the relevant relativistic fluxes within the adiabatic treatment of the inspiral. With the solution of the scalar perturbation equation in the frequency domain, the resulting fluxes are presented through the Chebyshev interpolants in order to have the efficient inspiral evolution across the parameter space considered. We quantify the impact of scalar field mass and scalar charge on the orbital evolution and GW signal through phase shifts and waveform mismatches relative to both GR and the massless-scalar scenario. We find that massive scalar radiation can generate significant GW dephasing that increases with orbital eccentricity; however, the scalar flux is suppressed as the scalar field mass is becoming larger. Using a Fisher information matrix (FIM) analysis, we estimate the ability of Laser Interferometer Space Antenna (LISA) to measure or constrain the scalar charge and scalar field mass. Our results indicate that eccentric EMRIs can place meaningful constraints on massive scalar fields and provide a promising as well as important avenue for testing scalar-tensor extensions of gravity in the region of a strong gravitational field.

We report on numerical simulations of critical phenomena in the collapse of axisymmetric vacuum gravitational waves, adopting families of initial data that, to the best of our knowledge, have not been used in this context before. Like Teukolsky waves, the data are based on linear wave solutions to the Einstein equations. We follow Nakamura's construction and encode the wave content in the extrinsic curvature rather than the spatial curvature, which leads to several simplifications when the data are "dressed up" so that they satisfy the nonlinear constraint equations. We are able to fine-tune these data to the onset of black hole formation slightly better than in our previous simulations, allowing us to observe and examine one more echo in the approximately self-similar threshold solution. Our findings are consistent with earlier studies: while we find threshold solutions that are approximately discretely self-similar, the self-similarity is not exact, and we find no evidence for a unique critical solution. We discuss common features between the different threshold solutions, including the appearance of alternating maxima in the direction of the poles and the equator.

Photonic quantum metrology has demonstrated advantages in precision and resource efficiency for a wide range of applications, with several schemes approaching the fundamental quantum Cramér-Rao precision bound (QCRB). However, the intrinsic incompatibility of quantum measurements represents a hurdle in extending these advantages to the simultaneous estimation of multiple parameters. In this paper, we present an experimental protocol approaching the QCRB simultaneously in two polarisation parameters, across a wide range of the parameter space, with as few as $\sim 200$ photon pairs, offering advantages for polarimetric sensing for dim sources such as in X-ray astronomy or photosensitive samples.

Multimessenger observations of neutron star mergers are unique opportunities to constrain the properties of dense matter and the production site of heavy nuclei. To leverage these observations, we require reliable models of the electromagnetic signals powered by mergers. An important limitation to our ability to develop such models is the use of approximate neutrino physics in simulations. Here, we present simulations using an improved version of our Monte Carlo transport algorithm specifically designed to allow for more advanced on-the-fly calculations of reaction rates that use the simulated energy distribution of neutrinos, including in blocking factors, while still relying on approximations for the angular distribution of neutrinos. We use these new methods to include in simulations inelastic scattering of neutrinos on electrons, and to improve our treatment of neutrino-antineutrino pair annihilation. We find that, without increasing the cost of simulations, we can marginally get to the point when the addition of a single packet represents a change $Δf_ν<1$ in the angle-integrated distribution function, at the cost of increased shot noise in the coupling to the fluid. With inelastic scattering and a better treatment of pair processes, we find a reduction in the average energy and total luminosity of heavy-lepton neutrinos, and an increase in the amount of mass ejected -- here by $50\%$, although on a relatively low amount of total ejected mass $<0.005M_\odot$. In a separate set of simulations varying the total mass of the binary away from its prompt collapse threshold, we find rapid variations in the amount of ejected matter and in the geometry and composition of the outflows with the total mass of the system. Finally, we use the simulations with our more advanced transport scheme to study in more detail the energy spectrum of neutrinos across the merger remnant.

Non-Hermitian Hamiltonians are a fascinating class of many-body models that describe the effective dynamics of quantum systems interacting with the environment through particle, energy, or information exchange. Although their theoretical framework is well established, the controlled engineering of such Hamiltonians in the context of quantum simulations remains challenging, even more so when the non-Hermitian part describes a $k$-body interaction. Qudit quantum simulators offer a compelling framework to implement such models. We theoretically investigate the dynamics of a one-dimensional chain of qudits undergoing hybrid unitary-projective evolution, where suitably designed measurements constrain the dynamics to a Zeno subspace. As we illustrate for the case of qutrits, within the Zeno subspace the dynamics is governed by an effective non-Hermitian Hamiltonian for an ensemble of pseudo-spins $1/2$, which can inherit non-Hermitian two-body interactions with the same connectivity as the full qutrit chain. We derive an analytical relation linking the monitored qutrits' evolution to a desired target non-Hermitian Hamiltonian and validate the effective description through numerical simulations of a representative model. Our scheme provides a constructive route for the realization of a large class of interacting non-Hermitian many-body Hamiltonians in experimentally relevant multilevel quantum platforms, including trapped ions and superconducting circuits.

Infrared singularities of gapless Goldstone modes preclude magnetic long-range order at finite temperature in conventional two-dimensional systems. By studying the spin-$S$ Heisenberg model on regular tilings of the hyperbolic plane, we show that this obstruction is absent in negatively curved space. Using spin-wave theory, we find that the zero-energy collective modes required by symmetry carry vanishing local spectral weight and are separated from the thermodynamic bulk magnon continuum by a finite gap in the bulk local spectral density. As a result, local transverse correlations remain short ranged, with a finite correlation length, despite the presence of Goldstone modes associated with the broken SO(3) spin-rotation symmetry. Stronger negative curvature is found to suppress quantum fluctuations in bulk thermodynamic quantities, pushing the ordered state toward "mean-field-like" behavior. We further estimate the ordering temperature from the thermal spin-wave correction to the ordered moment. These results establish hyperbolic geometry as a route to finite-temperature magnetic order that circumvents the Mermin-Wagner obstruction without breaking or modifying the continuous symmetry.

Because massive particles asymptote to timelike rather than null infinity, current flat space holographic proposals such as celestial or Carrollian holography struggle to describe scattering processes with massive external states. We take a step toward addressing this limitation by studying how information about massive scattering amplitudes is carried to the late-time limit of null infinity by soft graviton radiation. We show that continuity between the boundaries of timelike and null infinity implies that the late-time limit of the Bondi mass aspect naturally acts as a detector operator for massive outgoing radiation. We further relate in-in correlation functions of the Bondi mass aspect at $\mathscr{I}^+_+$ to weighted sums of scattering cross sections, implying that an observer at $\mathscr{I}$ can extract information about massive scattering processes at late times. Finally, we interpret the Bondi mass aspect as a Carrollian stress-tensor component, and study Ward identities to constrain its two-point functions.

The gravitational-wave sky is expected to contain a rich variety of sources across a very broad range of frequencies. Much like in electromagnetic astronomy, exploring new gravitational-wave frequency bands therefore has the potential to unlock powerful new insights into the Universe. In this chapter, we investigate the prospects for using high-precision timing of binary millisecond pulsars with the Square Kilometre Array Observatory (SKAO) to search for gravitational waves in the microhertz ($μ$Hz) frequency band by targeting resonant gravitational-wave perturbations to the orbits of these binaries. Using only a handful of known systems, we show that SKAO observations can achieve unprecedented sensitivity to microhertz gravitational waves, with the potential to detect inspiralling massive black hole binaries in this band. These searches are complementary to conventional pulsar timing array analyses, adding a new dimension to the gravitational-wave science achievable with the SKAO.

Molecular ground-state energies help determine conformer rankings, reaction energetics, and electronic effects in computational drug discovery, but accurate calculations become difficult when strong correlation or large active spaces are important. Variational quantum eigensolvers estimate these energies by optimizing a parameterized quantum state, making ansatz design central to both accuracy and cost. We study a fixed-topology Givens-exchange ansatz that avoids architecture search. The circuit starts from the computational-basis state with the lowest diagonal Hamiltonian expectation and applies local RY rotations with two ordered all-pair Givens exchange blocks. Parameters are optimized using Hamiltonian expectation values, while exact diagonalization is used only after optimization to compute errors and fidelities. Across six fixed seeds, coefficient-verified LiH-6 and H2O-8 Hamiltonians, together with a BeH2-6 public-specification candidate, are chemically accurate in every run. The corresponding six-seed mean errors are 0.000000124 Hartree, equivalent to 0.000124 milli-Hartree; 0.000128558 Hartree, equivalent to 0.128558 milli-Hartree; and 0.000002152 Hartree, equivalent to 0.002152 milli-Hartree, respectively. On LiH-6 and H2O-8, these mean errors are lower than the published point errors of the compared quantum-architecture-search methods, while the ansatz uses a larger pre-compilation macro budget. The method is therefore an accurate, reproducible, and search-free reference template for molecular variational eigensolvers.

A quantum battery charged through a lossy intermediate state faces a structural trade-off between charging speed and dissipation. We show that an exact algebraic cancellation removes it in a driven three-level cell: the radiatively decaying state is fed by a single bright amplitude, and a counterdiabatic field annuls the lone residual source that drives it, holding the lossy state identically empty. Charging is then lossless -- not one photon is emitted through the bridge -- at any one-photon detuning, coupling, linewidth, and speed down to the rotating-wave limit, with no adiabatic elimination, so the charging power is bounded by the drive amplitude (a quantum speed limit) rather than by dissipation. Crucially, this losslessness is independent of the reservoir: because the dark sector never engages the system-bath coupling, the emission vanishes exactly for an arbitrary spectral density, Markovian or not, as an exact damped-pseudomode treatment confirms to machine precision across all memory times. The entire non-Hermitian structure -- a Markovian second-order exceptional point that reservoir memory promotes to a third-order one, and the attendant dissipation phase diagram -- lives in the bright sector, from which the protocol is by construction exempt. This inverts dissipation-engineered charging, where an exceptional point or reservoir memory is a resource; here the lossy sector is never populated at all. The same dark-state structure protects the stored charge, converting fast radiative self-discharge into the slow metastable lifetime, with residuals quadratic in the control error. We detail experimental requirements and representative parameters for neutral alkaline-earth atoms, trapped ions, transmons, and defect centers.

Arguments for emergent spacetime require that quantum typicality, the generic absence of inter-subsystem correlations, persists on the physical Hilbert space of a gauge theory, where non-Abelian constraints could in principle inject geometry-supporting entanglement. Using $SU(2)$ lattice gauge theory on two-dimensional tori ($d_{\mathrm{phys}}$ up to $4{,}193$), we show that it does: the typical mutual information between strictly disjoint links matches an exact parameter-free analytical prediction combining a microcanonical baseline with Haar-random fluctuations. The Kogut-Susskind Hamiltonian generates correlations from states of definite geometry (such as the electric vacuum), while generic states show only regression to the mean, establishing that the arrow of correlation growth requires a non-generic initial condition.

We study the near-threshold scattering and bound-state structure of the $NΩ$ system by solving the Lippmann-Schwinger (L-S) equation within the framework of the meson exchange model and the Pomeron exchange model. The numerical results indicate that after incorporating the Pomeron exchange mechanism, the observables of the ${^5}S{_2}$ channel, such as the binding energy, scattering length, and effective range, agree better with the experimental measurements. In addition, The Pomeron exchange can provide an extra attractive interaction to make the hadronic state more compact. We also predict the scattering behavior of the ${^3}S{_1}$ channel and confirm that a weak quasi-bound state exists in this channel. Future experimental measurements on the ${^3}S{_1}$ channel will provide an important criterion for verifying the dynamic role played by the Pomeron exchange mechanism within the $NΩ$ system.

We provide the first mapping of the critical surface in (2+1)-flavor QCD in the full $(T,μ_B,μ_Q,μ_S)$ space, anchored on lattice QCD results at vanishing chemical potentials and obtained within an expansion along contours of constant entropy density. In the pure $μ_B$ direction, this framework yields a critical point at $(T_c,μ_{B,c}) \simeq (114,\, 602)$ MeV. Here we extend the construction to arbitrary directions in the three-dimensional chemical-potential space, parametrized by spherical coordinates $(μ,θ,\varphi)$, with the radial expansion truncated at $\mathcal{O}(μ^2)$. The resulting two-dimensional surface carries a direction-dependent critical temperature $T_c(θ,\varphi)$ and baryochemical potential $μ_{B,c}(θ,\varphi)$, which quantify the shift of the critical point relative to the pure $μ_B$ direction. We find that $μ_{B,c}$ increases by 40-100 MeV along the approximately strangeness neutral direction [$μ_S \approx (0.15$--$0.33)\, μ_B$, $μ_Q \approx 0$] relevant for heavy-ion collisions, while the critical temperature stays essentially unchanged. In the charge-neutral, weak-equilibrium direction~[$μ_Q \approx -(0.05$--$0.1) \,μ_B$, $μ_S = 0$] relevant for neutron star mergers, the critical point, and the associated first-order phase transition, remain present at essentially the same location in the $(T,μ_B)$ plane. We find no evidence for a critical point at large isospin densities, $|μ_Q| / μ_B \gtrsim 1$, relevant for cosmic trajectories in the early Universe, nor along the pure electric-charge or strangeness directions, at least outside the regions where pion or kaon condensation may occur.

We study the classical tree-level scattering amplitudes of deep-water surface gravity waves using the methods of high-energy physics. For scattering in one horizontal dimension and in the two-negative-wavenumber sector we obtain a closed formula for $n$-wave scattering. Up to a kinematic prefactor, the amplitude is the volume of a classic polytope -- a box sliced by a hyperplane, which we dub the hydrotope, whose purpose in life is simply to organize the sign patterns of the "chambers" characterizing all the different regions of the two-minus kinematic space. The general formula was discovered by Claude Opus 4.6 working under our guidance, beginning with our earlier discovery of a one-term expression valid in the "simplest" kinematic chamber. Our results resolve the puzzle raised by Y.V. Lvov's 1997 computation of the five-wave amplitudes, unifying and extending it to all multiplicities.

Symmetry-protected seesaw models can accommodate light-neutrino oscillation data while keeping heavy neutral leptons (HNLs) within collider reach. In these models, the smallness of the light-neutrino masses is protected by an approximate lepton number (LN)-like symmetry that is broken only by small parameters. We study the minimal scenario in which the new states form one pseudo-Dirac HNL pair. The exact LN-conserving Dirac limit is diagonalised without expanding in the active-sterile mixing, and the small LN-violating entries are then included perturbatively. This yields a symmetry-protected flavour reconstruction of the active-heavy interaction matrix. The rank-two light-neutrino mass matrix fixes the normalised active-flavour direction, while the remaining high-energy information is a single complex light-heavy amplitude whose phase defines a CP-odd light-heavy invariant. For the normalised leading active-heavy interaction weights, this amplitude and the heavy-sector rotation cancel, leaving an ellipse in the flavour simplex determined by light-neutrino oscillation data and the Majorana phase. We also identify how the linear LN-violating terms enter coherent heavy-neutrino oscillations and the neutrinoless double beta effective mass.

We propose a new formulation of continuous-spin particles (CSP) with the help of a single two-component spinor to build asymptotic states. This formulation allows to write down amplitudes in a straightforward way, similar to the massless spinor-helicity approach. The helicity states naturally emerge upon decomposing into components of a fixed homogeneity degree. We show that CSP amplitudes can be understood as the infinite-spin limit of amplitudes of massive particles and, similarly to the scattering of black holes, a certain universal factor exponentiates. In analogy with the recent discovery of collinear amplitudes in self-dual theories, we find nontrivial CSP collinear amplitudes where the collinear fractions are constrained by the dimensionful parameter characterising the various continuous-spin particles.

Reliable nuclear matrix elements (NMEs) are essential for interpreting double-beta-decay experiments and for connecting measured or constrained half-lives to the underlying weak-interaction physics. The two-neutrino mode ($2νββ$) is allowed by the Standard Model and has been observed in several nuclei, whereas the neutrinoless mode ($0νββ$) remains the key experimental signature of lepton-number violation and Majorana neutrino masses. Recent statistical shell-model studies indicate a strong correlation between the $2νββ$ and $0νββ$ NMEs, making accurate and efficient calculations of the former especially useful for assessing the latter. Direct evaluations of $2νββ$ NMEs usually require summing over many $1^+$ states in the intermediate odd-odd nucleus, a procedure that becomes expensive and may converge slowly in large model spaces. We present and test an improved strength-function method based on Lanczos iterations that avoids full diagonalization while preserving the accuracy of explicit summation where such benchmarks are possible. The method is applied to several experimentally important emitters and to different effective Hamiltonians. We also show that the same framework can be used for the higher-order NMEs entering Taylor-expanded phase-space treatments of $2νββ$ and related decay modes.

