Peer Revenue Now accepting positions

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Saturday 7 November

Optics

Precision-informed Fourier compression of refractive index tomograms with voxel-wise error control

Significance: Holotomography (HT) generates large refractive index (RI) volumes and time series, creating a need for compact storage with controlled numerical fidelity. Aim: We sought to reduce storage by directly encoding reconstructed three-dimensional RI volumes in Fourier space with voxelwise validation against an assigned error budget. Approach: Quantized Fourier coefficients represent volumetric structure, while spatial residual quantization uses a 1% safety margin without storing separate corrections for individual voxels. We evaluated cellular and organoid data, concurrent processing, and extensions to dielectric-tensor and absorption tomography. Results: At a budget of 0.0001 RI units, complete packets achieved compression ratios of 7.89-9.50 relative to float32 arrays and were 28-30% smaller than BitRound-Pcodec packets satisfying the same error limit. Concurrent processing reduced encoding and decoding wall times by factors of 5.14 and 4.81 without changing packets. The fixed margin passed all primary-budget tests but failed at one smaller budget in both precision-sweep inputs. Conclusions: Precision-informed volumetric Fourier coding reduces storage of holotomograms with verified voxel-wise fidelity at the primary budget, at additional computational cost.
Read the full abstract

All authors Dohyeon Lee, Juheon Lee, Sehyeon Lee, Juyeon Park, YongKeun Park

High Energy Astrophysical Phenomena

Self-bound quantum dark fermion stars: saturation, a coupling-capped maximum mass, and tidal signatures of a vector dark sector

I show that a dark sector whose fermions couple to both a scalar and a vector mediator saturates in close analogy with nuclear matter: the energy per particle develops a minimum at finite density, cold dark matter of this kind is self-bound, and its stars, computed here in full general relativity, possess the sharp surfaces of strange quark stars. Saturation makes the maximum stellar mass a property of the Lagrangian rather than an inheritance of the Chandrasekhar scaling: it rises eightfold across the saturation window, which a critical vector coupling closes, so the theory caps its own maximum mass, at $3.9\,M_\odot$ for a fermion of $1\,\mathrm{GeV}$ and the fiducial saturation density, a ceiling shared by the whole family of couplings rather than by any single parameter set. Metastable couplings support two disconnected stable families, one dilute and gravitationally bound, the other dense and self-bound. The quantum-pressure correction, confined by saturation to a surface layer a few Compton wavelengths thick, governs the smallest dark nuggets whilst shifting stellar masses by less than one part in $10^{21}$. A $1.4\,M_\odot$ member of the fiducial family has a radius of $13.3\,\mathrm{km}$, a compactness of $0.155$, and a tidal deformability of $1.6\times 10^{3}$, almost an order of magnitude above the measured neutron-star value, so at a given particle mass and couplings a single well-measured inspiral distinguishes these stars from neutron stars and from their scalar-only counterparts, whilst the deformability surviving at the maximum mass separates them, in principle, from black holes.
Read the full abstract

General Relativity and Quantum Cosmology

Black-hole ringdown at an exceptional point

Black holes ring in damped oscillations. At an exceptional point, photon-sphere and near-horizon modes merge, and their response can acquire a linear time factor. We compute its amplitude from constrained Einstein-Maxwell data for a rotating, charged black hole in de Sitter space. With seven competing modes fixed, the late-time curvature error is 0.00383$\%$ for the pair and 0.01401$\%$ for a refitted exponential. We suppress the linear term by tuning the initial disturbance at fixed geometry. Hence, near coalescence, changing only the initial disturbance can shift the frequency inferred from a short ringdown fit.
Read the full abstract

Computer Vision

Dynamics-Informed Action Learning for Goal-Conditioned Latent Planning

Latent world models enable visual control by planning action sequences in compact representation spaces, but a gap remains between how latent dynamics are learned and how they are used for planning. First, teacher-forced training does not expose the dynamics model to the recursively predicted latent states encountered during planning. Second, sampling-based planners typically initialize action search from goal-agnostic proposals, without exploiting the action-conditioned dynamics learned by the world model. We introduce Dynamics-Informed Action Learning (DIAL) to bridge these two gaps. During training, rollout alignment exposes the forward model to its own autoregressive latent predictions, while multi-step inverse supervision trains a lightweight inverse dynamics head to infer actions from latent-state transitions. We further apply the inverse head to model-generated transitions with its parameters held fixed, encouraging predicted rollouts to retain action-relevant information. At test time, DIAL autoregressively couples the inverse and forward dynamics to construct a dynamics-informed action sequence toward the goal, which initializes the mean of the Cross-Entropy Method (CEM) proposal distribution. The subsequent CEM search then refines this initialization using the standard sampling and updating procedure. Experiments across four visual control benchmarks demonstrate that DIAL outperforms existing latent world models and planning baselines in both standard and long-horizon scenarios.
Read the full abstract

