Computation and Language
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Sunday 6 September
An LLM judge evaluates outputs at scale. Experts should label only where it is least sure. Its natural escalation signal conflates two uncertainties: aleatoric, real disagreement in the expert pool, which labels cannot reduce, and epistemic, the judge's ignorance, which labels do reduce. A small Bayesian model separates them: a regression on labels already collected learns how far to trust a black-box judge's prediction. Both components follow as simple formulas, with no sampling or further judge calls. The components isolate on a real LLM judge against exactly known truth, and stated confidence is no guide to its actual error. On real human disagreement (ChaosNLI) the epistemic ranking removes 83% more error than total uncertainty for the same expert labels, though simply escalating the least-labelled items does as well there. We demonstrate we can estimate where a judge is ignorant rather than where experts genuinely disagree, and propose using this to direct expert labelling. Code and data are available at https://github.com/composo-ai/judge-uncertainty-decomposition.
Programming Languages
Expressive power of one-shot control operators and coroutines
Control operators, such as exceptions and effect handlers, provide a means of representing computational effects in programs abstractly and modularly. While most theoretical studies have focused on multi-shot control operators, one-shot control operators—which restrict the use of captured continuations to at most once—are gaining attention for their balance between expressiveness and efficiency. This study aims to fill the gap. We present a mathematically rigorous comparison of the expressive power among one-shot control operators, including effect handlers, delimited continuations, and even asymmetric coroutines. Following previous studies on multi-shot control operators, we adopt Felleisen's macro-expressiveness as our measure of expressiveness. We verify the folklore that one-shot effect handlers and one-shot delimited-control operators can be macro-expressed by asymmetric coroutines, but not vice versa. We explain why a previous informal argument fails, and how to revise it to make a valid macro-translation.
Read the full abstract
Combinatorics
A proof of the maximum Laplacian energy conjecture for connected graphs via a sharp eigenvalue-sum bound
Let $S_k(G)$ denote the sum of the $k$ largest Laplacian eigenvalues of a connected graph $G$ of order $n$ and size $m$. Write $\mathrm{PA}_{n,\omega}$ for the graph obtained from an $\omega$-vertex clique by attaching $n-\omega$ pendant vertices to one of its vertices, and set \[ M_{n,k}:=\binom{k+1}{2}+n-k-1, \] the number of edges of $\mathrm{PA}_{n,k+1}$. For $n/2<k\le n-2$, we prove the sharp bound \[ S_k(G)\le \frac{2k}{n}m+ \frac{2(n-k)}{n}M_{n,k}-(n-k-1), \] with equality attained by $\mathrm{PA}_{n,k+1}$. This bound is complementary to Brouwer's inequality and is strictly stronger when $m<M_{n,k}$. Combining our bound with Brouwer's inequality, we resolve and strengthen a conjecture of Vinagre, Del-Vecchio, Justo, and Trevisan: for every $n$, the pineapple $\mathrm{PA}_{n,1+\lfloor2n/3\rfloor}$ maximizes the Laplacian energy among all connected graphs of order $n$; moreover, for $n>4$, it is the unique maximizer.
Read the full abstract
Information Theory
Random Algebraic Geometry Codes Approach the Half-Singleton Bound for Insertions and Deletions
In this paper, we study the performance of algebraic geometry (AG) codes against adversarial insertion-deletion (insdel) errors. The half-Singleton bound states that an $[n,k]_q$ linear code can correct at most $n-2k+1$ insdel errors. It was recently proven that random Reed-Solomon codes approach this bound. However, these constructions require the field size $q$ to grow linearly with the code length $n$. We overcome this barrier by extending the probabilistic analysis of general linear insdel codes to AG codes. We demonstrate that curves with many rational points allow for nearly optimal codes over significantly smaller alphabets. We prove the following main asymptotic results: (1) For general smooth complete curves of fixed genus, random AG codes are nearly optimal, that is, they can correct $(1-\varepsilon)n-2k$ insdel errors with high probability over linear-sized fields ($q=\Theta(n)$). (2) By utilizing Hermitian curves, we achieve this optimality over sublinear fields of size $q=\Theta(n^{2/3})$, breaking the linear field size barrier. (3) Using asymptotically optimal García-Stichtenoth towers, we prove the existence of random AG codes that approach the half-Singleton bound with high probability over fields of size $q=2^{O_R(1/\varepsilon^2)}$, independent of $n$.
