arXiv:2605.21642v1 Announce Type: new
Abstract: Vision-language models (VLMs) are increasingly augmented with continuous or latent non-textual tokens intended to support "visual thinking." Despite improved task accuracy, this alone does not show that models actually use these tokens for reasoning -- gains may arise from confounds such as added context length, special-token anchoring, or training-time regularization. We formalize a diagnostic principle, Ablate-to-Validate, for testing whether latent-token content is genuinely utilized, and instantiate it as the Token Replacement Test (TRT), a standardized suite of content-replacement ablations. TRT holds the prompt, image, token budget, and decoding fixed while replacing intermediate tokens with zero, random, first-repeat, or oracle alternatives, isolating whether performance depends on token content or merely on token presence. As a controlled testbed, we study relative depth reasoning with LLaVA-13B and Qwen2.5-VL-3B, training models to predict and consume continuous or discrete depth spans across multiple frozen encoders (SigLIP2, CLIP, DINOv2) and token budgets. We additionally apply TRT to three off-the-shelf visual-thinking systems (Mirage, Mull-Tokens, CoVT) on BLINK, VSP, and CV-Bench. Across all settings, accuracy gains are a misleading proxy for latent-token reasoning: VLMs retain most improvement even when token content is corrupted or replaced, revealing a persistent gap between having a latent channel and using it as an information bottleneck. We recommend TRT as a standard diagnostic alongside accuracy for any method introducing continuous thought tokens.
Science Journals
arXiv:2605.21653v1 Announce Type: new
Abstract: AI text detectors amplify a pretrained typicality axis; they do not construct an AI-vs-human boundary. On raw encoders before any task supervision, projecting onto centroid(AI)-centroid(HC3) achieves NYT-vs-HC3 AUROC 0.806/0.944/0.834 across three architectures (86-106% of the fine-tuned discrimination ceiling: on RoBERTa-base, raw projection exceeds fine-tuning); on RoBERTa-base, full fine-tuning reduces discrimination below raw on both fluent-formal populations tested. The same axis inverts on non-native ESL writing (AUROC 0.06-0.20) -- a falsifiable prediction unique to the typicality reading. A 24-example frozen probe matches full fine-tuning (0.900 vs 0.895). A closed-form Jacobian predictor parameterises axis-manipulating interventions with R^2 = 1.000 universal, lifts ELECTRA-CE deployment TPR from 0.000 to 0.904 at FPR = 1%, and transfers to three independently-trained third-party RoBERTa detectors at 16/16 oracle-equivalence (57% NYT-FPR reduction on the OpenAI detector). Scope: encoder family; mechanism magnitude HC3-anchored; population-level shared axis with per-text mechanisms varying across architectures. Three operationally distinct probes -- text-surface caps_rate residualisation, geometric signed-epsilon ablation, closed-form text-pair predictor -- agree at cos 0.74/0.81/1.00 across three architectures, confirming observer-invariance. Under matched-TPR-0.90 evaluation, the published intervention zoo (CC, dealign-f2c) is calibration-equivalent across 27 cells (|Delta AUROC| <= 0.0081), and >= 97% of the LoRA->full-FT bias gap on ELECTRA is calibration shift, not learned representation -- the central claim's prediction confirmed.
arXiv:2605.21654v1 Announce Type: new
Abstract: Reinforcement learning substantially improves pretrained language models, but it remains understudied why critic-free methods such as PPO and GRPO work as well as they do, and when they should provide the largest gains. We develop a value-gradient perspective of critic-free RL for LLM post-training. First, under a differentiable rollout and additive-noise parameterization, we show that the actor update is value-gradient-like in expectation: the backward pass propagates costates whose conditional expectation equals the value gradient. Second, for discrete transformer policies, we show that autodifferentiation through attention produces empirical costates that approximate this value signal, with an error controlled by the sampling gap and policy entropy. These results motivate a decomposition of RL impact into value gradient signal and reachable reward headroom, yielding a criterion for when RL should be most effective along a pretraining trajectory.