All authors Yinuo Jiang, Guangzhi Wang, Yiran Wang, Junfu Pu, Cheng Cheng, Honghao Chen, Cheng Cheng, Ying Shan

Audio and Speech Processing

Pandemonium: Longform Target-Speaker Speech Recognition with Cross-Speaker Interactions

Diarization-conditioned target-speaker ASR (TS-ASR), in particular Diarization-conditioned Whisper (DiCoW), is an accurate and data-efficient approach to multi-talker transcription. It has, however, three structural limitations: the model transcribes each speaker independently, operates on short (30 s) windows, and optimizes its diarization and TS-ASR components separately. We address the first two. We integrate the diarization conditioning into a longform Parakeet backbone (DiCoP) and add cross-speaker attention, which lets the model attend across the speaker dimension so that speakers are no longer decoded in isolation, and an STNO mask refiner that re-estimates the diarization-derived conditioning. As conversational data with genuine multi-speaker interaction is scarce, we scale training with a pseudo-labeling pipeline that mines 5.4k hours of in-the-wild multi-speaker audio from YODAS2. Combining these with synthetic and real data yields state-of-the-art results across several meeting benchmarks. We open-source the resulting models, DiCoP and Pandemonium, and the pseudo-labeled data.
Read the full abstract

All authors Alexander Polok, Dominik Klement, Ivan Medennikov, Antoine Laurent, Petr Pálka, Samuele Cornell, Shinji Watanabe, Sanjeev Khudanpur, Jagadeesh Balam, Jan Černocký, Lukáš Burget, Matthew Wiesner

Artificial Intelligence

The Delta Gap: Cellular World Models for Perturbation Response Prediction Are Far Harder Than Global Correlation Implies

Cellular world models aim to predict how cells respond to genetic perturbations, analogous to world models in reinforcement learning. Perturbation response models are evaluated almost exclusively by global Pearson correlation between predicted and observed transcriptomes. We show this metric is dominated by thousands of non-responsive genes and severely inflates apparent performance. On the Replogle K562 screen (509 unseen perturbations, 5000 genes), GEARS achieves global Pearson 0.977, but its delta Pearson is only 0.076. A trivial baseline predicting mean training delta achieves 0.267, which is 3.5x better than GEARS. Our primary contribution is methodological: we identify and quantify this delta gap as a systematic evaluation artifact, and propose delta-space evaluation as the primary benchmark. We provide two post-hoc corrections (RWR on co-expression graphs and kNN matching) and show that 2.8x of the remaining gap lies in perturbation representation geometry, not decoder capacity. We further show the gap is partially addressable end-to-end: delta-aligned training (lambda = 3.0) raises delta Pearson to 0.179 on Replogle (2.35x over GEARS) and surpasses the trivial baseline on direction accuracy (0.680 vs 0.567) and gene recovery (1.52 vs 1.19); combined with kNN it reaches 0.217. An amplitude-calibration variant (eta = 0.3) further raises gene recovery to 2.16 while reaching 0.218. Critically, these representations transfer zero-shot across cell lines: from Replogle K562 to unseen Norman A549, the delta-aligned model reaches 0.328 on 741 shared genes, 4.7x a trivial baseline and 2.1x standard GEARS. This is the first systematic quantification of the delta gap, and the first diagnostic framework localizing the residual gap to perturbation representation geometry via an oracle upper bound.
Read the full abstract

Artificial Intelligence

MMGraph-Agent: Agentic Multimodal RAG via Cache-Inspired Multimodal Knowledge HyperGraphs

Driven by increasingly complex real-world applications, Retrieval-Augmented Generation (RAG) has evolved from textual pipelines to multimodal settings, comprising two complementary dimensions: knowledge organization and retrieval enhancement. However, existing methods still face challenges of high construction cost, context window limitations, and the lack of unified training across offline and online knowledge sources. To address these challenges, we propose MMGraph-Agent, a unified multimodal RAG framework based on cache-inspired multimodal knowledge hypergraphs. MMGraph-Agent enables extremely lightweight hypergraph construction with zero API cost and dynamic memory-based retrieval that alleviates long-context issues. Experiments across offline, online, and joint retrieval settings demonstrate consistent improvements in performance, efficiency, and architectural unification. Our software and data are publicly available at https://github.com/LHRLAB/MMGraph-Agent.
Read the full abstract