Read the full abstract
Differential Geometry
The topological Ax–Lindemann theorem for the complex hyperbolic space
In the case of the complex hyperbolic space, we establish a version of the hyperbolic Ax–Lindemann Theorem for the topological closure rather than the Zariski closure. Our approach combines classical differential geometry with homogeneous dynamics.
Algebraic Geometry
The Constant in Thomae-Type Formulas for Eight Points on the Complex Projective Line
We consider the family of cyclic fourfold covers $C(x)$ of the complex projective line branched at eight points $x=(x_1,\ldots,x_8)$. It is known that the period map identifies the configuration space $X(2,8)$ of the branch points with a Zariski open subset of a quotient of the five-dimensional complex ball, and that the inverse of the period map is expressed projectively by $105$ automorphic forms $f_J$, which are proportional to the signed branch-point polynomials $\hat x_J$ with a common scalar factor. We normalize an element $\omega$ of the one-dimensional eigenspace of the deck transformation group acting on the holomorphic differentials on $C(x)$. For the period $\eta$ of $\omega$, we determine the common scalar factor: it is the product of a constant expressed in terms of the gamma function and the square of a quadratic form in $\eta$.
Read the full abstract
Computation and Language
InsightChain: Optimized Chain-of-Insight Analytics for LLM-driven Data Visualization
Large language models (LLMs) are increasingly used for automated data visualization, yet existing approaches often frame visualization generation as a single-step mapping from user query to figure or code, overlooking the iterative analytical reasoning process of expert analysts. We present InsightChain, a four-stage visualization prompting pipeline (Explore–Focus–Test–Present) that emulates expert analytical workflows, together with VG-COPRO, a vision-guided automatic prompt optimization (APO) method adapted to jointly optimize such multi-stage, executable pipelines. To address the evaluation gap for complex data visualization, we introduce the Insight Progression Metric (IPM), a rubric combining four text-based dimensions with a vision-based dimension. We assess IPM through a 100-chain human pilot and an expanded 300-chain agent-based evaluation spanning all ten domains. Experiments on public datasets show that InsightChain consistently outperforms competing prompting baselines. Existing APO methods fail to yield consistent gains on this multi-stage task, whereas VG-COPRO improves performance in both in-domain and cross-domain settings.
Read the full abstract
Computer Vision
Large-Scale Pretraining for Improving Deep Learning-Based Geometric Distortion Correction of Diffusion-Weighted Imaging
Diffusion-weighted imaging (DWI) is widely used in clinical settings but remains vulnerable to geometric distortion. Conventional correction methods often require additional acquisitions or vendor-specific solutions, limiting their feasibility in high-throughput, resource-constrained settings. This study investigates whether large-scale pretraining strategies can improve deep learning-based distortion correction for single-phase-encoding DWI. We formulate the task as image reconstruction, and compare a non-pretrained baseline against a self-supervised and a generative pretrained model, evaluated using both quantitative image-similarity metrics and qualitative expert assessment. The best-performing model was further tested for transferability on data collected in an LMIC setting with acquisition shift. Pretrained models outperformed the non-pretrained baseline, with cWDM achieving the strongest results across both quantitative and qualitative evaluation. However, application to LMIC data revealed transferability challenges, including contrast alteration and over-reliance on T1-weighted anatomical structure. Registering images to a common standard space improved predictions, suggesting that harmonized preprocessing may enhance cross-domain deployment.
