Peer Revenue Play credits. Real papers.

Papers

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

All of arXiv

Tuesday 29 September

Computer Vision

LazySloth: Bounded LLM-based Lazy Tree Search for Fast Long Video Comprehension

Arka Mukherjee, Kaleen Shrestha, Larissa Zhu et al. (+1 more)

Modern vision-language models (VLMs) have shown promising results in long-video understanding due to the rich semantic information they can capture. However, most methods focus on coarse captioning of extracted image frames that are computationally inefficient and require models with large context windows. While past work has explored efficient methods through multimodal retrieval-augmented generation (RAG), they rely on lossy embeddings that lose temporal context and fine-grained detail. Few works to date have investigated how VLM-based query-relevant information retrieval can be optimized. We introduce LazySloth, an efficient tree-based search method that speeds up video comprehension and retrieval tasks 2.9-8.3x (compared to existing agentic methods) through bounded captioning of portions of the video considered irrelevant by a VLM of the video. Compared to contemporary specialized video-understanding VLMs and RAG-based methods, LazySloth achieved similar or better final task accuracy across two recent open-source base VLMs–Gemma 4 31B and Qwen3.6 27B–across four benchmarks. LazySloth reduced the gap between the base open-source model and a closed-source model, GPT-4o. Ablations showed that replacing VLM scene understanding with CLIP-based retrieval cost 8.8-19.9% in accuracy, while lazy tree construction matches eager construction at a fraction of the captioning cost. With LazySloth, we demonstrate the possibility of faster long-video comprehension without substantial loss in performance.
Read the full abstract

All authors Arka Mukherjee, Kaleen Shrestha, Larissa Zhu, Maja Matarić

Optimization and Control

A stochastic subgradient method with optimal failure exponent

Bart P.G. van Parys

Fix a target accuracy $\varepsilon$, a gradient-noise level $s$, and a horizon $N$. We wish to design algorithms which minimize the probability of observing a suboptimality gap which exceeds the target accuracy, i.e., $\mathcal{E}_N = -\log \sup_{f,P} \mathbb{P}_P(f(x_A) - f_\star \ge \varepsilon)$, with the noise law $P$ known only to be sub-Gaussian. We single out a uniformly averaged schedule which is harmonic (of the form $h_k = R^2/(\varepsilon (N+m-k))$) and prove, via an optimized exponential supermartingale argument, that it attains the optimal exponent $\mathcal{E}_N^\star = \varepsilon^2 N (1+o(1))/(2R^2 s^2)$. Optimality is certified by a matching impossibility result: under Gaussian noise, a gradient-masking change of measure caps the exponent of every algorithm at the same leading order. In a small-noise limit, our setting degenerates into the adversarial-error model of Gösgens and van Parys (2025) for subgradient methods.
Read the full abstract

Machine Learning

Simultaneous Neural Optimal Transport

Milena Gazdieva, Kirill Sokolov, Jiawei Chen et al. (+2 more)

Optimal Transport (OT) provides a principled framework for learning transformations between probability distributions from unpaired samples. In many applications, however, a single transformation must map several source distributions to a common target distribution. For example, image restoration might require handling different types of degradation without knowing the degradation of each input at inference time. Simple approaches of pooling the source distributions only encourage alignment with the target at the aggregate level and may leave individual sources misaligned. In our paper, we consider the simultaneous OT problem which formalizes the task of learning a shared transport map that minimizes the average transport cost while aligning each source distribution with a prescribed target. We propose a neural method for solving the simultaneous OT problem by learning a shared transport map that minimizes the average transport cost while aligning each source distribution with a prescribed target. We derive a max-min formulation for learning this map. We illustrate its application to image restoration, where a single model handles multiple degradation types using a common collection of clean target images.
Read the full abstract

All authors Milena Gazdieva, Kirill Sokolov, Jiawei Chen, Evgeny Burnaev, Alexander Korotin

