High Energy Physics (Phenomenology)
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Saturday 12 September
Unraveling the partonic origin of the proton spin requires precise determinations of polarized parton distribution functions (PDFs), which depend on comparably precise theoretical predictions for polarized scattering, particularly in view of the high-precision measurements anticipated at the future Electron-Ion Collider. We present the first next-to-next-to-next-to-leading order (N$^3$LO) QCD predictions for longitudinally polarized semi-inclusive deep-inelastic scattering (SIDIS) in a fully differential form, together with next-to-next-to-leading order predictions for the hadron transverse-momentum spectrum. These results are obtained by extending the two-dimensional transverse-momentum subtraction framework to the spin-dependent cross section. Together with the corresponding unpolarized calculation, these results enable consistent N$^3$LO predictions for longitudinal double-spin asymmetries and provide a precision baseline for future analyses of helicity PDFs and transverse-momentum-dependent helicity distributions.
Computer Vision
Hyper-LLaVA: Hyperbolic Uncertainty-aware Modality-Balanced Routing for Multimodal Continual Instruction Tuning
Multimodal Continual Instruction Tuning (MCIT) aims to exploit the incrementally accumulated knowledge to process multimodal inputs of diverse tasks, where parameter routing plays an important role. State-of-the-art methods rely on sample-to-task center similarity and cross-modal fusion with equal weight during routing. However, such solutions face two fundamental flaws: (1) Within each modality, the sample-to-task center distance is sub-optimal for routing since the abundant intra-task diversity information is underleveraged. (2) Different modalities exhibit varying reliability across tasks, where the modality with inter-task ambiguity can easily misguide the routing result. To address these problems, we propose Hyperbolic Uncertainty-aware Modality-Balanced Routing (Hyper-LLaVA) to improve parameter routing capacity based on cross-modality task feature uncertainty modeling. Specifically, to improve intra-modality task matching, Hyper-LLaVA accesses the sample-to-task distribution similarity in the Hyperbolic space. Besides, to alleviate the degradation brought by unreliable modalities, Hyper-LLaVA quantifies the task matching ambiguity within each modality to achieve adaptive balancing between task matching across modalities. Based on the complementary intra- and inter-modality task matching enhancement, our Hyper-LLaVA outperforms state-of-the-art approaches by large margins. Our source code is available at https://github.com/zhoujiahuan1991/ICML2026-Hyper-LLaVA
Read the full abstract
Commutative Algebra
The structure of automorphism groups of zero-dimensional monomial algebras
Let $A$ be a zero-dimensional monomial algebra over an algebraically closed field of characteristic zero, that is, a finite-dimensional quotient of a polynomial ring by a monomial ideal. Its automorphism group $G$ is a linear algebraic group, described through the homogeneous nilpotent derivations of $A$. We analyze the structure of $G$ in detail. Its identity component $G^0$ is a semidirect product of its unipotent radical and a reductive subgroup isomorphic to a product of general linear groups, and for each root degree we characterize when the associated derivations give rise to an additive root subgroup, and determine its dimension. Using the Lie brackets of these derivations, we then give an explicit algorithm that produces, out of the minimal monomial generators of the ideal, a family of root subgroups generating $G^0$ together with a maximal torus. Such a family is minimal in the generic case. We also show that the component group $G/G^0$ can be arbitrary: every finite group arises as the component group of the automorphism group of some zero-dimensional monomial algebra. Finally, we apply these results to the algebras $\mathbf{k}[\mathbf{x}]/\mathfrak{m}^d$, showing that the subgroup generated by a maximal torus and the outer root subgroups is exactly the subgroup of automorphisms with constant Jacobian determinant, and we deduce from this a new proof of Anick's theorem on the density of the tame automorphisms of $\mathbf{k}[\mathbf{x}]$.
Read the full abstract
Statistical Methodology
Optimal Cohort Staircase Designs
Staircase cluster randomized trials restrict outcome collection to a limited window around each cluster's transition from control to intervention. This can reduce the burden of repeated measurement, but the allocation of clusters and participants across treatment sequences requires careful consideration. We develop an optimal design framework for closed-cohort staircase trials with continuous outcomes, allowing cluster sizes to differ across sequences. Under a linear mixed model, we express treatment-effect precision through sequence-specific information contributions and establish convexity of the variance criterion in the cluster allocation proportions. An equivalence theorem characterizes optimal allocations, and reversal symmetry yields analytical solutions for several three- and four-sequence designs. We also consider joint optimization of sequence allocation and cluster sizes under a fixed participant total. Numerical investigations show that the value of unequal cluster sizes depends on the observation window and correlation structure: gains are often modest but can be substantial in selected settings. Comparisons with complete stepped-wedge designs demonstrate a trade-off between repeated measurements and additional clusters, with staircase designs achieving lower costs in some settings at comparable precision. Illustrations motivated by psychosocial cancer care and trial recruitment distinguish hypothetical design adaptations from the original studies. The results support choosing sequence allocation, cluster sizes, and observation windows jointly when repeated measurement is a major constraint.