Read the full abstract
Computer Vision
PLSR: Progressive and Localized Super-Resolution of 3D Objects via Localized Latent Voxel Diffusion
High-resolution 3D asset generation is vital in various 3D applications. Existing state-of-the-art diffusion-based models remain constrained by fixed resolutions, limiting their ability to produce details. In this paper, we tackle the challenge of generating more detailed, higher-resolution 3D objects by introducing a 3D super-resolution (SR) framework built on existing 3D generative foundation models. To this end, we design PLSR, a progressive and localized super-resolution solution to achieve this goal effectively and memory efficiently. Technically, given a coarse geometry from a pretrained 3D generator, we decompose the global SR task into localized sub-tasks via an associative input decomposition scheme, adapt a flow-based 3D generator into a localized super-resolution model through low-cost finetuning, and unify them in an iterative patch-wise denoising pipeline for seamless high-resolution output. Experiments on challenging objects show that our approach is able to generate 3D details with new strong fine-detail fidelity while significantly reducing the computational cost, offering a new and practical solution for high-resolution 3D asset generation.
Read the full abstract
Metric Geometry
Equilateral dimension of the Heisenberg group $\fH^n$
For every $n\geqslant1$, the equilateral dimension of the Heisenberg group $\fH^n=\C^n\times\R$ with the Korányi metric is $2n+2$.
Human-Computer Interaction
Can People Distinguish Human and AI Agency in Humanoid Teleoperation? A Preliminary Study of Agency Perception
Can people distinguish between human and AI agency in humanoid teleoperation? To explore this question, we developed Ghost-in-the-Loop, a teleoperation framework that supports both human-operated and AI-generated control of a robot's voice, facial expressions, and gestures while maintaining a consistent embodiment. We conducted a preliminary online study ($N=50$) in which participants viewed short interaction clips generated by either a Human Operator or an AI Control and judged the perceived source of control. Results suggest that participants often struggled to distinguish between the two conditions in brief interactions. Qualitative responses indicate that judgments were primarily influenced by perceived naturalness, temporal coordination, and consistency across speech, facial expression, and gesture. These findings provide initial insights into agency perception in embodied human–AI communication and motivate future investigations of blended human–AI telepresence systems.
Read the full abstract
Robotics
Collision Snapshot Guided Time-Reversed Safety-Critical Scenario Generation
The generation of safety-critical traffic scenarios is essential for training and evaluating autonomous vehicles. Prior approaches typically perturb the trajectories of existing agents in a traffic scenario using simplified adversarial objectives to induce safety-critical interactions, which can limit the plausibility and diversity of the generated scenarios. Although inserting new adversarial vehicles can alleviate this limitation, determining when and where to introduce them in a scenario-specific manner remains challenging. In this work, we introduce \underline{CO}llision \underline{S}napshot guided \underline{T}im\underline{E}-\underline{R}eversed safety-critical scenario generation (COSTER), a framework that leverages learned traffic priors to determine plausible collision times and locations. COSTER first constructs a collision snapshot by inserting a new vehicle in contact with the target vehicle at the identified collision state within a traffic scenario. Starting from this collision snapshot, a conditional variational autoencoder is used to perform a time-reversed rollout, reconstructing the trajectory of the inserted vehicle backward toward earlier timesteps. Experiments show that COSTER outperforms existing methods in plausibility, diversity, and data efficiency. Moreover, agents trained on COSTER-generated scenarios reduce collision rates by 31% on safety-critical scenarios from the Waymo Open Motion Dataset while also improving ego task completion. The project website is available at https://anonym-121.github.io/COSTER/.
Read the full abstract
Information Theory
From Tsallis to KL: Convergence and Error Estimates for Tsallis-Regularized Optimal Transport
We study the Tsallis-to-Kullback–Leibler (KL) limit for entropy-regularized optimal transport with nonnegative bounded continuous costs. Fixing the regularization parameter $\varepsilon > 0$, we first derive an exact variational reformulation of Tsallis-regularized optimal transport in terms of the Tsallis information projection onto the set of couplings. The formula isolates an explicit correction term and thereby explains why, unlike in the KL case, the regularized transport problem and the corresponding information projection problem do not coincide exactly. We also establish existence and uniqueness for the Tsallis information projection. We then prove, with respect to the narrow topology, the $\Gamma$-convergence of the Tsallis-regularized functionals to the KL-regularized functional as $q\downarrow1$, together with narrow convergence of their unique minimizers. Finally, we obtain explicit error estimates of order $O(q-1)$ for both the regularized optimal transport values and the associated information projection values. These results quantify the passage from Tsallis regularization to the classical KL setting and clarify the relation between entropic regularization and information projection for $1 < q \leq 2$.