All authors Haoran Luo, Ziyue Zhu, Shangyang Wu, Linhao Luo, Yikai Guo, Qika Lin, Fangzhi Xu, Jiapu Wang, Xiaobao Wu, Yifan Zhu, Luu Anh Tuan

Functional Analysis

Sharp Constants and First-Order Asymptotics in a Bourgain–Brezis–Mironescu Formula for Riesz Potentials

We investigate sharp constants and asymptotic properties of a pointwise inequality of Bourgain–Brezis–Mironescu type in dimension \(n\geq2\), involving Riesz potentials and Spector's nonlinear fractional differential operator \(\mathcal D^\alpha\) of order \(0<\alpha<1\). For every \(n\geq3\) and \(0<\alpha<1\), we determine the optimal constant without symmetry restrictions and prove that it is attained by smooth radially monotone functions centered at the evaluation point. The proof combines geodesic-flow invariance, kernel monotonicity, and circular rearrangement. In dimension two, we establish the leading-order asymptotic behavior of the optimal constant and derive an explicit formula in the radial class. We further refine the pointwise limit established by A. Claros and C. Pérez [Bourgain–Brezis–Mironescu formula for Riesz potentials, Adv. Math. 503 (2026), Art. 111226] by obtaining an explicit first-order correction with a quadratic remainder uniform in the spatial variable. Finally, we strengthen their convergence-in-measure result on compact subsets for functions in \(W^{1,1}(\mathbb R^n)\) by establishing almost-everywhere convergence as \(\alpha\to1^-\), without passing to a subsequence.
Read the full abstract

Machine Learning

STABLE: A Continual Learning Optimizer with Adaptive Drift Control

Continual learning suffers from catastrophic forgetting, leading to the loss of past knowledge and unstable adaptation across task streams. Existing approaches typically address this issue through replay buffers, architectural changes, or additional regularization terms. We argue that forgetting is a consequence of unstable learning dynamics, which may benefit from an optimizer-level solution. The main challenge is that learning dynamics are time-dependent, as demonstrated by empirical phenomena such as the stability gap. First, we formally prove that, for distance-based classifiers, bounded parameter drift guarantees bounded margin erosion under a local score-stability condition. Based on this insight, we propose STABLE, a primal-dual optimizer with adaptive drift control. STABLE enforces a drift budget by increasing the dual variable when the budget is violated and decreasing it otherwise. This results in high stability without penalizing plasticity. Experiments on long-horizon unimodal streams and CLIP-based multimodal streams show that vanilla PEFT baselines trained with STABLE match or outperform specialized continual learning methods. In the multimodal setting, STABLE further preserves zero-shot transfer, matching or exceeding dedicated multimodal continual learning methods. Code is available at https://github.com/atlas-luiss/STABLE.
Read the full abstract

All authors Xiaoyan Li, Lanpei Li, Massimo Coppola, Vincenzo Lomonaco, Antonio Carta

Combinatorics

A Lie-theoretic approach to the type $C_n^{(1)}$ partitions of Capparelli, Meurman, Primc and Primc

Capparelli, Meurman, Primc and Primc (CMPP) extended the frequency conditions in Gordon's partition theorem to paths in coloured arrays. Their conjecture for type $C_n^{(1)}$ identifies the generating function of admissible arrays with a principally specialized character. We prove this conjecture for arbitrary boundary data by constructing an ordered Poincaré–Birkhoff–Witt (PBW) basis of the corresponding integrable highest weight module. A decomposition of the admissible arrays and a tensor product argument establish linear independence of the corresponding monomial vectors. To prove spanning, we use relations from annihilating fields and PBW bases for modules over finite Levi subalgebras. Principal specialization then yields the conjectured product. The same basis identifies the number of parts with the degree in the associated graded module for the PBW filtration.
Read the full abstract

Computer Vision

FG-DSA: Frequency-Guided Depthwise Separable Attention for Efficient Dense Prediction

Dense prediction and generative modeling remain challenging in resource-constrained environments, particularly because repeated downsampling can lead to the loss of spatial details and less accurate object boundaries. Although attention mechanisms have been widely adopted to address semantic ambiguity, preserving high-frequency information while maintaining computational efficiency remains an important challenge. To address this issue, we propose Frequency-Guided Depthwise Separable Attention for Efficient Dense Prediction (FG-DSA), a framework that integrates frequency-domain information into lightweight vision architectures for dense visual understanding and generation. Specifically, FG-DSA uses a 2D Fast Fourier Transform (FFT)-based high-pass filtering module to guide spatial attention, aiming to retain fine-grained features and improve object-boundary representation. Experimental results show that FG-DSA can improve boundary-related performance in semantic segmentation under the evaluated setting. Beyond discriminative tasks, we also examine its application in a generative U-Net for single-step image restoration. The results suggest that incorporating FG-DSA can enhance reconstruction quality and help preserve high-frequency details.
Read the full abstract