Analysis of PDEs

Interface-Shifted Minnaert Resonances

Huaian Diao, Long Li and Mourad Sini

We develop a unified asymptotic theory for Minnaert resonances generated by a small bubble $D_\varepsilon$ placed near an interface $\Gamma$ of material discontinuity in an acoustic background. The bubble has characteristic size $\varepsilon>0$, with its density and bulk modulus depending on the same small parameter through the Minnaert high-contrast scaling. The analysis is organized by the scaled separation \[ \Theta_\varepsilon := \frac{\operatorname{dist}(D_\varepsilon,\Gamma)}{\varepsilon}, \] and covers the three natural regimes \[ \Theta_\varepsilon\to+\infty, \qquad \Theta_\varepsilon\to\Theta_*\in(0,\infty), \qquad \Theta_\varepsilon\to0: \] 1) In the far-separation regime, the interface is invisible at leading order and the classical full-space Minnaert law is recovered. 2) In the critical regime, the rescaled geometry retains both the bubble and the interface at finite separation, and the Newtonian capacitance is replaced by a flat-interface capacitance determined by the limiting tangent transmission problem. 3) In the near-contact regime, this family converges to a contact-limiting capacitance, yielding the same leading-order Minnaert resonance for all gap scalings satisfying $\operatorname{dist}(D_\varepsilon,\Gamma) = o(\varepsilon)$. These three regimes provide a unified description of how the resonance frequency changes as the bubble approaches the interface, from the classical isolated-bubble limit to the contact limit. The analysis relies on a uniform reduction of the resonance problem to a scalar equation and on a variational characterization of the interface-dependent capacitances; in particular, Mosco convergence of the associated transmission energies yields convergence of the capacitary potentials and minimum energies up to the singular contact configuration.
Read the full abstract

High Energy Astrophysical Phenomena

Search for dark matter annihilation in the Sun using the full ANTARES dataset

A. Albert, S. Alves, M. André et al. (+133 more)

Weakly Interacting Massive Particles (WIMPs) are among the most popular candidates for particle dark matter. In certain scenarios, WIMPs could be gravitationally captured into massive celestial objects and then annihilate into Standard Model particles, possibly leading to the emission of neutrinos. The ANTARES neutrino telescope, located in the Mediterranean Sea, is sensitive to neutrinos with energies ranging from a few tens of GeV to the TeV scale, making it well suited to search for the annihilation of WIMPs with masses below a few TeV/c$^2$. One of the closest potential astrophysical sources of neutrinos from WIMP annihilations is the Sun. In this work, a search for high-energy neutrinos originating from the Sun, using the full dataset recorded by ANTARES from 2007 to 2022 was conducted. The analysis yielded no significant evidence for a neutrino excess over the expected background, and upper limits on the WIMP-proton interaction cross section have been obtained, for both spin-dependent and spin-independent processes. These limits are valid for dark matter masses ranging from 35 GeV/c$^2$ to 10 TeV/c$^2$ for the annihilation benchmark channels WIMP$+$WIMP $\longrightarrow$ $b\bar{b},\tau^+\tau^-, W^+W^-$, assuming a 100$\%$ branching ratio for each of these channels. These results improve on those previously obtained by the ANTARES Collaboration by a factor of two-three and are consistent with those reported by other experiments in direct and indirect searches for dark matter.
Read the full abstract

All authors A. Albert, S. Alves, M. André, M. Ardid, S. Ardid, J.-J. Aubert, J. Aublin, B. Baret, S. Basa, Y. Becherini, B. Belhorma, F. Benfenati, V. Bertin, S. Biagi, J. Boumaaza, M. Bouta, M.C. Bouwhuis, H. Brânzaş, R. Bruijn, J. Brunner, J. Busto, B. Caiffi, D. Calvo, S. Campion, A. Capone, L. Caramete, F. Carenini, J. Carr, V. Carretero, T. Cartraud, S. Celli, M. Chabab, T. Chiarusi, M. Circella, J.A.B. Coelho, A. Coleiro, R. Coniglione, P. Coyle, A. Creusot, A. F. Díaz, B. De Martino, C. Distefano, I. Di Palma, C. Donzaud, D. Dornic, D. Drouhin, T. Eberl, A. Eddymaoui, T. van Eeden, D. van Eijk, S. El Hedri, A. Enzenhöfer, P. Fermani, G. Ferrara, F. Filippini, L.A. Fusco, S. Gagliardini, J. García-Méndez, C. Gatius Oliver, P. Gay, N. Geißelbrecht, H. Glotin, R. Gozzini, R. Gracia Ruiz, K. Graf, C. Guidi, L. Haegel, H. van Haren, A.J. Heijboer, Y. Hello, L. Hennig, J.J. Hernández-Rey, J. Hößl, F. Huang, G. Illuminati, B. Jisse-Jung, M. de Jong, P. de Jong, M. Kadler, O. Kalekin, U. Katz, A. Kouchner, I. Kreykenbohm, V. Kulikovskiy, R. Lahmann, M. Lamoureux, A. Lazo, D. Lefèvre, E. Leonora, G. Levi, S. Le Stum, S. Loucatos, J. Manczak, M. Marcelin, A. Margiotta, A. Marinelli, P. Migliozzi, A. Moussa, R. Muller, S. Navas, E. Nezri, B. Ó Fearraigh, E. Oukacha, A.M. Păun, G.E. Păvălaş, S. Peña-Martínez, M. Perrin-Terrin, P. Piattelli, C. Poirè, T. Pradier, N. Randazzo, D. Real, G. Riccobene, A. Romanov, A. Sánchez Losa, A. Saina, F. Salesa Greus, D. F. E. Samtleben, M. Sanguineti, P. Sapienza, F. Schüssler, J. Seneca, M. Spurio, Th. Stolarczyk, M. Taiuti, Y. Tayalati, B. Vallage, G. Vannoye, V. Van Elewyck, S. Viola, D. Vivolo, J. Wilms, S. Zavatarelli, A. Zegarelli, J.D. Zornoza, J. Zúñiga