Read the full abstract
Symplectic Geometry
A contact-geometric variant of Smale's contraction
Motivated by Smale's contraction of the compactly supported diffeomorphism group of the two-disc, we study contact forms on an open rotationally symmetric Darboux ball that agree with the standard contact form outside a compact set and whose Reeb flows have no trapped orbits. Such forms are called vertically convex. In dimensions three and five, we identify the quotient of their space by compactly supported diffeomorphisms with a contractible monodromy space and prove an equivariant product splitting. Consequently, the orbit of the standard contact form is a strong deformation retract. It follows that the space of vertically convex contact forms is contractible in dimension three and homotopy equivalent to the compactly supported diffeomorphism group, hence connected, in dimension five. The analogous subspace of forms defining the standard contact structure has the homotopy type of the corresponding compactly supported contactomorphism group in dimension five and is contractible in dimension three.
Read the full abstract
Machine Learning
Stability and Generalization of Straight-Through Estimators for Training Two-Layer Quantized Neural Networks
We study the identity straight-through estimator (STE) for training a two-layer binary-activation network with hinge loss from the perspective of Statistical Learning Theory (SLT). Our central question is whether algorithmic stability can explain the statistical generalization of the estimator produced by the discontinuous STE training rule. In the saturated-output regime, the zero-initialized samplewise STE recursion is exactly the stochastic subgradient descent on the convex latent loss $(-yu^\top x)_+$. This representation makes a stability analysis possible. We derive an exact distance identity for two coupled updates and prove approximate non-expansiveness of the common-example map, with a quadratic defect only when the two latent margins straddle zero. We then obtain explicit $\ell_2$ on-average model-stability and generalization bounds, transferring stability isometrically from the latent vector to the full first-layer matrix. Combining stability with a standard optimization bound yields an explicit excess induced-risk guarantee and the rate $O(n^{-1/2})$ when $T=n^2$. Under margin separability, a complementary argument gives the optimal-order $O(R^2/(\gamma^2n))$ expected excess misclassification error for a randomized one-pass STE iterate and a corresponding majority-vote bound.
Read the full abstract
Signal Processing
Active STAR-RIS-Aided RSMA IoT Systems: Performance Analysis, Model-Based and Data-Driven Resource Allocation Frameworks
STAR-RISs have emerged as a promising technology for achieving full-space signal coverage with low hardware complexity and energy consumption. In parallel, RSMA offers a flexible interference management mechanism that enhances spectral efficiency and user fairness. Integrating STAR-RIS with RSMA provides strong potential for robust and energy-efficient communications; however, the resulting double-fading cascaded channels and the inherent imbalance between near and far users pose reliability and fairness challenges. This study presents comprehensive information-theoretic and optimization frameworks for active STAR-RIS-assisted RSMA Internet-of-Things systems. Closed-form expressions for the OP and EC are derived under both the presence and absence of direct links. Based on these results, asymptotic OP and EC, diversity order, and array gain are analyzed to characterize system behavior. In addition, throughput-based and spectral-based energy efficiency metrics are investigated to quantify the tradeoffs between transmission performance and power consumption. To enhance user fairness, a fairness-oriented RA framework is developed to jointly optimize power allocation and rate-splitting coefficients using successive convex approximation and block coordinate descent techniques. To further improve scalability and real-time applicability, we introduce a data-driven RA framework based on DMNN, CMNN, and MXGB. This framework approximates the optimization-based solutions with substantially reduced computational complexity. Numerical results show that the proposed MXGB model achieves the shortest execution time while maintaining performance comparable to ground-truth-based benchmarks. Extensive Monte Carlo simulations validate the analytical results and demonstrate the effectiveness of the proposed frameworks.
