arXiv:2607.16306v1 Announce Type: cross
Abstract: Hybrid beamforming (HB) with quality-of-service (QoS) provisioning per stream in millimeter waves is indispensable in 5G/6G networks. HB includes baseband and radio frequency (RF) beamforming, and requires error-free channel state information (CSI), which is erroneous in practice. So there is a need for efficient, feasible, robust, and QoS-aware HB. To achieve this, we mitigate CSI uncertainty via baseband beamforming, and we steer the RF beamformer by using the estimates of the channel's eigenvectors. In doing so, we consider the effective channel's uncertainty region instead of the uncertainty region of the channel itself, as the former is smaller than the latter, requiring less transmit power to satisfy the QoS constraint. We also detect and eliminate the infeasible data streams. Our iterative scheme (which is based on the cutting-set method) for baseband beamforming satisfies the mean-squared error (MSE) constraint per stream, where a limited number of constraints are considered instead of infinitely many constraints. In our low-complexity scheme, we derive a simple sufficient condition to check the feasibility of each stream, we diagonalize the effective channel at the baseband precoder, and we use minimum MSE combining at the baseband combiner. Extensive simulations validate our formulations and theoretical derivations.
Science Journals
arXiv:2607.16721v1 Announce Type: new
Abstract: The strongest open-weight coding models are mixture-of-experts (MoE) networks: most of their size comes from large pools of "expert" subnetworks, of which only a few act on any token. That pool is why these models do not fit on the machines most developers own, yet for a user who only wants coding help, most experts encode abilities that will never be invoked. We ask how many experts can be removed, and which, by pruning two recent open-weight MoE models from different families (Qwen3.6-35B-A3B and Gemma-4-26B-A4B) under five selection strategies, judged the way a user would: by whether the model still writes correct code. Half the experts can be removed from either model with no statistically detectable loss on the primary code benchmark, and the damage lands almost entirely on abilities outside coding, the intended trade. But the winning strategy flips between the two models, so a recipe validated on one family cannot be assumed to work on another. We further show that perplexity, the metric much of the pruning literature leans on, can rate a broken model above an intact one; that a lightweight fine-tune recovers about half of what aggressive pruning loses; and that against quantizing the full model to the same memory, pruning wins only where quantization would have to drop below 3 bits per weight. Five attempts to overturn that crossover, with failure criteria fixed in advance (better calibration, guarded selection, causal expert importance, failure attribution, and an agentic evaluation letting each model repair its failures from execution feedback), all leave it standing; the last shows single-shot benchmarks overstate compression penalties broadly, as one repair turn erases the 2-bit quantization penalty entirely. Expert pruning works, but it demands per-model validation on the task the model will actually serve.
arXiv:2607.16363v1 Announce Type: cross
Abstract: A large body of Semi-supervised Learning~(SSL) algorithms encounter the threshold $\tau$ to select pseudo-labels. The value of $\tau$ across different SSL algorithms can vary depending on the learning perspective, yet they may achieve similar performance. It motivates us to establish a unified theoretical framework to explain the role of $\tau$ in SSL. We statistically explained that the unsupervised loss is affected independently by correct and incorrect pseudo-labels, while $\tau$ adjusts their numbers to balance the corresponding error term. This inherent trade-off indicates that SSL can reach the same loss with varying $\tau$, precise optimal values of $\tau$ during training may be unnecessary. With this, we treat $\tau$ as an updatable parameter and optimize it via differentiation; the new policy is named \textbf{Meta-Thresholding Semi-Supervised Learning (MTSSL)}. Extensive experiments demonstrate the superior performance of MTSSL. We observe that the accuracy curves of SSL algorithms can overlap completely even when the values of $\tau$ differ significantly, which supports our theoretical framework and indicates that the selection of $\tau$ can be relaxed in the future design of SSL algorithms.