Exactly Solvable and Integrable Systems

Fredholm determinant solutions to the modified KP hierarchy and 2D Toda lattice: isospectral and non-isospectral reductions

Dan Dai and Han-Han Sheng

We study Fredholm determinant solutions to the first two positive and the first negative members of the modified Kadomtsev-Petviashvili (mKP) hierarchy, together with the bilinear two-dimensional Toda lattice (2DTL) equation. By applying isospectral reductions, we explicitly construct Fredholm determinant solutions for the modified Korteweg-de Vries (mKdV), the modified Camassa-Holm, and the focusing nonlinear Schrödinger equations. Furthermore, we demonstrate that in the non-isospectral case, the reduction from the mKP and 2DTL equations to the non-isospectral negative KdV equation yields solutions characterized by the deformed Bessel determinant.
Read the full abstract

Algebraic Geometry

Some applications of the moduli space of $U\oplus D_8(-1)$-polarized K3 surfaces

Adrian Clingher, Andreas Malmendier and Brandon Williams

We study K3 surfaces polarized by the lattice $U\oplus D_8(-1)$ through elliptic fibrations and orthogonal modular forms. Using an alternate genus-one fibration and its relative Jacobian, we construct an explicit map from Vinberg's coefficient space to the Hashimoto–Ueda family. We also explain the geometric meaning of certain character forms associated with these moduli spaces and extend the relative-Jacobian construction to the exceptional lattices $U\oplus E_8(-1)\oplus A_k(-1)$ with $k=1, 2$.
Read the full abstract

Robotics

Towards Spatial Perception for Heterogeneous Robot Collaboration in Subterranean Mining Environments

Mario Alberto Valdes Saucedo, Akash Patel, Christoforos Kanellakis et al. (+1 more)

The autonomous extraction of deep mineral deposits in abandoned underground mines is fundamentally a multi-agent integration problem. No single platform simultaneously offers the mobility to traverse kilometers of degraded drifts and the sensing payload required to characterize an ore body. This article presents the onboard perception pipeline that bridges two heterogeneous agents within the PERSEPHONE autonomous mining mission. Which consist of a lightweight Explorer robot that maps an unknown mine and generates a 3D scene graph of inspection targets, by running a zero-shot, vision-language semantic segmentation stack that detects mineral deposits directly from natural-language prompts. The map and the graph are then handed to a second Inspector robot, which carries an advanced sensing payload and uses them to plan close-range inspection viewpoints. We detail the complete pipeline, with emphasis on the geometric abstraction that turns raw detections into actionable inspection targets, spanning per-view bounding-box generation, cross-view box merging, plane fitting, and polygon extraction, and we report an extensive field validation in a subterranean test facility and in an active magnesite mine, covering both iron-vein and magnesite mineralization under realistic, perceptually degraded conditions.
Read the full abstract

All authors Mario Alberto Valdes Saucedo, Akash Patel, Christoforos Kanellakis, George Nikolakopoulos

Mesoscale and Nanoscale Physics

Nonlinear optical responses of semiconducting graphene nanoribbons: Strong constraint from a hidden symmetry

Jiawei Ruan, Weichen Tang, Yang-Hao Chan et al. (+3 more)

Nonlinear optical (NLO) responses are powerful tools for probing material structures, as they are highly sensitive to spatial symmetries. For instance, it is well known that having inversion symmetry forbids even-order NLO effects. Here, we show that nonspatial symmetries can also significantly constrain NLO responses. In graphene nanoribbons (GNRs) with bipartite-lattice structures, we identify the approximate bipartite-lattice particle-hole (BLPH) symmetry strongly suppresses even-order NLO effects, even for systems with no inversion symmetry. Importantly, we demonstrate that this approximate symmetry enables the optical generation of nearly pure spin shift currents in ferromagnetic GNRs, with the charge shift currents strongly suppressed. The spin orientation of these photo-induced DC spin currents can be easily tuned by an external magnetic field, providing distinct advantages over previous schemes for generating pure spin shift currents in antiferromagnetic systems. We illustrate our findings using a recently synthesized ferromagnetic Janus GNR. Our results underscore the significance of an overlooked symmetry aspect of NLO responses and highlight magnetic GNRs as promising candidates for spintronic devices.
Read the full abstract

All authors Jiawei Ruan, Weichen Tang, Yang-Hao Chan, Chen Hu, Xiaoxun Gong, Steven G. Louie