Read the full abstract
Category Theory
A two-categorical Snake Lemma
We prove a Snake Lemma for 2-categories. Working in a 2-category with a strong bizero object, we develop 2-kernels and 2-cokernels, 2-monomorphisms and 2-epimorphisms as fully faithful and cofully faithful 1-cells, short 2-exact sequences and normal image factorisations, and we prove a two-dimensional Normal Short Five Lemma. We then introduce the dinversion of an antinormal pair and the notion of a homologically self-dual 2-category, characterised equally by the self-duality of homology, by a Pure Snake Lemma and by a Third Isomorphism Property. Two-dimensional di-exactness implies homological self-duality. Our main result is the Snake Lemma in a 2-di-exact 2-category: a ladder of 2-exact rows with normal verticals induces a 2-exact six-term sequence, with a connecting 1-cell that is 2-natural in the ladder. Di-exactness can be traded for two hypotheses that are not self-dual: that dinversion preserve normality, and that normal 2-epimorphisms compose. Everything specialises, on passing to a locally discrete 2-category, to its classical counterpart. We close by exhibiting three models: a 2-di-exact 2-category of abelian categories containing $\mathsf{Coh}(X)$ for every noetherian scheme $X$; the locally ordered 2-category of complete modular lattices, in which 2-di-exactness amounts to Dedekind's transposition principle; and the 2-category of Hilbert lattices, which is not 2-di-exact but satisfies the non-self-dual hypotheses, by a theorem of Mackey on pairs of closed subspaces.
Read the full abstract
Networking
The LQ Digital Mode Family: Protocol Architecture and Reference Specification for Weak-Signal Communications
This paper introduces LQ8, an open-source weak-signal digital mode for amateur radio that completes structured two-way contacts in 4 transmissions (60 s), achieving a 1.50x speedup over canonical 6-transmission exchanges (90 s) and 1.25x over 5-transmission RR73 exchanges (75 s) while preserving complete bidirectional exchange of Maidenhead grid locators, signal reports, and mutual acknowledgments. By replacing fixed-width message-type headers with variable-length prefix codes matched to physical-layer bit budgets, LQ8 packs the answering station's grid locator and measured signal report into a single 77-bit reply. In multi-station pileup operations, LQ8 confirms up to two answering stations simultaneously within a single-carrier 50.0 Hz transmission with a 0.0 dB RF power-splitting penalty, achieving a peak rate of 240 QSOs/h in continuous pileup (4x over single-carrier FT8; 480 QSOs/h dual-carrier) while retaining -21.0 dB SNR sensitivity. The protocol family also defines three extended profiles (LQ16, LQ4, and LQ2) tailored for extreme sensitivity, VHF/UHF contesting, and fast-burst channels.
Read the full abstract
Machine Learning
Sparse Oblique Rule Boosting for Simpler Additive Rule Ensembles
Small additive ensembles of symbolic rules offer interpretable prediction models. Traditionally, these ensembles use rule conditions based on conjunctions of simple threshold propositions $x \geq t$ on a single input variable $x$ and threshold $t$, resulting geometrically in axis-parallel polytopes as decision regions. While this form ensures a high degree of interpretability for individual rules and can be learned efficiently using the gradient boosting approach, it relies on having access to a curated set of expressive input features so that a small ensemble of axis-parallel regions can describe the target variable well. Absent such features, reaching sufficient accuracy requires increasing the number and complexity of individual rules, which diminishes the interpretability of the model. Here, we extend classical rule ensembles by introducing logical propositions with learnable sparse linear transformations of input variables, i.e., propositions of the form $\mathbf{x}^T\mathbf{w} \geq t$, where $\mathbf{w}$ is a learnable sparse weight vector, enabling decision regions as general polyhedrons with oblique faces. We propose a learning method using gradient boosting based on a weighted logistic regression. Empirical results across 14 regression and classification tasks demonstrate that the proposed method achieves lower model complexity than competitive baselines while maintaining similar or better predictive accuracy. Hence, the approach provides a favorable trade-off between interpretability and accuracy and reduces the reliance on manual feature engineering.