arXiv:2605.21676v1 Announce Type: new
Abstract: Stochastic cyber-physical systems (CPS) permeate critical infrastructure, from autonomous vehicles to medical devices. Yet, tools for runtime verification of such systems capturing the probabilistic dynamics in stochastic systems remain generally absent despite theoretical foundations established nearly a decade ago. In this paper, we present SENTIL, a novel runtime verification tool with provable statistical guarantees for the runtime monitoring of requirements expressed as Probabilistic Signal Temporal Logic (PrSTL). SENTIL combines an efficient Rust core with universal ecosystem integration, delivering performance exceeding existing deterministic monitors while providing rigorous probabilistic guarantees through statistical model checking, sequential probability ratio testing, and adaptive rare event estimation. SENTIL employs streaming algorithms for incremental robustness computation, parallel Monte Carlo sampling, and a language-agnostic C-ABI enabling seamless deployment across ROS, Apollo, MATLAB Simulink, and AUTOSAR platforms, and direct integration in C, C++, Python, and Java. To validate the effectiveness of the proposed tool, we validate SENTIL across various scenarios spanning autonomous vehicle monitoring, medical device validation, and biological networks, demonstrating 10-1,000$\times$ performance improvements over existing tools while maintaining provable confidence intervals. SENTIL is open source (\href{https://github.com/sedislab/SENTIL}{\texttt{sedislab/SENTIL}}) and it positions probabilistic runtime verification as a deployable infrastructure for all real-world safety-critical stochastic systems.
arXiv:2605.21677v1 Announce Type: new
Abstract: Flaky tests pass and fail non-deterministically when run on the same version of code. Although many techniques have been proposed to detect, debug, and repair flaky tests, reproducing their failures remains a major challenge due to their inherent nondeterminism. Many flaky test datasets exist to help researchers study them, but these datasets are often composed of disjoint sets of flaky tests, where each dataset provides unique information, such as flaky tests of different categories, failure logs of flaky tests, or flaky tests reported by developers vs. flaky tests found by automated tools. In this work, we aim to create a reproducible dataset of flaky tests, curated from both developer issue reports and a popular dataset of flaky tests. Compared to prior flaky test datasets, our dataset is the first to provide (1) a reproducible environment to compile flaky tests, (2) scripts to reproduce failures, (3) scripts to automatically apply flaky test fixes and ensure that the tests are no longer flaky, and (4) execution logs of flaky test passing and failing. We present ReproFlake, a dataset of 1115 reproducible flaky tests across four flaky test categories. We create guidelines to help others contribute to this reproducible dataset, and demonstrate how to use our dataset to understand challenges in reproducing flaky test failures (e.g., challenges researchers may face when using prior flaky test datasets), the characteristics (e.g., location of the fix and its correlation with the flaky test category), and difficulties researchers may face in using our dataset to collect additional information (e.g., code coverage) about flaky tests. Our findings show that error information helps identify flaky test categories and guide repairs, that unresolved compilation failures highlight challenges in building legacy projects, and knowing typical fix locations can help prioritize repair efforts.
arXiv:2605.21679v1 Announce Type: new
Abstract: Component-wise accurate algorithms for computing the principal square root of an M-matrix are designed in terms of triplet representations. A triplet representation of an M-matrix $A$ is the triple $(P, {\bf u},{\bf v})$, where the matrix $P$ is such that $p_{ij}=-a_{ij}$ for $i\ne j$, $p_{ii}=0$, and ${\bf u}>0$, ${\bf v}\ge 0$ are two vectors such that $A{\bf u}={\bf v}$. It is shown that if $A$ is an M-matrix representable by a triplet, then its principal square root exists and is an M-matrix represented by a triplet as well. New versions of the Cyclic Reduction and the Incremental Newton iterations are provided in terms of triplets, to compute the principal matrix square root of $A$. It is shown that these algorithms are component-wise numerically stable independently of the singularity of $A$ and of its condition number. Numerical experiments are shown to confirm the component-wise stability.
arXiv:2605.21683v1 Announce Type: new
Abstract: Developing AI systems with a human-like understanding of everyday concepts is a key step towards developing safe, reliable systems whose behavior makes sense to humans. When probing concept understanding, asking questions about plausible category members (e.g., "Is a car a vehicle?") is likely to recall patterns in the model's vast training data. We pursue an alternative strategy, characterizing the boundaries of conceptual categories by asking about implausible category members (e.g., "Is an olive a vehicle?") to probe the kind of concept-level knowledge we take for granted in fellow humans. We characterize concept boundaries for a set of fundamental concepts by studying AI systems' assignments of objects to superordinate categories from a classic psychological study by Rosch and Mervis, as well as their assignments of the same objects to mismatched superordinate categories. We compare these assignments to those made by human participants on the full range of within-category and cross-category assignment tasks. Our results reveal a range of concepts for which which models differ in meaningful and surprising ways from humans, including treating "words" as belonging to categories like "vehicles" and "clothing," identifying several "vegetable" category members as "fruit," and assigning exemplars from non-weapon categories to the "weapons" category. We also demonstrate how these instances of concept misalignment translate into problematic downstream behavior with implications for AI safety.