Computational Complexity

Robust Approximation and the Arity Barrier at Width Two

Nathan Benedetto Proença, Koppány István Encz and Monaldo Mastrolilli

Zwick's algorithm for Horn satisfiability shows that constraints of unbounded arity can admit a robust approximation guarantee independent of the arity. For finite constraint languages, Barto and Kozik proved that robust approximability is characterized by bounded width. We ask whether arity-independent robustness persists beyond width one and show that it fails already at width two. For Majority-closed Boolean linear constraints of maximum arity $k \ge 2$, we give a randomized polynomial-time algorithm that, without knowing $\varepsilon$, violates an expected $O(\sqrt{\varepsilon \log k})$ fraction of the constraint weight on $(1-\varepsilon)$-satisfiable instances. Under the Unique Games Conjecture (UGC), a matching NP-hardness lower bound of $\Omega(\sqrt{\varepsilon \log k})$ holds in an explicit parameter regime. Under this assumption, the bounded-width characterization therefore does not extend uniformly to constraints of unbounded arity. The algorithm rounds a degree-eight Sum-of-Squares (SoS) relaxation with a single Gaussian threshold. Since polynomial size alone does not make SoS solvable in polynomial bit complexity, we construct complete truncated Groebner bases for the soft Majority ideal and show that, at every fixed degree $2d$, the relaxation can be optimized to any rational accuracy in polynomial time with exactly feasible solutions, and degree-$2d$ SoS proofs can be found after an additive perturbation at degree at most $4d+10$. The lower bound combines Raghavendra's gap-to-hardness theorem with an integrality gap on a Gaussian star, analyzed via the Isaksson-Mossel theorem that parallel halfspaces maximize the joint membership probability of exchangeable Gaussians.
Read the full abstract

Machine Learning

Scale Sensitivity in Low-Bit Post-Training Quantization: Curvature of the Quantization Error Landscape

Jonas von Berg, Massimiliano Datres, Carlo Kneißl et al. (+1 more)

Post-training quantization (PTQ) methods in the GPTQ family minimize a layer-wise reconstruction error on a uniform grid whose scale must be chosen; the common max-based choice degrades sharply at low bit-widths. We study how sensitive this objective is to the scale. For a layer with i.i.d. Gaussian weights and calibration activations of sufficiently large effective rank, we prove that, as the width grows, the normalized round-to-nearest loss converges with high probability, uniformly over all scales, to the mean-squared error of a uniform quantizer applied to a standard Gaussian; we verify the effective-rank condition for wide, randomly initialized MLPs with odd Lipschitz activations and isotropic Gaussian calibration data. The limiting objective has a unique nondegenerate minimizer, whose scale decreases strictly with the number of levels and whose curvature with respect to relative scale errors decays approximately exponentially with the bit-width. GPTQ experiments on five LLMs show the same trend: the scale rule changes perplexity substantially at 2–3 bits and negligibly from 6 bits on, and a local measure of GPTQ scale sensitivity decreases with bit-width in line with the Gaussian curvature. The Gaussian-optimal scale fails on raw weights; after Hadamard incoherence processing it matches the best searched rule at 3 bits and above without any search, but remains clearly worse at 2 bits.
Read the full abstract

All authors Jonas von Berg, Massimiliano Datres, Carlo Kneißl, Gitta Kutyniok

Signal Processing

A Low-Complexity Message-Passing Receiver for FTN-WAFT under Nonlinear Distortion

Xianle Dai, Qu Luo, Jianguo Li et al. (+4 more)

This letter proposes a faster-than-Nyquist (FTN) weighted affine Fourier transform (WAFT) transmission scheme, termed FTN-WAFT, which combines the peak-to-average power ratio (PAPR) flexibility of WAFT with FTN time compression to improve spectral efficiency and power amplifier (PA) efficiency over high-mobility channels. However, operating the PA at a reduced input back-off (IBO) inevitably introduces nonlinear distortion, which further couples with FTN-induced inter-symbol interference (FTN-ISI) and doubly selective channel dispersion. To address this practical impairment, a multilayer message-passing (MLMP) receiver aided by PA interference cancellation (PA-IC), termed PA-IC-MLMP, is proposed. The proposed receiver reconstructs the dominant PA distortion from soft symbol estimates, cancels the channel-propagated nonlinear interference, and performs MLMP over the WAFT, FTN, channel, and constellation layers. Simulation results show that the proposed receiver improves the bit error rate (BER) and effective throughput under nonlinear PA constraints while maintaining low complexity.
Read the full abstract

All authors Xianle Dai, Qu Luo, Jianguo Li, Bowei Xue, Kai Yang, Luciano Leonel Mendes, Pei Xiao

Biomolecules

Leveraging secondary-structure information for accurate nucleic acid structure prediction with OFoldNA