A first global analysis of unpolarized Transverse-Momentum-Dependent (TMD) distributions based on a neural-network (NN) parametrization is presented. Drell-Yan (DY) and semi-inclusive deep inelastic scattering (SIDIS) data are simultaneously included at next-to-next-to-next-to-leading logarithmic (N$^3$LL) accuracy. The results indicate that the inclusion of SIDIS data leads to broader unpolarized TMD PDFs compared to a DY-only NN extraction. The associated uncertainties are reduced with respect to the DY-only case, while remaining larger than the ones obtained using traditional models. These results demonstrate the potential of flexible NN parametrizations in reducing model dependence and provide guidance for future high-precision measurements at Jefferson Lab and the Electron-Ion Collider.

We investigate the analytic structure of the quark propagator in the Landau gauge by dynamically coupling the standard gap equation to the non-perturbative quark-gluon vertex. Employing the full vertex basis, we demonstrate that for sub-GeV time-like momenta, the proper inclusion of the underlying dynamics leads to a pair of real poles with opposite-sign residues. In particular, in stark contradistinction to the results obtained in widely used approximations, we see no sign of complex conjugate poles. This distinctive analytic structure evades conceptual shortcomings frequently associated with complex conjugate poles while remaining fully compatible with the aspects of color confinement related to positivity violation. Crucially, this novel behavior is governed by a dominant triplet of vertex form factors: the tree-level component, the anomalous chromomagnetic moment, and a component we label as "spin-momentum curvature". By gradually tuning the individual strengths of these components, we demonstrate that while they contribute in distinct ways to the quark propagator, their joint action is vital for stabilizing the system. Together, they place the low-lying poles onto the real axis while producing a robust constituent quark mass of $350$ MeV.

We bootstrap quantum anomalies using on-shell techniques in the simplest setting: at three points. The necessary field-theoretic input includes the local or global classical symmetry, which the anomaly may break, and the symmetries of the effective action -- though not its explicit form. We use massless helicity spinors together with an on-shell representation of the discrete C, P, and T transformations. Our approach determines the Weyl, chiral (global and local), diffeomorphism, and Lorentz anomalies up to real constant prefactors. In particular, we recover gauge-anomaly cancelation conditions, as well as find that the Weyl anomaly should not involve Pontryagin densities.

Bayesian inference of the neutron star (NS) equation of state (EOS) generally assumes that improved observations primarily reduce posterior uncertainties while leaving inferred EOS parameters unchanged. Using mock measurements of the radius of a canonical $1.4\,M_\odot$ NS with identical central values but varying observational precisions, we show that the inferred posterior means of EOS parameters can shift systematically as the measurement uncertainty changes. We demonstrate that this behavior originates from previously unidentified nearly universal inverse mappings between the NS radius $R_{1.4}$ and empirical EOS parameters. Across a broad range of observational precisions, posterior samples collapse onto nearly unique functions. These mappings are largely independent of observational precision and define a low-dimensional EOS manifold underlying Bayesian inference. We show that the precision dependence of inferred EOS parameters arises from nonlinear filtering of the posterior radius distribution through these mappings. In the narrow-distribution limit this effect reduces to a Jensen-type correction proportional to the local curvature of the inverse mapping, while for presently realistic uncertainties the full nonlinear-filtering relation accurately reproduces the posterior means. Our results reveal a geometric origin of precision-dependent inference in NS EOS studies and provide a new framework for connecting astrophysical observations directly to microscopic nuclear many-body theories.

Probing the partonic structure of the pion and the kaon provides essential insights into the non-perturbative dynamics of QCD, yet their parton distribution functions (PDFs) remain poorly constrained due to the scarcity of high-precision experimental data, especially for the gluon distributions. We present an improved determination of pion and kaon PDFs within the dynamical parton model combined with the maximum entropy method (MEM) framework. Our analysis features two key advancements: firstly, we employ an infrared-safe QCD evolution scheme, allowing the evolution to be reliably extended down to very low $Q^2$, approaching the hadronic scale; secondly, we incorporate pion- and kaon-induced $J/ψ$ hadroproduction data as crucial constraints in the global fit. We find that our approach yields a good description of the available Drell-Yan data, deep-inelastic scattering structure functions ($F_2$), and $J/ψ$ production across various energies and targets. The results provide significantly improved constraints on the gluon distributions at moderate and large $x$ in both the pion and the kaon, offering a more complete picture of their internal structure.

Spectral densities link experimental measurements to dynamical properties of a quantum field theory which, in turn, can be resolved non-perturbatively from the Euclidean time-dependence of correlation functions. By making extensive use of integral transforms, we present analytic formulae to carry out the inverse Laplace transform so as to extract spectral densities from either the continuum or the discrete sampling of correlation functions in the Euclidean time. Formulae extend to regulated and/or smeared spectral densities as well. We explicitly show that the proposed lattice solution tends to its continuum counterpart up to $O(a^2)$ effects in the lattice spacing $a$ if the lattice correlator is $O(a)$-improved. In practical computations, lattices have necessarily a finite Euclidean temporal extent, a lack of knowledge which suggests to introduce incomplete integral transforms and the corresponding incomplete smeared spectral densities. The contribution from the unknowns to a smeared spectral density can then be rigorously bound and kept under control if the integral transform of the smearing function decays fast enough with the conjugate variable. Conversely, the bound can be used to plan lattices so as to achieve a given target precision on the reconstructed spectral density of interest. The formulae presented here in the context of lattice field theory can be easily applied or extended to other areas of research.

We investigate the configurational temperature estimator in interacting matrix models and Gaussian random-matrix ensembles. The estimator follows from an exact Schwinger--Dyson identity and may be expressed in terms of the gradient and Hessian of the action. We study the Gross--Witten--Wadia model, a quartic double-well matrix model, and the Gaussian Orthogonal, Unitary, and Symplectic Ensembles. In all cases, the estimator satisfies the exact Schwinger--Dyson identity, $β_{\rm config} = 1$, within statistical uncertainties. Separating the estimator into isotropic and anisotropic parts, we find that the leading finite-$N$ corrections satisfy the approximate relation $β_{\rm iso} - 1 \simeq - β_{\rm aniso}$. We also show that the configurational temperature estimator provides a sensitive diagnostic of Monte Carlo simulations.

The electron neutrino degeneracy parameter, $ξ_{ν_\mathrm{e}} = μ_{ν_\mathrm{e}} / T$, is tightly constrained by Big Bang Nucleosynthesis (BBN), while the degeneracy parameters of the other neutrino species, $ξ_{ν_\mathrm{x}}$, remain weakly constrained by cosmological observations alone. In this manuscript we shall compute up-to-date bounds on $ξ_{ν_\mathrm{e}}$ and $ξ_{ν_\mathrm{x}}$ assuming that either they are constant free-parameters along the cosmic history or that they are redshift dependent quantities. In the latter case we employ a model-independent reconstruction approach based on the Piecewise Cubic Hermite Interpolating Polynomial (PCHIP) formalism with four nodes, located at $z\simeq$ 10, 100, 1000 and $10^8$. We shall also consider two scenarios for neutrinos, specifically three degenerate neutrinos ($ξ_{ν_\mathrm{e}}$ = $ξ_{ν_\mathrm{x}}$) and the case in which we actually differentiate between $ξ_{ν_\mathrm{e}}$ and $ξ_{ν_\mathrm{x}}$. We perform a cosmological analysis combining CMB data from Planck, SPT, and ACT with BAO measurements from DESI, showing the impact of including BBN observables from either EMPRESS results, which allow for a non-zero chemical potential, or from LBT observations, compatible with the standard $ξ_ν$ = 0 prediction. We explicitly show that the BBN data, via the change in neutron-to-proton interconversion rates, mostly constrain $ξ_{ν_\mathrm{e}}$, parameter for which we observe a preferred non-zero positive value at $95\%$ C.L. in the non-degenerate neutrino case at the BBN period. Since the Hubble constant is correlated with $ξ_ν$, through $N_{\rm eff}$, a larger value of $H_0$ is allowed within these models, making them really interesting scenarios where to test non-standard physics models.

The QED scattering amplitude in a chiral medium characterized by a constant chiral chemical potential $μ_5$ and chiral magnetic conductivity $b_0$ is analyzed. We show the emergence of the resonant behavior in $1\to 2$, $2\to 2$, and $2\to 3$ processes. We compute the rates of paradigm $1\to 2$ processes that determine the widths of quasi-stationary fermion and photon states in the medium. We elucidate the origin of these resonances, the conditions of their emergence, and the physical principles of their regularization.

We present a comprehensive review of regularization schemes for magnetized dense quark matter within effective models of quantum chromodynamics, focusing on the Magnetic-Field-Independent Regularization (MFIR) and the Medium Separation Scheme (MSS) at finite chemical potential and magnetic field. In nonrenormalizable frameworks such as the Nambu-Jona-Lasinio model, the treatment of ultraviolet divergences is crucial, particularly in magnetized and dense environments where conventional regularization procedures may introduce unphysical artifacts. We show that MFIR consistently isolates divergent vacuum contributions from finite magnetic-field-dependent terms, while MSS extends this separation to the medium sector, ensuring that only vacuum quantities are regularized. Within this unified framework, we analyze the thermodynamics of cold and dense quark matter, including color-superconducting phases, and demonstrate that the superconducting gap remains finite at large chemical potentials, even in the presence of strong magnetic fields. In contrast to results obtained with traditional regularization schemes, we find no evidence for a transition to a normal phase at zero temperature, highlighting the importance of a proper separation between vacuum and medium contributions. These results eliminate spurious oscillations and other nonphysical artifacts, leading to a more robust and physically consistent description of strongly interacting matter under extreme conditions relevant to compact stars and heavy-ion collisions.

We present a systematic procedure for constructing brane brick models corresponding to abelian orbifolds of toric Calabi-Yau 4-folds, extending the orbifold construction beyond the well-studied case of abelian orbifolds of C^4. Given a parent brane brick model corresponding to a toric Calabi-Yau 4-fold M, we show that the action of an abelian orbifold group Gamma on the generators of M induces an action on the chiral and Fermi fields as well as the J- and E-terms of the associated 2d (0,2) supersymmetric gauge theory. The requirement that the orbifolded brane brick model remains consistent with the closed paths associated with the J- and E-terms, together with the chiral cycles formed by their products, precisely reproduces the Calabi-Yau condition on the orbifold action. This procedure yields an explicit formula for the J- and E-terms of the orbifolded brane brick model in terms of those of the parent theory. We apply our construction to the brane brick models corresponding to Q^{1,1,1} and D_3, and present the resulting families of 2d (0,2) quiver gauge theories. We also present explicit expressions for generating functions that count distinct abelian orbifolds of Q^{1,1,1} and D_3.

We present a first-principles lattice-QCD determination of Mellin moments of the unpolarized pion and kaon parton distribution functions using matrix elements of boosted mesons coupled to nonlocal operators containing a straight Wilson line. The calculation is performed on an $N_f=2+1+1$ ensemble of maximally twisted-mass fermions with a clover term, with lattice volume $32^3\times64$, lattice spacing $a=0.0934$ fm, and pion mass $m_π=260$ MeV. Matrix elements are computed for hadron momenta $P_3=0$, 0.41, 0.83, 1.25, 1.66, and 2.07 GeV and analyzed within the short-distance factorization framework. We investigate the dependence of the extracted moments on the truncation of the operator-product expansion, the coordinate-space fit window, and the perturbative accuracy of the Wilson coefficients, comparing next-to-leading-order and next-to-next-to-leading-order results. We also perform an RG-improved analysis as a consistency check of the perturbative treatment. Our final results are obtained from combined fits in $(P_3,z)$ space at next-to-next-to-leading-order and are quoted at $μ=2$ GeV. We also study the SU(3) symmetry-breaking effect and reconstruct the valence PDFs from the moments.

By utilizing novel lobster-eye optics, the Wide-field X-ray Telescope (WXT) onboard the Einstein Probe (EP) satellite achieves an unprecedented combination of a large instantaneous field-of-view (FoV) and high sensitivity for monitoring the dynamic X-ray sky. In this paper, we present the in-orbit calibration results of the WXT during its first two and a half years of operations. By conducting observations of standard celestial sources--including the Crab Nebula, Scorpius X-1, and Cassiopeia A--we systematically characterized key instrumental properties. Our analysis demonstrates that the in-orbit performance of the WXT agrees with prelaunch ground calibrations well. The spatial resolution, denoted by the full width at half maximum (FWHM) of the focal spot, typically ranges from $3'$ to $6'$ across $\sim$90% of the FoV, with a median of $\sim 4.3'$. The post-calibration source positioning accuracy achieves $1.3'$ (at the 90% confidence level). The in-orbit effective area is consistent with model predictions and ground measurements, exhibiting an overall systematic uncertainty of $\lesssim 10\%$ (90% C.L.) in the 0.5-4 keV band. While the vast majority of the detectors remain highly stable, a noticeable long-term degradation at the low-energy end ($\sim30\%$-$40\%$, 0.4-0.6 keV) is observed in a few specific modules. Furthermore, spectral evaluations using Cas A confirm the stability of the energy scale and spectral resolution of the focal-plane Complementary Metal-Oxide Semiconductor (CMOS) detectors. All derived calibration products have been incorporated into the WXT calibration database (CALDB). These results comprehensively verify the instrumental capabilities of the WXT, providing a solid foundation for the reliable analysis of scientific observations.

Femtoscopic analysis can shed light on hadron production in pp collisions. In this paper, proton-proton correlations measured in collisions at $\sqrt{s}=13.6$ TeV recorded with the ALICE detector at the LHC are presented. The analysis is based on the minimum bias dataset collected in 2022 following the upgrade of the ALICE detector and corresponds to an integrated luminosity of $19.3$ pb$^{-1}$. The increased integrated luminosity allows us, for the first time, to simultaneously measure the multiplicity and transverse-mass ($m_{\rm T}$) dependence of the size of the hadron-emitting source. Precise knowledge of the femtoscopic source size in pp collisions is a crucial ingredient for using femtoscopy to study the residual strong interaction among stable and unstable hadrons at the LHC. In this light, the source radius was determined from the measured correlation functions by assuming several state-of-the-art models of the nucleon$-$nucleon interactions. The consistency among the extracted radii demonstrates the robustness of the measurement with respect to interaction model assumptions. A comparison to femtoscopic radii measured in Pb$-$Pb collisions at $\sqrt{s}=5.02$ TeV reveals a markedly different multiplicity dependence in similar $m_{\rm T}$ intervals, providing new insight into the system-size dependence of particle emission dynamics.

In this work, we study the diquark-diquark-antiquark type decuplet $qssc\bar{c}$ pentaquark states with the QCD sum rules comprehensively, and obtain the mass spectrum of the lowest decuplet $qssc\bar{c}$ pentaquark states with the quantum numbers $IJ^{P}=\frac{1}{2}{\frac{1}{2}}^-$, $\frac{1}{2}{\frac{3}{2}}^-$ and $\frac{1}{2}{\frac{5}{2}}^-$, and suggest to search for those exotic states in the processes $Ξ_b^{\prime0} \to P_{css}^0\,φ\to J/ψΞ^{*0} φ$ and $Ω_b^{-}\to P_{css}^-\, \bar{K}^0 \to J/ψΞ^{*-}\, \bar{K}^0$. As a byproduct, we can examine classifications of the light baryons by studying the pentaquark decays.

We investigate the phase transition of Quantum Chromodynamics (QCD) with three degenerate quark flavors at zero baryon chemical potential. Using Möbius domain-wall fermions as the lattice fermion formulation, we ensure excellent chiral symmetry preservation. Our simulations are performed at three different temporal lattice extents, $N_{t}=6, 8, 12$, with a fixed lattice spacing $a=0.1361(20)$ fm, corresponding to temperatures of 242(4), 181(3), and 121(2) MeV, respectively. We explore a range of quark masses and spatial volumes with aspect ratios $N_{s}/N_{t}$ spanning from 2 to 4. By analyzing the mass and volume dependencies of the plaquette, plaquette susceptibility, chiral condensate, chiral susceptibilities, and Binder cumulant, we identify the pseudocritical transition quark masses from our largest lattice volumes. For $N_t=6$, this is 184(10) MeV (determined from the plaquette susceptibility). For $N_t=8$ and 12, the transition points vary slightly depending on whether the total or disconnected chiral susceptibility is used, yielding ranges of 36(1)-39.1(9) MeV and 3.5(3)-3.7(2) MeV, respectively, in the $\overline{\text{MS}}$ scheme at a scale of $μ=2$ GeV. The negligible volume dependence at $N_t=6$ and 8, combined with finite-size scaling analysis at $N_t=12$ revealing volume growth significantly weaker than expected for a first- or second-order phase transition, points to a continuous crossover at these specific quark mass points. Additionally, we study the effects of residual chiral symmetry breaking on the chiral condensate and chiral susceptibilities using two different values of $L_s$.