arXiv:2607.16424v1 Announce Type: cross
Abstract: Let $G=(V,E)$ be a finite simple graph of order $n\geq 1$, and let $\ell:V\to\{0,1\}$ be a prescribed parity labeling. A set $S\subseteq V$ is called $\ell$-admissible if $d_S(v)\equiv \ell(v)\pmod 2$ for every $v\in S$, where $d_S(v)=|N_G(v)\cap S|$. Let $h_\ell(G)$ be the maximum order of an $\ell$-admissible set and let $f_{\rm oe}(G)=\min_\ell h_\ell(G)$. For $x\in\mathbb R$, define the weighted counting polynomial $$
M_{\ell,x}(G)=\sum_{S\in {\cal A}_\ell(G)}x^{|S|}, $$ where ${\cal A}_\ell(G)$ is the collection of all $\ell$-admissible sets in $G$. For $R\subseteq V$, let $z_\ell(R)$ be the number of vertices $v\in V\setminus R$ for which $d_R(v)\equiv\ell(v)\pmod 2$. We prove the exact identity $$
M_{\ell,x}(G)
=2^{-n}\sum_{R\subseteq V}
x^{|R|}(2+x)^{z_\ell(R)}(2-x)^{n-z_\ell(R)-|R|}. $$ If $G$ has no isolated vertices, then, for every $\ell$ and every $x\in(0,2)$, $
M_{\ell,x}(G)>x^{n/2}(4-x^2)^{n/4}. $ Combining this estimate with a binary-entropy upper bound and optimizing $x$ gives $$
f_{\rm oe}(G)>c_*n>\frac{2n}{21}, $$ where $c_*\approx0.095862615$. Ferber and Krivelevich (Adv. Math. 2022) proved that $h_{\mathbf{1}}(G)\ge 10^{-4}n$, where $\mathbf{1}$ is the all-one labeling. Since $h_{\mathbf{1}}(G)\ge f_{\rm oe}(G)$, our result improves coefficient in their bound by almost three orders of magnitude, and does so simultaneously for every labeling.
arXiv:2607.16916v1 Announce Type: new
Abstract: Bladder cancer treatment requires personalized and adaptive decision-making, particularly for recurrent disease, where treatment effectiveness changes across successive clinical episodes. Conventional clinical decision support systems typically rely on static treatment guidelines or single-step predictive models, limiting their ability to capture disease progression over time. This paper presents a recurrent patient state-transition simulation framework for bladder cancer treatment planning that integrates predictive state-transition modeling with a Markov Decision Process (MDP) and a Deep Q-Network (DQN) reinforcement learning environment. The predictive module estimates changes in tumor characteristics following treatment, while the reinforcement learning agent sequentially optimizes treatment decisions by interacting with simulated patient trajectories. This framework enables dynamic, patient-specific treatment planning by continuously adapting recommendations to evolving clinical states. It also generates interpretable treatment trajectories and detailed simulation logs to improve transparency and support clinical decision-making. The proposed framework was evaluated against existing reinforcement learning-based treatment planning approaches. It achieved a cumulative reward of 63,918.87, an average training loss per episode of 0.0056, and a policy improvement score of 6.62%, demonstrating effective sequential learning and robust treatment optimization in a simulated recurrent treatment environment. These findings highlight the potential of recurrent patient state-transition simulation with reinforcement learning as a flexible decision-support framework for personalized bladder cancer treatment planning and AI-assisted precision oncology.
arXiv:2607.17935v1 Announce Type: new
Abstract: Automated fact-checking remains a challenge for Large Language Models (LLMs) due to "query brittleness" in traditional retrieval systems. We propose DeLIVeR (Decomposed Learning for Information-grounded Veracity Recognition), a framework that treats evidence retrieval as a reinforced strategic exploration task. DeLIVeR utilizes a Planner LLM to decompose complex claims into targeted question sets, which are used to traverse structured Knowledge Graphs (KGs) for high-precision evidence. We optimize the Planner's policy using Group Relative Policy Optimization (GRPO) with a reward system prioritizing structural diversity and verdict accuracy. Our evaluation on LIAR, FEVER, and PolitiFact shows that DeLIVeR significantly outperforms state-of-the-art baselines. Using Qwen2.5-7B, our framework achieved peak F1-scores of 83.73, 84.57, and 79.70 respectively, representing a 10-15% improvement over HippoRAG2. By shifting to a reinforced question-planning strategy, DeLIVeR effectively bridges multi-hop reasoning gaps and provides an auditable, transparent path for verifiable misinformation detection.