Read the full abstract
Machine Learning
HarnessBandit: Joint Learnability-Transferability Scheduling for Multi-Harness Agentic Reinforcement Learning
Language-model agents are increasingly deployed through diverse harnesses that differ in system prompts, tool schemas, control loops, and trajectory formats. The same model can perform unevenly across these interfaces, making robustness to harness variation an important objective. A natural approach is to train a shared policy through multiple harnesses, but doing so introduces a scheduling problem: each training step should favor a harness that currently provides a useful learning signal while also producing an update that benefits the other harnesses. We develop HarnessBandit, an online scheduler that selects one harness per optimizer step. After a group-relative policy optimization (GRPO) update, it observes learnability – the mean absolute advantage on the batch – and transferability – the cosine between a low-dimensional gradient sketch of the current harness and exponential moving averages of the remaining harnesses. The two signals are fused after pooled sliding-window min-max normalization and sampled with a visit-dependent bonus and an explicit exploration floor. We train Qwen3.5-2B across six harnesses on ClawGym and evaluate on PinchBench (held-out tasks, in-distribution OpenClaw) and ClawEval (held-out tasks and harness). HarnessBandit improves over mixed-batch multi-harness training on both benchmarks, while training diagnostics indicate that learnability and transferability provide distinct, evolving signals.
Read the full abstract
Analysis of PDEs
Gradient estimates and Liouville-type theorems for the semilinear elliptic equation involving the nonlinear gradient source
We study local and global properties of positive solutions to the equation $-\Delta u=u^p+M|\nabla u|^q$ in a domain $\Omega$ of $\mathbb R^N$, where $p,q$ are parameters and $M>0$. By constructing a linear operator, we establish the differential inequality containing an auxiliary function. By selecting appropriate auxiliary functions over various regions and employing the maximum principle, we derive the local gradient estimates for all $(p,q)\in \mathbb R^2$, and further establish Liouville-type theorems. As an application, we acquire universal estimates for local solutions of elliptic equations with general nonlinearities. Our results extend partial conclusions established in Bidaut-Véron, Garcia-Huidobro and Véron [Math. Ann. 378 (1-2) (2020) 13-56].
Read the full abstract
Computation and Language
ForeSight: Enhancing Risk Monitoring via Early Safety Signal Distillation
As large language models (LLMs) are increasingly deployed, the generation of harmful content has become a critical safety concern. Existing safeguards operate at the input, output, or streaming-generation stages, while early-risk methods that rely on surface tokens or output logits may suffer from weak initial signals, and internals-based detectors using dense representations may retain highly entangled and redundant safety-irrelevant information. It therefore remains unclear whether the earliest post-generation hidden states already contain reliable signals about final-response harmfulness. To address this gap, we propose ForeSight, a first-token output-risk forecasting framework that distills weak and redundant early safety signals into compact, layer-aware risk representations. Experiments on five safety benchmarks and two target models demonstrate that ForeSight achieves superior and efficient early-risk forecasting while relying solely on first-token hidden states. The code is available at: https://github.com/Scabbards1500/Foresight
Read the full abstract
Statistical Methodology
Adjusting for Many Covariates in Randomized Clinical Trials with GLMs: Bias Reduction by Jackknife and Practical Guidance
Adjusting for baseline covariates has become standard practice in analyzing randomized clinical trials. In the low-dimensional setting, it is well understood that covariate adjustment through a parametric working model can sometimes be more efficient than the unadjusted difference-in-mean estimator. However, when the number of adjusted covariates is large relative to the sample size $n$, a naïve adjustment may introduce excessive bias, leading to invalid statistical inference. The current literature that tries to resolve this issue is either limited to linear working models or relies on sample splitting, which may raise concerns about the replicability of RCT analyses. In this paper, we devise a novel jackknife-based approach to covariate adjustment through generalized linear models (GLMs), which we term as JAckknife Score-based Adjustment (JASA), together with its calibrated version JASACal. By employing a nuanced jackknife strategy, JASA and JASACal avoid sample splitting and make full use of the data, while ensuring that the bias of JASA or JASACal is still negligible even when the number of adjusted covariates is large compared to $n$. JASA also encompasses state-of-the-art adjusted estimators through linear working models as a special case. Through extensive simulation experiments and a real data analysis, we demonstrate that JASA or JASACal can adjust for a much greater number of covariates than existing benchmarks. These empirical results also shed some new light on practical guidance for covariate adjustment with GLMs. Both JASA and JASACal have been incorporated into our R package HOIFCar available from CRAN. The package HOIFCar is developed to serve as a user-friendly option for covariate adjustment in RCTs, in particular when practitioners hope to adjust for a large number of covariates.