We construct simple analytical solutions of the RG equations for the running couplings α_s and y_{top} in the asymptotic regime. These solutions have an explicit form, contain only logarithms and no special functions, and subsequently sum up the leading, subleading, etc logarithms in all orders of PT. While the effect of y_{top} on the running of α_s happens to be negligible, the role of α_s on the running of y_{top} is essential, the account of higher orders gives a noticeable contribution.

We present a direct reconstruction formula for planar loop integrands from $D$-dimensional generalized unitarity cuts in any colored theory. The reconstruction combinatorics is separated from the theory-dependent tree amplitudes entering the cuts: for the $L$-loop $n$-point color-ordered amplitude, the integrand is expressed as a sum over admissible non-scaleless scalar graphs dressed by corresponding cuts in $D$ dimensions; the coefficients are given by the universal Möbius-inversion formula of the refinement poset, or equivalently one minus the Euler characteristics of associated complexes. As an application we write down closed-formulas for loop integrands in pure Yang--Mills theory, where the required cuts are generated by gluing $D$-dimensional tree amplitudes and summing over internal gluon states. We also use the two-loop five-point case as a validation, comparing with known integrand data and after integration-by-parts reduction, with known integrated helicity amplitudes. The same framework also produces compact cut-organized data for larger examples, including the two-loop six-point and three-loop four-point cases. We also describe the corresponding simplification in maximally supersymmetric Yang--Mills theory, where the absence of bubble and triangle subgraphs reduces the relevant cut poset substantially.

We construct the traceless components of the energy-momentum tensor on the lattice for QCD with $N_f=3$ flavours, such that their correlation functions satisfy the appropriate Ward identities in the continuum limit. To carry out this program, we define the theory on the lattice by the Wilson-plaquette and the $O(a)$-improved Wilson actions for gluons and quarks respectively. The discretization of the space-time entails that (i) the irreducible nonet representation of the SO($4$) group splits into a triplet and a sextet irreducible representations of the hypercubic group, and (ii) for each multiplet non-perturbative determinations of the the gluonic and fermionic renormalization constants are required. The bare gluonic components of the energy-momentum tensor are defined via the clover discretization of the field strength tensor, while the fermionic ones are discretized by appropriate combinations of symmetric covariant derivatives. Either for the triplet or the sextet representations, the two independent renormalization constants are then fixed non-perturbatively by imposing discretized versions of continuum Ward identities for one-point correlation functions in the presence of shifted boundary conditions and an imaginary chemical potential. The non-perturbative calculation is then carried out by Monte Carlo simulations, and the resulting renormalization constants are determined with a final accuracy of a few percent for values of the bare coupling constant squared in the range $0 \leq g_0^2\leq 0.96$.

Microstates of BPS AdS black holes have been studied from the classical cohomologies of the maximal super-Yang-Mills theories, but their quantum natures have been conjectural. It was recently found that a classical black hole (fortuitous) cohomology in the $SO(7)$ theory is lifted by 1-loop corrections. We show that such lifts also happen in the $SU(2)$ theory, presenting both lifted/unlifted examples. In particular, the lightest fortuitous cohomology and its `hairy' versions are unlifted, while many heavier `core' fortuitous ones are lifted. We argue that the entropy of classical cohomologies in the Cardy limit is larger than the indicial entropy of strictly protected states by at least $\approx 1.2 \%$.

We introduce a two-dimensional $O(N)$ nonlinear sigma model with random Gaussian $p$-body interactions. The model combines the structure of a two-dimensional bosonic SYK-type quantum field theory with the stabilizing spherical constraint of the nonlinear sigma model. At large $N$ we derive the Schwinger-Dyson equations on a torus and analyze the solutions using both analytical approximations and numerical methods. We find a phase diagram qualitatively similar to that of the one-dimensional quantum spherical $p$-spin model, including a low-temperature transition to a spin glass phase. This phase is characterized by a finite Edwards-Anderson order parameter, while the dynamical part of the two-point function displays an approximate scaling regime. These results provide a tractable setting for studying approximate conformal behavior and glassy physics in a two-dimensional relativistic field theory with disorder.

Light antinuclei, such as antideuteron ($\bar{\mathrm{d}}$) and antihelium (${}^{3}\overline{\mathrm{He}}$,${}^{4}\overline{\mathrm{He}}$), provide a link between collider physics and indirect Dark Matter searches. Despite extensive studies of antinucleus production in high-energy collisions, $\overline{\mathrm{d}}$ production from beauty-hadron decays remains experimentally unconstrained and has not yet been quantitatively predicted. In this work, we present the first phenomenological study of $\overline{\mathrm{d}}$ production from $\overlineΛ_{b}$ baryon and B$^{-}$ meson decays, providing the first estimates of the corresponding branching ratios. Beauty-hadron decays are simulated with PYTHIA using realistic input kinematics and three hadronization scenarios. Antideuteron formation is modelled through a quantum-mechanical coalescence approach based on an $\bar{\mathrm{d}}$ wave function derived from the Argonne $v_{18}$ nucleon-nucleon potential. Depending on the adopted hadronization model, we estimate inclusive branching ratios to be $(5.68 \pm 0.02)\times10^{-4} < BR(\overlineΛ_{b} \rightarrow \overline{\mathrm{d}}+X) < (1.408 \pm 0.004)\times10^{-3}$ and $(7.4 \pm 0.3)\times10^{-6} < BR(B^{-} \rightarrow \overline{\mathrm{d}}+X < (4.34 \pm 0.07)\times10^{-5}$. The predicted rapidity- and transverse-momentum-differential yields populate the kinematic region where $\overline{\mathrm{d}}$ can be identified by the ALICE experiment, motivating dedicated searches for these decay channels. These results provide a quantitative benchmark for $\overline{\mathrm{d}}$ production from beauty-hadron decays and establish a phenomenological framework to support future experimental searches, with potential implications beyond collider physics.

Vector-like leptons (VLLs) are well-motivated candidates for physics beyond the Standard Model. We investigate the sensitivity of the $6$ TeV muon collider to the neutral doublet VLL (denoted as $N$) via its pair production within a general VLL framework. Taking vector-like muon as a case study, we study the subsequent decay $N\rightarrow W^+ μ$ and analyze two representative signals, namely $4j2μ$ and $2\ell2μE\mkern-10.5 mu/$, arising from the hadronic and leptonic decays of the $W$ boson, respectively. The signal and background events are simulated within a complete Monte Carlo framework, and a cut-based analysis is performed at the $6$ TeV muon collider with an integrated luminosity $\mathcal{L}=4$ ab$^{-1}$ and beam polarizations $(P_{μ^+},P_{μ^-})=(-1,1)$. We consider VLL masses in the range of $1300-3000$ GeV and evaluate the corresponding search sensitivity. Our results show that the future $6$ TeV muon collider can effectively probe neutral doublet VLLs through the $4j2μ$ and $2\ell2μE\mkern-10.5 mu/$ signals, with statistical significances exceeding $5σ$ over a broad mass range. These results demonstrate that the future $6$ TeV muon collider has excellent potential to search neutral doublet VLLs.

We consider the supersymmetric $λ$-deformation of $\text{AdS}_3 \times \text{S}^3 \times \text{T}^4$ superstrings and compute its perturbative bosonic tree-level worldsheet S matrix in the light-cone gauge. For generic values of $0 \leq λ< 1$, we show that the worldsheet scattering remains compatible with integrability due to a non-trivial cancellation of non-elastic scattering processes. By contrast, the S matrix becomes ill-defined for $λ\to 1$, despite the fact that this limit reproduces the non-Abelian T-dual geometry up to an analytic continuation. This suggests that the $λ\to 1$ limit does not capture the full worldsheet dynamics of the T-dual theory.

One generation of standard-model fermions can be realized on the complexified octonions through the Clifford algebra $\mathcal{C}l(6)$; the octonionic unification programme extends this to three generations, with generation transport implemented by $G_2$ automorphisms or by rotors built from the ladder operators. We prove a localization theorem for the CP-violating phases of this structure, using only the $\mathcal{C}l(6)$ construction and the stated three-generation representatives, independently of the wider programme. For quarks, the first-to-second generation step is the occupation flip of one ladder mode, with the up and down species coupling to conjugate ladder directions; a conjugation theorem forces $A_d=A_u^*$ for every real transport, and the most general rung-generated rotor yields the exact one-parameter law $φ_{12}=-2χ$: the $(1,2)$ transport phase is twice one Yukawa orientation angle. The programme's geometric rotor sits exactly at the quadrature-balanced point $|φ_{12}|=π/2$; the companion analysis reproduces the Cabibbo \emph{magnitude} $|V_{us}|$ with a single real tilt, leaving the rung near quadrature, but it does not extract a CKM CP phase, so the quark Dirac phase is fixed only once the underlying Yukawa orientation is computed. For leptons we prove a reality theorem: every charged-lepton and every neutrino transport amplitude is exactly real for every $G_2$ automorphism and every rotor that does not mix the identity line $\mathbb C\cdot1$ with the lepton--flavor plane $\mathrm{span}(e_7,e_5,e_2)$ a class that contains the entire quark-rung family--and identity--flavor mixing across that plane is the unique possible source of a leptonic phase. [Truncated]

The maximally transcendental, or most complicated, terms of gauge-theory scattering amplitudes have long been singled out, following Lipatov and collaborators, as those parts of a QCD amplitude that most closely mirror maximally supersymmetric Yang--Mills theory. We report on a programme that turns this observation into a practical computational tool. We show that the rational prefactors multiplying the highest-weight special functions of planar QCD amplitudes are governed by four-dimensional leading singularities, which can be classified and evaluated using on-shell diagrams. The resulting prefactors are manifestly conformally invariant and admit compact spinor-helicity expressions that hold for arbitrary multiplicity. Combining this input with the recently established two-loop six-particle function space, we set up a symbol bootstrap and determine, for the first time, the maximal-weight symbol of the planar two-loop six-gluon amplitude in massless QCD, first for the ${-}{-}{+}{+}{+}{+}$ helicity configuration and subsequently for all MHV configurations. The answer is fixed uniquely by physical consistency conditions, requires a reduced alphabet of only $137$ symbol letters, and yields as a byproduct previously unknown two-loop triple-collinear and double-soft splitting functions. We summarise the method and the results, and outline the directions they open up.

We present HydroGrav, a C++ code used to construct self-similar fluid profiles, using the exact equation of state determined directly from the effective potential, for any particle physics model capable of producing a first-order electroweak phase transition. HydroGrav also supports the bag and $μν$ (or improved bag) equations of state and includes an implementation of the sound shell model for computing the corresponding gravitational wave spectra. Using this framework, we compare the fluid profiles and gravitational wave spectra for the simplified (bag and $μν$) and exact equations of state for a $\mathbb{Z}_2$-symmetric extension of the Standard Model. Furthermore, we perform a scan across the parameter space of this model to identify regions where the simplified and exact equations of state differ in peak amplitude and spectral shape. Finally, we estimate the effect of using the exact equation of state on the signal-to-noise ratio across the parameter space, as measured by LISA after a 4-year mission.

The quark parton model grew out of deeply inelastic scattering experiments. The parton model developed into a full theory, quantum chromodynamics, QCD. This article explains some of the physics issues encountered in connecting the parton model and QCD.

We present a phenomenological study where we probe the sensitivity to invisible dark matter (DM) mediators produced in association with a $t\bar{t}$ pair at the Large Hadron Collider (LHC). Building on previous work focused on scalar mediators, we extend the analysis to include spin-1 mediators, $Y_1$, with both vector and axial-vector couplings to top quarks. The mediator mass is fixed to 5 GeV. Signal samples of $pp \rightarrow t\bar{t}Y_i$ ($i = 0, 1$) are generated using a MadGraph5_aMC@NLO simplified DM model. Only dileptonic final states of the $t\bar{t}$ system are considered, and the reconstruction is performed through a kinematic fit without explicitly reconstructing the invisible mediator. All relevant Standard Model backgrounds are included. We consider several exclusion scenarios to assess the sensitivity to the presence of a spin-1 mediator, as well as the ability to distinguish a pure vector or axial-vector mediator from alternative hypotheses with different spin and CP properties. We find that the analysis is sensitive to light spin-1 mediators and that CP-sensitive angular observables provide discrimination power between vector, axial-vector, scalar and pseudoscalar scenarios. These results highlight the potential of $t\bar{t}$ final states not only to search for invisible particles, but also to characterize their spin and parity properties in case of discovery.

Two recent studies of radiative corrections to bound state properties are discussed. The change of the decay rate of a muon bound to a light nucleus has been calculated for several light nuclei $4\leq Z \leq 9$ with high precision, resolving a long-standing discrepancy between analytical and numerical results for oxygen ($Z=8$). The decay of parapositronium into three photons has been calculated including effects of the $Z$ boson. The resulting rate is many orders of magnitude smaller than previously estimated.

CIPro is a Mathematica package for constructing and analyzing complete intersections in products of projective spaces and line bundles over such varieties. It computes properties of complete intersections, such as Chern classes, Hilbert series, GV invariants and symmetries, as well as properties of line bundles on complete intersections, including their cohomology groups. The package also consolidates a number of data sets available in the literature into a single system, including the lists of complete intersection Calabi-Yau three- and four-folds. This short tutorial introduces the package, provides a brief discussion of some of the mathematical background underlying its computations, and gives a series of examples to illustrate its use. These tools are of utility for many computations in string compactifications, especially for Calabi-Yau geometries appearing in Heterotic and Type II constructions. Many tools apply beyond the Calabi-Yau context, including for example, almost Fano bases in F-theory.

We present model calculations to reveal the nature of the newly found $Ω^-(2109)$ by the BESIII Collaboration, and show that the state has a strong correlation with the $\bar K^*(892)Ξ$ system. Our study is based on solving scattering equations in a coupled channel approach, which involves $K^-Ξ^0$, $\bar K^0Ξ^-$, $K^{*-}Ξ^0$, and $\bar K^{*0}Ξ^-$. We obtain the lowest order amplitudes for different spin and isospin cases and find that an isoscalar state with spin-parity $1/2^-$ is generated with precisely the same mass as $Ω^-(2109)$. We do not find any state with total spin 3/2, nor do we find any state in the isovector sector. We determine correlation functions to encourage such an experimental study and confirm the nature of $Ω^-(2109)$.

We compute the amplitude for the scattering of a gravitational wave off of a spinless point particle at fourth order in Newton's constant, using the worldline quantum field theory formalism. A decomposition of our master integrals incorporating Murua coefficients allows us to entirely bypass the cut subtraction needed to convert the scattering amplitude into the Magnusian, the latter being desirable as it maps directly onto the scattering phase shift in partial wave space. This is then matched to the prediction from black hole perturbation theory, proving that point-particle worldline quantum field theory accurately describes Schwarzschild black holes up to $\mathcal{O}(G^4)$. Elliptic functions appear in momentum space for the first time for this process at this order.

We establish a correspondence between the free-field construction and the minimal-model construction of the Calabi--Yau sector of the four-dimensional heterotic string compactified on Berglund--Hübsch type Calabi--Yau manifolds and their orbifolds. For Fermat-type polynomials the Calabi--Yau vertex operators expressed in terms of free fields are shown to correspond to products of primary fields of $\mathcal{N}{=}2$ minimal models. Using this correspondence we verify modular invariance of the free-field construction and extend it to Berglund--Hübsch Calabi--Yau orbifolds, deriving the conditions on complete vertex operators that parallel those of the minimal-model construction.

Recent results from the LHC on oxygen-oxygen (O-O) and neon-neon (Ne-Ne) collisions open a new window for investigating the interplay of cold nuclear matter (CNM) and quark-gluon plasma (QGP) effects in small collision systems. Building upon recent theoretical work on particle production dynamics in heavy-ion reactions, we present an updated study of light and heavy hadron modification relative to the proton-proton baseline in these systems for selected centralities. Our analysis combines perturbative QCD and hydrodynamic simulations to quantify initial-state effect, collisional energy loss, and medium-induced radiative corrections. We give theoretical predictions at both midrapidity and forward rapidity that can be confronted with ALICE, ATLAS, CMS, and LHCb measurements. Through comparison to the available data, we discuss the relative importance of CNM and QGP effects in O-O and Ne-Ne systems and the role of the heavy quark mass. Our analysis aims to clarify the onset of collective and deconfined behavior in small systems and to provide new insights into the transport properties of matter. We further argue that investigation of other observable such as energy correlators and quarkonia can lead to a more complete picture of QGP formation in these collisions.