arXiv:2607.16430v1 Announce Type: cross
Abstract: The integration of distributed energy resources (DERs) into the power grid has introduced new challenges to AC optimal power flow (AC-OPF) problems. Traditional OPF optimize consider transmission systems, treating distribution networks as static loads. However, the growing presence of DERs makes accurate distribution system modeling crucial for grid operations. Consequently, efficiently solving the resultant large-scale, nonconvex transmission and distribution (T&D) AC-OPF problem remains a significant challenge. This paper proposes a Smoothed Two-Stage Decomposition Optimizer (StsDOpt) to address these complexities by decomposing the T&D AC-OPF problem into a master-subproblem(s) structure, enabling parallel solving. Unlike traditional methods, StsDOpt does not rely on approximations or relaxations. It uses a smoothing technique to render the subproblems responses differentiable with respect to the master problem, leveraging the barrier problem properties inherent in primal-dual interior point methods. This approach is crucial for accurately modeling and solving distribution systems, which are multiphase, unbalanced, and nonlinear, distinguishing StsDOpt apart from other methods. Integrated into the PowerModelsITD framework, StsDOpt has been validated through numerical experiments, demonstrating reduced wall-clock solve time and increased scalability. Results highlight its efficacy as a robust, scalable solution for large-scale T&D AC-OPF problems, facilitating the reliable integration of DERs into complex T&D systems.
arXiv:2607.17230v1 Announce Type: new
Abstract: This paper describes TalTech's submissions to the Beyond Transcription Challenge (BeTraC), which requires generating SOAP notes directly from long doctor-patient conversation recordings, without intermediate transcription. After screening open-weight speech LLMs for long-audio robustness, we adapted Voxtral Mini (lightweight track) and Voxtral Small (heavyweight track) with LoRA supervised fine-tuning followed by DAPO reinforcement learning that uses the challenge metric, Open Medical Concept F1, as its reward. Our systems ranked first in both tracks, and an independent LLM-as-a-judge evaluation showed the lowest hallucination rate among all submissions, indicating that reinforcement learning against a concept-matching metric need not compromise factual reliability. We also find that fine-tuning on text transcripts transfers well to speech input and appears to improve robustness on out-of-domain real recordings.
arXiv:2607.17476v1 Announce Type: new
Abstract: Autonomous flight of aerial robots in narrow space remains challenging due to strong aerodynamic disturbances and limited flying space. Existing approaches mainly address aerodynamic disturbances at the control level, while motion planning typically relies on geometric constraints and fixed speed limits, leading to conservative or unsafe behaviors in confined environments. This paper presents a disturbance-aware planning and control framework (DAPCF) that integrates online disturbance estimation into the planning-control loop for quadrotor flight in narrow space. First, the dual-loop observers estimate 6-degree-of-freedom disturbance forces and torques in real time based on odometry and motor speed measurements. Then, a disturbance risk function is introduced that adaptively modulates the reference speed of the planner based on disturbance estimation, reducing velocity when disturbances exceed a threshold and restoring it under low-disturbance conditions. Finally, a motor-dynamics-based nonlinear model predictive controller (MDNMPC) with disturbance compensation is designed to ensure robust trajectory tracking under perturbed conditions. Experiments demonstrate that a quadrotor with a diagonal length of 0.39~m can traverse straight, sloped, and curved tunnels as narrow as 0.6~m, outperforming human pilots in both success rate and flight efficiency.