Read the full abstract
Number Theory
Finiteness of an Isogeny Class via Equidistribution
In a very recent, beautiful paper, Junyi Xie and Ziquan Yang gave a new proof of Faltings' isogeny theorem over number fields by reducing to Tate's theorem over finite fields. We show that the same ideas can be adapted to prove finiteness of an isogeny class over number fields from the finiteness of an isogeny class over finite fields. Hence we obtain a new proof of the Shafarevich conjecture over number fields. The proof in this paper was obtained through the use of GPT6 Astra Ultra.
Read the full abstract
Computation and Language
PolicyMem: Geometric Policy Memory for LLM Governance
Effective LLM governance extends beyond detecting unsafe responses to guiding their correction through closed-loop assessment, revision, and verification. Learning-based guards provide strong semantic discrimination but often leave policy knowledge implicit in trained parameters, limiting its explicit reuse across stages. Programmable guardrails make policies explicit through prompts and workflows, but rely heavily on manual engineering to operationalize them. This motivates explicit policy representations that are both learnable and reusable across governance operations. To this end, we introduce PolicyMem, a geometric policy memory that compiles natural-language policies into learnable low-rank subspaces for reuse across governance stages. A policy assessor projects query-response representations onto these subspaces, producing evidence that directly mediates safety judgments and policy attribution. A response rewriter uses the implicated policies to revise unsafe responses, with the same assessor re-evaluating the revised responses. Beyond inference, the same memory constructs training data from model-generated candidates to supervise the rewriter without external teacher calls. Across five widely used benchmarks, PolicyMem achieves the best average unsafe-behavior detection performance among representative baselines while supporting effective policy attribution, rewriting, and post-intervention assessment through the shared policy memory.
Read the full abstract
Computer Vision
FFVO: A Feedforward Pose Decoder for Long-Horizon Visual Odometry
Stable and reliable 4D spatial understanding is fundamental for autonomous driving systems. While feedforward reconstruction networks can estimate camera motion and 3D structure in one pass, pose estimation over long videos remains challenged by computational cost, long-context ambiguity, and temporal instability. To address these challenges, we propose Feedforward Visual Odometry (FFVO), a pose-specialized adaptation of joint reconstruction architectures for efficient and temporally stable camera-pose estimation. FFVO uses (i) a compact camera-token representation for computationally efficient temporal aggregation, (ii) a hierarchical local-to-global temporal decoder that mitigates geometric ambiguity by separating short-range motion aggregation from sequence-level integration, and (iii) intermediate trajectory supervision that promotes temporal stability. Extensive evaluation on the Waymo Open Dataset (WOD), KITTI, and a large-scale proprietary benchmark demonstrates that our method performs favorably against existing feedforward approaches, and greatly reduces jitter and drift. These results support FFVO as an effective feedforward camera-pose decoder in long-horizon visual odometry settings.
Read the full abstract
Functional Analysis
Relevant sampling of non-decaying signals in Orlicz and mixed-norm Orlicz spaces defined on a locally compact group
We study the sampling problem for non-decaying signals defined on a locally compact group. The signals are modeled as elements of suitable subspaces of weighted Orlicz and weighted mixed-norm Orlicz spaces, thereby allowing controlled growth at infinity. Our principal focus is on image spaces of idempotent integral operators. Under suitable oscillation estimates on the associated integral kernels, we establish deterministic sampling theorems that guarantee stable reconstruction from pointwise samples. We further prove an average sampling result in the same setting. In addition, we prove random sampling theorems for functions whose norm is essentially concentrated on a compact subset of the group. In particular, we show that, with high probability, sampling is possible from $\mathcal{O}\!\left(\mu(K)\log \mu(K)\right)$ randomly chosen sample points, where $K$ denotes the compact set on which the signal, belonging to Orlicz space, is essentially norm concentrated. In the same spirit, a random sampling result is proved for signals in mixed-norm Orlicz space. As applications of the abstract theory, we obtain sampling theorems for weighted shift-invariant spaces and Orlicz modulation spaces.