Compact star cores reach extreme densities and may contain exotic dense-matter phases. Information about the exotic interiors of rapidly rotating pulsars can be inferred from r-mode oscillations, whose stability is governed by viscous dissipation. In this work, we model a compact star containing a possible mixed phase of hadronic and quark matter and employ a hybrid statistical framework based on Bayesian inference to infer the dissipation time scales associated with the hybrid phase. Using low-mass X-ray binaries (LMXB) timing observations together with mass-radius constraints from the Neutron Star Interior Composition Explorer (NICER) mission, we estimate the shear and bulk viscosity contributions to r-mode damping for a hybrid star of two layers. Our inference yields shear and bulk viscous dissipation time scales of $τ_s=(4.99^{+0.49}_{-0.52}) \times 10^8 T^{\frac{5}{3}}$s and $τ_B=(2.150^{+1.23}_{-0.60}) \times 10^{19} (T^4 10^{-12}+T^2 10^{-6})^{-1}Ω^{-2}$s respectively. The timescales thus obtained can be implemented to obtain the minima of the star's rotation frequency at $Ω=451.87$ Hz at temperature $T=0.259$ MeV for a hybrid star of mass $1.5$ $M_{\odot}$ and $Ω=517.47$ Hz at $T=0.234$ MeV for $M=1.75 M_{\odot}$. We find that the instability window obtained through the inference framework effectively explains the observed stability of millisecond pulsars in both the radio and LMXB populations, particularly for XTE J0929-314 and XTE J1807-294, J0437-4715, J2124-3358, respectively. These results demonstrate that Bayesian inference combined with r-mode phenomenology provides a powerful and observationally consistent framework for constraining the transport properties of dense hybrid matter.

In this work, we propose a novel definition of the one-point charge correlator (QC) adapted to the Breit frame in deep-inelastic scattering (DIS). We demonstrate that this observable is infrared and collinear (IRC) safe, ensuring its perturbative calculability. Utilizing soft-collinear effective theory (SCET), we systematically analyze the QC in both the forward and back-to-back limits. In the forward limit, we introduce the nucleon charge correlator as a novel non-perturbative object that encodes the multi-dimensional microscopic structure of the nucleon. In the back-to-back limit, the QC establishes a direct correspondence with transverse momentum-dependent distributions (TMDs), enabling its description within the standard TMD factorization formalism. The singular distributions are derived within SCET and are verified by the full QCD calculations up to $\mathcal{O}(α_s^2)$. The corresponding collinear logarithms are resummed to all orders with the accuracy of NLL (${\cal{O}}(α_s^n L^{n-1})$), while the transverse momentum-dependent logarithms are resummed to all orders with the accuracy of $N^3$LL for the unpolarized distribution and N$^2$LL for the Sivers asymmetry.

Characterising fully hadronic events is a difficult task at hadron colliders. Signal jets from the hard process are mingled with an arbitrary number of ISR and FSR jets, leading to a large combinatorial background. This also poses a challenge for machine-learning analyses, where the number of input features is fixed while the jet multiplicity fluctuates from event to event due to QCD radiation. In this work, we explore the use of the two-point correlation spectral function as an input feature for machine-learning analyses of such events. The spectral function maps the transverse-momentum data of an event into a one-dimensional function of the angular distance, encoding the event information modulo collider isometries and jet permutations, and is defined independently of the jet multiplicity. As a concrete benchmark we apply the method to discriminate gluino-pair production followed by $\tilde{g} \to t \bar{t} \tildeχ_1^0$ against the fully hadronic $t \bar{t}$ background. With $139~{\rm fb}^{-1}$ of $\sqrt{s} = 13$ TeV $pp$ collision data, a dense neural network supplied with spectral-function features improves the expected reach in gluino-mass by roughly 150 GeV relative to a recent ATLAS analysis, and by roughly 250 GeV relative to the same network trained on jet kinematics alone.

The 21-cm absorption line of neutral hydrogen has been a long hypothesized observational feature of the spectra of Fast Radio Bursts (FRBs). The difficulties associated with detector noise in extracting HI absorption have been previously studied. We test the role that scintillation plays in the HI absorption line's detectability, and characterize the regimes where a realistic FRB may yield the 21-cm line. We build an efficient model to simulate diffractive scintillation arising from FRB passage through a thin scattering screen. We find that the absorption profile is detectable in a scintillation-dominated high signal-to-noise spectrum if the scintillation decorrelation bandwidth differs significantly in scale from the width of the absorption profile. Active repeaters also enable favorable conditions as the absorption signal improves when repeat bursts are stacked. Repeat bursts must be separated in time by more than the diffractive scintillation timescale, otherwise flux modulations with frequency are correlated. By cross-referencing repeating FRB positions with an observational catalog of Milky Way molecular clouds detected in CO, we find that the sightline to FRB 20180916B may intersect a Galactic molecular cloud. For currently operating and planned sensitive telescopes, the presence of both scintillation and noise requires $\gtrsim 1000$ bursts to be stacked to detect the HI absorption line at a $5σ$ significance. Improvement in detector sensitivities will help probe HI clouds intersected by FRBs in the host or intervening galaxies, or in high-redshift minihalos.

Active galactic nuclei (AGN) are thought to have dusty outflows; however, unlike the gas phase, measuring the kinematics of dust is challenging. We present the detection and analysis of the kinematics of dust in five type 2 quasars (QSO2s) at $z\sim0.1$ from the Quasar Feedback (QSOFEED) sample observed with JWST/MIRI spectroscopy. We use Principal Component Analysis (PCA) tomography to produce velocity maps of Polycyclic Aromatic Hydrocarbon (PAH) features, which are the smallest carbonaceous dust particles. We are then able to compare velocity maps of the PAHs with emission lines of ionised and molecular gas. We are able to produce velocity maps of the 11.3 $μ$m PAH feature, which traces large and neutral PAHs, for three out of the five objects where all three show the presence of an outflow in the PAH kinematics. This becomes particularly clear after subtracting disk kinematics, where the H$_2$ rotational transitions also show residuals consistent with an outflow. Compared to previous work with Seyfert galaxies, this work suggests that dusty outflows are more common at higher Eddington ratios, $λ_{\rm Edd}\gtrsim0.1$, in agreement with previous suggestions, although the sample size is small. We are unable to produce velocity maps for the 6.2 $μ$m PAH, which traces ionised PAHs, potentially due to differences in the intrinsic profile and/or suppression of the feature in AGN, which was seen previously in Seyfert galaxies. This reflects studies of PAH band ratios where AGN outflows have more neutral PAHs. This work demonstrates that dusty outflows may be common, particularly at high Eddington ratios, and therefore play a key role in the evolution and life cycle of AGN.

Non-standard inflationary models often predict oscillatory features in the primordial power spectrum. We present constraints on general oscillatory templates for primordial power spectra, including those that vary linearly and logarithmically with wavenumber, as well as oscillations induced by inflationary particle production. We utilize the Planck 2018 and Atacama Cosmology Telescope Data Release 6 cosmic microwave background data as well as large-scale structure data from the Dark Energy Spectroscopic Instrument Data Release 2. To efficiently explore the multimodal posteriors for these models as well as performing mode comparisons, we integrate the preconditioned sequential Monte Carlo sampler, pocoMC, into the widely used sampling code, Cobaya. We find that the combined dataset tightens the 95% CL upper bounds on the general oscillation amplitudes to $A_{\text{lin}} < 0.021$, $A_{\text{log}} < 0.022$, and $A_{\text{log rf}} < 0.023$, restricting the amplitude to $\sim 2\%\, A_s$. For the inflationary particle production model, our analysis places a maximum a posteriori constraint on the coupling constant of $g = 0.034$. While these models all provide an improved fit to the data compared to the concordance $Λ$CDM model, the Bayesian evidence still reveals a moderate preference for $Λ$CDM compared to models with general oscillations and is inconclusive regarding the particle production model, suggesting that the added complexity of these models beyond the standard model is not statistically justified by current data.

Active Galactic Nuclei (AGN) are among the brightest sources in the Universe, and those that are also jetted are uniquely valuable at the earliest epochs, because their relativistic outflows can regulate the gas supply of their host galaxies, potentially affecting both early star formation and the rapid growth of supermassive black holes (SMBHs). Their compact, high-brightness-temperature radio cores provide the sharpest beacons for very long baseline interferometry (VLBI), enabling direct constraints on Doppler boosting, jet duty cycles, and jet$-$environment coupling at extreme redshifts. In this White Paper, we discuss how the SKA-VLBI will provide sub-$μ$Jy sensitivity together with milliarcsecond (mas) angular resolution to image and characterise jetted AGN at $z>6$ across SKA-Mid and SKA-Low frequencies. These observations can directly test SMBHs ($>10^6$ M$_{\odot}$) formation/evolution models (including jet-assisted super-Eddington phases) and infer the geometry of the Universe, directly probing the cosmological framework at high precision. Synergies with current and next-generation multi-band facilities will also be crucial to fully understand their host galaxies and their environment, providing an unprecedented panchromatic knowledge of the first jetted AGN.

We present a catalogue of 131 compact $(\lesssim 0.1 \,arcsec)$ sources detected at 823 MHz via their Interplanetary Scintillation (IPS). These measurements were made with the ASKAP telescope, across its full field of view of 35 square degrees, utilising all 36 Phased Array Feed (PAF) beams. To bypass ASKAP's standard correlator's minimum integration limit of 10 s, we used the CRAFT data capture system (CRACO), with visibilities sampled every 110 ms. Here we present the data processing steps, the sources detected, and their IPS-inferred properties. ASKAP IPS cleanly separates two populations: compact hot spots embedded in extended lobes and IPS-unresolved sources which are AGN or CSO sources associated with the galactic nucleus. We also compare these results with the results from observations of IPS at 162 MHz with the Murchison Widefield Array, providing the spectra of compact components between 162 MHz and 888 MHz. These measurements further re-enforce the dominance of peaked-spectrum SEDs in the compact-source population at frequencies below 1 GHz. This pilot study using test data is a pathfinder for a more comprehensive ASKAP IPS survey which is underway.

Pair-instability supernovae (PISNe) are among the most luminous transients in the Universe. However, they have never been confidently observed. Solving this puzzle would have key implications for several astrophysical topics, including galaxy chemical enrichment, the interpretation of gravitational waves from binary black hole mergers, and the nature of red dropout sources seen by JWST. With this aim, we present the first in-depth study of PISN occurrence in binary stars, both in isolation and in dense star clusters. We employ the SEVN code, with PARSEC stellar tracks, to evolve a suite of 35 synthetic binary populations, including variations on formation channels, cluster properties, and upper limit of the stellar initial mass function. We find that binary interactions can boost the PISN rate by up to threefold, relative to single stars, whereas binary hardening can either enhance or suppress PISN production, depending on whether the progenitors are primordial or dynamically formed. Moreover, we showcase how our comprehensive framework for the cosmic PISN rate can be used to constrain uncertain aspects of stellar and galaxy evolution models, via comparison with observations, including the recipes for stellar-wind mass loss in very-massive stars, and the galaxy metallicity distribution throughout the Universe.

The origin of the most massive gravitational-wave sources remains elusive. We show that gas accretion can be understood as a transport process in mass space, causing compact objects to migrate through a population at rates determined by the underlying growth law. Using a continuity-equation framework, we demonstrate that population evolution is governed primarily by the mass dependence of the accretion rate, $\dot m \propto m^β$. Accretion laws with $β>1$ naturally produce divergent evolution and generate extended high-mass tails, whereas $β<1$ leads to convergent evolution and compresses the population toward a narrower range of masses. We apply this framework to physically motivated accretion regimes and explore their consequences using analytical calculations and Monte Carlo population models. We show that sustained gas accretion can substantially broaden compact-object mass distributions, populate the high-mass end of gravitational-wave catalogs, and alter the mass-ratio distribution of compact-object binaries. In particular, collective accretion within compact binaries drives their mass ratios toward unity. Our results suggest that gaseous environments act as transport media that continuously reshape compact-object populations, providing a natural pathway toward the formation of massive mergers such as GW231123 and the high-mass tails increasingly revealed by gravitational-wave observations.

We examine published JWST observations of the core-collapse supernova (CCSN) remnant SNR 0540-69.3 and identify a point-symmetric morphology in its inner ejecta. Within the framework of the jittering jets explosion mechanism (JJEM), we interpret this morphology as evidence that the ejecta were shaped by two, and likely three or more, pairs of jets during the explosion process. Both visual inspection and a recently developed quantitative symmetry-identification method for astrophysical imaging reveal an approximate rotational symmetry between the northwestern redshifted ejecta and the southeastern blueshifted ejecta. Each side contains clumps (knots) surrounding a previously identified cavity, with the best quantitative correspondence obtained for a rotation of 189°. We further identify a symmetry center that is offset from the current pulsar position, strengthening an earlier claim for a pulsar kick. We interpret the pair of cavities and their surrounding clumpy structures as having been shaped by multiple jet-launching episodes. In addition, we identify a pair of opposing nozzles at a large angle to the cavities, which we attribute to another jet pair. Guided by the similarities to point-symmetric planetary nebulae shaped by jets and by recent three-dimensional hydrodynamical simulations of the JJEM, we conclude that the inner ejecta were shaped by at least three jet pairs launched by the neutron star after it acquired its kick velocity, consistent with the JJEM.

Gamma-ray bursts (GRBs) are the most luminous electromagnetic explosions in the Universe, and offer unique laboratories for studying relativistic jets, compact-object formation, particle acceleration, and the high-redshift Universe. The early optical emission of GRBs, particularly within seconds to minutes after the burst, carries crucial information about the central engine, jet magnetization, bulk Lorentz factor, and circumburst environment. We present a comprehensive review of the early optical phenomenology of GRBs and the essential role played by ground-based robotic optical telescopes to observe the fleeting early-time phenomena through rapid, automated responses to real-time GRB alerts and high-cadence photometry. We examine the key early optical features of GRBs, including prompt optical emission coincident with the $γ$-ray phase, bright reverse shock optical flashes, the onset of external forward shock afterglow, and superimposed optical flares, plateaus, and discuss the diagnostic power of each in constraining jet physics. We discuss the physical mechanisms underlying these phenomena and their implications for GRB physics (e.g., estimating the initial Lorentz factor $Γ_0$, magnetization, and the density profile). Early optical observations have constrained the initial bulk Lorentz factor $Γ_0 \sim 100$--$1000$, weak-to-moderate ejecta magnetization for events with prominent reverse shocks, the circumburst density profile, and the geometry of the magnetic field in the ejecta through polarimetry. We also provide the technical capabilities and landmark contributions of major robotic facilities, and discuss future prospects in the era of SVOM, Einstein Probe, Rubin/LSST, ULTRASAT, TeV observatories, and multi-messenger alerts.

The energy deposition of fast electrons in interstellar molecular gas is considered. We use the rotationally resolved cross sections for electron-impact excitation of H$_2$ molecule that were calculated using the adiabatic-nuclei molecular convergent close-coupling method. The initial electron energy distribution is assumed mono-energetic, and the differential equation for electron energy distribution is solved. We compare calculated energy deposition parameters with the results of similar studies in which the Monte Carlo approach was used. It is shown that about 11 per cent of the initial energy of fast electrons goes into direct ro-vibrational excitation of energy levels of H$_2$ molecule including pure rotational excitation in neutral molecular gas. About 7 per cent of initial electron energy goes into the excitation to $v=1$ vibrational state of H$_2$ molecule, most of this energy eventually converts into emission of transitions at near-infrared wavelengths. For ro-vibrational levels with $v \geq 3$, the electron-impact excitation to electronic states followed by downward radiative transitions to the ground electronic state is the dominant mechanism of excitation. The yields for excitation to vibrational states via radiative cascading from excited electronic states are found to be $1.5-2$ times higher than were obtained in previous studies.

We present new Gemini/GNIRS near-infrared spectroscopic observations of eight low-redshift ($z < 0.1$) blue low-excitation radio galaxies (BLERGs), a rare subset ($\sim 2.5\%$ of low-excitation radio galaxies; LERGs) that complicate the classical jet-mode AGN picture by combining radio activity with star-forming, gas-rich hosts. These star-forming BLERGs exhibit significant warm H$_2$ emission traced via ro-vibrational transitions at $T \sim 2000$--$4000$ K. We find that BLERGs span a broad range of mass-normalized warm H$_2$ luminosities ($L_{\rm H_2}/M_\star$), comparable to radio-emitting early-type galaxies, yet without a clear positive dependence on radio power. Instead, the strongest H$_2$ emission preferentially occurs in morphologically disturbed and advanced-merger systems, while compact radio sources ($\lesssim 20$ kpc) remain plausible sites of localized jet-ISM interaction. Together, these results suggest that merger-driven processes, including tidal shocks, gas inflows, and disturbed interstellar medium conditions, are the dominant drivers of warm molecular gas excitation in BLERGs, although localized jet-driven heating may contribute in individual systems. The compact radio morphologies, gas-rich hosts, and rarity of BLERGs are consistent with a short-lived evolutionary phase in which radio AGN activity coexists with an interaction-driven, molecular-rich interstellar medium prior to the onset of large-scale maintenance-mode feedback. Spatially resolved spectroscopy and higher-resolution radio imaging will be essential to disentangle the relative roles of mergers and jets in regulating the molecular gas of jet-mode AGN.

