Papers

Compression schemes provide a general framework for deriving data-dependent risk bounds for learning algorithms. For learning rules based on preferent compression, existing results upper-bound the risk in terms of a sample-dependent complexity $\boldsymbol{k}$, defined as the cardinality of the compression augmented by the training examples on which the reconstructed hypothesis is inappropriate. These results, however, yield a margin of order $O(\sqrt{\ln N/N})$ above $\boldsymbol{k}/N$. In this paper, we establish that the logarithmic factor is not intrinsic. We prove that, for every fixed confidence level $1-\delta$, the risk is bounded with probability at least $1-\delta$ by $$ \frac{\boldsymbol{k}}{N} + O\left(\frac{1}{\sqrt{N}}\right), $$ uniformly over all possible values of $\boldsymbol{k}$. Thus, risk estimates for learning algorithms based on preferent compression attain the same law-of-large-numbers rate as the empirical estimate of the probability of a fixed event. We also show that the leading term $\boldsymbol{k}/N$ is necessary and investigate its relation to the empirical risk. In particular, we identify regimes in which the empirical risk retains the $O(1/\sqrt{N})$ convergence rate, as well as regimes in which it converges more slowly or ceases to be consistent, while the complexity-based estimator preserves the $O(1/\sqrt{N})$ rate.

We prove that the dimension of the deformation cone of any polytope combinatorially equivalent to a 2-cubical zonotope is bounded above by the number of generators of the zonotope. This bound is tight, and we show that it cannot be improved to an identity. We further extend our results to any 2-cubical polytope whose normal fan is combinatorially isomorphic to a pseudohyperplane arrangement, but we show that it cannot be extended to arbitrary 2-cubical polytopes.

Recent accent classification systems can achieve high accuracy in standard in-corpus evaluations. However, when speakers overlap between the training and test sets, such results may overestimate model robustness compared with real-world deployment scenarios. This study revisits the robustness of DistilHuBERT-based accent classification by using L2-ARCTIC and multiple external English accent datasets. The comparison covers file-level and speaker-disjoint splits, along with balanced external-data training and teacher-student pseudo-labeling. The large performance gap between in-corpus and external evaluations suggests that file-level evaluation substantially overestimates performance, while speaker-disjoint evaluation can provide a more realistic estimate of the generalization to unseen speakers. In this study, models trained on external data alone do not reliably generalize to L2-ARCTIC. Among the tested configurations, balanced external-data training offers the most promising supervised setting. The pseudo-labeling experiments further indicate that confidence-based filtering alone is insufficient under domain mismatch. Overall, this study suggests that robust accent classification evaluation should simultaneously include speaker-disjoint evaluation and cross-corpus testing.

Collective Autonomous Vehicle (AV) route choices in real-world traffic networks can create externalities for Human-Driven Vehicles (HDVs), as routes that benefit AVs may increase travel times for other road users. To address this, we introduce a social component, the marginal cost, to the selfish travel time AV reward. Marginal cost captures an AV's impact on other road users by comparing their travel times with those in a counterfactual simulation in which the AV is removed. We study AV route choice in mixed-autonomy traffic settings using Multi-Agent Reinforcement Learning (MARL), where AVs learn routing policies through repeated interactions with the traffic environment. Experiments on the real-world networks of Ingolstadt and Provins show that incorporating marginal cost into the AV reward not only limits the increase in HDV travel times and shortens AV travel times relative to selfish AV routing, but also improves overall system performance. In Ingolstadt, selfish routing of 40% AVs increases the travel times of other vehicles by 1.7%, and this drops to 1.1% when the social reward is included. Interestingly, not only does overall system performance improve, but the AVs are also better off when acting socially.This effect is consistent across tested cities, penetration rates, and replications, yet when the weight of the social component in the reward increases, the system fails to reach the desired performance. For a more scalable approach to social AV routing in mixed-autonomy traffic, we compute the marginal cost only for the AVs whose route choices most affect the traffic system. This reduces the number of counterfactual simulations and achieves similar performance at a significantly lower computational cost. Our results suggest that AV routing should go beyond selfish rewards and, for the benefit of all traffic participants, include a social component in its objective.