arXiv:2605.21692v1 Announce Type: new
Abstract: Characterizing precisely the asymptotic generalization error of neural networks using parameters that can be estimated efficiently is a crucial problem in machine learning, which relies heavily on heuristics and practitioners' intuition to make key design choices. In order to mitigate this issue, we introduce the Representation Gap, a metric closely related to the generalization error, but admitting better-behaved asymptotic dynamics. Focusing on equivariant diffusion models and leveraging results from optimal quantization and point-process theory, we derive a precise asymptotic equivalent of the Representation Gap and show that it is governed by a single parameter, the \textit{intrinsic dimension} of the task, which is easy to interpret, efficient to estimate, and can be linked to the equivariances of common neural network architectures. We show that this asymptotic dynamic also extends to a broader range of tasks and training algorithms. Finally, we demonstrate empirically that our asymptotic law and intrinsic dimension estimation are accurate on a wide range of synthetic datasets, where these quantities are known, as well as on more realistic datasets, where we obtain results consistent with the related literature.
arXiv:2605.21704v1 Announce Type: new
Abstract: Locomotion in microgravity often relies on sparsely and irregularly arranged anchors, motivating grasp-based mobility with multiple limbs. In this setting, dynamic locomotion is feasible only through deliberate regulation of both anchored interactions and whole-body coordination under coupled dynamic and kinematic constraints. This paper presents design insights for grasp-based dynamic locomotion with multi-limbed robotic systems in microgravity, targeting scenarios that require 6D limb manipulation to establish contacts with candidate anchors. The investigated design parameters include gait pattern, stride length, locomotion speed, and nominal posture. A parameterizable locomotion planning framework is proposed to support variations of these parameters and to evaluate the resulting locomotion performance in terms of stability and actuation demand. Two representative quadruped morphologies are adopted for evaluation in physics-based simulation. The results demonstrate that enlarging the feasible contact wrench space and attenuating impulsive whole-body dynamics improve locomotion performance. These findings inform strategies for contact configuration selection and whole-body coordination in microgravity locomotion with multi-limbed systems.
arXiv:2605.21708v1 Announce Type: new
Abstract: This paper investigates the control synthesis for continuous-time uncertain systems under nested Signal Temporal Logic (STL) specifications containing nested temporal operators. Control Barrier Functions (CBFs) are utilized herein to encode STL formulas into system constraints. However, traditional CBF designs fail to encode nested STL formulas, whereas recent reachability analysis-based methods capable of handling such formulas are inapplicable to uncertain systems and suffer from a severe computational burden. To overcome these challenges, a novel recursive CBF design procedure based on a modified STL tree (sTLT) is proposed to yield explicit parameterized CBFs. Within this framework, sliding window variables are introduced to capture complex temporal relationships. Crucially, satisfying the resulting CBF constraints is proven to guarantee the fulfillment of the STL specifications. To render the proposed recursive CBF design applicable to systems subject to uncertain disturbance, a novel controller based on reconstructed CBF using quadratic programming (QP) is proposed, ensuring strict CBF constraint satisfaction under disturbances. In contrast to existing methods, the proposed reconstructed CBF approach requires no prior knowledge of the disturbances while relaxing initial safety assumptions. Simulation results validate the efficacy of the proposed approach.