Read the full abstract
Artificial Intelligence
Trustworthy Agentic AI: A Comprehensive Cybersecurity and Systems Survey on Threat Landscapes, Defense Architectures, and Open Challenges
The transition from passive foundation models to autonomous, goal-directed agentic AI systems has introduced unprecedented capabilities by coupling recursive cognitive reasoning loops, persistent memory architectures, live tool execution planes, and multi-agent collaboration topologies. However, granting probabilistic neural cores execution authority across filesystems, networks, and cloud infrastructure dissolves classical security perimeters: natural language simultaneously serves as input data, internal control code, and communication protocols, exposing a Turing-complete blast radius where untrusted data represents executable instructions. This survey delivers a comprehensive systems-security reference framework for trustworthy agentic AI, synthesizing 206 foundational studies and regulatory standards. We formalize the general agent architecture as a stateful 5-tuple and establish a 6-dimensional trustworthiness taxonomy covering security, safety, privacy, explainability, fairness, and accountability. We systematically analyze threat surfaces across intra-execution loops and interaction planes, formulate a multi-layered zero-trust defense-in-depth architecture integrating Dual-LLM isolation, Capability-Based Access Control, kernel eBPF probes, and sandboxed runtimes, review standardized evaluation benchmarks, and map technical controls to international AI governance frameworks.
Read the full abstract
Machine Learning
JumpStart Your Policy Learning with Lessons from 160,000 Training Runs
Reliable progress in offline policy learning depends on careful reporting, well-tuned baselines, and evaluation across diverse conditions. Prior work has shown that results can be sensitive to reporting choices, hyperparameter tuning, and dataset properties, but these sources of variability have not been systematically investigated together at the scale needed to understand how they shape conclusions. To address this gap, we present a large-scale empirical study of offline reinforcement and imitation learning, training over 160,000 policies across 114 datasets. At this scale, no algorithm dominates: aggregate performance among the strongest methods is often close, but the leaders differ substantially across environments. We find that proper hyperparameter tuning frequently reshuffles perceived algorithm rankings and that benchmark composition can produce conflicting conclusions. We also study hyperparameter sensitivity and transfer across environments, identifying a simple strategy for deriving strong default configurations. We use our findings to develop a dataset-conditioned recommender that provides task-specific algorithm recommendations for practitioners. Finally, we release JumpStart: a resource suite containing every trained policy, per-model scores and hyperparameters, strong baselines across all environments, training and evaluation code, and an extensible website for retrieving, analyzing, and contributing results. Together, these resources aim to make offline policy-learning research more reliable and enable future work beyond the scope of this study.
Read the full abstract
Machine Learning
A Variational Optimal Transport Operator on Incompressible Flow
We present the Variational Incompressible Optimal Transport (VIOT) operator, a generative neural operator for amortized incompressible density transport. Given a new source-target density pair, VIOT predicts a divergence-free velocity field and generates the full transport trajectory by feed-forward inference, replacing the hour-scale per-pair optimization used by adjoint fluid solvers and differentiable simulation baselines. The system consists of three components: a stream-function or vector-potential representation that enforces incompressibility by construction, a regularized incompressible transport objective that balances endpoint accuracy and flow smoothness, and a Fourier Neural Operator backbone that amortizes the solve across new pairs and grid resolutions. Together, these components make incompressible transport a reusable neural operator that facilitates various transport processes. Further, the generative capability extends beyond the training distribution, with VIOT producing incompressible transports for user-drawn source-target pairs in a real-time interactive system. We demonstrate VIOT on 2D and 3D density-transport benchmarks. Both 2D and 3D rollouts complete in seconds per pair, while per-instance baselines in our 2D comparisons optimize each new pair from scratch and require on the order of an hour, a roughly $10^4\times$ online speedup.