Computer Vision

SHIN: Scalable Hierarchy for INcremental 3D Gaussian Splatting

3D Gaussian Splatting (3DGS) provides high-quality real-time novel-view synthesis, but deployment budgets vary across devices and network conditions. Most compact methods target one operating point, so another budget requires a separate representation. SHIN is a scalable 3DGS hierarchy with a reusable Base and additive enhancement layers, so each higher-quality operating point extends rather than replaces the decoded representation. Constructing this hierarchy requires deciding which Gaussians each stage adds and how much spherical-harmonic (SH) appearance each retains, because these choices remain fixed in later prefixes. SHIN derives one shared Gaussian order from multi-scale position-scale sensitivity and divides each stage budget between adding Gaussians and retaining SH bands using calibrated gradient-based scores per byte. Lightweight opacity updates adjust compositing while transmitted geometry and SH coefficients remain unchanged. Across three benchmarks, SHIN progressively improves one cumulative representation and achieves the best PSNR-size efficiency in every overlapping comparison with the multi-rate and progressive baselines.
Read the full abstract

Popular Physics

The Physics of Interstellar Travel

How can we send an uncrewed spacecraft to the nearest stars within a human lifetime? What propulsion methods are available? Will it survive impacts in the interstellar medium? What science will it do at its destination? How can it send back data over enormous distances? In this short semi-popular article, I examine various aspects of the physics of interstellar travel, its challenges and possible solutions. I focus on nuclear rockets, laser sails, dust impacts, and communication. These, as well as other issues such as navigation and onboard power, are explored in more detail in my book, The Physics of Interstellar Travel (2026).
Read the full abstract

2611.10651 · 7 Nov · This is an earlier version of a later-edited article that will appear in Physik Journal in December 2026 (https://pro-physik.de/zeitschriften/physik-journal/). It is based on my book The Physics of Interstellar Travel (2026), published by Cambridge University Press, available open access from https://doi.org/10.1017/9781009700429 and https://bailer-jones.www3.mpia.de/pit.html

Robotics

Trajectory Inversion for a Rudderless Dual Vectored-Thrust Flying Wing

A rudderless dual vectored-thrust flying wing combines vertical takeoff and landing with efficient for- ward flight in a compact tailless configuration. Without aerodynamic control surfaces, vectored thrust couples translation and pitch control. Therefore, generating dy- namically compatible trajectories remains an open chal- lenge. To address this challenge, this study proposes a trajectory generation method that jointly optimizes path geometry and timing. Trajectory inversion yields state and control histories that satisfy the coupled equations of motion. The optimization balances curvature-induced acceleration against aerodynamic support, lowering thrust- vectoring demand while retaining the defining maneuver features. Representative looping and rolling sequences are tracked in closed-loop simulations. These results provide a basis for planning autonomous aerobatic maneuvers with rudderless vectored-thrust aircraft.
Read the full abstract

All authors Weicheng Di, Qiqi Liu, Jinwu Xiang, Daochun Li, Zhan Tu

Machine Learning

MACaQ: Fast, Model Agnostic Calibrated Quantization for Low-Precision Weights and Activations

Weight-activation quantization can reduce the memory and compute cost of neural-network inference, but recent methods often require expensive optimization and architecture-specific propagation of basis transforms across layers. We introduce MACaQ, an MXFP4 W4A4 post-training quantization method that treats every linear layer independently. MACaQ applies a per-channel scale and a block-Hadamard transform online to each layer's activations, while absorbing the corresponding inverse transforms into its weights. Retaining these transforms online avoids cross-layer dependencies and enables a simple streaming implementation of the quantization process, in which only one layer must reside in memory at a time. To improve over round-to-nearest quantization, we introduce Flip-Refinement, a Hessian-aware coordinate-descent procedure that greedily moves individual FP4 codes to adjacent representable values. Despite executing the activation transforms at inference time, MACaQ adds only 2.3% end-to-end overhead over pure MXFP4 on Qwen3-32B in vLLM. MACaQ reaches lower perplexity than DuQuant++ and MR-GPTQ, and already matches their best perplexity on Llama-3-8B in 11 times less quantization time. We further apply MACaQ without architecture-specific modifications to an image diffusion model, illustrating the generality of its layer-local design and show quantization of a 122B parameter model in 64 minutes, using only 32.5GB GPU RAM. Code is available at https://github.com/huawei-csl/MACaQ .
Read the full abstract

All authors Lorenz K. Muller, Philippe Bich, Chiara Boretti, Hyun-Min Chang, Jiawei Zhuang, Lukas Cavigelli