Reinforcement learning (RL) has emerged as a key paradigm for improving large language model (LLM) agents, enabling stronger reasoning, planning, and tool-use in interactive environments. However, existing agent RL methods remain fundamentally constrained by an on-policy reasoning bottleneck: when tasks exceed the model's current capability boundary, effective trajectories become extremely sparse and training often collapses. We introduce Rollout-Time Intervention (RTI), a technique designed to elicit successful trajectories beyond the policy's intrinsic exploration capacity. RTI operates by leveraging reward-side evidence (i.e., evaluation rubrics) during rollout to convert erroneous trajectories into successful ones, providing effective learning signals for tasks that the policy rarely solves on its own. Although intervention trajectories are inherently off-policy, we find that they can effectively guide policy optimization when trajectories with excessive distributional shifts are filtered out. Extensive experiments across five benchmarks demonstrate that RTI consistently outperforms RL baselines, enabling the post-trained policy to approach the performance of substantially stronger contemporary LLMs and, more importantly, transcend its original capability boundary. The code is available at https://github.com/Alibaba-NLP/qqr.

Anomaly detection, the identification of collision events that deviate from Standard Model expectations, is one of the most promising model-agnostic strategies for new-physics searches at the Large Hadron Collider. Searching for rare events in learned low-dimensional representations removes the need to commit to hand-crafted observables a priori, but sacrifices the immediate physical interpretation those observables provide. For anomaly detection to enter a discovery pipeline, flagged events must therefore be diagnosed locally, both to distinguish potential new physics from mismodeling or detector effects and to identify observables for targeted follow-up searches. We evaluate supervised and self-supervised neural embedding strategies on COLLIDE-2V, a 750-million-event dataset simulated under realistic High-Luminosity LHC conditions, using the reconstruction error of an autoencoder trained in the learned latent space as the anomaly score. Across the configurations and benchmark signals considered, supervised objectives, which exploit simulation process labels during training, generally outperform their self-supervised counterparts and, in their best configurations, improve on an autoencoder trained directly on kinematic features, raising the signal efficiency at 10% background acceptance on all held-out processes. We further introduce a local diagnostic pipeline that fits a Gaussian Mixture Model to the SM latent space, profiles each component through high-level observables without process labels, localizes the flagged events, and ranks observables by their local discriminating power. The ranking recovers the defining kinematics of the held-out processes, while correlating the anomaly score with the same observables quantifies how closely the score tracks them, revealing cases where the most discriminating observable is not the one the score follows most closely.

Building upon our preliminary study, this paper proposes a standalone beam-management framework for mobile MIMO-OFDM systems that integrates Blind Initial Beam Acquisition (BIBA) with Clock-Assisted Re-Acquisition (CARA). BIBA is a blind and asynchronous protocol that requires neither prior synchronization nor an out-of-band feedback channel, so the Base Station (BS) and User Equipment (UE) can independently identify the beam pair maximizing received signal strength. To handle subsequent link drops, the newly introduced CARA protocol exploits the clock synchronization established after BIBA to reduce the re-acquisition time by approximately 55% compared with BIBA. The operational reliability of both protocols is bounded by a probabilistic feasibility analysis under random UE mobility trajectories. Protocols are validated on a 28 GHz mmWave testbed under practical hardware constraints, including beam-switching and Time Division Duplex (TDD) latencies, and measured Software-Defined Radio (SDR) software processing delays. Finally, we formulate a unified Finite State Machine (FSM) model for both BIBA and CARA to ensure robust integration into practical wireless systems