arXiv:2607.16937v1 Announce Type: new
Abstract: In this study, we characterize the permittivity of various commercially available materials commonly used in 3D printing over the millimeter-wave frequency range of 70-110 GHz. The open-waveguide extraction method is employed to efficiently determine the permittivity of these 3D-printed materials, and key strategies for improving its accuracy are discussed. This methodology can be readily applied to the characterization of similar materials and extended to higher frequency ranges. The resulting permittivity data are expected to support advances in 6G and beyond wireless communication technologies, in which 3D-printed materials and millimeter-wave frequencies are anticipated to play an increasingly important role in the development of novel antennas and metasurfaces.
arXiv:2607.16757v1 Announce Type: new
Abstract: We introduce a new branching bisimulation for probabilistic processes, which induces a more refined equivalence relation than any known equivalence that abstracts from unobservable actions, with a rooted version that is a congruence for a language of probabilistic process with the usual static as well as dynamic constructs including recursion.
arXiv:2607.16436v1 Announce Type: cross
Abstract: We consider a formal model of quantum circuit description languages (QCDLs) in which semantically meaningful programs correspond to computable unitary matrices. We show that any semantically universal QCDL -- that is, any QCDL able to describe all computable unitary matrices, which in turn form the set of matrices we can meaningfully represent on digital hardware -- cannot have a semi-decidable set of semantically meaningful descriptions. In particular, no such language admits a compiler that reliably recognizes all valid program descriptions. This result stands in contrast to classical programming languages. While compilation in languages such as C or C++ may itself involve non-terminating computations, the set of semantically meaningful programs remains recursively enumerable, since successful compilation provides a witness of validity. The essential difference lies in the nature of the semantic domains: classical languages describe partial recursive functions, whereas QCDLs describe total unitary operators. Our analysis establishes a fundamental limitation of quantum circuit description languages and highlights a structural distinction between classical and quantum models of computation at the level of formal language theory.
arXiv:2607.16466v1 Announce Type: cross
Abstract: Molecular materials that enable coherent control over an electron's spin state at room temperature are promising candidates for quantum technologies, including quantum sensors and ultra-low noise microwave amplifiers, known as masers. Host-guest molecular crystals enable independent control of spin-active guests and their local environments to enhance molecular spin properties and so improve device performance. Using electron paramagnetic resonance and optically-detected magnetic resonance, we demonstrate the ability to tune triplet population, depopulation, and spin-lattice relaxation by modulating host-dependent lattice rigidity and vibrational coupling to significantly reduce the operating requirements for building useful masers. Importantly, the most rigid host, picene, reveals the ability to slow spin-lattice relaxation without lengthening triplet lifetime, though at the cost of strain-induced line width broadening and reduced triplet spin polarisation. We also find that deuteration reduces the triplet resonance line width and vibrationally-mediated triplet depopulation. Consequently, we find that perdeuterated pentacene in perdeuterated p-terphenyl is the most viable candidate for building a continuous wave maser. This work demonstrates host-guest engineering as an important and practical method for tuning the spin-dependent performance of room-temperature molecular quantum technologies.
arXiv:2607.17712v1 Announce Type: new
Abstract: Detecting high-level semantic concepts like negation across modalities remains a challenge for current multimodal systems. We analyze this as a fundamental representation learning problem, providing the first evidence that negation does not form a linearly or non-linearly separable class in the latent spaces of standard vision-language models (VLMs). We demonstrate that pretrained embeddings primarily encode modality-specific features, lacking a generalizable negation signal. To overcome this, we propose a novel cross-modal attention architecture that explicitly models inter-modal dependencies, achieving performance gains of up to +7.03% F1 over unimodal baselines. Our analysis reveals a key asymmetry: while textual negation often appears independently, visual negation is semantically dependent on linguistic context, a finding validated through our statistical analysis of 3,222 political video-text pairs automatically annotated via \textsc{Qwen2.5-VL}. By combining this analysis with self-supervised video representations (JEPA2), we advance the modeling of temporal negation. This work provides new methods and insights for learning robust, semantically-aligned representations in multimodal systems.