Hardware Architecture

PENSA: A Spatial Accelerator for BitNet b1.58 Inference on an HBM-equipped AMD Alveo FPGA

Ternary large language models (tLLMs) replace general multiplication in dominant linear layers with sign-controlled accumulation, shifting hardware bottlenecks toward packed-weight delivery, reduction, and operator occupancy. PENSA is an end-to-end FPGA accelerator for the BitNet b1.58 2B4T model on an AMD Alveo U50. Two row-partitioned ternary matrix engines consume packed TQ1_0 weights from eight independent high-bandwidth memory (HBM) pseudo-channels, each processing 1,280 ternary weights per cycle through sign selection and integer reduction. The sequencer, attention, and vocabulary-projection units remain resident on the FPGA, while two-stream interleaved scheduling overlaps sequences across the persistent pipeline to form the Penta-Unit. At 104.1~MHz, PENSA achieves 58.09 tok/s at context 128 and 24.60 tok/s at context 2048, corresponding to 2.14-3.14$\times$ the throughput of a same-model Xeon E-2356G baseline. The resulting system demonstrates end-to-end FPGA execution of 2B-scale BitNet topology without modifying its model structure.
Read the full abstract

Robotics

Which Taxels, Not How Many: Tactile Sensor Placement for Dexterous Policies

Policies for dexterous manipulation depend on design choices made before training: which sensors enter the observation, where they are placed, and how their readings are arranged for the encoder. These choices are rarely examined. For tactile sensing, simulation studies place taxels uniformly over the hand, and this placement has become a default without being tested. We test this default under a single controlled protocol. The protocol holds the teacher, the training procedure, and the candidate taxels fixed and varies only the sensor configuration and the representation. We evaluate it on in-palm rotation and grasping with three robot hands. A small set of taxels selected from contact statistics matches a far larger uniform set. Taxels at rarely contacted sites are equivalent to no tactile input. With all candidate taxels, their layout in the input tensor does not affect performance, even for a convolutional encoder. Uniform placement therefore overestimates the number of taxels a task requires. We distill these results into lessons for designing tactile policies and into a minimal set of design choices that a tactile policy study should report.
Read the full abstract

All authors Taeyun Kim, Jinsoo Shin, Taehwa Kim, Taeyoon Kim, Jinyoung Choi

Computation and Language

On the Geometric Bottlenecks of Cross-Lingual Knowledge Editing

Transformer-based autoregressive multilingual language models (LMs) are often assumed to store factual knowledge in an interlingua: a language-agnostic latent space. If this assumption holds, a parameter update applied in a source language during knowledge editing should inherently propagate to its semantic equivalents in target languages. However, empirical benchmarks show that cross-lingual propagation of edits often degrades substantially, a degradation that prior work treats largely as an algorithmic symptom rather than a structural bottleneck. In this work, we investigate whether representational anisotropy creates a structural bottleneck for source-only associative-memory editing, and perform a geometric dissection of the cross-lingual propagation gap. Using associative memory-based "locate-then-edit" methods, including ROME, MEMIT, and AlphaEdit, as analytical frameworks, we derive a closed-form decomposition of this gap into key-space propagation, parallel residual mismatch, and orthogonal residual mismatch. Our empirical findings show that cross-lingual propagation is primarily bottlenecked by a large orthogonal residual component, indicating that source-language edits often fail to span the target-language update directions required for successful transfer. Furthermore, we show that post-hoc correction using only source-language residual statistics encounters a large residual-prediction barrier. Finally, we perform a structural realignment via a targeted contrastive alignment objective and find that it substantially improves key-space propagation and partially improves cross-lingual edit transfer, while leaving value-space mismatch as a remaining bottleneck.
Read the full abstract

All authors Aastha A K Verma, Anwoy Chatterjee, Sparsh Jain, Tanmoy Chakraborty

Group Theory

Automorphism of reduced finitary power monoids

Let $H$ be a multiplicatively written monoid with identity element $1_H$. The family $\mathcal{P}_{\mathrm{fin},1}(H)$ of all finite subsets of $H$ containing $1_H$ is a monoid with identity $\{1_H\}$ under the setwise multiplication induced by $H$. We prove that $\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},1}(H)) \cong \operatorname{Aut}(H)$ whenever $H$ satisfies either of the following two conditions: (1) $H$ is commutative, cancellative, and contains an invertible element of infinite order; (2) $H$ is an abelian group which is not isomorphic to the Klein four-group.
Read the full abstract

Computer Vision

Agent2BIM: A Systematic Evaluation of Reality-Capture Configurations and Prior Models for As-Built Modeling