Photoplethysmography (PPG) is increasingly recorded by wearable and clinical devices at enormous scale, and a growing number of PPG foundation models (FMs) have been developed to compute reusable representations for various downstream tasks. These models are difficult to compare: different evaluations use different datasets, task definitions, preprocessing, downstream heads, hyperparameter tuning, and baselines. We present PPG-Bench, a unifying open-source benchmark evaluating five public PPG FMs across nine open-access datasets and 26 tasks spanning cardiovascular state estimation, activity, sleep, screening, and identification. We evaluate models under four downstream regimes of increasing adaptability (linear probing, random forests, frozen MLP probes, and end-to-end finetuning), with model-specific hyperparameter optimization and subject-disjoint cross-validation. We also evaluate against three controls: random/constant baseline, a deterministic pulse-rate-variability (PRV) feature set, and a generic time-series FM (MOMENT). Across all 68 task-regime combinations, 35.3% show no meaningful FM advantage over the random/constant baseline, and in 83.8% no PPG-specific FM outperforms both PRV and MOMENT. We also find that added downstream head capacity generally improves performance, but can cause significant degradation if the model is performing at chance/constant level. Current PPG FMs therefore cannot yet claim to provide generalised and transferrable representations of PPG signals, and future evaluation in this field should include checks against simple, non-learned controls. Code for the benchmark and instructions to evaluate more models can be found here - https://github.com/mannas-ai/ppg-bench.

Saturn's icy moon Enceladus is a prime target in the search for life beyond Earth. Abundant H2 and CH4 in Enceladus's south-polar plume have been shown to be consistent with hydrogenotrophic methanogens feeding on products of long-lived serpentinization. However, thermo-permeability models make sustained core-wide H2 production by serpentinization uncertain. We therefore explore whether an alternative metabolism based on fermentation-driven heterotrophy could reproduce Cassini H2 and CH4 escape fluxes under limited serpentinization. We develop a coupled ecosystem model of fermentation and methanogenesis in a hydrothermal-ocean-plume framework, using glycine as a proxy for amino-acid-like fermentable organics. With effectively no abiotic H2 and CH4 input, purely heterotrophic communities with fermenters and methanogens can reproduce both Cassini-derived flux intervals given sufficient hydrothermal glycine-equivalent supply, while the estimated organic stock could sustain the required H2 production on gigayear timescales if fully available. We then test a scenario with input ranges for fermentable organic matter, H2, and CH4 informed by asteroid-sample measurements and geochemical estimates of limited serpentinization. Simulations within both Cassini-derived flux intervals are concentrated in communities with acetoclastic methanogens; Bayesian model comparison favors the same communities. This holds even allowing for greater abiotic H2 and CH4 inputs that reduce the fermentable-organic supply required to reproduce the plume. We therefore identify fermentation heterotrophy as a testable alternative to purely hydrogenotrophic, serpentinization-dominated biotic interpretations and motivate future plume searches for co-exported fermentation products, including acetate and ammonium, together with measurements of methane carbon isotopes.

This paper addresses the grid quantization error of deep-learning (DL)-based direction-of-arrival (DOA) estimation. Classification- based DOA methods discretize the continuous angular space into a grid, so that with a peak-based read-out, the accuracy and resolution are limited by the grid step and by the width of the label peak. To address this issue, we propose a sigmoid label that encodes the sources' DOAs as the sum of angle-shifted monotonic sigmoid functions. Differentiating the output of a CNN-Transformer network trained with the sigmoid label produces sharp spectral peaks at the source DOAs. Combined with a moderately refined grid step of 0.25 degrees, the CNN-Transformer with the proposed sigmoid label attains sub-degree accuracy and a 1-degree resolution, breaking the limit of the grid quantization error. Under the unified CNN-Transformer backbone, we compare the proposed Sigmoid label with the Gaussian label, the 0-1 hard label, and the binary cross-entropy (BCE) baseline. Numerical results demonstrate that the Sigmoid label attains the highest accuracy and superior resolution compared to other labels.