A wide variety of Galactic and extragalactic sources are known to emitradiation across the entire electromagnetic spectrum, including both transient and steady-state phenomena. A few hundred of these sources (~300) have been detected even at the highest energies, in the TeV range. The number of known TeV emitters is expected to increase substantially in the coming years with the operation of current and next-generation Cherenkov detectors, such as the Large High Altitude Air Shower Observatory (LHAASO) and the Cherenkov Telescope Array Observatory (CTAO). These sources typically exhibit broad, non-thermal, spectral energy distributions. Explaining such emission requires efficient particle acceleration mechanisms (e.g. Fermi processes, shock acceleration) and radiative processes involving magnetic fields (e.g. synchrotron and inverse Compton radiation), often accompanied by polarization signatures. However, the relative contribution of these emission mechanisms and the underlying physical processes are still debated. In this work, we present an overview of the scientific potential arising from the synergy between the Square Kilometre Array (SKA) and current and upcoming gamma-ray facilities. Combined observations across these energy bands will provide crucial insights into the physical mechanisms driving emission from GeV-TeV sources of both Galactic and extragalactic origin. These include transient events (e.g. gamma-ray bursts, supernovae, fast radio bursts, tidal disruption events, neutrino and gravitational-wave counterparts), variable sources (e.g. blazars, active galactic nuclei), and steady emitters (e.g. the Galactic centre, supernova remnants, radio galaxies, and galaxy clusters). We discuss the prospects for coordinated SKA-gamma-ray observations, including wide-field surveys, monitoring of variable sources, and target-of-opportunity follow-ups.

The appearance of galaxies is strongly dominated by two physical phenomena: star formation and accretion of material onto compact objects, primarily supermassive black holes. Nearby galaxies offer a unique window to study these processes in detail and with high spatial resolution. The $μ$Jy sensitivity, sub-arcsecond angular resolution and broadband coverage provided by SKA AA4 configuration will enable us to separate morphologically and spectrally the contribution from SMBH accretion and star formation processes, from the most compact nuclear region out to the most diffuse, galaxy-wide components. Disentangling thermal and non-thermal emission using multi-scale spectral index maps (traced by SKA-Mid) and absorption processes (traced mainly by SKA-Low) will allow us to account for the present and past star-formation rate, the AGN activity and the ISM properties. SKA observations of a wide range of galaxy types in the nearby Universe, including quiescent, AGN-dominated and starburst galaxies, will provide a broad view of how star formation and accretion regulate galaxy evolution, while time-domain and spectral-variability information further isolate compact accretion and transient phenomena.

We present a comprehensive forecasting framework to assess the detection of primordial oscillatory features by exploiting the synergy between future neutral hydrogen (HI) intensity mapping (IM) surveys and cosmic microwave background (CMB) measurements. Focusing on next-generation single-dish (SKAO) and interferometric (HIRAX) radio configurations, we perform a joint analysis of the redshift-space power spectrum and bispectrum, consistently incorporating scale-dependent bias, redshift-space distortions, and non-Gaussian covariance. We investigate phenomenological templates with linear and logarithmic primordial oscillations, together with a physically motivated sharp-feature model. We find that including the large-scale structure bispectrum improves marginalised constraints on feature amplitudes by $30$--$40\%$ relative to the power spectrum alone and helps break parameter degeneracies. In several cases, the bispectrum contains more information on the power-spectrum feature parameters than the power spectrum itself. Much of this gain arises from the late-time gravitational contribution, which inherits the oscillatory structure of the primordial feature signal and acts as an independent source of information. While the CMB angular power spectrum is crucial for constraining low oscillation frequencies, joint interferometric IM analyses ($P+B$) outperform CMB amplitude constraints by up to $75\%$ in the linear regime. We also show that, despite non-Gaussian covariance degrading the independent constraining power of the bispectrum by $55$--$90\%$, the combined HI+CMB probe achieves percent-level precision on the frequency of primordial features, providing a powerful test of non-slow-roll inflation.

A major question in galaxy evolution is how galaxies acquire sufficient gas to sustain their star formation rates. HI observations with high angular resolution and sensitivity to very low column densities are some of the important observational ingredients that are currently still missing. Answers to these questions are necessary for a correct interpretation of observations of galaxy evolution in the high-redshift universe and will provide crucial input for the sub-grid physics in hydrodynamical simulations of galaxy evolutions. In this chapter we discuss the progress that has been made over the past years, describe the various processes that lead to inflow and outflow of gas, and discuss how SKA-Mid AA4 observations can contribute to further understanding these important aspects of galaxy evolution using deep observations of nearby individual disk and dwarf galaxies.

Loop-like patterns between spectral parameters are frequently interpreted as evidence of hysteresis in time-dependent astrophysical emission processes. Such patterns have been reported in hardness-intensity diagrams (HID) of accreting black-hole X-ray binaries during state transitions, in the radio-to-X-ray correlation plane during outbursts, in solar activity indices over the solar cycle, and in the spectral energy distribution of active galactic nuclei during flaring episodes. HID often exhibit apparent loops, whose orientation encodes the relative timescales of particle acceleration and radiative cooling. Visual inspection, however, does not provide a statistically controlled detection method. We develop a statistically robust and empirically calibrated framework for detecting, quantifying, and classifying hysteresis patterns in ordered two-dimensional data with measurement uncertainties. The framework provides the normalised signed area $A_\mathrm{norm}$ enclosed by chronologically ordered points in the plane as the primary detection statistic, computed using the shoelace formula. We define open and closed area estimators, introduce cancellation diagnostics for multi-loop structures, and propagate measurement uncertainties via Monte Carlo sampling. Statistical significance is assessed using null ensembles generated by time-order randomization, physically motivated autoregressive surrogate models, and Fourier phase-randomized surrogates. We validate the method on synthetic blazar flare trajectories, and demonstrate its application to an XMM-Newton observation of Markarian~421 during a December 2023 flaring episode, where we confirm a CCW hysteresis loop with $A_\mathrm{norm} = +0.64$ that is robust against measurement noise but does not reach formal significance against stochastic null models, possibly due to the open trajectory geometry.

In a magnetic field with a complex topology, as can be the Galactic magnetic field, cosmic ray transport cannot simply be described by diffusion parallel and perpendicular to magnetic field lines, because the gradients and curvature of the large-scale magnetic field induce drift motions. These effects become especially important at high energies. Here we revisit the possibility that the competition between diffusion and drifts may lead to a knee in the cosmic ray spectrum. We carry out test-particle simulations of cosmic ray transport in a mock Galactic magnetic field made of a regular large scale component, with a non-trivial topology and a homogeneous and isotropic turbulent magnetic field, with a spectrum that is assumed to be Kolmogorov-like in the basic setup. These simulations are used to infer the escape time and the grammage accumulated by cosmic rays with energy in the TeV--10 PeV energy range. In the case of a large scale magnetic field with a purely azimuthal structure, the drift due to the curvature of magnetic field lines produces a knee in the PeV range, but the model fails to reproduce the grammage, due to the exceedingly low value of the perpendicular diffusion coefficient. If the large scale magnetic field acquires a component perpendicular to the Galactic disc, the parallel diffusion coefficient becomes quickly dominant in terms of particle escape, and drifts are unable to compete. A knee structure does not appear in such a scenario. However, if the parallel diffusion coefficient becomes energy independent at $E\gtrsim 1$ TeV, a knee may arise around PeV energies due to drift dominance. We discuss two cases in which this situation may occur.

This report summarizes potential studies of the Hi-ReX Science Analysis Group for an Ultra-High Angular Resolution X-ray Observatory. Many potential science cases are presented, spanning a range of astrophysical topics from solar system planets and exoplanets to supermassive black holes and cosmology. These cases demonstrate the value of X-ray imaging beyond current capabilities with an aim toward milli-arcsecond and micro-arcsecond angular scales. There is an imaging gap between the current capability of the best X-ray imaging telescope, the Chandra X-ray Observatory, compared to that of instruments across the electromagnetic spectrum. Next-generation, ultra-high angular resolution X-ray imaging will close this gap and transform our understanding of the universe and fundamental physics.

The [O II]$λ\lambda3726,3729$ and [S II]$λ\lambda6716,6731$ doublets are widely used as low-ionization electron-density diagnostics in galaxy spectra and are often treated as interchangeable when only one of them is accessible. We test this assumption using the DESI DR1 Emission Line Catalog. For star-forming galaxies with fiducial emission lines, [O II] yields systematically higher electron densities than [S II], with a median offset of 0.228 dex. The binned median calibration is $\log n_e({\rm OII})=(0.752^{+0.182}_{-0.097})\log n_e({\rm SII}) +(0.832^{+0.231}_{-0.422})$. The offset is larger in galaxies with higher stellar mass, H$α$ star-formation rate, dust attenuation, and $N2\equiv\log([{\rm NII}]\lambda6583/{\rm H}α)$, an empirical gas-phase metallicity proxy, and smaller in galaxies with higher \(\log O_{32}\equiv\log\{[{\rm OIII}]\lambda5007/ ([{\rm OII}]\lambda3726+\lambda3729)\}\), an ionization proxy; no significant trend is found with specific star-formation rate. These trends are consistent with [O II] and [S II] sampling different low-ionization gas phases in integrated spectra, with [S II] more strongly weighted toward lower-density diffuse or outer gas. Our results show that [O II]- and [S II]-based densities should not be mixed without empirical calibration in studies of ISM pressure, nebular density, and their evolution across galaxy samples and redshift.

The Einstein Probe (EP) has detected several Fast X-ray Transients (FXTs) associated with broad-lined Type Ic supernovae (SNe), including EP240414a and EP250108a. The observations reveal common features among these FXTs, but the corresponding physical origin remains debated. By comparing the FXTs with low-luminosity gamma-ray bursts (e.g., GRB 171205A), we propose a unified model that explains the common features in these events. In this model, a rapidly spinning magnetar generates a collimated Poynting flux-dominated jet and an isotropic wind. As the jet propagates through the stellar envelope, it generates a hot cocoon. In addition, a pulsar wind nebula (PWN) is formed during the interaction of the wind and the ejecta. As the surrounding cocoon gradually becomes transparent, the emission from the PWN escapes and is observed. This model provides a unified explanation for the observations: (1) Early thermal emission originates from the cocoon; (2) Mid-term non-thermal emission comes from the PWN; (3) Late-term emission originates from SNe driven by $^{56}$Ni radioactive decay and magnetar. (4) The X-ray afterglows originate from the structured jet. Our research thus provides a natural explanation for the observed thermal-to-nonthermal evolution in such FXTs and reveals their shared physical origin with some GRB-SNe.

Cross-correlations between the thermal Sunyaev-Zel'dovich (tSZ) effect and galaxy surveys provide a sensitive probe of hot gas in low-mass halos, but on the angular scales of greatest astrophysical interest they are also highly vulnerable to residual foreground contamination. We analyze the cross-correlation of two low-redshift unWISE galaxy samples with Planck PR4 and ACT DR6 microwave temperature sky maps directly in harmonic space, fitting simultaneously for correlated tSZ, cosmic infrared background (CIB), and radio emission. Using the nine Planck bands to perform model selection, we find that a three-component model consisting of tSZ, radio emission, and a CIB amplitude term is strongly preferred over a model that omits radio contamination, with a significance of $9.5σ$ for the unWISE Low-z sample and $11σ$ for the unWISE Mid-z sample respectively. For the unWISE Low-z sample ($\bar{z}=0.14$) the preferred \textit{Planck} fit gives an effective CIB emissivity index $β_0=1.79\pm0.29$, an effective CIB dust temperature $T_0=22.0\pm5.1 K$, and a radio spectral index $β_r=-2.18\pm0.24$; for the unWISE Mid-z sample ($\bar{z}=0.23$) we find $β_0=1.41\pm0.13$, $T_0=28.0\pm4.2 K$, and $β_r=-2.62\pm0.26$. We apply this radio-inclusive recipe to the ACT+Planck 90, 150, and 220 GHz maps to obtain the best small-scale measurements. This removes the apparent negative galaxy-tSZ cross-correlation seen in the fiducial ACT DR6 NILC reconstruction and in no-radio fits. The cleaned tSZ cross unWISE spectra remain positive to $\ell \simeq 6000$ and are broadly consistent with the Planck-only reconstruction on overlapping scales. We show that these radio-cleaned spectra are well described by a conventional halo model calibrated to the galaxy population and a two-parameter generalized NFW electron pressure profile.

Gamma-ray-emitting narrow-line Seyfert 1 ($γ$-NLSy1) galaxies represent an enigmatic class of active galactic nuclei (AGNs) bridging populations of radio-quiet and radio-loud AGNs. Here we report the multi-wavelength investigation of a rare $γ$-ray flaring episode of an NLSy1 galaxy, 1H 0323+342 ($z=0.063$), using data from Fermi-Large Area Telescope, Swift, and Nuclear Spectroscopic Telescopic Array. The source exhibited a significant enhancement in $γ$-ray flux along with rapid variability on $\sim$sub-hour timescales in both $γ$-ray and hard X-ray wavelengths, hinting that the energy dissipation is happening from a compact region close to the central supermassive black-hole. The time-resolved X-ray spectral study revealed a transition between a jet-dominated and a mixed jet+corona emission state. Reproducing the broadband spectral energy distribution with a one-zone leptonic model suggested that the high-energy emission is mainly produced by external Compton scattering of the accretion disc and broad line region photons. The inferred jet power reaches values of the order of $10^{46}$ erg s$^{-1}$ comparable to those of powerful flat-spectrum radio quasars, suggesting that even low black-hole mass systems can occasionally produce powerful $γ$-ray outbursts.

Cosmological observations are placing increasingly stringent bounds on the sum of neutrino masses, approaching the lower limits implied by neutrino oscillation experiments. Recent studies have suggested that dynamical dark energy may alleviate this apparent tension. However, these conclusions generally rely on the assumption that dark energy remains smooth, neglecting its perturbations. In this work we investigate the robustness of cosmological neutrino-mass constraints by consistently incorporating dark-energy perturbations. Using CMB, BAO, RSD, and supernova data, we show that the commonly reported alleviation of the neutrino-mass tension in dynamical dark-energy models is not generic. While smooth dark energy substantially relaxes the neutrino-mass bounds, allowing dark energy to cluster shifts the preferred neutrino mass toward smaller, and even more negative, effective values. We demonstrate that this behavior originates from a degeneracy between neutrino free-streaming and dark-energy perturbations in structure-growth observables. Different combinations of neutrino mass and dark-energy clustering can provide similarly good fits to current data while yielding significantly different neutrino-mass constraints. Our results show that cosmological neutrino-mass measurements are inherently model dependent and that reliable neutrino-mass inference requires a consistent treatment of dark-energy perturbations.

The origin of high-energy cosmic rays remain one of astrophysics' greatest unsolved mysteries. SKA-Low will be able to measure air showers initiated by cosmic rays with unprecedented precision in the PeV - EeV energy range, covering the critical transition region between Galactic and extragalactic sources. SKA-Low's densely instrumented core and broad bandwidth will allow for measurements of individual air showers with a level of detail unmatched by any existing or planned detector. The depth of shower maximum, the primary mass-sensitive observable, will be reconstructed with a resolution of better than 8~g/cm$^2$, a significant improvement over existing methods. Additionally, new reconstruction methods are expected to enable full air shower reconstruction across a wide energy range, down to PeV levels. At these energies, efficient photon/hadron separation may offer an opportunity to measure PeV gamma-ray air showers. Furthermore, SKA-Low opens a window into studying high-energy hadronic interactions, including via the unique channel of anomalous air showers. This combination of measurements provides a unique opportunity to investigate the origins and physics of high-energy cosmic rays. A dedicated particle detector array will provide triggered readout of raw antenna-level voltage buffers, enabling fully commensal cosmic-ray observations alongside regular operations. We outline our science case and discuss the observational strategy, signal properties and detector design underpinning these measurements. We also summarize the accompanying book chapters, which address composition measurements in the Galactic-to-extragalactic transition region, next-generation interferometric reconstruction techniques, hadronic interaction physics through anomalous air showers, the prospects for detecting PeV gamma-rays from Galactic sources, and the related project of imaging lightning using SKA-Low.