AI agents can generate as-built Building Information Models (BIMs) from reality-capture data, but the evidence required for accurate reconstruction remains unclear. This paper presents Agent2BIM, an agent-based Scan-to-BIM framework, and evaluates seven capture configurations in a research laboratory and a mechanical, electrical, and plumbing (MEP) scene. Inputs include 360{\deg} panoramas, smartphone video, smartphone LiDAR, terrestrial laser scanning (TLS), TLS with posed panoramas, and imagery supplemented by prior IFC models. Outputs are compared against TLS-verified reference models for component detection, geometric accuracy, and MEP connectivity. In the laboratory, component recall ranged from 84% to 100%, but only 9% of components reconstructed from panoramas were within 20 cm of their reference positions, compared with 86% using smartphone LiDAR. In the MEP scene, scale-free inputs recovered 11 - 43% of components, versus 75 - 93% for metric inputs. Prior IFC models improved MEP reconstruction but could propagate inaccuracies. The results highlight the complementary roles of visual and metric evidence in agent-based Scan-to-BIM.
Read the full abstract

Probability

Iteration-Clock Selection for Euler Schemes of Degenerate Diffusions

We study Euler–Maruyama approximation of the driftless one-dimensional SDE $dX_t=\sigma(X_t)\,dW_t$ when $\sigma$ vanishes at the origin and weak uniqueness fails because solutions may remain at zero for an arbitrary time. The central object is the logarithmic coefficient profile $q(Q)=-\log\sigma(e^{-Q})$. Under a two-sided logarithmic increment condition, with both indices in $(0,1/2)$, we show that weak-solution selection is governed by a deterministic iteration clock: if $N_q(Q)$ is the number of iterates of $q$ needed to reach a fixed reference depth and $hN_q(\log(1/|x_h|))\to d$, then the Euler scheme started from $x_h$ converges to the fundamental zero-sojourn diffusion delayed by $d$. This single theorem yields the common-index generalized-power law $d=\theta/|\log\alpha|$ when $h\log\log(1/|x_h|)\to\theta$, an effective translation-number formula for periodically modulated logarithmic profiles, a two-index delay window, and a sharpness construction in which, for one coefficient and one initialization family, the cluster set is the entire delay window. Thus the macroscopic selection mechanism is the iteration count of the logarithmic coefficient profile; the explicit double-logarithmic law of the homogeneous power case is its common-index closed form.
Read the full abstract

Computer Vision

VisProve: Learning Typed Visual Provenance for Scientific Document Question Answering

Visual citations show where an answer is supported, but not how the cited content contributes to it. In scientific document question answering, a model may read a value directly, compress several observations, or combine evidence to infer a conclusion. We study typed visual provenance, where a system generates an answer together with source regions and explicit support relations. We introduce ReViE, a dataset of 3,941 questions over complete scientific documents, with page, section, and region annotations labeled as Quotation, Compression, or Inference. We further propose VisProve, which augments joint answer and provenance generation with differentiable Spatial Evidence Reading and Evidence-Answer Relation Feedback. VisProve conditions next-token prediction on localized visual evidence and its estimated support relation through a gated norm-bounded residual. On the document-disjoint ReViE test set, VisProve improves Typed-F1 from 3.52 to 5.48 and Section-F1 from 45.99 to 53.24 over joint supervised fine-tuning with the same backbone, while Page-F1 changes only marginally. These results suggest that coupling visual grounding with support relation modeling mainly improves fine-grained provenance, while the low absolute Typed-F1 reveals a substantial gap between answer generation and precise visual evidence attribution.
Read the full abstract

All authors Jingxuan Wei, Zhengyang Wang, Zheng Guo, Xiaoming Wu, Fuqiang Wang, Xin Wang, Bihui Yu, Yining Wang, Junnan Zhu

Physics and Society

Same Contacts, Different Rooms:Meeting Format Shapes LLM Group Influence Beyond Pairwise Contact Counts

Simulations of LLM agent collectives usually summarise interaction as a pairwise contact network. We test that summary with a contact-matched intervention: the members, their initial stances and every pairwise contact count stay fixed while the meetings are reorganised, and each member gives a one-step support readout after reading the arguments of everyone present. Across six open-weight LLMs and 14 debate topics, merging triads into five-person meetings (resize) moves group support more than reshuffling who meets in triads (regroup), although both move the same 60 dyadic exposures. The root-mean-square change is 4.36 versus 2.59 percentage points (pp), larger for resize in all six models and in 81 of 84 model-topic cells. Resizing also has a consistent direction: it amplifies the initial majority of the merged members by 1.72 pp, in every model, whereas regrouping has no direction (<= 0.03 pp). A contact network cannot represent regrouping at all, yet pairwise rules applied meeting by meeting explain up to 76% of its effect, and with identical weights they raise resize R^2 by 0.51-0.58 for two models. We show that the gap between the two placements is an exact Jensen gap of the sigmoid readout, bounded by how much a member's predicted logits vary across meetings; this dispersion tracks the gap with log-log r=0.97 and tells a modeller when contact counts are a safe summary. Meeting format is a first-class variable for LLM social simulation.
Read the full abstract