arXiv:2607.18066v1 Announce Type: new
Abstract: The value alignment of large language models (LLMs) is crucial for ensuring responses align with human intention and value preferences. However, most evaluations of value alignment focus on Western or universal values, while assessments grounded in the value systems of specific countries remain scarce. In this paper, we introduce Pancasila-Dilemmas, an evaluation dataset of 1,834 questions derived from Indonesian news, classified by 5 values of Pancasila: Religion, Humanity, Unity, Democracy, and Social Justice. This dataset reflects daily life in Indonesia, making it suitable for measuring the value alignment of LLMs deployed for Indonesia. To ensure a more rigorous evaluation, we choose scenarios containing dilemmas. The dataset is proofread by native speakers and answered by 5 diverse Indonesian citizens. We evaluate 50 closed- and open-source LLMs on our dataset. Results reveal that all evaluated LLMs achieves less than 0.5 Probability Match Score (PMS) and 0.72 Max-Vote Agreement Score (MVAS). Compared by each values, LLMs mostly struggle in Religion and Unity dilemma cases. This highlights a significant gap in capturing Indonesian values. The dataset is publicly available at https://github.com/tjunlp-lab/Pancasila-Dilemmas.
arXiv:2607.16773v1 Announce Type: new
Abstract: The classical Ingleton inequality is known to hold for entropic points under specific exact conditional independence constraints. Recently, Matveev and Romashchenko (2026) investigated the stability of these implications, quantifying the extent to which the Ingleton inequality can be violated when a group of conditional mutual information terms is small but non-zero. While their proofs relied fundamentally on the complex framework of tropical probability spaces, we revisit these stability results using a completely tropicalization-free approach. By developing an alternative framework, we significantly streamline the underlying concepts and proofs, derive explicit error terms, and improve some estimates. Furthermore, we resolve an open problem posed in prior work by exhibiting a new infinite family of entropy inequalities that establishes the stability of the sum of two Ingleton expressions.
arXiv:2509.14357v4 Announce Type: replace
Abstract: The Freeze-Tag Problem (FTP) asks for the minimum time needed to activate a swarm of robots, starting from a single active robot. When an active robot reaches a frozen robot, the latter becomes active; both robots may then move independently and activate further robots. We prove that FTP is strongly NP-hard in the plane under every fixed rational $L_p$ distance, $1 \le p < \infty$, and under $L_\infty$. The geometric argument also applies to every fixed real $p > 1$ for which the metric admits an effective specification. For $L_1$ and $L_\infty$, the integer-coordinate decision problems are strongly NP-complete. The reduction starts from Numerical 3-Dimensional Matching with distinct integers and also yields NP-completeness for unweighted planar grid graphs.
arXiv:2607.18069v1 Announce Type: new
Abstract: As more capable AI models are increasingly integrated into critical computer systems, the lack of control over AI intent motivates safety mechanisms. Existing software safeguards impose only behavioral constraints that can potentially be bypassed by sufficiently intelligent models. While hardware-level safety enforcement has been recognized as an essential last line of defense, few mechanisms have been proposed beyond policy regulations on unauthorized accesses or coarse full-chip shutdown. What is missing is a fine-grained, dynamic intervention mechanism at the architecture level.
In this paper, we introduce a set of microarchitecture knobs which dynamically control the available hardware resources to limit AI performance at runtime. We evaluate candidate knobs spanning the GPU memory subsystem, across capacity, bandwidth, latency and frequency dimensions, and narrow down to four strong candidates: L2 size, L2 latency, L2 bandwidth, and shared memory port access rate. To minimize new logic and extra design cost, we build all four mechanisms from well-established microarchitectural primitives: cache way masking, credit-based rate limiting, latency insertion, and bank arbitration. We show that these knobs achieve high performance sensitivity (up to 80% performance cut at 1/8 resource availability), negligible implementation cost (<~10K flip flops), fast stabilization after dynamic throttling (5-80K cycles), and minimal collateral impact on the rest of the chip. Further, multi-knob analysis reveals combinations of knobs that amplify the performance degradation beyond the effect of each knob individually, which enables a broader range of performance targets.
arXiv:2607.18074v1 Announce Type: new
Abstract: Equivariant graph neural networks repeatedly apply edge-conditioned tensor-product convolutions over graph edges. Conventional implementations materialize edge-specific weights, messages, and adjoints, causing tensor-product workspace and memory traffic to grow rapidly with graph size and operator width. This limits feasible workloads and can prevent larger problems from fully utilizing the GPU.