Read the full abstract
Software Engineering
TyPatch: Transforming Patches into Typestate Rules for Kernel Bug Detection
Historical Linux kernel patches capture defect knowledge that applies beyond their original repair sites. Recent work has shown that large language models (LLMs) can generate static-analysis checkers from historical patches and use them to uncover new kernel bugs. However, complete-checker generation requires the model both to recover the defect semantics expressed by a patch and to implement sophisticated program-analysis machinery, including object tracking, alias analysis, path-state maintenance, and interprocedural propagation. Coupling these responsibilities in a single end-to-end code-generation task can turn a simple defect rule into an unstable and expensive analyzer-implementation problem. To address this problem, we present TyPatch, which decouples patch-specific defect semantics from analyzer implementation. An LLM translates each patch into a typestate rule specifying its tracked object, actions, guards, transitions, and violations. A shared backend then executes these rules, binding their actions to program events, tracking object identity across aliases, propagating typestate along program paths, and producing reports for all rules. On Linux v6.16, TyPatch finds 559 distinct bugs, 121 of which have been confirmed by kernel developers. In a matched 38-patch comparison with the state-of-the-art complete-checker construction workflow, TyPatch uses 88.3-90.1% fewer generation tokens, while its initial report pools achieve 3.42-14.95$\times$ the precision of those produced by that workflow.
Read the full abstract
Human-Computer Interaction
"What Can I Do for You”: How Should AI Companions Provide Assistance to Players in Virtual Reality Games
Recent advances in artificial intelligence (AI) have expanded the capabilities of non-player characters (NPCs), enabling them to perceive game states, perform in-game actions. In immersive virtual reality (VR) games, such assistance is not limited to providing hints or interface-level support: an AI companion can appear as a co-present character, share the player's spatial environment, and visibly act on game objects. This raises a design question for VR gameplay: how can AI companions best assist players while preserving their active participation in the virtual world? To explore this question, we developed a VR puzzle game for Apple Vision Pro featuring an AI companion across four gameplay modes: no assistance, command-based assistance, discussion-based assistance, and autonomous agent play. A within-subjects study with 24 participants showed that AI assistance significantly reduced players' workload. However, autonomous agent play, despite producing the lowest workload, substantially diminished player experience by reducing challenge, autonomy, immersion, and enjoyment. Qualitative analysis further showed that players evaluated the companion not only by its usefulness, but also by whether it felt like a tool, a teammate, or an integrated character in the game world. We categorised participants into four player types and summarised their expectations of AI companions. These findings provide design implications for AI companions as embodied participants in VR games.
Read the full abstract
Operator Algebras
Some entropic inequalities for primitive KMS symmetric quantum Markov semigroups
In this note, we prove that every primitive KMS-symmetric quantum Markov semigroup on a finite-dimensional matrix algebra satisfies a modified logarithmic Sobolev inequality (MLSI). We also construct a primitive quantum Markov semigroup without KMS symmetry that fails MLSI, and a primitive KMS-symmetric quantum Markov semigroup that fails complete MLSI (CMLSI). The latter construction also provides a non-primitive KMS-symmetric quantum Markov semigroup without MLSI. For the graph-based KMS-symmetric quantum Markov semigroups studied here, we prove MLSI and CMLSI when the underlying graph is connected and has at least three vertices. Finally, we establish CMLSI for a class of primitive bimodule KMS-symmetric quantum Markov semigroups arising from fermionic systems.
Read the full abstract
Artificial Intelligence
IBBench-Light: A Paired Evaluation of Task-Conditioned Responses to External Directives
An external record may contain a procedure to apply or text to read, depending on the user's request. IBBench-Light tests both uses against the same record. Twelve semantic bases yield 144 matched pairs per model; four quantized instruction models produced 1,152 archived greedy responses. Paired exact-contract accuracy (PECA) requires both members to satisfy their output contracts. Qwen succeeds on 132 execute and 109 process prompts, but only 97 complete pairs, showing what marginal averages omit. We audit literal-target exposure and case normalization, then add 1,722 logged CPU generations to test directive-absent controls, twelve additional semantic bases, within-base wording changes, and generation stopping. In the pinned Phi rerun, changing the end-of-sequence (EOS) set changes exact paired success from 0/144 to 62/144. A bounded IHEval comparison uses the same SmolLM2 checkpoint and output budget while preserving its published instruction roles and scorer. The benchmark measures conditional task and output-contract success. Its task margins and paired count need to be read together with the stopping policy.