Valhalla Team

Recent advances in biomolecular structure prediction have enabled accurate modelling of increasingly complex molecular systems. However, nucleic acid structure prediction remains challenging because of conformational flexibility and the limited availability of high-quality 3D structural data. Secondary structure (SS) provides a more readily available layer of structural information that captures base-pairing relationships and folding topology. Here we present OFoldNA, an all-atom diffusion model that incorporates SS information into nucleic acid folding and protein–nucleic acid co-folding. Without external SS information, OFoldNA achieved leading performance on FoldBench for both nucleic acid monomer folding and protein–nucleic acid co-folding, with particularly strong performance on DNA monomers and protein–DNA interfaces involving longer nucleic acid chains. When accurate base-pairing information was provided, OFoldNA-SS2TS further improved both folding and co-folding accuracy, while partial SS information also yielded consistent gains. The same auxiliary branch can also be used for RNA SS prediction as OFoldNA-SS, which achieved the best out-of-distribution performance on CHANRG. Together, these results show that intermediate structural information such as nucleic acid SS can be leveraged to improve all-atom 3D modelling, providing a general direction for incorporating complementary structural modalities into molecular structure prediction and design.
Read the full abstract

Instrumentation and Detectors

Quantitative Interleaved-Pulse Acquisition Extends the Range of Isotope-Ratio Measurements in ToF-SIMS

Anton V. Ievlev

Isotope ratios underpin tracer studies of transport in oxides, battery materials, and biological systems, yet time-of-flight secondary ion mass spectrometry (ToF-SIMS) struggles to quantify ratios whose members differ in intensity by orders of magnitude: a primary-ion pulse long enough to count the minor isotope precisely can saturate the major isotope, whereas a pulse short enough to keep the major isotope linear yields too few minor-isotope counts. We show that multi-pulse-width acquisition yields quantitative isotope ratios when a dose factor measured within each crater links the phases and each isotope is taken only from phases in which it remains linear; we call this quantitative reconstruction interleaved-pulse acquisition. Pulse widths are selected from measured transfer curves and peak shapes. On thermal SiO2, a tuned (6, 35) ns pair achieves the same per-layer precision as short-pulse acquisition with a 13-fold reduction in the number of analysis frames. On 18O-enriched WOx films, interleaving extends the quantifiable range at the low-fraction end by about 14-fold. Across isotope ratios spanning approximately 1:500 to 1:1, the interleaved reconstruction agrees with the corresponding linear reference measurements, showing no measurable bias introduced by interleaving, and natural-abundance ratios of oxygen, strontium and titanium are reproduced within several percent. The method requires no hardware additions and enables depth-resolved isotope-ratio measurements over a wide range within a single crater.
Read the full abstract

Econometrics

Debiased Inference for Bounding Wage Inequality with Many Controls

Yaroslav Korobka and Vira Semenova

We study estimation and inference for a partially identified parameter whose identified set depends on a first-stage nuisance parameter that must itself be estimated. Combining the criterion-function approach with the theory of Neyman-orthogonal moments that underlies double/debiased machine learning, we propose a two-step procedure: the point-identified nuisance is estimated by flexible machine-learning methods, and the set-identified target is recovered as a level set of a sample criterion built from orthogonal moment inequalities with cross-fitting. When the contour level is bounded, we show that the resulting set estimator converges in Hausdorff distance at the parametric rate of the infeasible criterion built on the true nuisance. We further develop a subsampling procedure that delivers asymptotically valid coverage, provided the product of the first-stage estimation errors is $o(N^{-1/2})$. We illustrate the method on bounds for the wage distribution and the interquantile range under selection into employment and on the gender wage gap with an interval-censored wage. The empirical application studies the gender wage gap using the March supplement of the 2015 Current Population Survey.
Read the full abstract

Mathematical Physics

A posteriori controllability of non-linear remodelling in linear elastic continua

Salvatore Di Stefano and Marta Zoppello

We investigate the a posteriori controllability of remodelling within the framework of Geometric Control Theory applied to continuum systems. Remodelling is described through an isochoric Bilby–Kroner–Lee decomposition, where the internal structural transformation is represented by a remodelling tensor and its evolution is governed by a stress-driven constitutive law. Assuming infinitesimal deformation and remodelling stretches, while retaining finite rotations of the principal remodelling directions, we derive a nonlinear state-space representation in terms of principal remodelling stretches and orientation variables. This formulation allows the remodelling process to be interpreted as a finite-dimensional nonlinear control system parametrised by the material position. Equilibrium configurations are characterised, and under-actuated controllability problems are studied by means of the tools of Geometric Control Theory. The three-dimensional setting shows that the nonlinear coupling between remodelling stretches and orientations enables complete steering of the internal state through a reduced number of independent controls, while the planar reduction reveals a loss of controllability despite the presence of nonlinearity. The results provide a first systematic investigation of remodelling controllability and further support the development of Continuum Control Theory as a framework for the analysis and design of controlled internal structural transformations in continuous media.
Read the full abstract