Extragalactic jets have long prompted the question of how far relativistic outflows can extend, with some radio sources reaching 5 - 7 Mpc in length. These great extents motivate investigations into their ages, propagation dynamics, stability, and impact on the environment. We perform 3D high-resolution numerical simulations of two jet configurations involving continuous injection at different powers propagating in low-density regions of the cosmos (static and laminar), investigating the conditions for jet collimation versus disruption at extreme scales. We show that the combined effects of higher jet thrust (enhanced kinetic power), improved collimation (suppression of transverse distortions), and magnetic stabilization (strengthened poloidal field) can sustain a laterally confined flow, enabling such a jet to reach 5 Mpc in just 15 Myr (injecting a total energy of $2.3 \times 10^{61}$ erg into the environment). In contrast, a jet lacking these conditions dissipates more rapidly, forming lobe-like morphologies and reaching only $\sim 3$ Mpc over $\sim35$ Myr (injecting total energy of $8.1 \times 10^{60}$ erg). Pinch and kink MHD instabilities are identified as the primary drivers of transverse distortions; their suppression allows the persistence of a fast spine alongside a slower, dissipative head (location of maximum environmental interaction). We find that the jet-head propagation shows two regimes: one with speed $\sim0.5 c$; the other with speed from $\sim 0.2 c$ to $\sim 0.05 c$. We consider a proxy of synchrotron emission and find that radiation is concentrated in regions of enhanced compression and magnetic amplification, primarily near the first recollimation shock (producing a bright radio spot) and at the jet-head interaction zone (producing the radio termination lobe). Such jets facilitate the transport of substantial energy and magnetic flux into underdense cosmic regions.

Meteorite analysis shows that the older solids of the solar system, such as the calcium-aluminum-rich inclusions (CAIs), have isotopic inhomogeneity. This indicates that the isotopic inhomogeneity could originate from parental molecular clouds. We investigate the evolution of the isotopically heterogeneous molecular cloud cores formed from filament fragmentation using the smoothed particle hydrodynamics method. We show that the effect of the variation of isotopic ratio along the minor axes of the filament is smaller than that along the longitudinal axis of the filament due to the filament geometry. Our results also suggest that isotopic inhomogeneities remain in the resulting cores, although the amounts of initial inhomogeneities are reduced by a factor of 100 from those over the initial filament length of 1 pc. This fraction corresponds to 1-10% of the maximum isotopic ratio that the core can acquire from the filament in each model. The origin of the isotopic inhomogeneity of the shells could be attributed to the initial difference in the center-of-mass of shells caused by the turbulent velocity field. Our model indicates that the isotopic inhomogeneity could survive even in the circumstellar disk.

In this series of lectures, we seek to describe the evolution of the cosmic large-scale structure. We will discover the cosmic web - the large-scale skeleton of matter traced by galaxies. It arises from the interplay of the gravitational pull of dark matter and the expansion driven by dark energy. Major large-scale galaxy surveys map the distribution of matter and galaxies across most of the sky, spanning over 10 billion years of cosmic history. I will guide you through some of the principles and challenges behind predicting the statistical properties of the matter and galaxy distribution in vast cosmic volumes. In particular we discuss the underlying nonlinear physics and resulting non-Gaussian statistics that need to be predicted to extract fundamental physics from observational data.

Precision cosmology assumes that cosmic reionization was a single smooth transition. We show that this assumption is in tension with current observations: the Planck optical depth, patchy kinetic Sunyaev--Zel'dovich (PkSZ)limits from SPT and ACT, and the Ly$α$ forest endpoint cannot be simultaneously reproduced by any viable monotonic ionization history. Non-parametric reconstructions and Planck EE polarization provide independent support for an additional ionization component at $z \gtrsim 12$. Incorporating this early phase relaxes the upper bound on the summed neutrino mass to $\sum m_ν< 0.39$~eV ($95\%$ CL) and shifts $σ_8$ toward values preferred by weak-lensing surveys, both consequences of the standard $A_s$--$τ_e$ degeneracy. These shifts arise from relaxing a hidden prior on reionization shape rather than from new physics, and identify reionization shape as an implicit prior in cosmological inference. Next generation of CMB and 21~cm experiments will be able to test this directly,which will convert what has been an unflagged systematic on $σ_8$ and $\sum m_ν$ into a quantified statistical uncertainty.

The Square Kilometre Array's low frequency telescope (SKA-Low) will enable inference of astrophysical and cosmological parameters from the redshifted 21 cm signal, probing the Cosmic Dawn and Epoch of Reionisation. While the power spectrum is the primary target for initial detection, the inherently non-Gaussian nature of the 21 cm signal, driven by the patchy evolution of ionised regions and spin temperature fluctuations, encodes rich information accessible through higher-order statistics and morphological measurements. Extracting these constraints requires diverse inference tools, encompassing both sophisticated modelling frameworks (analytical, semi-numerical, numerical, and emulators) used to predict the 21 cm signal, and advanced inference techniques (Bayesian, simulation-based, field-level) to connect statistics to the underlying physics. This chapter reviews these tools and explores the constraining power of different statistical probes accessible with SKA-Low, including the power spectrum, statistics beyond order two, moments of the signal distribution, and morphological measures. Combining these complementary statistics is crucial for breaking parameter degeneracies and unveiling the properties of the early Universe. We specifically assess the potential of the initial SKA-Low configuration (AA*) to measure galaxy and IGM properties, demonstrating its capability for early science results. This chapter forms part of a comprehensive set detailing the Epoch of Reionisation and Cosmic Dawn science case for the SKA-Low telescope.

We present rotation curves and inner mass distributions for 30 star-forming galaxies at $0.5<z<1.7$, observed with JWST/NIRSpec as part of the MSA-3D Cycle 1 survey. Combining spatially resolved ionised-gas kinematics with JWST/NIRCam imaging, we constrain baryonic and dark matter contributions through forward dynamical modelling for galaxies extending down to stellar masses of $\sim10^{9}M_\odot$. For the 23 galaxies in our primary statistical sample, we find predominantly rotationally supported disks with intrinsic dispersions $σ_0\sim31$-65 km s$^{-1}$ and a wide range of dark matter fractions, $f_{DM}(R_e)\sim0.1$-0.9, with a median of 0.63 and substantial galaxy-to-galaxy scatter of $\sim0.2$ dex. These results are supported by a complementary consistency check using stellar mass maps and SFR-derived gas profiles. Among the 19 galaxies reaching $\gtrsim2R_e$, we identify six rising, six flat, and seven falling rotation curves. These classes define an observed ordering from rotationally dominated, dark-matter-rich disks ($V_{rot}/σ_0\approx4$, $f_{DM}\gtrsim0.7$) to more dispersion-supported systems with centrally concentrated baryonic mass distributions ($V_{rot}/σ_0\approx2$, $f_{DM}\lesssim0.55$). The stellar Tully-Fisher relation lies close to the local relation evolved under the adopted self-similar $Λ$CDM scaling. A simplified seeing-degradation test shifts the inferred normalisation by ~0.2 dex at fixed $V_c$, suggesting that spatial resolution contributes to, but does not fully explain, differences among high-redshift Tully-Fisher measurements. Overall, MSA-3D provides a high-resolution extension of previous surveys toward lower stellar masses, spanning $9.0 < \log(M_\star/M_\odot) < 11.2$, and reinforces that star-forming disks near $z\sim1$ span a broad range of dynamical states and inner mass distributions.

Recent gravitational-wave (GW) observations have revealed binary black hole (BBH) mergers with both extreme mass ratios and large effective spin parameters ($χ_{\rm{eff}}$). GW241011 is a notable example that shows these properties. Although hierarchical mergers (second-generation + first-generation BHs) can naturally produce high spins, they rarely produce such an extreme mass ratio ($\sim 0.3$), and are further limited by gravitational recoil kicks that can eject the second-generation BH from the host environment. Moreover, recent studies have argued against a dynamical origin for GW241011. Here, we investigate the formation of GW241011-like systems through the mass-ratio reversal (MRR) channel in isolated binary evolution. By quantifying the probability of producing such systems across a range of binary-evolution models, we identify the key dependencies on stellar-evolution and binary-interaction physics. Our results demonstrate the conditions under which the MRR channel can provide a viable alternative to hierarchical mergers and place constraints on the physical processes governing binary evolution.

Ultraluminous X-ray sources (ULXs) are non-nuclear X-ray emitters exceeding the Eddington luminosity, offering key insights into extreme accretion physics. Their origin is attributed to either sub-Eddington accretion onto intermediate-mass black holes (IMBHs) or supercritical accretion onto stellar-mass compact objects in binary systems. A subset of ULXs shows radio jets and surrounding bubbles, indicative of strong mechanical feedback, but no polarization detection has been reported to date because of limited sensitivity and angular resolution at multi-megaparsec distances. Next-generation facilities such as the Square Kilometre Array (SKA) will enable polarimetric studies of ULX outflows and bubbles, providing vital information on their magnetic field structure and jet-ISM interactions. Using Holmberg II X-1 (3.39 Mpc) as a representative system, we evaluated the detectability of its polarized jet and bubble with the SKA-Mid AA4 configuration. Assuming a fractional polarization of 20% and Band 2 observations centered at 1310 MHz, the expected polarized intensities are 260 $μ$Jy beam$^{-1}$ for the jet and 20 $μ$Jy beam$^{-1}$ for the bubble. A Holmberg II X-1-like jet is detectable within 10 hours even at $\sim$ 10 Mpc, while bubble polarization requires $\sim$ 100 hours. Although resolving fine structures is feasible only for nearby ULXs, future SKA-VLBI capabilities will help overcome this limitation and enable detailed mapping of magnetic field morphology. These results indicate that the SKA will enable the first detection of polarized emission from ULXs, advancing our understanding of magnetic field geometry, jet formation, and feedback in super- and sub-Eddington accretors.

Supernova remnants (SNRs) strongly influence the physical and chemical properties of the molecular clouds (MCs) with which they interact. We carried out a high-resolution observation toward W28F, a chemically rich MC interacting with SNR W28, with the Atacama Compact Array (ACA) in Band 7. Significant emission (> 10 sigma) of CO, CH3OH, p-H2CO, SiO and SO is detected. We reveal the clumpy structures of the shocked MC, with different spatial distributions between CH3OH and SiO. We select six molecular clumps to conduct spectral decomposition and non-local-thermodynamic-equilibrium analysis with the CH3OH and p-H2CO lines. The best-fit results show a H2 density of nH2 ~ (1-3) * 10^5 cm^-3 and a gas temperature of Tgas ~ 50-170 K in most of the fitted components. The H2 density and gas temperature show a clear anti-correlation across different regions, with the thermal pressure consistent with that of the adjacent X-ray-emitting hot plasma. This is consistent with the picture that the SNR shocks propagate into multi-phase gas, with a pressure balance existing between different phases. We propose that the high abundance ratio between E-CH3OHand A-CH3OH (> 0.9) suggests extra gas-phase processes to enhance this ratio, such as proton exchange with H3+ and HCO+. The chemical segregation between CH3OH and SiO, in both the spatial and spectral regime, can be explained by the fact that CH3OH traces slow shocks while SiO traces fast shocks.

We observed the flat-spectrum radio quasar 4C 29.45 in the BVRI optical bands for 39 nights from February 2022 to July 2022 with the T60 telescope at TÜBİTAK National Observatory (TUG) in Turkey. In this study, we aimed to study flux, color, and spectral variability on short timescales. The object was in an active (bright) phase with an average optical R-band brightness of 14.7 mag and was variable in the BVRI bands throughout the monitoring period. We analyzed the flux variability during our observation period, and the variability amplitudes in V, R, and I bands were determined to be 220%, 208%, and 209%, respectively. Optical spectral energy distributions of 4C 29.45 were derived from the observational data of 33 nights, indicating spectral indices ranging from 1.032 to 1.573. We found modest correlations between optical light curves, and between R-band light curve and spectral indices, suggesting that the time lag ranges from several hours to days. Investigation on relation between spectral and color indices versus R-band magnitude revealed achromatic trend during the bright phase of 4C 29.45 in the first half of 2022. Our periodicity search suggested that the periodicity would be larger than 100 days, and no significant signal for periodicity was found for short timescales.

Studying the morphology and coherence scale of the magnetic fields in synchrotron emitting halos of galaxy clusters through detection of polarized synchrotron emission is important in order to understand how they distribute relativistic plasma, their contribution to pressure balance, and how they may affect gas content in galaxies and set up initial conditions during non-linear collapse to eventually form galaxies in the intracluster medium (ICM). Using synthetic maps over broad-bandwidths, generated from high resolution magnetohydrodynamic simulations of fluctuation dynamo, we study the efficacy of SKA-Low and Mid in Array Assembly AA4 in our quest for detecting polarized synchrotron emission from the intracluster medium (ICM). Fluctuation dynamo action in the ICM are expected to generate ubiquitous filamentary and sheet-like magnetic field structures. The associated synchrotron emission projected in the plane of the sky appear highly filamentary that can span hundreds of kiloparsecs. Such filaments can be robustly identified and quantified through high angular resolution ($\lesssim 1^{\prime\prime}$), sensitive observations (about $\rm 0.25\textrm{--}1\,μJy\,beam^{-1}$) above about 3\,GHz, a niche for the SKA-Mid in Band\,5a. Statistical properties of polarized emission and morphological measures, such as, Minkowski functionals, can then be used to infer the turbulence driving scales in the ICM with deep observations of galaxy clusters using SKA-Mid. Availability of a Band\,4 receiver covering 2--4\,GHz would be a major boost in studying the polarization properties of the ICM.

Open clusters are the primary birthplaces of stars in the Milky Way disk, yet their neutron star progeny are rarely found within them, presumably due to supernova-induced kicks that eject them at birth. Here we report the arcsec-level localization of the pulsar PSR J1921+3745 to the tidal tail of NGC 6791, one of the oldest and most massive open clusters. Our N-body simulation shows that more than 95% of neutron stars formed in such clusters have been ejected. This pulsar's location in the tidal tail indicates it was retained for billions of years before being stripped by Galactic tides. This long-term retention requires low natal kicks, consistent with formation via electron-capture supernova. Our findings capture a rare snapshot of a neutron star transitioning into the Galactic field, identifying tidal stripping of ancient clusters as a verified source of the Galactic neutron star population.

The Fe K $α$ fluorescence line at 6.4 keV has long been used to probe the relative geometry between photoionizing X-ray sources and surrounding cold material in a wide range of astrophysical systems. With the advent of the X-ray microcalorimeter XRISM/Resolve, Fe K $α$ lines with equivalent widths down to $\sim 5$ eV-previously inaccessible-are now detectable, and even non-detections can place upper limits of a few eV, making non-detections themselves valuable for constraining the geometry. Considering that Fe K $α$-based geometric diagnostics are entering a new stage in the microcalorimeter era, we present Fe K $α$ equivalent-width maps computed with the three-dimensional Monte Carlo radiative-transfer code SKIRT for a generalized configuration consisting of a spherical reflector of radius $R_{*}$ and a point source located at a height $h$ above the surface. The equivalent-width maps exhibit two characteristic features: (1) an increase toward the center of the projected surface of the sphere; and (2) an overall decrease with increasing $h/R_{*}$. The key point is that we confirm these features for equivalent widths of $< 40$ eV, a regime that has become accessible for the first time thanks to the improved detection threshold from $\sim 50$ eV with Chandra/HETG to $\sim 5$ eV with XRISM/Resolve. As an illustrative application, we compare the maps with XRISM/Resolve spectra of three RS Canum Venaticorum-type stars (GT Muscae, $σ$ Geminorum, and HR 1099) and constrain the locations of the flare loop and coronal bright points in these systems. Because the maps are constructed for a highly generalized point-source--spherical-reflector geometry, they are readily applicable to many other objects, including X-ray binaries and cataclysmic variables.

Fast radio bursts (FRBs) are brief, coherent radio pulses of extragalactic origin. They typically last from microseconds to milliseconds and have energies large enough to be visible over cosmological distances. Since FRBs interact with free electrons along their paths, the original burst is dispersed (Dispersion Measure, DM) and broadened (scattering). Furthermore, the burst's polarization is altered by Faraday rotation. Consequently, FRBs are excellent probes of the cosmological distribution of baryons, the expansion of the Universe, magnetic fields, and minuscule effects of fundamental physics that accumulate over vast distances. This chapter is the second of a trilogy of FRB chapters and discusses FRBs as a standalone probe. We first introduce the foundation of FRB observables related to those questions. Next, we lay the groundwork for forecasting SKA's potential by describing the method to simulate the expected FRB population observable with the SKA. These synthetic FRB catalogues are then used to investigate the SKA's potential to probe the Universe's expansion rate and fundamental physics, such as the equivalence principle and the existence of massive photons. Furthermore, we investigate the possibility of tracing cosmic magnetic fields and investigating different dark matter candidates.