Read the full abstract
Soft Condensed Matter
Lagrangian approach to origami vertex analysis: Multistability
Studying the multistability of origami structures presents challenges due to the nonlinearity of their kinematics and the high-dimensional configuration spaces that are difficult to visualize and explore exhaustively. To address this, we utilize the Lagrangian framework for origami to exploit symmetries and obtain reduced-dimensional slices of the configuration space. Our analysis of degree-6 vertices with reflection symmetry reveals topological transitions in their kinematic space as sector angles are varied, with implications for the number of symmetry-constrained minima and the emergence of metastable regions. These lower-dimensional slices are amenable to exhaustive search and visualization. A subsequent full-space stability analysis shows that 18 of 41 degree-6 and 14 of 45 degree-8 symmetry-constrained minima remain minima when all locally compatible perturbations, including those that break symmetry, are admitted. The metastable regions, which would likely be overlooked by numerical optimization alone, are influenced by the interplay between the boundaries of admissible kinematic space and crease mechanical properties. We extend our analysis to cone-like vertices with higher symmetry and one-degree-of-freedom kinematics, exploring symmetry-breaking phenomena, combinatorial structure, and their consequences for branchwise stability. The stability landscapes uncovered have potential applications in mechanical metamaterials, mechanical computing, origami-based robotics, and structures designed to self-deploy and retain their shape.
Read the full abstract
Strongly Correlated Electrons
Layer- and Orbital-Selective Mott Physics Driving Dimensional Crossover in Nickelate Superlattices
The design of quantum materials to manipulate dimensionality and electronic correlations is a central focus in the exploration of emergent phenomena. We propose a new correlated quantum material platform for realizing a layer- and orbital-selective Mott transition (LOSMT) in a nickelate heterostructure by combining bilayer and monolayer LaNiO$_3$ separated by a monolayer LaAlO$_3$ spacer under compressive strain from the LaAlO$_3$ substrate. The LOSMT emerges from the layer-dependent dimensional confinement of the apical Ni($d$)-$z^2$ orbital and strain-induced lifting of the orbital degeneracy. The monolayer LaNiO$_3$ exhibits an orbital-selective Mott insulating phase with a Mott gap in the Ni($d$)-$z^2$ orbital, whereas the bilayer LaNiO$_3$ exhibits a multi-orbital metallic phase, triggering the orbital-selective Mott transition upon cooling $via$ inter-layer hybridization of the Ni($d$)-$z^2$ orbitals. Calculations within the density functional theory plus dynamical mean-field theory (DFT+DMFT) framework reveal that the LOSMT drives a dimensional crossover by enabling out-of-plane electronic transport through the metallization of the localized apical Ni($d$)-$z^2$ orbital, while the planar Ni($d$)-$x^2-y^2$ orbital remains itinerant across the transition. The emergence of the LOSMT provides a new pathway for engineering correlated quantum materials toward unconventional superconductivity and Mott-transistor functionalities.
Read the full abstract
Analysis of PDEs
Gradient self-improvement for mixed local and nonlocal parabolic equations
We introduce a new method for studying gradient higher integrability for mixed local and nonlocal parabolic equations. More precisely, for $p>2d/(d+2)$ and $s \in (0,1)$, we prove that if the inhomogeneity $F \in L^{p(1+\sigma)}_{\mathrm{loc}}$ for some $\sigma>0$, then every weak solution satisfies $\nabla u \in L^{p(1+\eps)}_{\mathrm{loc}}$ for some $\eps>0$, together with quantitative local estimates. The proof relies on stopping time arguments and an intrinsic Calderón-Zygmund-type covering decomposition in which the inhomogeneity and nonlocal energy contributions are treated by classical and fractional maximal estimates.
Read the full abstract
Computation and Language
Scaling Hindi Quantum Natural Language Processing through Automatic Pregroup Supertagging
Quantum Natural Language Processing (QNLP) uses pregroup grammars to translate grammatical structure into diagrammatic representations and quantum circuits. Recent Hindi QNLP work has shown that Hindi-specific pregroup grammars can support grammar-sensitive compositional models, but grammatical type assignment is still largely manual, limiting scalability. This paper formulates automatic Hindi pregroup supertagging as a token-level classification task. Using a manually annotated corpus of 380 Hindi sentences, we evaluate lexical, contextual, prompting-based, lexical-repair, and suffix/morphology-aware methods. Results show that simple lexical and contextual models are strong in this low-resource setting: contextual backoff achieves the best completed accuracy of 64.56%, while raw Qwen2.5 prompting reaches only 11.65%. Lexical repair raises LLM-assisted prediction to 64.08%, demonstrating the value of constraining generative outputs with symbolic grammar knowledge. Diagnostic analysis further shows that seen and unambiguous tokens are much easier than unseen tokens, and suffix/morphology features improve karaka-token accuracy but not overall performance. These results show that automatic Hindi pregroup assignment is feasible and can reduce reliance on manual annotation in future multilingual QNLP pipelines.