Combinatorics

A counterexample to Nagamochi's scoring lemma and a new rectangle packing bound

Hakan Karakuş

Let $s(N)$ denote the smallest side length of a square containing $N$ unit squares with arbitrary orientations and pairwise disjoint interiors. Nagamochi's Packing Unit Squares in a Rectangle (2005) states a rectangle packing bound from which he deduces two infinite families of exact values: $s(k^2-1) = k$ and $s(k^2-2) = k$ for every integer $k \geq 2$. We construct a family of counterexamples, local to a corner of the container, to the scoring assertion in Nagamochi's Lemma 1. These counterexamples show that the published proof of the rectangle bound is incomplete, but do not disprove the bound itself. We then give an independent proof of a weaker rectangle bound using a strip measure. This recovers $s(k^2-1) = k$ for every integer $k \geq 2$ and yields an explicit lower bound for $s(N)$ that improves strictly on the area bound for every nonsquare integer $N \geq 8$. Our argument does not establish Nagamochi's full rectangle bound or the identity $s(k^2-2) = k$.
Read the full abstract

Plasma Physics

Global Phase Locking in Superradiant Chorus Emissions

Xin Tao, Fulvio Zonca and Liu Chen

Coherent wave amplification in inhomogeneous systems is usually described in terms of resonance and nonlinear phase locking, but these local phase conditions do not determine how the wave amplitude evolves when the resonance condition changes during wave propagation. Here we identify a global phase-locking principle, in which the stable resonant phase remains nearly unchanged as the resonance detuning evolves. This condition requires the wave amplitude to track the evolving detuning and leads to linear amplitude growth with distance, producing the quadratic intensity scaling characteristic of superradiant emission. We demonstrate this mechanism using self-consistent kinetic simulations of whistler-mode chorus waves, one of the most important coherent electromagnetic emissions in Earth's magnetosphere. The simulated chorus amplitude and effective growth rate follow the predicted superradiant scalings, rather than exponential growth. Direct phase-space analysis further shows that the resonant electron trapping island remains centred at an approximately fixed phase during amplification. These results identify global phase locking as the dynamical origin of superradiant chorus growth and reveal a general phase-organization principle for coherent radiation in evolving resonant systems.
Read the full abstract

Computation and Language

Look What You Made Us Cluster: Hate Narrative Extraction from Reddit Discourse

Annabelle K. L. Chua, Forster J. Khoo, Joel C. R. Tan et al. (+8 more)

Narrative extraction allows us to identify online hate narratives, supporting the construction of rigorous detection systems. Existing computational approaches, however, are limited in precision as they rely on semantic representations, which tend to capture only surface-level meaning. To detect more precise and interpretable narratives, we present an extraction pipeline that represents narratives as entity-evaluation pairs. Narratives are extracted using a Large Language Model (LLM) reasoning process that extends Aspect-Based Sentiment Analysis, identifying the aspect, classifying its judgement type as the basis for evaluation, and deriving the evaluation accordingly. Extracted narratives are then clustered using Leiden, following which clusters are resolved to an intended level of granularity through an LLM-guided refinement process. We illustrate this narrative pipeline with English Reddit comments from 2024 that criticize Taylor Swift, analyzing a representative cluster that exhibits hate speech patterns to demonstrate its interpretive value.
Read the full abstract

All authors Annabelle K. L. Chua, Forster J. Khoo, Joel C. R. Tan, Huey Ting Ang, Kheng Hwee Tan, Joel Y. A. Sim, Shirley W. H. Ow, Ria Mundhra, Elsie C. K. Toh, Youfeng Xu, Lynnette H. X. Ng

Computer Vision

Complementary Retrieval-Augmented Prompting for Consistent Long-Form Video Generation

Xianghan Wei, Xiaoda Yang, Zhi Wang et al. (+7 more)

While recent video foundation models excel at generating high-quality short videos, long-form video generation remains a critical challenge, where a major bottleneck lies in conditioning independently generated shots to preserve consistent characters, scenes, and objects throughout a story. Existing training-free approaches typically condition target shots using retrieved historical visuals. However, these references often suffer from severe informational mismatch, either introducing irrelevant contextual redundancy or failing to provide the full combination of required elements for the target shot. To resolve this, we present Complementary Retrieval-Augmented Prompting, an agentic framework that strategically aggregates a compact set of mutually supportive historical references to achieve complete and targeted conditioning for long-form video generation without retraining or modifying the underlying generator. Specifically, our framework explicitly models the visual elements required by each target shot by parsing the narrative script into a text-grounded visual element registry that tracks characters, objects, scenes, and their shot-level states. A VLM-annotated keyframe library further maps these elements to past visual observations. Guided by the required elements, our agent retrieves complementary references that maximize target-element coverage while minimizing historical noise. Finally, the retrieved references, structured element states, and grounding instructions are assembled into a unified prompt for the frozen video generator. This element-aware process provides comprehensive conditioning while remaining fully interpretable. Quantitative and qualitative evaluations on multi-shot story generation demonstrate that our method consistently outperforms recent-frame, memory-based, and entity-level retrieval baselines in cross-shot consistency and text-controllability.
Read the full abstract