Quantum Physics

A Grover-Amplified Oracle for Maximum Clique on Graph, Minimum Hitting set and Dualizations on Hypergraphs Built on Quantum Phase Estimation

In this paper, we use a quantum circuit called the Graph Encoding Operator which, applied to the complement $\bar G$ of any graph $G$, yields an oracle for the maximum clique problem suitable for amplification via Grover's algorithm. Analogously, we devise a Hypergraph Encoding Operator which, when applied to a hypergraph $\mathcal H$, yields an oracle for the minimum hitting set suitable for amplification via Grover's algorithm. Interestingly, the latter oracle provides, when measured, an approximation of the number of hitting sets of the hypergraph, thus solving the dualization problem for positive Boolean functions represented in a positive irredundant disjunctive normal form. We give the full construction of both operators, prove their correctness, analyze their complexity, and provide an implementation on a quantum simulator. We discuss the resulting complexity, contrast them with existing quantum computer algorithms based on Grover search and we will show that our algorithms outperform, to the best of our knowledge, all of the previous algorithms on space computational complexity by requiring only $O(N)$ number of qubits with respect to the number of vertices $N$, for representing both graphs and hypergraphs in the encoding operator and, consequently, results in an oracle for both graphs and hypergraphs that uses only $O(N)$ qubits even when the hypergraph contains an exponential (on the number of vertices) number of hyperedges.
Read the full abstract

All authors Francisco Piñero, Fernando L.\Pelayo, Hernan I. de la Cruz, Jose J. Paulet, Mauro Mezzini, Fernando Cuartero

Analysis of PDEs

Total Variation Regularized Barycenters in Wasserstein Space

We study total variation regularization of Wasserstein barycenters on bounded connected Lipschitz domains. We prove existence for arbitrary prescribed probability measures and uniqueness when at least one is absolutely continuous. Our main results are attained dual formulations in terms of relative perimeters and vector fields with zero normal trace. These yield necessary and sufficient optimality conditions and a variational characterization of superlevel sets. We also establish a maximum principle on convex domains, stability of minimizers, and an exact reduction to a single TV-JKO step in one dimension. The existence, uniqueness, and perimeter duality results extend to costs $|x-y|^p$, $p>1$, under the same assumptions.
Read the full abstract

Artificial Intelligence

SLA-Aware Task Offloading in the IoT Edge-Cloud Continuum: Class-Weighted Deep Reinforcement Learning Against an Exact Optimum

Deep reinforcement learning (RL) is often used to decide where Internet of Things (IoT) tasks with different latency limits run in the edge-cloud continuum, yet some methods give every missed task the same penalty and few compare their agents with an optimal policy. This benchmark study builds four offloading systems small enough to solve exactly by dynamic programming and measures reward and learner against the optimum. A bonus for a met limit makes each miss cost its penalty plus the bonus, which shifts the class weights of published per-class rewards. A class-blind penalty stays 65% to 118% above the optimal cost even with a perfect learner, while a corrected per-class penalty comes within 0.13% of the optimum. With a per-class reward, tuned proximal policy optimization (PPO) agents come within 9% to 11% of the optimum in the smallest system, and the corrected one beats a tuned reservation rule in 9 of 10 seeds. In larger systems the agents stay 41% to 76% above the optimum, and a tuned rule beats their mean. Training ten times longer does not close this gap, and a PPO-Lagrangian agent with reachable targets does no better than the per-class reward.
Read the full abstract