Read the full abstract
General Relativity and Quantum Cosmology
Thermodynamic Representations of Algebraically Modified FLRW Dynamics at the Apparent Horizon
We investigate the thermodynamic description of a general class of algebraically modified Friedmann dynamics in which the physical matter obeys the standard conservation equation and the modification is encoded through a function \(g(\rho)\). By introducing a thermodynamic matter representation \(\chi(\rho)\), we construct the corresponding matter variables, heat flow, work term, and horizon entropy, and show that both the Clausius relation and the first-law balance can be consistently maintained. A continuous family of representations is then defined by \(\chi_\lambda(\rho)=(1-\lambda)\rho+\lambda g(\rho)\) with \(0\leq\lambda\leq1\), and the physical-matter and effective-fluid descriptions as the two endpoints. The associated thermodynamic quantities transform consistently between these limits. We further show that apparent departures from equilibrium can arise from mixing quantities belonging to different representations, rather than from a genuine non-equilibrium entropy-production mechanism. These results identify a continuous freedom of thermodynamic representation within this class of modified FLRW dynamics.
Read the full abstract
Cryptography and Security
PatchRisk: Forecasting Future Vulnerability Exposure in Open-Source Dependency Networks
Open-source software ecosystems are web-scale dependency networks. A downstream package can become exposed to security risk not because its own source code changes, but because one of its transitive dependencies later receives a vulnerability advisory. Existing vulnerability-detection work often focuses on whether code is currently vulnerable or whether a known vulnerable dependency is already present. We study a different problem: future transitive vulnerability exposure. Given a package-version dependency graph observed at release time, the task is to predict whether any non-root dependency will receive a vulnerability advisory within a future horizon. This problem is important for Web intelligence and software supply-chain security because it supports proactive dependency triage before future exposure is visible. It is also easy to evaluate incorrectly: current vulnerable dependencies can leak the label, and different versions of the same root package can create package-level memorization across train and test splits. We therefore construct PatchRisk, a leakage-aware benchmark from Open Source Vulnerability advisories and deps.dev dependency graphs for npm and PyPI. The benchmark uses filtration-aware labels, package-disjoint evaluation, temporal testing, and nested 1K, 3K, 5K, and 10K sampling scales. The largest cleaned setting contains 9,007 root package-version graphs spanning 4,157 root packages. We evaluate three feature families: TimeOnly, GraphStruct, and HistoryGraph. On the 10K temporal-test benchmark, HistoryGraph improves AUPRC over the strongest TimeOnly baseline from 0.351 to 0.640 for 90-day forecasting and from 0.473 to 0.813 for 365-day forecasting. The improvement remains stable across smaller scales and package-group shuffle robustness tests.
Read the full abstract
Human-Computer Interaction
Oops, Not Now: PEARL, a RAG-Based Support Agent for Gameplay and What Players Want from AI Help
AI-powered gameplay support agents hold promise for game-based learning, yet grounding generative models in structured game data remains an open challenge. We present PEARL (Parallel Education Agent for Reflection and Learning), a dual-component Retrieval-Augmented Generation (RAG) system that combines semantic knowledge retrieval with structural board-state matching to deliver contextualized scaffolding in Parallel, a puzzle game for learning parallel programming. PEARL operates on two input streams (natural language queries and board topology), retrieving both conceptual explanations of gameplay moves and peer-generated board states as evidence: capabilities unavailable to a standard Large Language Model (LLM) with game state access alone. In a qualitative evaluation (N=10) comparing PEARL against an existing community-based Open Player Model (OPM) visualization system, participants preferred the visualization system on perceived usefulness and reported higher frustration with PEARL; five of ten minimized or abandoned the AI tool during play. Proactive delivery, generic responses, and trust deficits drove disengagement, while a subset of four participants found PEARL's grounded explanations complementary to visualization in specific contexts where they initiated the interaction. We position PEARL as a deployed design probe whose failure modes inform a concrete design agenda for AI gameplay support, captured as seven open problems for the community.