All authors Xianghan Wei, Xiaoda Yang, Zhi Wang, An Pan, Daoan Zhang, Huayi Zhang, Yan Zhang, Wei Xu, Zishun Liao, Jianwen Lou

Materials Science

Topological Hall Effect Induced by Chiral Spin Textures at the Ferroelectric/Ferromagnetic Interface

Jingkuan Xiao, Yaqing Han, Jianfeng Guo et al. (+25 more)

Chiral spin textures, largely driven by the Dzyaloshinskii-Moriya interaction, offer significant potential for next-generation computing technologies due to their chirality and topological stability. Ferroelectric/ferromagnetic van der Waals heterostructures are particularly appealing because they can combine interfacial inversion-symmetry breaking and spin-orbit coupling to promote interfacial Dzyaloshinskii-Moriya interaction, while switchable ferroelectric polarization provides a nonvolatile tuning knob. This study investigates interfacial chiral spin textures in few-layer Fe$_3$GeTe$_2$/$\alpha$-In$_2$Se$_3$ heterostructures. Two groups of topological Hall signals are identified just below and above the coercive field, and thickness-dependent transport reveals a notable reduction in critical temperature with increasing Fe$_3$GeTe$_2$ layer thickness. Low-temperature magnetic force microscopy images reveal two types of magnetic bubbles with opposite magnetic contrasts near the coercive field, each associated with distinct topological Hall signals. Together with atomistic spin-dynamics simulations and first-principles calculations, these results support the formation of interfacial DMI-stabilized chiral spin textures. Switching the ferroelectric polarization of the $\alpha$-In$_2$Se$_3$ layer further enables nonvolatile modulation of both anomalous and topological Hall effects. The resulting ferroelectric and magnetic bistabilities generate four distinguishable Hall resistance states programmable by electric and magnetic fields. These findings highlight the potential of van der Waals interfaces for advanced device applications.
Read the full abstract

All authors Jingkuan Xiao, Yaqing Han, Jianfeng Guo, Renjun Du, Jiawei Jiang, Baoshan Cui, Runnong Zhou, Siqin Wang, Siqi Jiang, Fuzhuo Lian, Di Zhang, Guodong Ma, Jiabei Huang, Zhaochen Qu, Wanting Xu, Kenji Watanabe, Takashi Taniguchi, Alexander S. Mayorov, Jinsheng Wen, Haifeng Ding, Gong Chen, Ahmet Avsar, Hongxin Yang, Lihong Bao, Hong-Jun Gao, Shiyu Zhu, Lei Wang, Geliang Yu

Software Engineering

Complexity-Aware Evaluation of LLM Comprehension

Ali Mohammadi Esfahani, Nafiseh Kahani and Samuel A.Ajila

Large language models (LLMs) are increasingly used for software engineering tasks that require understanding existing source code, including behavior prediction, function explanation, debugging, and code review. However, aggregate benchmark accuracy can conceal how model reliability changes as source code becomes structurally more complex. This paper presents a complexity-aware framework for evaluating LLM code comprehension using cyclomatic complexity, nesting depth, branching factor, and Halstead volume. We evaluate DeepSeek-Coder-V2 and Llama through two complementary tasks: automatic input-output prediction over 300 Python functions and manually assessed semantic comprehension over a balanced subset of 60 functions. The functions are grouped into Low-, Medium-, and High-complexity bands. DeepSeek-Coder-V2 achieves an overall automatic accuracy of 78.33%, compared with 70.33% for Llama. However, accuracy decreases substantially from Low to High complexity, from 93.52% to 52.78% for DeepSeek-Coder-V2 and from 87.04% to 47.22% for Llama. Incorrect predictions are consistently associated with higher values of all four complexity metrics, and correlation and logistic-regression analyses confirm broadly comparable negative associations between structural complexity and correctness. Manual semantic comprehension shows the same degradation pattern, with accuracy decreasing from 100.00% to 75.00% for DeepSeek-Coder-V2 and from 90.00% to 60.00% for Llama. These findings demonstrate that complexity-aware evaluation provides a more diagnostic assessment of LLM code-comprehension reliability than aggregate accuracy alone.
Read the full abstract

Computational Physics

Geometry-Controlled Chern transfer and Flat Band Reconstruction in distorted kagome lattices