SKA-VLBI is poised to revolutionize our understanding of the Galactic structure through its unprecedented astrometric precision and sensitivity. As a next-generation facility, it will answer long-standing questions about the Galactic structure by mapping its entire spiral structure in detail, spanning from the solar neighborhood, through the Galactic Center, to the far side of the Milky Way. Its access to the Southern sky will allow us to obtain more precise 3D parameters of the Galactic bar, reveal the nature of the 3-kpc Arm, and clarify the dynamical coupling between the bar and the spiral arms. By leveraging high-precision astrometry of numerous celestial objects with SKA-VLBI, the Galactic fundamental parameters such as the Solar motion and the Galactic rotation curve can be constrained more precisely. These advancements will not only elucidate the structure of our Milky Way, but also provide benchmarks for understanding barred spiral galaxies in general. Furthermore, they are important for advancing our knowledge of cosmological structure formation. The capabilities of SKA-VLBI will open a new era of high precision Galactic astrometry.

NGC 1275 showed two TeV $γ$-ray outbursts between November 2022 and January 2023, as detected by the Large High Altitude Air Shower Observatory (LHAASO). The source was also active in the X-ray and GeV bands during the TeV outburst period. Very-long-baseline radio observations reported a sudden acceleration and deflection of a jet knot in late 2022, before the main TeV activity. Motivated by this sequence, we examine whether magnetic reconnection triggered by the interaction between the jet and the ambient medium can explain the TeV flares. In this picture, reconnection produces many plasmoids, and a large ``monster'' plasmoid becomes the main flare region. We model the low-state emission with a multi-zone stochastic-dissipation component and add a compact reconnection-powered region for the flaring state. We then compare leptonic and hadronic interpretations. The leptonic model explains the enhanced X-ray emission as electron synchrotron radiation and the TeV emission mainly as inverse-Compton radiation. A pure proton--proton model can also produce TeV photons if dense target gas is present, but it requires a compact cloud with a density above the values directly inferred from free--free absorption and a large proton power. These requirements are demanding, but they do not by themselves exclude the hadronic interpretation, because the gas may be compressed by the jet or may contain denser cloud cores. Our results show that magnetic reconnection in the parsec-scale jet is a viable origin of the 2022--2023 TeV activity of NGC 1275, while better constraints on the gas density and jet power are needed to distinguish between leptonic and hadronic radiation channels.

We present a spectroscopic atlas of 887 broad-line active galactic nuclei (AGNs) in the Southern sky, spanning redshifts $z < 0.1$, declinations $δ< 0$ deg and Galactic latitudes $|b| > 10$ deg. The sample aims at being a largely complete census of nearby broad-line AGN. The atlas is constructed from observations with $R\sim3000$ using the integral-field Wide Field Spectrograph at the Australian National University 2.3 m telescope. Spectra are extracted in a 6.7 arcsec aperture and have a median signal-to-noise ratio of 13 and 23 per Angstrom in the blue and red arms, respectively. Each spectrum is accompanied by a spectral decomposition that models the AGN continuum and host galaxy, along with fits to the emission lines H$γ$, H$β$, H$α$, He II, He I, [O III], [O I], [N II] and [S II], including broad Balmer and Helium components. The data products are publicly available and designed to support studies of population demographics and AGN variability in conjunction with future time-domain surveys.

Context: Westerlund 1 (Wd1), the most massive young star cluster in the Milky Way, is an excellent laboratory for studying star formation and early stellar evolution in a starburst-like environment. However, high extinction restricts studies of its stellar content, and focus on high-mass stars limits our knowledge of the full spatial extent of the cluster. Aims: We characterize the near-infrared (NIR) variability of the stellar population of Wd1, filling the mass gap between massive stars traced by Gaia and very low-mass stars from previous Extended Westerlund 1 and 2 Open Clusters Survey (EWOCS) studies, to provide a more complete view of cluster membership across solar and super-solar masses.} Methods: We exploited data from the VISTA Variables of the Vía Láctea survey and its extension (VVVX), using NIR point spread function (PSF) photometry and astrometric solutions from its latest data release, namely the VIRAC2 catalogs, mainly in the Ks band. Their large spatial coverage enables study of both the central regions and outskirts of the cluster. We applied HDBSCAN clustering algorithm in a 6D parameter space to differentiate cluster members from field contaminants, assessing robustness through Monte Carlo simulations. Variable sources along the line of sight were also identified and characterized. Results: We identify 1286 high-probability candidate members (12 < J < 18 mag) spanning $\sim 1.5$--$20\,M_{\odot}$, adopting both PARSEC 5 and 6 Myr isochrones ($A_{K_{\rm{s}}}=0.6\,\rm{mag}$, $d=4.23\,\rm{kpc}$). A considerable fraction (34%) shows statistically significant flux variations. We present, for the first time, a parametric analysis of variability modes of Wd1 candidate members in the Ks band, providing a membership catalog suitable for future kinematic studies.

Recent low-frequency MeerKAT observations of radio galaxies have revealed an unexpected population of thin, highly collimated synchrotron threads, whose numbers continue to grow with increasing survey depth and sensitivity. These intricate structures display a remarkable diversity of morphologies -- appearing as narrow filaments linking jets and lobes, as well as ring- or ribbon-like features embedded within the jets and radio lobes. Despite their ubiquity, the physical origin and stability of these collimated synchrotron threads remain poorly understood. Proposed mechanisms include shock compression and interactions with the magneto-ionic intracluster medium, magnetic flux tube formation, and reconnection-driven magnetic filaments. In this work, we investigate the formation and evolution of such magnetic filaments using three-dimensional, two-temperature general relativistic magnetohydrodynamic (GRMHD) simulations of realistically launched jets from supermassive black holes. From these simulations, we compute synthetic synchrotron emission maps and polarisation signatures, allowing us to predict the observable characteristics including morphology, brightness profiles, and polarisation patterns. Finally, we assess the detectability and diagnostic potential of these signatures with the Square Kilometre Array Observatory (SKAO), outlining how upcoming SKA observations can distinguish between competing physical models and illuminate the magnetic origin of the collimated synchrotron threads revealed by MeerKAT observations.

Fast radio bursts (FRBs) provide a glimpse of high-energy astrophysical phenomena in other galaxies. They point the way to extreme conditions that are currently undetectable by any other known means. These coherent radio flashes have timescales of microseconds to milliseconds, and inferred energies that are comparable to those of the most extreme bursts seen from Galactic neutron stars. However, the nature of FRB sources remains an open question in astrophysics. Magnetically powered neutron stars known as `magnetars' are a leading candidate for explaining the FRB phenomenon, but other plausible progenitors include magnetically interacting neutron-star binaries or accreting black holes. The diversity of FRB burst types and their galactic environments hint that multiple mechanisms and progenitor types may be responsible. Here we discuss the ways in which the SKA can uncover the nature of FRBs. In particular, we focus on the key advantages of the SKA: its Southern Hemisphere location and hence overlapping sky coverage with the Vera C. Rubin Observatory, its high sensitivity compared to existing wide-field FRB surveys, its fast search timescales down to tens of $μ$s, and its broad spectral coverage with bands from 50 MHz to 15 GHz. With these capabilities, the SKA will excel in detecting FRB sources across new frequency ranges and timescales. This will aid in a better understanding of the fundamental astrophysics behind FRBs, which will in turn also contribute to their use as cosmological probes, as explored in companion chapters.

We present the first spectroscopic mineralogical analysis of the dust coma of an interstellar object (ISO) from JWST mid-infrared spectroscopy of 3I/ATLAS (3I). 3I exhibits a strong 10-micron emissivity feature commonly seen on asteroids, comets, disks, and the interstellar medium. Characterization of this 10-micron emissivity maximum reveals that 3I's dust composition is dominated by amorphous silicates, and that 3I is unlike Solar System comets, which show significant crystalline silicate dust. Instead, 3I's dust composition is more similar to circumstellar transition disks and the interstellar medium. We suggest 3I may have formed in a distant part of its home system out of interstellar medium-like material, without substantial incorporation of silicates condensed near its host star, unlike the mixing scenarios commonly hypothesized for Solar System comets. Alternatively, 3I's original crystalline silicates may have been amorphized during its Gyr-long journey, although we find this alternative less likely due to 3I's mass loss rate and distinct 10 micron feature as opposed to observed Solar System comets.

We present the results of an effort to derive interstellar gas-phase C II abundances along the lines of sight toward rho Oph A and B. Our analysis is based on high-resolution NUV and FUV archival spectra acquired with the Space Telescope Imaging Spectrograph on the Hubble Space Telescope. Column densities of C II are derived both from fits to the weak C II] 2325 intersystem transition and from fits to the damping wings of the strong C II 1334 line. We find that the results from the weak-line and strong-line determinations agree with each other remarkably well for both sight lines, demonstrating the reliability of the f-values of the C II transitions. Furthermore, the gas-phase C abundance that we obtain for rho Oph A is in very good agreement with previous determinations of interstellar C abundances from measurements of the weak C II] 2325 line. By demonstrating the reliability of the damping wing fitting technique for the C II 1334 line, our analysis opens the door to future surveys of interstellar C abundances using the same methodology.

We report the discovery of an extremely narrow and highly collimated gaseous trail-like structure crossing the eastern region of the Veil Nebula (Cygnus Loop). The feature was first identified in deep narrow-band H-alpha images obtained by a small telescope and was confirmed in more than a dozen publicly available amateur and professional images in the last two decades, rejecting the possibility of being an artifact (e.g., artificial satellite trail). To characterize its structural and luminosity parameters, we applied a model-fitting code originally developed for the analysis of extragalactic stellar tidal streams. The structure is only detected in H-alpha emission, with no detectable counterpart in [SII] or in visible observations, and shows an almost constant brightness and width along its extension. It exhibits a surface brightness of 22.32+/- 0.13 H-alpha mag arcsec^2. Model fitting yields a median width of 1.63", which corresponds to a physical scale of approximately 1200 UA, assuming it is at the same distance (2400 light-years) as the Veil Nebula remnant. We discuss several potential scenarios for the origin of this feature: a Herbig-Haro-like jet, the trail of a high-velocity object, or a non-radiative shock associated with the supernova remnant. We rule out the Herbig-Haro scenario, given the absence of a driving stellar source and the lack of [S II] emission. A high-velocity-object wake cannot yet be excluded, particularly if the driver is a compact remnant, but the absence of an obvious source or bow-shock apex and the extremely small, nearly constant width of the structure make a normal runaway-star origin unlikely. The current evidence instead favours a Balmer-dominated non-radiative shock associated with the Cygnus Loop, generated as the SNR blast wave encounters a dense material layer or magnetic structure viewed nearly edge-on.

Massive stars and stellar clusters shape galactic evolution through powerful feedback mechanisms including radiation pressure, photoionization, stellar winds, and cosmic ray acceleration. However, their impact remains poorly understood due to observational challenges: they are rare, distant on average, and deeply embedded within dense, dusty environments. Radio observations provide a unique window into these processes, as radio emission penetrates obscuring material and traces both thermal free-free emission from ionized gas and non-thermal synchrotron emission from shocks and particle acceleration. The Square Kilometre Array (SKA) will revolutionize massive star studies through unprecedented sensitivity and angular resolution. SKA observations will enable detailed characterization of hierarchical structures within HII regions, measurements of physical conditions through hydrogen, helium, and carbon radio recombination lines (RRLs), and detection of non-thermal emission from cosmic ray acceleration in star-forming regions. SKA will permit systematic measurements of stellar wind mass-loss rates, studies of photoionized gas kinematics and dynamics, and exploration of photodissociation regions surrounding ultracompact HII regions. Additionally, magnetic field strengths can be probed through Zeeman effect observations of RRLs. This chapter discusses the current understanding of massive stars and stellar clusters and their feedback processes. We highlight how SKA observations will advance our knowledge of massive star formation, stellar winds, hierarchical structures in HII regions, cosmic ray acceleration, and magnetic field regulation of star formation - providing crucial insights into feedback mechanisms governing the structure and evolution of the Milky Way and galaxies.

Despite its remarkable success on large scales, the standard $Λ$CDM paradigm faces persistent small-scale challenges that have motivated alternative models for the dark sector. Self-interacting dark matter (SIDM) offers a compelling possibility, in which dark matter particles can scatter off each other. Stellar bars are a ubiquitous feature of disk galaxies across cosmic time. Bars are dynamically coupled to their host galaxy's dark matter halo, and therefore their properties provide a powerful probe of the nature and distribution of dark matter. In this paper, we use idealized, high-resolution $N$-body simulations and analytic calculations based on kinetic theory to study bar formation and evolution in disk galaxies embedded in SIDM halos. We find that compared with collisionless CDM, SIDM produces bars that form earlier and grow to larger amplitudes, even for modest self-interaction cross sections. In several cases, disks that remain stable in CDM, including kinematically hot and dark-matter-dominated disks, develop strong bars in SIDM. This accelerated bar growth occurs because self-interactions broaden the bar-halo resonances and enhance angular momentum transfer from the stellar disk to the halo. We explicitly show that this phenomenon is not related to core formation in SIDM halos. At late times, gravothermal core collapse can raise the central dark matter density enough to weaken or dissolve the bar. These results suggest that the abundance, strength, and redshift evolution of barred galaxies offer a promising observational route to constraining dark matter self-interactions, particularly in light of the growing sample of high-redshift bars revealed by JWST.

This chapter investigates the low-mass frontier of galaxy formation through two complementary populations: the starless Reionization-Limited HI Clouds (RELHICs) that trace the ``dark'' Universe, and the faint, gas-rich galaxies that define the ``dim'' Universe. RELHICs offer pristine laboratories for probing the distribution of DM on sub-galactic scales, providing a direct test of the Lambda Cold Dark Matter ($Λ$CDM) model predictions. The dim Universe provides statistical constraints on cosmology, galaxy formation and evolution, as well as baryoninc physics through key observables including the low-mass end of the neutral-hydrogen mass function (HIMF), the neutral-hydrogen velocity function (HIVF), and the low-mass end of the baryonic Tully-Fisher relation (bTFR). This chapter outlines core science questions that can be tackled leveraging radio observations of both the dark and dim Universe. Additionally, it outlines strategies to identify RELHICs amid tidal or pressure-confined contaminants, while providing observational predictions for the dim Universe. The Square Kilometre Array (SKA) in its mid-frequency Array Assembly 4 (AA4) configuration will, for the first time, resolve the internal gas structure of nearby RELHICs and build deep, wide-area datasets that definitively constrain the HIMF, HIVF, and bTFR down to masses of $10^{6}~\msol$ -- offering a complete observational framework to test the $Λ$CDM paradigm and the baryonic processes that shape the faint end of galaxy formation.

We present JWST/MIRI spectroscopy of 14 disks in the older (~5-10 Myr) Upper Scorpius (USco) association and use slab of gas in local thermal equilibrium to infer basic gas properties. We find that half of these disks are molecular rich, with detections of H$_2$O, CO$_2$, HCN, C$_2$H$_2$, and H$_2$, while the other half are molecular poor, showing no molecular emission other than H$_2$. We further combine this sample with 10 other USco disks from the AGE-PRO program and compare the combined older sample to young (~1-3 Myr) JDISCS Cycle~1 systems, which are analyzed in a similar manner. We find that USco disks have lower detection rates of major molecular species but a significantly higher detection rate of rarer C-bearing molecules such as C$_4$H$_2$. At a given accretion luminosity, molecular line luminosities are systematically lower in USco than in young disks, and the scaling relations with accretion luminosity differ between the two populations. Moreover, we find that about half of the older disks, preferentially the millimeter faint, and likely more compact disks, have observable mass ratios of C- to O-bearing molecules that are higher than the maximum values in the young sample. These results point to reduced inner-disk molecular gas masses, cooler emitting layers, and higher inner gas C/O ratios in older disks, the latter being consistent with pebble drift. Taken together, our findings provide evidence for chemical evolution of inner disk gas from young to older systems, with important implications for the accretion of primordial planetary atmospheres.

We present JWST/NIRSpec spectroscopy of the low-mass, metal-poor galaxy SDSS~J160135.95+311353.7 (J1601), selected for its extreme mid-infrared colors and compact nuclear emission, placing it within widely used WISE color diagnostics for active galactic nuclei (AGNs). Despite this selection, we find no evidence for coronal lines, X-ray emission, or variability typically associated with accretion activity. We compare J1601 to SDSS~J120122.30+021108.3 (J1201), a similar but lower-mass, more metal-poor system studied previously (Doan, 2025). Both galaxies host compact nuclear starbursts but differ in their stellar populations and dust properties: J1601 shows CO bandhead absorption indicative of red supergiants, weak nuclear Wolf--Rayet features, and a circumnuclear PAH ring, consistent with a more developed recent starburst, while J1201 is more dust-enshrouded and chemically primitive. Despite these differences, neither system shows evidence for AGN activity, indicating that the absence of accretion is not simply due to evolutionary timing. Photoionization models show that the weakness of high-ionization emission cannot be explained by low metallicity alone, implying a genuine deficit of hard ionizing photons. Crucially, the red mid-infrared colors in both systems originate from compact, unresolved nuclear emission confined to the nuclear star cluster. These results demonstrate that compact nuclear starbursts can mimic AGN-like mid-infrared colors without accretion, and that commonly used AGN diagnostics may not uniquely identify accreting black holes in metal-poor dwarf galaxies. Our findings suggest that such systems may not provide the conditions required for efficient black hole growth and/or may lie near or below the regime where black hole seeds can form.