Read the full abstract
Artificial Intelligence
MANAS-2: Constrained Reconstruction for EEG Foundation Models
Masked reconstruction is widely used for EEG foundation models, but optimizing reconstruction on low-SNR waveforms does not necessarily produce the most useful latent representation. We introduce MANAS-2, a new EEG foundation model that combines a Raw-Band Hybrid (RBH) masked autoencoder with Constrained Reconstruction (ConRec), a physics-motivated regularizer. RBH jointly reconstructs temporal waveform patches and compact spectral-band targets, while ConRec acts only on the temporal decoder output, penalizing differences in RMS energy between adjacent short windows of the reconstructed waveform. ConRec is intended to shape the encoder by biasing it toward the organization of oscillatory-envelope information. Across seven held-out EEG datasets, adding ConRec to an otherwise identical RBH model increases frozen ridge recovery of six-band spectral power from mean R^2=0.860 to 0.906 and recovery of inter-patch band-energy dynamics from R^2=0.283 to 0.354, while temporal waveform information remains highly recoverable from the frozen latents. Applied to a temporal-only masked autoencoder, ConRec also improves frozen downstream transfer and frequency-dependent latent geometry despite receiving no spectral targets: i.e., the effects of ConRec are architecture-independent. MANAS-2 also outperforms leading EEG Foundation Models on most downstream knowledge-transfer tasks. From the effects of ConRec, we see that a physically motivated constraint imposed through the decoder can make for a more spectrally organized and transferable latent space. MANAS-2 therefore provides a new EEG foundation model built around constrained reconstruction as a mechanism for shaping representation–rather than reconstruction–quality.
Read the full abstract
Artificial Intelligence
JaxAHT: A JAX-Based Library for Ad Hoc Teamwork
As autonomous agents are increasingly deployed, they must be able to coordinate with previously unseen agents, such as humans. This challenge has been formalized as Ad Hoc Teamwork (AHT), and has attracted a substantial body of research. However, progress has been hindered by the prohibitive computational cost of the AHT research lifecycle, lack of standardized benchmark implementations, and the absence of a diverse, validated evaluation teammate suite. This work introduces JaxAHT, the first open-source, JAX-based library designed to accelerate and standardize the AHT research lifecycle. Leveraging JAX's hardware acceleration and massive parallelization capabilities, JaxAHT provides a unified framework for teammate generation, ego agent training, and evaluation against unseen teammates, achieving approximately 95x wall-clock speedup over existing PyTorch counterparts. Along with the library, we contribute a diverse suite of evaluation teammates across Level-Based Foraging, Overcooked, and Hanabi. Using JaxAHT, we conduct a large-scale, compute-controlled benchmark study of AHT algorithms, finding that no algorithm consistently performs best, that agent modeling primarily offers benefits in role-based scenarios, and that best response returns are underestimated by reinforcement learning from scratch.
Read the full abstract
Trading and Market Microstructure
Event-Time Order-Flow Memory, Operational-Time Impact, and Subordinated Market Observables
We consider two canonical market-microstructure regularities: the long-memory of trade signs and the square-root law of meta-order impact. The point is not to propose new empirical laws, but to separate the clocks on which existing laws are defined. The sign-memory law is an event-time statement about the ordering and fragmentation of hidden orders. The square-root impact law is an operational-time statement about front motion in a locally linear latent order book. Starting from a discrete-time random-walk bid/ask reaction–diffusion order book with a separate event clock, we identify an event-time imbalance reduction, derive the operational-time front and impact equations in the locally linear regime, and then subordinate both sign and impact observables to calendar time. Fractional or tempered clock effects enter through this event-to-calendar projection, not through a different operational-time impact mechanism. This gives a compact clock-aware framework in which event-time sign persistence, operational-time square-root impact, and anomalous calendar-time effects appear as distinct but compatible consequences of a common order-book representation.
Read the full abstract
Machine Learning
Certifying Model Upgrades with Slice-Wise Non-Regression and Incumbent Fallback
An updated model can improve an aggregate metric while degrading a slice that matters to a downstream user. We study checkpoint selection subject to non-regression tolerances relative to a retained incumbent. The central distinction is between failing to detect harm and certifying non-inferiority: the former can release harmful updates with high probability when evaluation is noisy. We give a reproducible release procedure that separates candidate search from independent, paired evaluation and returns the exact incumbent when certification fails. Applying established intersection-union and Learn-then-Test principles, we state finite-sample guarantees for one frozen candidate, a finite candidate library, and a prespecified testing order. A joint release decision does not require a slice-count Bonferroni penalty, although certification power can still decrease with the number of slices. In bounded-score simulations, a no-detected-harm gate releases a harmful candidate in 99.7% of trials in one 32-slice setting, compared with 2.6% for an exact non-inferiority gate at a 5% target. A constructed two-block family yields larger certified utility than a scalar path under matched candidate counts. Public digits experiments, including a subsequent continuation that improves average aggregate accuracy, return the incumbent in every run because certification is underpowered. These results establish an auditable protocol and its limitations; they do not establish benefits on foundation-model or multilingual translation upgrades.