Earnings call transcription presents a realistic continual learning scenario where Automatic Speech Recognition (ASR) systems are updated with new recordings but risk catastrophic forgetting of previously learned financial speech patterns. Adapting Whisper-small from a general-domain speech corpus to the Earnings22 benchmark reveals the opposite: all evaluated methods exhibit improvement on held-out financial speech from the same domain during adaptation rather than forgetting. This task complementarity makes regularization counterproductive. Nevertheless, when regularization is warranted, a layer-wise mini-Kronecker-Factored Approximate Curvature (mini-KFAC) approximation seems to achieve lower word error rates (WERs) and more stable optimization than standard diagonal Elastic Weight Consolidation (EWC). The results suggest that task complementarity should be assessed before applying continual learning constraints, and prefer mini-KFAC EWC over diagonal EWC when needed.

If seven AI advisors agree, what have we learned: that their forecast is reliable, or that the asset is easier to forecast? We examine 928 completed first-quarter outcomes for 232 assets across four historical iPulse AI batches. Ten Gemini configurations form seven equally weighted investment frameworks; no forecasts are regenerated. The council has mean absolute return error of 15.15 percentage points (pp), versus 14.99 for no change, with an unresolved paired difference of +0.16 pp. Error rises from 10.31 to 21.05 pp across disagreement quartiles, but the no-change baseline rises almost identically, from 9.89 to 21.15. In the pooled archive, selecting the lowest-volatility quarter produces lower error than selecting the least-disagreeing quarter: 7.97 versus 10.31 pp. Directional agreement introduces a second trap: its apparent value depends on the aggregation rule. When at least six of seven frameworks share a direction, the sign of their mean return is correct in 53.21% of cases, versus 44.11% with lower agreement. Scoring the modal direction instead changes the lower-agreement accuracy to 54.27%, removing that positive contrast. Recomputed-rank bootstraps, scale-aware losses, and period sensitivities qualify these descriptive results; four overlapping periods cannot establish future-regime performance. Agreement, visible asset risk, and aggregation-specific skill should be evaluated separately before consensus is presented as confidence.

Time series causal discovery — the process of detecting causation between time series — has its roots in signal processing research from over a half century ago. It has since received active and widespread interest, evolving into different forms across domains such as statistics, dynamical systems and machine learning. Entering this fragmented and constantly evolving research landscape can be daunting. This tutorial provides a unified overview of time series causal discovery across research communities. We introduce the key principles invoked by modern discovery methods, and we discuss their limitations and assumptions. We also discuss how discovery methods are evaluated, benchmarked and fine-tuned. We further discuss how causal discovery for time series is becoming crucial for downstream tasks in agriculture and Earth science, in healthcare, and in artificial intelligence systems. We believe that this tutorial will serve as a valuable resource in navigating the deluge of recent causality research, as well as an invitation to signal processing audiences to contribute to causal problems. To further help readers, we host a companion repository to link to useful code packages and resources at https://github.com/CHAI-UK/Awesome-Time-Series-Causal-Discovery.

For a finite group $G$ and a symmetric connection set $S$, we study the distance Laplacian ($D^L$) and distance signless Laplacian ($D^Q$) spectrum of the Cayley graph $\Cay(G,S)$ for $G$ a finite nilpotent group. Using the internal direct product decomposition over Sylow subgroups, we reduce every question to a single (possibly non-abelian) prime-power factor. In this setting, the required tools are furnished by ordinary representation theory rather than character theory. Based on this framework, we then derive a complete and explicit formula for every eigenvalue, with multiplicity, of the distance matrix, distance Laplacian, and distance signless Laplacian of $\Cay(G,S)$ for $G$ nilpotent: each eigenvalue is indexed by a tuple of irreducible characters of the Sylow factors. Finally, we use the explicit spectrum, together with Perron–Frobenius theory, to determine the eigenvalue-free interval of $D^L$ and $D^Q$.