The unprecedented statistical power of galaxy cluster catalogs from the SRG (Spectrum Roentgen Gamma)/eROSITA All-Sky Survey provides a unique opportunity to place stringent constraints on cosmological models through measurements of structure growth. Fully exploiting the potential of these large X-ray-selected cluster samples, however, requires robust statistical frameworks that accurately connect observable quantities to the underlying cosmological parameters. We develop and implement a simulation-based inference (SBI) framework for cosmological parameter estimation using a realistic mock-generation pipeline calibrated on eRASS1 simulations. Synthetic galaxy cluster catalogs are propagated through the survey selection function to produce mock eRASS1 observations that reproduce the data's statistical properties. At the core of the method lies a set-based neural network (GNN on sets) that encodes information from individual clusters and is coupled to a masked autoregressive flow for flexible posterior density estimation. This approach enables the use of the full cluster-level information content without compressing the observables into binned summary statistics. Our framework recovers the input cosmologies within the inferred uncertainties, and passes calibration tests, demonstrating robustness in the presence of realistic survey effects. We obtain mock constraints of 11.5% on $Ω_m$ and 4.4% on $σ_8$ averaged over a suite of simulated cluster catalogs matching the effective sample size of the data set (3,259 clusters). We achieve a precision comparable to that obtained with traditional MCMC analyses based on substantially larger cluster samples. The framework is readily extensible to more complex forward models and additional observables. This work highlights the potential of SBI methods for next-generation large-scale structure analyses with forthcoming X-ray cluster surveys.

Three-form fields provide a theoretically well-motivated framework for dark energy, arising in higher-dimensional theories and exhibiting a rich cosmological phenomenology. We investigate a minimally coupled three-form dark energy model with a Gaussian potential and constrain it using current cosmological observations, including CMB shift parameters, DESI DR2 baryon acoustic oscillation measurements, Pantheon+ supernovae with and without SH0ES calibration, cosmic chronometers, and gamma-ray bursts. Parameter estimation is performed within a Bayesian Markov-chain Monte Carlo framework, while model comparison relies on several information criteria and the Bayesian evidence, as well as tension statistics. We find that the three-form model provides a viable and competitive description of the expansion history of the Universe. It is mildly preferred over $Λ$CDM for the combination of early and late-time datasets that are heavily tensioned (CMB+BAO and Pantheon+SH0ES). This preference decreases to neutrality for the other, less tensioned combination of early and late-time data, while for individual early-time or late-time datasets analysed separately, the information criteria are neutral or favour $Λ$CDM. This suggests that the additional degrees of freedom of the three-form field may help accommodate cosmological observations of different origins within a common framework. The reconstructed dark energy dynamics exhibit a characteristic phantom phase at intermediate redshifts while approaching a cosmological-constant-like behaviour at early and late times, providing a distinctive observational signature. Although the model does not significantly alleviate the Hubble tension despite allowing higher values of $H_0$, it remains consistent with current observations and offers a well-motivated alternative to $Λ$CDM whose predictions can be tested by future cosmological surveys.

In 2017, the blazar TXS 0506+056 showed a remarkable gamma-ray outburst simultaneously with a high-energy neutrino detection by IceCube from the same sky region. The significance of this association was found to be on the order of 3 sigma thus providing a strong link between the neutrino emission and the blazar flare. The high-energy flare in TXS 0506+056 was followed by a delayed radio flare, peaking in $\sim2020$ in the aftermath of the neutrino event. We investigate the parsec-scale jet structure and dynamics of TXS 0506+056 using 15 GHz full-polarization VLBI observations obtained between 2009 and 2025 with the Very Long Baseline Array. Our kinematic analysis reveals moderate superluminal jet speeds of $\sim(1-2)$c and two quasi-stationary components. A new jet component with comparable speed was ejected contemporaneously with the 2017 IceCube neutrino event. The stacked polarization map is consistent with a stratified spine-sheath jet structure: an inner spine with EVPA aligned with the jet, surrounded by a sheath layer with perpendicular EVPA. Variability in total intensity and polarization further indicates interaction between these layers, particularly evident in characteristic linear polarization flares, associated with EVPA rotations of the quasi-stationary components. The multi-layered jet configuration is consistent with previous studies that were able to explain the neutrino emission in this TXS 0506+056 through a spine-sheath jet structure. We suggest that TXS 0506+056 represents an archetypal case and that similar polarization signatures and geometric light curve flares may be present in other neutrino-emitting sources, potentially offering a solution to the Doppler-crisis phenomenon observed in TeV-emitting blazars.

Radio observations of $γ$-ray bursts (GRBs) employing the very long baseline interferometry (VLBI) technique provide us with fundamental information on the dynamics and the geometry of the GRB blast wave. With its high angular resolution ($\sim$milli-arcsecond), VLBI allows us to measure the apparent superluminal expansion, to characterise the structure of the jet and to constrain the viewing angle and jet opening angle. While this information is crucial to understand these transient events, such studies have been possible only for three GRBs to date, owing to both the poor sensitivity of current radio facilities and the paucity of close and bright GRBs. In this chapter, we estimate the impact that the SKA-Mid will have on these studies, when included in a VLBI network. We performed a series of dedicated simulations of VLBI observations of GRBs, considering five VLBI networks and the SKA-Mid, both in its AA* and AA4 configurations. We show that including the SKA-Mid in a global-VLBI experiment will: (i) allow us to measure the size and the expansion of a GRB up to a redshift $z\simeq 0.25$ (at a confidence level of $3σ$); (ii) constrain the size $\gtrsim$2 times better than the current global-VLBI array; (iii) improve the localisation precision in Declination from 4 to 30 times; (iv) detect the apparent proper motion of GRBs seen slightly off-axis with a confidence level 3 times better than current VLBI networks. Ultimately, the SKA-Mid will open a new window on a portion of the GRB population that has been inaccessible so far.

Supermassive black holes launch powerful jets of plasma that can accelerate particles to extreme energies, but the physical conditions required to produce high-energy neutrinos remain unknown. In the blazar TXS 0506+056, the first source individually linked to a high-energy neutrino, radio images had seemed to reveal a jet too slow to sustain the extreme conditions required for neutrino production. Here we resolve this tension using long-term radio monitoring, particularly Very Long Baseline Interferometry (VLBI) imaging. We uncover a disturbance in emission propagating with an apparent speed of 21+-1 times the speed of light, that is masked by slower, radio-bright features that dominated earlier analyses. We interpret this as the signature of a stratified jet: an ultra-relativistic spine with Lorentz factor Gamma>20 embedded within a slower outer sheath. As the disturbance travels along the spine, it progressively illuminates the sheath, producing the delayed radio flare and naturally accounting for the years-long offset between neutrino and radio emission. The same pattern recurs in a second neutrino-associated event, pointing to a repeatable multi-messenger engine. These findings challenge the standard interpretation of VLBI jet speeds and establish a concrete, testable framework connecting structured jets to the sources of the Universe's highest-energy neutrinos.

We investigate the assembly history of early galaxies in the SEEDZ hydrodynamic simulations, to investigate the high inflow rates believed to be required for the formation of supermassive stars (SMSs), dense stellar clusters and subsequently heavy seed black holes. Using a heavy seed formation criteria of $>$1 M$_\odot$ yr$^{-1}$ flowing into 10 pc regions, we find that heavy seeds form in halos that grow rapidly compared to those halos that never meet the criteria. Halos with growth rates of $\gtrsim$1 M$_\odot$ yr$^{-1}$ at their virial radius (scales of a few hundred pc) are able to sustain a flow rate of 0.1 M$_\odot$ yr$^{-1}$ into the inner 1 pc of the halo, maintaining higher density environments within the central 10 - 100~pc. These halos continue to grow rapidly after their initial collapse, typically forming heavy seeds $\sim$100 Myr after forming their first stars and stellar mass black holes. By $z=10$, most heavy seeds form in regions of near-solar metallicity, although a minority of heavy seeds do continue to form in low metallicity (10$^{-2}$ Z$_\odot$) regions. Under the assumption that a SMS forms as the progenitor to a heavy seed if it forms in a region of low (10$^{-2}$ Z$_\odot$) metallicity, and can sustain high accretion rates above 0.02 M$_\odot$ yr$^{-1}$ throughout the SMS lifetime of 2 Myr, we find a number density of SMSs of 0.1 cMpc$^{-3}$, meaning that only a fraction of 10$^{-4}$ of these SMSs would need to be visible to JWST to account for the observed population of Little Red Dot galaxies.

We investigate the rotational properties of star-forming clusters at $z \sim 7.6$ in the high-resolution simulation MassiveBlackPS, focusing on two formation channels: clusters forming in galactic discs via gravitational instability and isolated circumgalactic systems, referred to as cosmic wallflowers, born out of cosmic filaments. Using stellar kinematics, we compare their rotational velocities, $v_{\rm rot}$, and rotational support, $v/σ$, to study whether formation environment leaves a clear dynamical imprint. We find a clear separation, wherein cosmic wallflowers systematically have lower rotational velocities and span a wide range in $v/σ$, whereas the identified disc clusters are strongly rotation-dominated and extend to higher $v_{\rm rot}$. When combined with stellar surface densities, a subset of the low-$v_{\rm rot}$ cosmic wallflowers lie surprisingly close to the observed globular cluster population in the Milky Way, whereas disc clusters remain offset. Within the cosmic wallflower population, we identify two regimes: lower-density, weakly rotating systems that overlap with these globular cluster properties, and denser, more rotationally supported systems that likely follow a different evolutionary pathway, possibly linking them to the origin of massive black hole seeds at high redshift. We further find that the gas content correlates with this behaviour, with gas-rich cosmic wallflowers preferentially occupying this low-rotation regime. This all suggests that environment and baryonic content together play a key role in setting the initial dynamical state and possible fate of clusters. In particular, weakly rotating, gas-rich cosmic wallflowers emerge as natural proto-globular cluster candidates, potentially evolving towards present-day systems through angular momentum loss and dynamical heating.

Lyman-alpha (Ly$α$) radiative transfer (RT) is important in many astrophysical environments and governed by multiple physical processes. In this paper, we provide analytical formulae/procedures for the solutions to Ly$α$ RT equations under three simple geometrical symmetries and investigate the effects of atomic recoil and gas bulk motion. We first study Ly$α$ spectra by solving Ly$α$ RT equations for a static, uniform gas cloud under cylindrical geometry. The solution is verified through Ly$α$ Monte Carlo RT simulations, and compared to those under slab and spherical geometries in literature. Second, to characterise the recoil effect, we empirically modify recoil-free Ly$α$ spectra. The method is motivated by Ly$α$ RT equations with recoil and justified by simulations. Finally, we account for constant velocity gradients in Ly$α$ RT equations and obtain series solutions for Ly$α$ spectra. The solutions demonstrate good agreement to Ly$α$ spectra from simulations for small velocity gradients (i.e. edge velocity $v_{\rm E}$ of a cloud being comparable to the thermal velocity $b$) but become less accurate for large ones. To characterise Ly$α$ spectra under large velocity gradients, we empirically extend the functional form of solutions and constrain them from fitting simulated Ly$α$ spectra. The resulting fitting formulae show significant improvement for large velocity gradients ($v_{\rm E}/b \sim 100$) under large optical depths. The analytical study of Ly$α$ spectra in this work completes the set of solutions under simple geometries, provides physical insights for Ly$α$ RT under recoil and velocity gradient, and develops analytical tools for theoretical studies that require inputs from Ly$α$ RT.

Polarimetric properties of blazars provide key constraints on the acceleration mechanisms powering their relativistic jets, the high-energy emission processes involved, and the composition of the jet itself. We present a multi-band polarimetric study spanning from soft X-rays to gamma-ray energies, considering several bands for current and future missions (0.275 keV, 2-8 keV, 0.5-10 keV, 6-35 keV, 0.2-5 MeV, and 1-100 GeV). Our sample is drawn from the RoboPol monitoring programme, a statistically complete gamma-ray sample including low-, intermediate-, and high-synchrotron-peaked blazars. Using spectral energy distribution fitting performed with the Bjet_MCMC tool, we give predictions on the flux expected for each source in the selected energy bands. We model the polarimetric signatures under three competing emission scenarios: leptonic, hadronic, and hybrid. The detectability of each source was assessed by accounting for the instruments' minimum detectable polarisation (MDP) and by computing the probability of the detection in a blind survey. We show that simultaneous multi-wavelength observations can effectively discriminate between competing emission models due to the difference in the expected polarisation degree. Finally, we derive sensitivity requirements for future gamma-ray polarimetric missions aimed at increasing the number of detectable sources.

The VLA Sky Survey (VLASS) is a radio continuum survey of the entire sky visible to the VLA (dec > -40deg) at ~2.5 arcsecond resolution over 2-4 GHz with full polarization. Data processing and quality assurance have proven challenging due to limited computing and staff resources, and require innovative approaches. Although "Quick Look" images in Stokes I continuum were produced within a few weeks of observation, high accuracy images that utilize the full potential of the dataset are computationally intensive and take much longer to produce. Half of the sky will need correction for w-terms to account for sky curvature, increasing the computational cost per image by ~100x. This problem will be tackled using GPUs and High-Throughput Computing (HTC) centers. Quality assurance of the tens of thousands of individual images has used a mix of algorithmic and machine learning techniques to automatically identify common artifacts and anomalies in the data. Archiving and dissemination of VLASS data products also pose a significant challenge.

Investigating the thermal and nonthermal processes in the interstellar medium (ISM) and intergalactic medium (IGM) is vital to understanding the evolution of galaxies over cosmic time. Resolved observations with SKA pathfinders show that the nonthermal processes, in which magnetic fields and cosmic rays are involved, can decelerate the formation of massive stars in strongly magnetized regions in nearby galaxies. They can also contribute to the onset of winds and outflows in galaxies. The effects of these processes are stronger at higher redshifts as a result of star formation activities. The SKA Observatory will allow a major breakthrough by mapping the thermal and nonthermal processes in distant universe galaxies, shedding light on the role of the ISM and IGM in the evolution of galaxies. We demonstrate this by simulating the radio continuum and HI emission from local galaxies back to high redshifts. Our simulations show that the AA4 surveys will make it possible to trace the thermal and nonthermal processes of the ISM in galaxies that are analogs to M51 and NGC6946, traced in continuum beyond cosmic noon (z=2-3) and the gas content traced by HI beyond z=1. Both simulations and precursor observations indicate the importance of nonthermal feedback at cosmic noon.

We provide an overview of the eight SKAO Science Book chapters that motivate the Epoch of Reionisation and Cosmic Dawn experiments with SKA-Low. We describe the individual SKA-Low experiments and expected sensitivity - power spectrum, tomography, 21-cm forest, cross-correlations, building on the broad observational plan laid out in the 2015 SKA Science Book. Finally, we outline features of the telescope that will be critical for the success of EoR/CD science, e.g., beam apodization, substations, and multi-beaming.

Star formation emerges from the complex interplay between gravity, turbulence, magnetic fields, and stellar feedback, all of which vary across spatial scales and Galactic environments. Over the past decades, extensive multiwavelength surveys of the Galactic Plane have progressively unveiled this complexity. Far-infrared and sub-millimetre surveys have identified and characterized tens of thousands of star-forming regions, revealing their mass, temperature, and evolutionary stage. Complementary molecular-line surveys, spanning several CO transitions and isotopologues, have mapped the gas kinematics from giant molecular clouds down to sub-parsec structures. The advent of interferometers such as ALMA has revolutionized this field, enabling systematic studies of gas dynamics, fragmentation, and collapse in dense clumps at scales of a few thousand astronomical units. At the same time, mid-infrared and radio surveys at frequencies 0.8 <= nu <= 5 GHz have traced ionised gas associated with the earliest and latest phases of massive-star evolution, including thermal radio jets, hypercompact and ultracompact HII regions, supernova remnants, planetary nebulae, and evolved massive stars. Yet, a uniform, Galaxy-wide census of ionised structures and feedback processes remains elusive. A transformational leap forward requires a sensitive, high-resolution radio survey of the Galactic Plane at 10-15 GHz, capable of resolving physical scales smaller than 0.05 pc at distances up to 20 kpc. This is precisely the goal of the SKA-Mid Galactic Plane survey, which will, with its unprecedented sensitivity, angular resolution, and mapping speed, provide the first panoptic view of ionised gas and stellar feedback across the Milky Way.