Read the full abstract
Differential Geometry
A complete classification of left-invariant Einstein metrics on $S^3\times S^3$
We complete the classification of compact simply connected homogeneous Einstein manifolds in dimension six by resolving the remaining cases for left-invariant Einstein metrics on $G=\mathrm{SU}(2)\times\mathrm{SU}(2)\cong S^3\times S^3$. Previous work leaves two cases for the isotropy group $K$, namely $K=\{e\}$ and $K\cong\mathbb Z_2$. We first rule out $K=\{e\}$. Then we show that any $\mathbb Z_2$ symmetry generated by an inner involution $\sigma$ with $\tr\sigma=-2$ necessarily extends to a $\mathbb Z_2\times\mathbb Z_2$ symmetry. Together with the previously known classification results, these theorems imply that every left-invariant Einstein metric on $G$ is, up to homothety and isometry, either the standard product metric $g_{\rm can}$ or the Jensen nearly Kähler metric $g_{\rm NK}$.
Read the full abstract
Robotics
OVMAN: A Task and Benchmark for Open-Vocabulary Motion-Aware Navigation
Homes change between a robot's visits. Navigation benchmarks pose their goals in the world the agent currently sees, and the two-visit benchmarks that exist score recall or rearrangement rather than navigation. None of them can express go to the chair that was moved or go to where the vase used to be. OVMAN is a task in which an agent tours a scene, returns after a scripted change, and must navigate to a goal specified by the change itself. Two of its six change relations answer with a place an object has left, where nothing remains to be detected. We release 219 two-visit episodes, each certified solvable by an oracle agent that completes it three times. Two released systems fail as predicted. A zero-shot object-goal navigator reaches the change-defined target in 12.3% of episodes and almost never reaches a vacated location, 0.026 on former and 0.050 on removed. When a self-maintaining open-vocabulary map is read as two visits rather than as one maintained map, overall success rises (0.396 to 0.479 on our mapping) but the past-position relations specifically do not recover, on our maps or on a released self-maintaining one; keeping a map current is therefore not the whole obstacle. A simple two-visit reference agent reaches 45.2% when navigating, against an embodied oracle of 99.5%. An error decomposition places the remaining difficulty in carrying an instance identity across visits rather than in naming it or in selecting the answer once positions are known.
Read the full abstract
High Energy Physics (Phenomenology)
A Mean-Field Approach to the Dielectric Response of Bulk Superconductors for Light Dark Matter Direct Detection
The dielectric function is central to describing many-body screening effects in dark matter (DM) direct detection with condensed matter targets. Current superconducting detector analyses employ the free-electron Lindhard dielectric function to model in-medium effects, an approximation whose validity in the superconducting state remains untested. We derive the electronic dielectric function for bulk superconductors within the Bardeen-Cooper-Schrieffer (BCS) framework, incorporating the full Bogoliubov quasiparticle coherence factors in the random-phase approximation. A systematic comparison with the Lindhard function for aluminum and tungsten silicide (WSi) reveals good agreement for energy depositions $\omega\gtrsim5\Delta$, establishing the Lindhard function as a robust approximation for superconducting DM detectors operating in this regime.
Read the full abstract
Statistical Finance
Asymmetric Long-Memory GARCH: Sign-Dependent Kernel Injection in a Two-Dimensional Markov Chain
We introduce ALM-GARCH, an asymmetric long-memory GARCH model in which positive and negative innovations enter conditional variance with different injection amplitudes and kernel offsets. These departures define testable level and memory channels relative to a nested symmetric benchmark. Positive Harris recurrence holds for interior configurations under a Foster-Lyapunov condition. Across five equity indices and Bitcoin, joint symmetry is rejected throughout, driven primarily by the level channel. The memory channel is supported for the Nikkei 225, KOSPI, and Bitcoin but is weakly identified when the positive branch is nearly inactive. Out-of-sample performance is broadly comparable to standard benchmarks.
Read the full abstract
Machine Learning
On BatchNorm Forward Modes in Value-Based Reinforcement Learning
Batch normalization (BN) substantially improves sample efficiency in continuous-control actor-critic methods such as CrossQ, yet recent studies report performance degradation in discrete-action value learning on Atari. These failures are surprising because discrete Q-networks lack the action-input distribution mismatch identified by CrossQ. We show for target-based C51 and target-free PQN that the simple choice between running and batch statistics at specific forward passes can reverse this degradation. In C51, switching the BN bootstrap forward to batch-statistic mode significantly improves performance over unnormalized and LayerNorm baselines and scales stably with update-to-data ratios up to 12. In PQN, using batch-statistics for both action selection and bootstrapping recovers performance from the failing running-statistic configuration. Across 26 Atari games at 400M frames, this configuration achieves a higher final aggregate score than PQN with LayerNorm. Our results show that carefully configured BN can substantially improve discrete-action value learning, and that its forward protocols are an essential part of the algorithm specification.