Computer Vision

VAComp: View-Aware Full Projector Compensation

Full projector compensation adjusts a projector's input image to compensate for both geometric and photometric distortions introduced by the projection surface. Existing methods generally treat the viewing condition as a static camera viewpoint. When applied to a moving viewer, existing methods face two key challenges: (1) view-dependent methods require model retraining or recalibration to compensate for novel viewpoints; and (2) NeRF/GS-based view-agnostic methods use structure from motion (SfM) to calibrate the projector-camera systems (ProCams) with an unknown global scale, while view-aware tracking requires calibrating a camera oriented opposite to the projector, which introduces an additional calibration challenge. To address these challenges, we introduce VAComp, the first view-aware framework for full projector compensation for a moving viewer in a naked-eye setting. Using a projector, VAComp enables the viewer to perceive a virtual object that visually replaces the real object or to observe view-consistent modifications of the real object's appearance and material properties. To resolve the SfM scale ambiguity, we use metric calibration as a reference to recover the global scale; to calibrate the oppositely facing camera, we use two checkerboards, one facing the RGB camera and the other facing the depth camera. To balance compensation quality and runtime efficiency, we develop two non-iterative hybrid methods, VAComp-SD and VAComp-SI. Both combine 2D Gaussian-based view-conditioned geometric correction with learned photometric compensation, predicting each compensation image through a single forward pass without per-frame iterative optimization. Real-world experiments show that both methods achieve competitive compensation quality, with VAComp-SD enabling real-time updates and VAComp-SI operating at interactive rates. Project page: https://realkaizhou.github.io/VAComp/.
Read the full abstract

Strongly Correlated Electrons

Strong Collective Dynamics Hidden behind a Nearly Free Single-Particle Spectrum

Single-particle spectroscopy has long served as a criterion for the strength of electronic correlations: strong collective motion is expected to shift and broaden spectral peaks and to renormalize the dispersion. Here we show that this correspondence can fail on the same microscopic orbital. The mechanism relies on local interference between two resonant side orbitals: one suppresses low-energy real transitions into the interacting environment, while the other suppresses spectral-weight backflow to the source orbital; virtual mixing nevertheless transfers the environmental spin polarization to the source. We prove a rigorous relative spectral bound that is uniform in environment size and interaction strength and holds for arbitrary stationary states; the protection becomes stronger as the relative frequency resolution improves. In a fully quantum two-cell device, a collective mode softens by 60%, whereas the particle intensity at the free peak changes by only 3.6 parts per million. Quantum Monte Carlo shows that the transferred spin response grows with spatial correlations and is quantitatively predicted by the free-band polarization. These results recast the meaning of a simple particle spectrum and provide a design principle for quantum probes that combine predictable particle spectra with strong collective sensitivity.
Read the full abstract

All authors Zhixuan Zhao, Sicheng He, Hanxiang Di, Nayun Chen, Jiabai Xiong, Guyixue Gao, Yanling Wu, Jun Li

Robotics

Model-Based Imitation Learning For Monocular Vision-Based Navigation

Reliable navigation with a single monocular limited field-of-view camera is challenging because the lack of depth measurements and limited spatial awareness complicate collision checking. Existing approaches to tackle these challenges fall into two categories. Model-based planners can incorporate explicit collision-avoidance guidance which leads to better generalization but are often computationally expensive and require manual tuning of the algorithmic parameters. In contrast, end-to-end approaches trained by imitation learning can be flexible but typically encode safety only implicitly and thus can suffer from high collision rates in novel scenarios. In this paper, we present a middle ground through model-based imitation learning. The model-based part is a learned collision model that can map a sequence of past images and a candidate control sequence to a distribution of worst-case obstacle clearance. This predictive distribution in turn is used to define a collision-risk cost in a trajectory optimization formulation. The imitation component uses human demonstrations to learn scene-dependent balancing of collision risk with other task costs such as goal progress. We also use it to learn the algorithmic parameters (e.g., a warm start) of our optimizer to accelerate the solution process. We show that our approach can demonstrate highly reliable navigation in cluttered environments on robots with different embodiments. We also demonstrate a substantial reduction in collision rate over recent end-to-end and model-based baselines.
Read the full abstract

All authors Basant Sharma, Prajyot Jadhav, Saqi Hussain Kalan, Arun Kumar Singh

Machine Learning

Coverage-Verified Sparse Attention: Closed-Loop Quality Control for Long-Context LLM Decoding

Sparse attention accelerates long-context decoding by reading a subset of the KV cache, but existing methods are open-loop: the kernel commits to its selection with no signal indicating whether enough attention mass was preserved and no recovery path when it was not. The needed signal already exists inside the kernel. The fraction of softmax mass on the selected tokens, which we call coverage, satisfies an exact algebraic equality with per-step output error and serves a dual role as the acceptance criterion and the interpolation weight for recovery. Coverage-Verified Sparse Attention (CVSA) turns coverage into a training-free, closed-loop verify-then-correct decode loop that accepts high-coverage drafts and recovers only the heads that fall short, with no sampling, no draft model, and no speculative-decoding infrastructure. Quality varies by only $0.14$ points across the threshold sweep; budget and scheduling ablations characterize the deployed operating point. On LongBench at 7B and 70B, CVSA comes within $1.13$ points of dense attention, statistically matching it at 70B, and it loses $0.8$ points on RULER; measured median decode-step speedup reaches $6.21\times$ at 128K context.
Read the full abstract