arXiv:2605.21710v1 Announce Type: new
Abstract: Behavior cloning for contact-rich bimanual manipulation remains challenging because diverse demonstrations are expensive to collect, and even small disturbances can push the system into off-manifold states where no recovery supervision is available. We propose PGDG, a data generation framework with zero-shot curation that expands a single demonstration into a compact dataset of physically plausible, successful, and diverse recovery behaviors without additional human labeling. PGDG iterates between a physics-grounded sampler and a dataset curator, where the curator selects informative, non-redundant, and recoverable behaviors to update the sampling distribution toward under-covered recovery modes, and the sampler draws physically plausible rollout candidates from this updated distribution and retains successful trajectories. To further improve data quality, PGDG applies short-horizon sampling-based control to relabel selected risky states with corrective actions. Across four bimanual manipulation tasks, PGDG consistently outperforms spatial-only augmentation in both simulation and zero-shot real-world transfer. On RotateBox-Pitch, success improves from 38% to 93% in simulation and from 35% to 82% in the real world. PGDG also enables effective foundation models fine-tuning such as GR00T, increasing success from 46% to 77%. Additional results are available in our website: https://cunxid.github.io/PGDG/.
arXiv:2605.21713v1 Announce Type: new
Abstract: How can we distinguish whether a peer review was written by a human or generated by an AI model? We argue that, in this setting, authorship should not be attributed solely from the textual features of a review, but also from the ideas, judgments, and claims it expresses. To this end, we propose Sem-Detect, an authorship detection method for peer reviews that operationalizes this principle by combining textual features with claim-level semantic analysis. Sem-Detect compares a target review against multiple AI-generated reviews of the same paper, leveraging the observation that different AI models tend to converge on similar points, while human reviewers introduce more unique and diverse ones. As a result, Sem-Detect is able to distinguish fully AI reviews from authentic human-written ones, including those that have been refined using an LLM but still reflect human judgment. Across a dataset of over 20,000 peer reviews from ICLR and NeurIPS conferences, Sem-Detect improves over the strongest baseline by 25.5% in TPR@0.1% FPR in the binary setting. Moreover, in the three-class scenario, we empirically show that LLM refinement preserves the semantic signals of human reviews, which remain distinct from the patterns exhibited by fully AI-generated text; as a result, fewer than 3.5% of LLM-refined human reviews are misclassified as AI-generated.
arXiv:2602.17517v2 Announce Type: replace
Abstract: Augmented reality (AR) guidance in laparoscopic liver surgery requires accurate registration of preoperative 3D models to intraoperative 2D video, but remains challenging due to partial visibility, specularities, and tissue deformation. Existing methods often rely on contour-based rigid initialization and finite-element (FE) models for deformable registration, increasing modeling and engineering complexity. We present a depth-augmented, FE-free 3D--2D registration pipeline that combines robust rigid initialization with patient-specific non-rigid refinement. For rigid alignment, we adapt the RefineNet module of FoundationPose to laparoscopic liver scenes by using multi-class contour maps and monocular depth for relative pose refinement. For deformable alignment, we construct a patient-specific statistical deformation model from non-rigid ICP (NICP) correspondences and optimize pose and shape parameters using a coarse-to-fine L-BFGS-B strategy. On a public clinical laparoscopic liver dataset, the proposed method achieves a mean target registration error (TRE) of 14.73\,mm under a controlled manual-contour setting designed to isolate registration performance. Ablation studies show that monocular depth improves rigid initialization over contour-only inputs, while tumor-mapping analysis indicates that good surface alignment does not necessarily translate into lower target localization error. On an external dataset without ground truth, the method produces visually plausible overlays for qualitative assessment. These results suggest that depth-augmented pose refinement and FE-free statistical deformation modeling provide a promising alternative to FE-based pipelines for controlled 3D--2D liver registration in surgical AR.
arXiv:2605.21714v1 Announce Type: new
Abstract: We present AVI-HT, an adaptive visual-IMU fusion approach for tracking 3D hand poses by jointly modeling the egocentric image with on-glove 6-DoF IMU signals. AVI-HT achieves significantly improved accuracy and availability, particularly in hand-object interaction (HOI) scenarios involving heavy visual occlusion. Two complementary ingredients underpin its success: (1) synchronized multi-modal training data pairing on-body vision-IMU sensor streams with ground-truth 3D hand poses from a motion-capture system, and (2) a cross-sensor deep attention mechanism that adaptively modulates the trust assigned to the vision and individual IMU sensors. To evaluate AVI-HT in real-world settings, we conduct extensive experiments on our DexGloveHOI dataset that consists of 100K+ pairwise vision-IMU samples with synchronized 3D annotated poses, in which users manipulate a variety of objects during daily tasks. We compare against multiple single- and multi-modal tracking approaches under two hand models (UmeTrack, MANO). The results show that AVI-HT reduces mean keypoint error by 16.1% and its wrist-aligned variant by 24.2% over the baselines. Ablation studies further reveal the per-finger contribution of IMU sensors across activity types, and the model's sensitivity to IMU noise and temporal misalignment in vision-IMU fusion.