We show that these edge-sized intermediates are artifacts of the execution schedule, not requirements of the equivariant operator. By reassociating radial projection, spherical-harmonic coupling, and graph aggregation, edge-local products can be consumed directly into bounded receiver-side state. The resulting streaming formulation preserves fully connected multiplicity mixing and extends through forward, backward, and double backward.
We implement this formulation in Sobek, a generated-CUDA backend, and evaluate it across edge-scaling regimes and varied feature structures. Across two operator families and all three differentiation orders, Sobek is faster in all 75 capacity-matched comparisons, with speedups ranging from $1.2\times$ to $49.7\times$, and reduces peak allocated memory by up to 99\%. It also executes workloads up to two orders of magnitude beyond OpenEquivariance's capacity while retaining near-peak throughput. These results show that edge-scaled tensor-product workspace is a property of the conventional schedule, not of equivariant convolution itself.
arXiv:2607.16789v1 Announce Type: new
Abstract: Real-world perception and decision making are inherently multimodal, integrating complementary signals across modalities. However, training multimodal models faces two main obstacles. First, collecting large-scale, well-aligned paired multimodal datasets is often impractical, making end-to-end multimodal training difficult. Second, existing multimodal representations frequently entangle information shared across modalities with modality-specific information, hindering interpretability and control. We introduce MultiLoReFT, an efficient and scalable low-rank representation fine-tuning framework for multimodal learning with pretrained unimodal models. MultiLoReFT extends low-rank adaptation to the multimodal setting and learns interpretable projection subspaces that decouple shared and modality-specific information. Across simulated and real-world benchmarks, it produces representations that support multimodal prediction while explicitly revealing how shared and modality-specific information is distributed across modalities.
arXiv:2607.17604v1 Announce Type: new
Abstract: Mobile robots in public spaces must ensure pedestrians' comfort, and yet empirical studies of walkers' subjective safety are rare. Many classical navigation algorithms do not distinguish the walkers from dynamic obstacles and do not explicitly model subjective human factors. Moreover, most studies focus on holonomic mobile robots, whereas applications demand Nonholonomic Mobile Robots (NMR). This paper develops socially aware algorithms for NMRs, proves the stability, verifies the performance experimentally, and statistically analyzes the reported comfort. We design a framework for NMRs using Social Force Model (SFM) and the projected Time-to-collision Social Force Model (TSFM). We formalize the NMR-pedestrians' and NMR-obstacles' interactions and prove the system's stability, assuming boundedly nonpassive pedestrians. Simulations calibrate the models by maximizing a hybrid cost function of comfort and speed. Pedestrian-robot interaction experiments compare SFM and TSFM to two remote-controlled baselines and collect walkers' reported comfort. Statistical tools analyze survey results collected during the experiments. Benchmarking the algorithms against previous studies highlights the proposed methods' advantage with respect to the studied metrics. Overall, the models are stable and improve pedestrian comfort when an NMR navigates through a pedestrian crowd.
arXiv:2607.16918v1 Announce Type: new
Abstract: We develop new adaptive numerical schemes for one- and two-dimensional hyperbolic systems of conservation laws. The methodology relies on the use of a smoothness indicator to automatically partition the computational domain into smooth and nonsmooth (``rough``) regions. We then follow the scheme adaption strategy recently introduced in [S. Chu, P. Feng, V. A. Kolotilov, A. Kurganov, and V. V. Ostapenko, Commun. Comput. Phys., accepted], but instead of the quasi-linear (QL) fifth-order finite-difference scheme used there, we employ the new QL seventh- and ninth-order schemes in the smooth regions. A series of numerical experiments for the Euler equations of gas dynamics demonstrates that the new adaptive schemes contain a smaller amount of numerical dissipation and achieve higher resolution compared with their counterpart that uses the QL fifth-order scheme in the smooth areas.