Self-consistent numerical simulations of free boundary instabilities, such as resistive wall modes (RWMs) and external kink modes, require special attention given their relevance on the path towards magnetic fusion based nuclear reactors. In this work, a new implementation and a verification benchmark are presented for the nonlinear three-dimensional (3D) full magneto-hydrodynamics (MHD) code SpeCyl, which self-consistently combines a free flow 3D treatment for the edge plasma velocity with double resistive wall thin shell magnetic boundary conditions, capable of capturing the essential features of common setups in experimental devices. Two coaxial resistive shells surround the plasma. Vacuum conditions are assumed between and outside the resistive shells, whose relative positions and time constants can be varied on purpose. The implementation is extensively verified against Shafranov's analytic theory, valid for a circular cross section large aspect ratio tokamak and assuming a flat current profile. The correctness of newly implemented Boundary Conditions (BCs) is confirmed by the good agreement obtained for the linear growth rates of (m=2,n=1) RWM and external kink, over a wide range of wall positions, resistive shell time constants, and for different initial reference equilibria. The (m=2,n=1) RWM growth rates are also benchmarked against previous different RWMs modelling and successful validation studies for RFX-mod tokamak plasmas. The new BCs make SpeCyl a flexible and robust tool to investigate free boundary MHD modes in large aspect ratio magnetic confinement configurations, such as those achieved with the RFX-mod machine and expected in RFX-mod2.

We report a soft-aperture Kerr-lens mode-locked Tm:Y2O3 ceramic laser stabilized by suppressing wavelength competition under high population-inversion conditions. The laser delivers an average output power of 0.5 W with a pulse duration of 112 fs at a repetition rate of 90 MHz and a central wavelength of 2.08 um. Further optimization for shorter pulses yields a pulse duration of 107 fs, through with a reduced output power of 0.22 W. These results demonstrate that spectral-control is an effective approach for achieving stable mode-locking in such laser with strong spectral competition.

Goal-oriented communication often struggles to sustain downstream task performance under variable and unpredictable bandwidth. We address this problem by proposing GO-OAE (Goal-Oriented Ordered AutoEncoding), which learns a progressively refinable quantized latent sequence, optimized directly for task utility rather than source reconstruction. Each component of the sequence is represented for transmission by its code index, and the learned ordering enables inference from any received prefix of the index sequence, with performance improving as more indices arrive. We prove that optimizing the goal-oriented distortion objective is equivalent to solving a weighted conditional information bottleneck formulation, and introduce an unbiased multi-prefix importance sampling (MPIS) estimator, to reduce GO-OAE training computational cost from O(BK) to O(Bm) with m << K, and reduces the estimator variance by a factor of 1/m. Experiments show that, on MiniImageNet, GO-OAE achieves over 96% of the lossless upper-bound classification accuracy using 324 bits per image (683X compression vs. raw RGB), and already attains over 94% of the upper bound with only 70% of the transmitted indices. It consistently outperforms reconstruction-optimized codecs and ordered autoencoders trained with uniform prefix sampling across all transmission budgets, demonstrating effective communication at extreme compression rates.

Reduced order modeling provides an efficient alternative to high fidelity simulations for the rapid prediction of fluid flows in many query settings. Among non intrusive approaches, Operator Inference (OpInf) enables the identification of reduced dynamical operators directly from data while preserving a prescribed mathematical structure. The present work extends physics structured OpInf to the construction of parametric Reduced Order Models for transient incompressible flows governed by the Navier-Stokes equations. The velocity and pressure fields are represented using separate global reduced bases, and their reduced operators are identified directly from temporal and parametric POD coefficients. The identification procedure is formulated as a sequence of uncoupled least squares problems, leading to an efficient offline stage while avoiding the explicit projection of the governing equations. The approach is first evaluated on a lid driven cavity flow in a periodic regime to assess temporal extrapolation and prediction at previously unseen parameter values. Its applicability is then demonstrated for transient flow through a porous medium with a square obstacle, for which the nonlinear terms are not directly amenable to a standard Galerkin projection. The results demonstrate that the approach can accurately reproduce both velocity and pressure fields while providing computationally efficient predictions for unseen parameter values.