arXiv:2603.02938v2 Announce Type: replace
Abstract: Graph-based tasks in the zero-shot setting remain a significant challenge due to data scarcity and the inability of traditional Graph Neural Networks (GNNs) to generalize to unseen domains or label spaces. While recent advancements have transitioned toward leveraging Large Language Models (LLMs) as predictors to enhance GNNs, these methods often suffer from cross-modal alignment issues. A recent paradigm (i.e., Graph-R1) overcomes the aforementioned architectural dependencies by adopting a purely text-based format and utilizing LLM-based graph reasoning, showing improved zero-shot generalization. However, it employs a task-agnostic, one-size-fits-all subgraph extraction strategy, which inevitably introduces significant structural noise--irrelevant neighbors and edges--that distorts the LLMs' receptive field and leads to suboptimal predictions. To address this limitation, we introduce GraphSSR, a novel framework designed for adaptive subgraph extraction and denoising in zero-shot LLM-based graph reasoning. Specifically, we propose the SSR pipeline, which dynamically tailors subgraph extraction to specific contexts through a "Sample-Select-Reason" process, enabling the model to autonomously filter out task-irrelevant neighbors and overcome the one-size-fits-all issue. To internalize this capability, we develop SSR-SFT, a data synthesis strategy that generates high-quality SSR-style graph reasoning traces for supervised fine-tuning of LLMs. Furthermore, we propose SSR-RL, a two-stage reinforcement learning framework that explicitly regulates sampling and selection operations within the proposed SSR pipeline designed for adaptive subgraph denoising. By incorporating Authenticity-Reinforced and Denoising-Reinforced RL, we guide the model to achieve accurate predictions using parsimonious, denoised subgraphs for reasoning.
arXiv:2605.21719v1 Announce Type: new
Abstract: In this work, we address the problem of multi-robot adaptive coverage, where teams of robots perform dynamic sampling by continuously adjusting their positions to collect data in an environment. This task can be challenging, particularly when robots must be efficiently allocated to new sampling locations over time. Ergodic search methods optimize robot trajectories by ensuring that the robots' time-averaged spatial distribution aligns with the spatial distribution of environmental information. While these methods promote effective exploration provided a target distribution, they often fail to account for unknown prior distributions of the environment. To overcome this limitation, we propose an adaptive coverage strategy that utilizes real-time feedback from an environmental model to adjust robot sampling behavior in response to unknown conditions. Our approach enhances traditional ergodic trajectory optimization by constructing a target spatial information distribution based on parametric models of the environment, which are updated online. This strategy assumes that the environment is either static or changes slowly compared to the robot's motion. Our framework allows robots to dynamically prioritize regions of high interest, improving coverage efficiency, synthesizing effective control policies for individual agents, and optimizing resource use in settings with unknown prior distributions. We validate our approach through simulations, demonstrating its effectiveness in enhancing coverage and resource allocation.
arXiv:2603.04383v2 Announce Type: replace
Abstract: YouTube has evolved into a powerful platform where creators monetize their influence through affiliate marketing, raising concerns about transparency and ethics, especially when creators fail to disclose their affiliate relationships. Although regulatory agencies like the US Federal Trade Commission (FTC) have issued guidelines to address these issues, non-compliance and consumer harm persist, and the extent of these problems remains unclear. In this paper, we introduce tools, developed with insights from recent advances in Web measurement and NLP research, to examine the state of the affiliate marketing ecosystem on YouTube. We apply these tools to a 10-year dataset of 2 million videos from nearly 540,000 creators, analyzing the prevalence of affiliate marketing on YouTube and the rates of non-compliant behavior. Our findings reveal that affiliate links are widespread, yet disclosure compliance remains low, with most videos failing to meet FTC standards. Furthermore, we analyze the effects of different stakeholders in improving disclosure behavior. Our study suggests that the platform is highly associated with improved compliance through standardized disclosure features. We recommend that regulators and affiliate partners collaborate with platforms to enhance transparency, accountability, and trust in the influencer economy.