arXiv:2607.18110v1 Announce Type: new
Abstract: Reinforcement learning (RL) on open-ended tasks compresses an LLM's rubric-based evaluation into a scalar reward, discarding rich textual feedback and conflating responses with distinct quality profiles. We propose Experiential Learning (EL), which repurposes the feedback model from an LLM-as-a-Judge into an LLM-as-a-Coach. The coach distills its assessment of each on-policy response into transferable experiential knowledge, which conditions a teacher model and is internalized by the policy through on-policy context distillation. Compared with scalar rewards, this higher-bandwidth feedback channel provides dense supervision and preserves fine-grained preferences among high-quality responses. Across two policy families, with feedback from the policy itself or a proprietary model, EL consistently outperforms rubric-based RL on held-out and unseen open-ended tasks. Notably, EL generalizes better beyond the training distribution, and mitigates reward hacking. These findings establish experiential knowledge as a richer and more generalizable learning signal for post-training on non-verifiable tasks.
arXiv:2607.16997v1 Announce Type: new
Abstract: Mathematicians distinguish proofs that explain, simplify, or introduce a nonstandard route, but these judgments are difficult to operationalize. We study a deliberately narrower construct: time-relative proof-route nonstandardness in formal mathematics. For a Lean theorem, PriorProof extracts the dependency footprint of its elaborated proof term and scores the weighted surprisal of that footprint under a retrieval-conditioned, hierarchically smoothed prior built only from an earlier quarterly snapshot of Mathlib. The method requires no hand-built technique ontology and no human labels: statement retrieval is learned from proof-derived contrastive pairs, while the scored object is read mechanically from proof terms. In a blinded topology study, 100 presentations collapse to 76 distinct underlying pairs: 12 canonical contrasts shown three times for consistency screening and 64 distinct stratified pairs. Against the majority of three retained domain raters, PriorProof agrees on 53/76 pairs (69.7%, Wilson 95% CI 58.7-78.9%), including 11/12 canonical pairs (91.7%, 64.6-98.5%) and 42/64 stratified pairs (65.6%, 53.4-76.1%). Score-gap quartiles are nonmonotone after repeat collapse; the endpoints are 12/19 (63.2%, 41.0-80.9%) in the smallest-gap bin and 16/19 (84.2%, 62.4-94.5%) in the largest, supporting an endpoint-calibration tendency rather than a resolved staircase. The best language-model condition agrees on 60/76 pairs (78.9%, 68.5-86.6%); on paired outcomes, PriorProof alone is correct on 8 pairs and the model alone on 15 (exact two-sided McNemar p = 0.210), so the difference is not established at this sample size. We therefore present PriorProof not as a replacement for expert or model judgment, but as a decomposable, time-anchored signal whose score gap provides an interpretable reliability indicator.
arXiv:2607.17745v1 Announce Type: new
Abstract: Large language models (LLMs) are increasingly considered for environmental enforcement, but their ability to produce traceable enforcement decisions remains unclear. We introduce WuYu-EnvLE-Bench, a benchmark built from real enforcement cases, regulatory standards, and expert review. It contains 2,521 benchmark instances, 14 tasks, and 12 pollution-medium subdomains across pre-enforcement, in-enforcement, and post-enforcement workflows. Using Absolute Environmental Enforcement Score (AES) and Intelligent Enforcement Index (IEI), we evaluate open-source and closed-source LLMs across capability, response quality, and resource efficiency. Results show that LLMs perform well on rule-bounded tasks but remain unreliable in evidence-chain construction, contradiction detection, multi-source integration, and procedural judgment. Model scaling also shows diminishing returns: medium-sized models approach leading models in structured tasks, while larger models do not reliably overcome evidence-reasoning bottlenecks. WuYu-EnvLE-Bench highlights the need for evidence-grounded, rule-aware, and task-adaptive enforcement reasoning.