Read the full abstract
Optimization and Control
Fast Rates and Strong Convergence of Tikhonov-Regularized Mixed-Order Primal-Dual Dynamics for Linearly Constrained Optimization Without Eventual Ball Conditions
In this paper, we study a Tikhonov-regularized mixed-order primal–dual dynamical system with implicit Hessian damping for linearly constrained convex optimization problems in finite-dimensional Euclidean spaces, where the primal equation is second order and incorporates the viscous damping term \(\delta\sqrt{\varepsilon(t)}\,\dot x(t)\), whereas the multiplier equation remains first order. By constructing a new class of energy functions, for a general Tikhonov regularization coefficient \(\varepsilon(t)\), we prove the strong convergence of the primal trajectory and derive fast convergence rates under the same parameter assumptions, without imposing any eventual inside/outside-ball condition. More precisely, the primal trajectory converges to the minimum-norm solution, and the multiplier converges to a compatible KKT multiplier, while the convergence rates of the Lagrangian gap, feasibility violation, and objective residual are \(o(\varepsilon(t))\), and the convergence rate of the velocity norm is \(o(\sqrt{\varepsilon(t)})\). For the critical case \(\varepsilon(t)=c/t^2\), in which the damping coefficient \(\delta\sqrt{\varepsilon(t)}\) reduces to \(\delta\sqrt{c}/t\), we establish the sharper convergence rates \(o(t^{-2})\) for the Lagrangian gap, feasibility violation, and objective residual, together with \(o(t^{-1})\) for the velocity norm, which improve the corresponding \(O(t^{-2})\) and \(O(t^{-1})\) decay estimates obtained in the related literature. Most importantly, when the proposed dynamical system is specialized to the finite-dimensional unconstrained setting, our analysis answers the open question on strong convergence in this critical regime posed by Attouch and László [Math. Methods Oper. Res., 99 (2024), pp.~307–347].
Read the full abstract
Computer Vision
Separating Capability from Confidence: Grounded Dual-State Calibration for GRPO-Trained Medical Vision-Language Models
Medical vision-language models (VLMs) require confidence that reflects both answer correctness and patient-specific visual evidence. Recent GRPO-based methods optimize verbalized confidence together with answer generation. However, this joint optimization may interfere with answer learning and drive confidence toward near-binary values. Verbalized confidence also provides no explicit assessment of visual support. We therefore separate capability learning from confidence estimation and propose DualRead. DualRead builds on the insight that reliability can be read from the actor's internal states at critical moments in the answering process. It freezes the GRPO-trained actor and combines pre-answer solvability with a post-answer assessment of the generated answer and its visual support. To further assess whether confidence reflects visual grounding, we introduce Counterfactual Confidence Grounding AUC (CCG-AUC). It measures whether confidence decreases when real-image substitution changes the actor from correct to incorrect. Across two VLM backbones and both in- and out-of-distribution medical VQA benchmarks, DualRead improves correctness discrimination and calibration over verbalized confidence while preserving answer accuracy. CCG-AUC reveals whether confidence responds to answer-relevant visual evidence rather than primarily to non-visual cues.