arXiv:2605.21726v1 Announce Type: new
Abstract: The generative nature of Large Language Models (LLMs) is reflected in the conditional probabilities they compute to sample each response token given the previous tokens. These probabilities encode the distributional structure that the model learns in training and exploits in inference. In this work, we use these probabilities to situate LLMs within the mathematical theory of stochastic processes. We use this framework to design a model-agnostic probabilistic token attribution measure, using Bayes rule to invert the next-token log-probabilities so as to capture the models internal representation of the distribution over token sequences. The representation is independent of the models computational structure. This representation yields the conditional probability of the response given the prompt, and of the response given the prompt with a token marginalized away. Our attribution score is the log of the ratio of these probabilities. We further compute the entropies of a single prompts token distributions, conditioned on the remaining context. The interplay between entropy and attribution score sheds light on LLM behavior. We evaluate 8 models across 7 prompts and investigate anomalies, token sensitivity, response stability, model stability, and training convergence, thereby improving interpretability and guiding users to focus on uncertain or unstable parts of the generation.
arXiv:2603.11642v2 Announce Type: replace
Abstract: Action chunking is widely used in generative visuomotor policies, yet the recurring execution discontinuities at chunk boundaries still lack a mechanistic explanation. This paper treats chunk-boundary artifact as an analyzable mechanism variable. We first show that successful and failed episodes separate stably on artifact metrics. We then show that, in stochastic action-chunked policies, fixing the observation context and changing only latent noise is sufficient to modulate artifact systematically. On the same Diffusion Policy checkpoint, comparisons among DDPM, zero-variance DDPM, and DDIM further show that this local controllability depends on whether the information path from initial noise to action output remains intact. Finally, from controlled interventions at fixed local execution states, we find that artifact changes can carry through to final outcome, and that the preferred direction can reverse even within the same task: some contexts achieve higher success under lower artifact, whereas others achieve higher success under higher artifact. In a representative high-artifact-favoring key context selected by held-out matched-continuation validation, success rate increases from 0.033 to 0.717. These results show that chunk-boundary artifact is not a mere execution-side by-product, but a variable in noise space that can be attributed, controlled, and mechanistically linked to task outcome.
arXiv:2605.21765v1 Announce Type: new
Abstract: The practical adoption of sampling-based inference (SAI) in Bayesian neural networks (BNNs) remains limited, partly due to persistent misconceptions about the feasibility and efficiency of sampling. This position paper argues that SAI has achieved computational parity with optimization-based methods and is at the verge of superseding such methods for effective and efficient inference in BNNs. This development should be in the interest of the whole community, promoting BNNs as a principled paradigm with its long-standing yet unfulfilled promise of providing principled uncertainty quantification for neural networks. SAI can even do more -- yielding superior prediction performance through model averaging, serving as the foundation for a plethora of possible downstream tasks, and providing crucial insights into the landscape of BNNs. In order to make such a change happen and unfold the potential of sampling, overcoming current misconceptions is a necessary first step. The next step is to realign research efforts toward addressing remaining challenges in SAI. In particular, the community must focus on two core problems: sufficient exploration of the posterior landscape and high-fidelity distillation of posterior samples for efficient downstream inference. By addressing conceptual and practical obstacles, we can unlock the full potential of SAI and establish it as a central tool in Bayesian deep learning.
arXiv:2602.17186v2 Announce Type: replace
Abstract: Large Vision Language Models (LVLMs) have achieved remarkable progress, yet they often suffer from language bias, producing answers without relying on visual evidence. While prior work attempts to mitigate this issue through decoding strategies, architectural modifications, or curated instruction data, they typically lack a quantitative measure of how much individual training samples or tokens actually benefit from the image. In this work, we introduce Visual Information Gain (VIG), a perplexity-based metric that measures the reduction in prediction uncertainty provided by visual input. VIG enables fine-grained analysis at both sample and token levels, effectively highlighting visually grounded elements such as colors, spatial relations, and attributes. Leveraging this, we propose a VIG-guided selective training scheme that prioritizes high-VIG samples and tokens. This approach improves visual grounding and mitigates language bias, achieving superior performance with significantly reduced supervision by focusing exclusively on visually informative samples and tokens.