The open-loop minimum-energy control (MEC) problem asks for control inputs that minimize the control energy while ensuring the terminal state reaches a desired value. In this paper, we study the open-loop MEC problem for stochastic linear time-invariant systems. We consider a variety of stochastic constraints that enforce the terminal error to be small, and show that Stochastic MEC (SMEC) problem can be formulated as a QCQP. We obtain the optimal solution for SMEC and show that it generalizes the well-known Deterministic MEC problem by introducing novel stochastic variants of controllability and Gramian matrices. Finally, we show that there exists a fundamental lower bound on the optimal control energy which cannot be overcome by increasing the time horizon of the problem.

We produce the first examples of singular isoperimetric boundaries with non-zero constant mean curvature in Riemannian manifolds. The main technical ingredient is establishing that, for constant mean curvature hypersurfaces containing isolated singularities modelled on strictly stable, strictly minimising regular tangent cones, weak strict stability implies local unique volume-constrained minimisation.

Chronic ischemic stroke lesions are difficult to segment on T1-weighted MRI. Their appearance overlaps with healthy tissue, and their volumes span four orders of magnitude, so voxel-level and lesion level metrics often disagree. We build an eleven-member nnU-Net ensemble that covers five design axes: a baseline dice + top-K loss, heavy augmentation, per-lesion volume loss re-weighting, stochastic weight averaging, and a sampling-and-architecture block that pairs cohort-balanced sampling with a residual-encoder backbone. Weighted soft-map fusion of the members feeds a tuned decision layer, SADL, that estimates lesion volume, applies an adaptive per-lesion-size threshold, prunes components by voxel count and peak probability, and adds a never-empty rescue. All fusion weights and per-bucket parameters are fitted on out-of-fold (OOF) predictions, so tuning never touches the held-out test set. On a site-disjoint 91-case holdout, SADL reaches Dice 0.7322, lesion-F1 0.7461 and PR-AUC 0.8430 when scored online with the challenge's released evaluation code. A paired non-parametric bootstrap against a reference single ensemble member confirms the lesion-F1 gain at p=0.024, and the gain is largest in the small-lesion regime, where the rescue step recovers cases that a fixed threshold would silently discard.

Concurrency bugs manifest only under specific thread interleavings, making them hard to detect and reproduce. Existing controlled concurrency testing techniques take over thread scheduling to explore the interleaving space. As the space grows rapidly with program scale, finding interleavings that can trigger bugs within a limited budget remains challenging. The program's concurrency semantics can reveal which variable accesses and synchronization operations under specific interleavings lead to unexpected program behavior, and thus help focus the exploration on bug-relevant interleavings. However, prior techniques have not fully exploited such semantic information. In this paper, we present REINS, an agentic approach for detecting and reproducing concurrency bugs. At its core, REINS guides thread scheduling with the agent's reasoning about the concurrency semantics and known concurrency bug patterns. REINS starts with runtime observations and combines them with static information to reason about the program's concurrency behavior and formulate hypotheses about potential bugs. A hypothesis translates into a set of constraints on event orders and variable states, guiding a controlled scheduler toward interleavings likely to trigger the bug. REINS draws conclusions from the execution results and keeps refining the constraints until the bug is triggered or the hypothesis is refuted. Compared with existing techniques on three public benchmarks, REINS detects more bugs within the same time budget. Furthermore, REINS found six previously unknown concurrency bugs in real-world projects, three of which have been confirmed by the developers.

Micro- and nanostructures fabricated by focused ion beam (FIB) milling invariably deviate from input designs due to systematic and stochastic effects such as ablation-rate fluctuations, and redeposition. As such, nominally identical iterations of a given process yield a distribution of non-identical fabrication outcomes, leading to variations in the physical (optical, mechanical) properties of fabricated structures. Existing models of the milling process typically predict a single deterministic fabrication result, neglecting this variability. Here we present a conditional diffusion model that learns the statistical distribution of process outcomes conditioned on input design, enabling it to predict a realistic distribution of possible output sample structures. These predictions accurately reproduce edge morphology, linewidth variations, and defect probabilities observed in fabricated samples, enabling statistical analysis of milling performance and the identification of dominant structural and process parameter-dependent deformation modes. The work establishes a probabilistic modelling framework for fabrication variability analysis and fabrication-aware design optimization.