Read the full abstract
Numerical Analysis
Sharp approximation rates for shallow ReLU neuralnetworks on critical Besov classes
Let $\mathbb D$ be the normalized ridge dictionary generated by $\operatorname{ReLU}^k$ on a bounded Lipschitz domain $\Omega\subset\mathbb R^d$. We determine the sharp algebraic rate of finite $n$-term approximation from $\mathbb D$ when the outer $\ell^1$ coefficient budget is independent of $n$. More precisely, let $d\ge3$, $k\in\mathbb N_+$, $0\le m\le k$, $0<p<1$, and $0<q\le1$. For the unit ball of the critical Besov space $B_{p,q}^{k+d/p}(\Omega)$, measured in $H^m(\Omega)$, the optimal algebraic exponent is \[ \min\left\{\frac{k-m+d/2}{d-1}, \frac{k-m+d/2+1/p-1/2}{d}\right\}. \] We prove two-sided estimates in the three regimes determined by the two branches of this minimum and give the corresponding logarithmic factors at and beyond the transition. The estimates coincide without logarithmic loss in the strict angular regime and at the transition when $q\le[1/2+(k-m+d/2)/(d-1)]^{-1}$. The upper estimate combines critical wavelet sparsity, localized Fourier–Radon representations, and a stable allocation of directions and biases. The lower estimates arise from two distinct obstructions, namely radial ridge approximation on the direction sphere and Gevrey localization in the joint direction–bias space. The critical smoothness is exactly the endpoint at which Besov regularity provides a uniform shallow-network variation bound without additional scale decay. The resulting representation exponent is strictly larger than the degree-limited exponent for fixed-degree isotropic finite elements under a parameter-count comparison. This last statement concerns best approximation, not training or computational complexity.
Read the full abstract
Materials Science
Machine learning for the design and prediction of soft-magnetic electromagnetic shielding FeCo-based alloys in laser cladding
Electromagnetic shielding materials play a pivotal role in both aerospace applications and daily life. However, their design and manufacturing still face persistent challenges. Machine learning demonstrates significant potential in accelerating material development and compositions optimization. Furthermore, laser additive manufacturing provides powerful technical support for fabricating multi-component, multifunctional electromagnetic shielding materials with tailored properties. In this study, the multiple machine learning strategies have been proposed, based on experimental derivation and soft magnetic material databases, to accelerate the design of multifunctional FeCo-based alloys for electromagnetic interference (EMI) shielding within an almost infinite compositional space. This work presents a novel approach for the rapid and automated discovery of multifunctional alloys with optimized EMI shielding effectiveness, as well as enhanced magnetic and electrical properties.
Read the full abstract
Algebraic Geometry
Monodromy Eigenvalues of Milnor Fibers for Line Arrangements
It is an important problem to know whether the monodromy on the cohomology of Milnor fibers associated to hyperplane arrangements is a combinatorial invariant. In this paper, we obtain a combinatorial vanishing criterion for certain eigenspaces of this algebraic monodromy in line arrangements. Combining with Hirzebruch inequality, which is a consequence of Bogomolov–Miyaoka–Yau inequality, we prove that for essential complex line arrangements, the eigenvalues of the monodromy have orders at most five. This is a partial progress toward Papadima–Suciu conjecture and proves Salvetti–Serventi connectivity conjecture. For essential complexified real line arrangements, the monodromy order is improved to at most four thanks to Shnurnikov's inequality. This confirms Papadima–Suciu conjecture for real line arrangements and also Yoshinaga's sharp pair conjecture.
Read the full abstract
Quantum Physics
After Theft: From Revocation to Neutralization in the Custody of Quantum Clones
Encrypted quantum cloning enables the creation of multiple encrypted clones of an unknown quantum state while allowing only one effective decryption, with the decryption resource being intrinsically consumed in the process. In this paper, we argue that this property supports a distinctive custodial security primitive for post-compromise response. We consider a threat model in which an adversary steals an encrypted quantum clone but does not yet possess the corresponding decryption key. In such a scenario, a legitimate custodian may unseal a different trusted clone, thereby exhausting the sole available unsealing opportunity and rendering the stolen clone permanently useless. We argue that this mechanism is not adequately described as mere revocation. Rather, it realizes a stronger form of post-theft response, which we call neutralization. We formalize this distinction, locate it within a broader post-theft response space, and introduce a temporal threat model. We compare the mechanism with its closest classical analogue, showing that the classical case can reproduce the policy outcome only through external procedural composition. We finally interpret encrypted quantum cloning as a primitive for post-compromise quantum custody in distributed preservation settings, with prospective relevance for Cyber-Humanities-oriented preservation architectures.