arXiv:2605.05658v2 Announce Type: replace
Abstract: We propose a mixed quantum-classical hydrodynamic framework to model short-time inertial effects in the non-adiabatic evolution of a quantum solute coupled to a classical polar solvent. Drawing upon the work of Burghardt and Bagchi [Chem. Phys. 329 (2006), 343], we employ the Hamiltonian approach to incorporate consistent backreaction and preserve quantum decoherence beyond standard Ehrenfest dynamics. The solvent is treated as an ideal polar fluid and the quantum solute state is coupled to both the position and molecular orientation coordinates of the liquid. This approach retains essential solute-solvent correlations while significantly reducing the computational complexity of previous approaches. We further incorporate dissipative terms to capture both inertial effects and polarization relaxation. After establishing the general setting for non-local dielectric continua, the Marcus local approximation is integrated into the model thereby extending traditional solvation theory to account for collective fluid sloshing on fast timescales.
arXiv:2605.21921v1 Announce Type: new
Abstract: We study the problem of sampling weighted partial triangulations of a convex polygon. We consider the distribution where each partial triangulation $\sigma$ is chosen with probability proportional to $\lambda^{|\sigma|}$, where $\lambda>0$ is a model parameter and $|\sigma|$ denotes the number of diagonals in $\sigma$. This model belongs to a broad class of weighted geometric partition problems that include lattice triangulations and dyadic tilings, and is closely related to several classical combinatorial structures, including the full triangulations of a convex polygon and the associated Catalan structures. While prior work has largely focused on Markov chain approaches, often only providing suboptimal mixing time bounds, we provide a direct efficient method for exact sampling. Our main result is a randomized algorithm that outputs an exact sample from the target distribution in expected time $O\big((n\sqrt{\lambda}+1)\log n\big)$ for all sufficiently large $n$. This provides a nearly optimal sampling algorithm for weighted partial triangulations, offering a compelling alternative to Markov chain-based techniques.
arXiv:2605.21924v1 Announce Type: new
Abstract: On-policy knowledge distillation has proven effective for language models, yet its application to vision-language models (VLMs) remains underexplored. We observe that standard on-policy distillation can improve a student's output quality while failing to strengthen its reliance on visual input: on vision-critical tokens, the student's predictions remain largely unchanged whether or not fine-grained visual detail is present, even though the teacher's predictions depend heavily on it.To make this difference observable, we introduce visual advantage (VA), the token-level log-probability difference when the teacher scores a student-generated rollout with versus without access to fine-grained visual detail. VA is concentrated in a small minority of tokens, and these high-VA tokens are the ones that actually carry the visual supervision signal. This motivates a distillation objective that treats them differently from language scaffolding, so their contribution is not diluted by the abundant surrounding language tokens.We propose Visual-Advantage On-Policy Distillation (VA-OPD), which uses VA at two granularities: rollout-level reweighting by trajectory-averaged VA, and token-level KL averaged within high-VA and low-VA groups separately. We train on two math datasets (Geometry3K and ViRL39K) and evaluate on eight benchmarks covering both mathematical reasoning and visual understanding, across three teacher sizes (4B, 8B, and 32B) on the Qwen3-VL family. VA-OPD improves over standard on-policy distillation on every benchmark, with the gain growing monotonically along both the teacher-size and data-scale axes, suggesting that these factors compound consistently.
arXiv:2605.21926v1 Announce Type: new
Abstract: Unidirectional electromagnetic absorption provides a powerful approach for controlling light and heat, yet broadband realization in the infrared spectral region remains experimentally unexplored. Here, we report a non-Hermitian multilayer structure that enables robust broadband infrared unidirectional absorption. By combining low- and high-loss materials and engineering their thicknesses using the transfer-matrix formulation, the structure exhibits nearly perfect absorption spectrally matched to the blackbody radiation at 373 K under forward illumination, while suppressing backward absorption below 30%. Spectral analysis indicates that the observed unidirectionality originates from non-Hermitian physics near an exceptional point. Notably, broadband unidirectional absorption is achieved even without strict exceptional-point condition. This indicates that the observed unidirectionality is governed by the combination effect of loss distribution and optical interference, rather than a singular condition, ensuring robustness against film thickness variations. Furthermore, thermal shielding experiments demonstrate that the structure enables unidirectional control of thermal radiation, resulting in a temperature difference of up to 21 0C between forward and backward configurations. These results establish a robust strategy for broadband directional control of infrared radiation, with potential applications in passive thermal management, including thermal smart windows and infrared heat-shielding devices.