Developmental dyslexia (DLX) affects 5-12% of the global population and impairs learning to read. Its biological basis, particularly the difficulties observed in oral phonological tasks, remains debated. Entrainment, the coupling of brain oscillations to speech rhythms, is essential for encoding phonological representations. We investigated whether brain entrainment to non-speech rhythms could predict DLX in 48 participants listening to white noise amplitude-modulated at rates matching Spanish speech units. Phase-amplitude coupling (PAC) analysis of electroencephalography (EEG) signals yielded accurate DLX prediction for the 40 Hz stimulus using the Heights Ratio PAC measure (balanced accuracy 0.929, p = 0.0069, in the best case), indicating a stronger phonemic entrainment deficit in dyslexic individuals. Regression analysis in control (CTL) participants showed a strong dependence between reading performance scores and greater entrainment in language-associated brain regions. In particular, participants with poorer phonemic awareness, reading comprehension, or rapid automatized naming showed greater right-hemisphere entrainment to the 40 Hz stimulus. This finding adds evidence for the hemispheric compensatory mechanisms hypothesized in the literature. In conclusion, our methodology identifies dyslexia-specific entrainment patterns near language-associated brain regions, together with compensatory entrainment in right-hemisphere areas.

Efficient simulations of grain growth are essential for investigating the process-structure-property relationships in polycrystalline materials. We introduce a two-dimensional von Neumann-Mullins-based graph grain growth simulation (VN-GraphGG), which couples the grain-scale kinetics with microstructure reconstruction. The polycrystalline microstructure is represented as a graph with grain and junction nodes. The grain-area evolution on the grain graph is governed by the von Neumann–Mullins (VN) law, while a graph neural network is trained to reconstruct the microstructure from the graph. This separation allows kinetic coefficients to be varied without retraining the neural reconstruction model. Topological changes during grain growth such as disappearance of grains and changes in neighbor relations are handled straightforwardly through reconstruction and postprocessing. Simulations demonstrate accurate predictions of VN kinetics and coarsening statistics, and significant speedup over the phase-field simulations. The proposed method provides a foundation for efficient grain growth simulation based on grain-scale kinetics, opening the way to incorporate a broader range of physical mechanisms. An application example is presented for abnormal grain growth, in which kinetic heterogeneity significantly influences the grain microstructure evolution.

Josephson junctions are the central nonlinear element of superconducting qubits, yet their standard description is reached through a chain of approximations. We propose a purely quantum-mechanical model of the junction, formulated directly in Hilbert space: a charged particle moves on a one-dimensional loop, and the junction is a point at which the wave function obeys self-adjoint boundary conditions, labelled by a coupling strength $\lambda$, which controls the transparency of the junction, and a phase jump $\delta$, which plays the role of the superconducting phase difference. For a loop with a single junction we solve the spectral problem and show that the ground-state momentum expectation, identified with the supercurrent, factorises exactly as $\langle\hat{p}_0\rangle = I_c(\delta,\lambda)\sin\delta$. For opaque junctions $I_c$ becomes independent of $\delta$ and the first Josephson relation $I=I_c\sin\delta$ is recovered; for strongly attractive coupling, where the ground state is bound to the junction, the $\sin\delta$ law is also recovered, but with an exponentially suppressed amplitude. For more transparent junctions, however, the model predicts deviations from the pure $\sin\delta$ law: $I_c$ acquires a dependence on $\delta$ already at order $\lambda^{-2}$, compatible with the higher Josephson harmonics recently found necessary to describe superconducting qubit spectra, and a fully transparent contact yields a current linear in $\delta$. When the phase varies in time, a gauge transformation shows that the voltage drop across the junction is proportional to $\dot\delta$, so that the second Josephson relation emerges without being imposed. The boundary-condition approach thus reproduces both Josephson relations from first principles and predicts corrections to them, offering a route towards higher-fidelity models of superconducting qubits.

Large language model (LLM) agents are becoming direct users of the open web. As they take on increasingly complex and information-intensive tasks, effective web access is becoming a core capability for agentic problem solving. Existing approaches have advanced agentic web access through different access models, including search-augmented agents, browser-based agents, and, more recently, filesystem-based agents. Nevertheless, as users expect agents to complete long-horizon, end-to-end tasks, they increasingly need to conduct broad information gathering, perform browser actions, and revisit acquired evidence within a single trajectory. Yet our preliminary analyses reveal that existing access models struggle to meet these demands. They primarily support either information gathering or browser interaction, while combining their capabilities can involve substantial costs or performance trade-offs. To this end, we present Web\$hell, an interface that unifies web access under a single access model. Using a common schema defined by WebContract, WebCompiler compiles live webpages into addressable information and action objects. These objects are maintained in WebCache and handled through a shared set of operations, providing agents with consistent, operable, and persistent web access. Through extensive experiments across web search, web navigation, and deep research benchmarks, we show that Web\$hell achieves strong task performance at lower cost. Our results further indicate the importance of interface design for effective web access, with Web\$hell opening new possibilities for more capable and efficient LLM agents on the web.

Author Name Disambiguation systems are commonly evaluated using pairwise and B$^3$ metrics, which summarize clustering quality through aggregate precision and recall but do not explicitly distinguish between lumping (merging publications from different authors) and splitting (fragmenting the publications of a single author). Because these errors can have asymmetric consequences in bibliographic applications, we propose a precision-driven evaluation framework that decomposes clustering quality into complementary instance-level and cluster-level signals. The framework introduces cluster-level Lumping Error and Splitting Error to separately quantify contamination and fragmentation, combines them with standard pairwise and B$^3$ metrics through a Composite Score, and uses the resulting measures for model selection. We characterize the diagnostic behavior of the framework through a controlled synthetic case study and assess the sensitivity of the Composite Score to its weighting scheme. The results show that the proposed structural measures distinguish error patterns that aggregate metrics can mask, while the sensitivity analysis identifies a stable region for the adopted weighting configuration. The framework is released as an open-source Python pipeline.

We establish a vector-valued extension of a theorem of Marciszewski. Let $K$ be a compact Hausdorff space, let $\Gamma$ be a set, and let $X$ be a nonzero Banach space containing no isomorphic copy of $c_0$. We prove that if $C(K)$ embeds isomorphically into $c_0(\Gamma,X)$, then $C(K)$ is isomorphic to $c_0(\Lambda)$ for some set $\Lambda$. Moreover, when $\Gamma$ is infinite, the optimal distortion of embeddings of $C(K)$ into $c_0(\Gamma,X)$ equals that of embeddings into $c_0(\Gamma)$. The classification result answers a question posed in an earlier paper.

The Filoche–Mayboroda (FM) localization landscape predicts where the spatially localized modes of a self-adjoint elliptic operator reside without diagonalizing it, but the construction fails for the complex, non-normal Green's matrix G of an open scattering array of electric dipoles. We propose three real-valued, positive landscapes built from the Feinberg–Zee hermitization of G: a regularised Gramian inverse (Gram), the diagonal Schur block that carries the Metz–Neri–Rogers spectral density (SY), and the full hermitized inverse contracted against a polarization-isotropic source pair (Herm). The three form a progression from spectrum-blind to fully hermitisation-faithful, agree in the limit $z\to 0$, and diverge where it matters: at nonzero spectral shift SY resolves modes that Gram collapses, and Herm resolves modes that SY collapses. We demonstrate the progression on random and deterministic aperiodic scattering arrays with a wide range of structural correlations, extend to the vectorial case of transverse electric polarization (TE), and provide rigorous justifications along with mathematical proofs and a complete numerical recipe in the appendices.