Yu Zhu, Claudia Felser and Xiaolong Feng

Flat band formation and nontrivial topology are central manifestations of kagome electronic structure, yet they are commonly discussed in idealized lattice geometries. In distorted kagome materials, structural deformation reorganizes electronic propagation pathways, but a microscopic understanding of how such distortions govern flat band dispersion and band topology is still lacking. Using a distorted kagome tight-binding model with rotated angle dependent long-range hopping and intrinsic spin-orbit coupling, we show that distortion reconstructs both the dispersion and topology of the kagome band manifold. The flat band descendant develops distinct bandwidth regimes associated with a redistribution of its extrema in momentum space. Simultaneously, symmetry-related band inversions generate quantized Chern number transfer whose parity is fixed by the multiplicity of the touching points, thereby determining the gap-resolved Z2 topology. The accompanying Berry curvature evolution produces characteristic anomalous Hall and Nernst responses. These results identify geometric deformation as a common microscopic origin of flat band and topological reconstruction in kagome systems.
Read the full abstract

Symplectic Geometry

The derived Picard group of Fukaya categories of closed surfaces

Dongjian Wu and Nantao Zhang

Let $\Sigma_g$ be a closed oriented surface of genus $g\geq2$, and let $\mathscr F_g$ be its split-closed, strictly unobstructed, two-periodic Fukaya category over the complex Novikov field $\Lambda$. We determine the derived Picard group \[ \operatorname{DPic}_{\Lambda}(\mathscr F_g) \cong \bigl(H^1(\Sigma_g;\Lambda^\times) \rtimes\pi_0\operatorname{Diff}^+(\Sigma_g)\bigr) \times\mathbb Z/2\mathbb Z. \] This proves the enhanced form of the conjecture of arXiv:2006.09689. As a corollary, we obtain that the derived Picard group itself determines the genus of the surface.
Read the full abstract

Artificial Intelligence

Routing Should Pay for Itself: Sparse Supervision for Economical LLM Routing

Guannan Lai, Gelin Bian, Hao-Xuan Ma et al. (+5 more)

Large language model (LLM) routing reduces serving cost by assigning each query to an appropriate model while preserving response quality. Learning such a router, however, often requires executing multiple candidate models on historical queries to collect query–model quality feedback, creating a nontrivial supervision cost before deployment. Existing work largely focuses on serving-time efficiency, overlooking whether the resulting savings are sufficient to recover this upfront expenditure. We further observe that routing quality often saturates well before all query–model feedback is collected, suggesting that dense supervision can be economically over-provisioned. We propose SaveRouter, a sparse-supervision routing framework that selectively acquires informative model feedback and shares capability information across related queries, while retaining query-level refinement for fine-grained routing. We evaluate routing by jointly accounting for supervision expenditure and subsequent serving-time savings. Across four routing benchmarks, the main setting uses only about 33–41% of available training feedback while maintaining competitive or better routing quality, and reduces the break-even deployment volume by approximately 1.9–9.5 times compared with the fastest conventional router. Further analysis shows that acquiring more supervision is not always economically preferable: the supervision level that minimizes serving cost can differ from the one that achieves the earliest payback. Our code is publicly available at https://github.com/LAMDA-Model-Reuse/SaveRouter.
Read the full abstract

All authors Guannan Lai, Gelin Bian, Hao-Xuan Ma, Jun-Peng Jiang, Long Chen, Jian-Dong Liu, Zhi-Hao Tan, Han-Jia Ye

High Energy Physics (Theory)

Quantum states in an AdS-invariant theory

Felix M. Lev and Valery E. Lyubovitskij

We note that the description of elementary particles in anti-de Sitter (AdS) quantum theory differs radically from standard approaches. Since the Cartan subalgebra of the AdS algebra has rank 2, it is necessary to consider only those elements that explicitly depend on the eigenvalues of operators in the irreducible representations (IRs) of the algebra. In the literature, AdS symmetry is widely invoked in various quantum applications, such as AdS/QCD. However, such frameworks typically do not take into account the unconventional properties of the AdS algebra IRs investigated in the present paper.
Read the full abstract

Computer Vision

BeatDance: Generating Beat-Consistent 3D Dance with Hierarchical Spatial-Temporal Modeling

Xiaojian Shen, Dahu Shi, Jianrong Zhang et al. (+7 more)

Generating realistic 3D dance from music is a challenging task that requires accurate synchronization with musical rhythms while capturing the spatial complexity of human motion. Although existing methods can generate physically plausible dance motions, they often struggle to achieve precise alignment with music, such as the beat. To address this limitation, we propose a novel diffusion-based framework, BeatDance, with two components: 1) We present a Hierarchical Decoupled Attention (HDA) module, which first disentangles the learning of human pose and temporal dynamics. A hierarchical structure is then employed to capture both short-term and long-term dependencies, thereby enhancing spatial-temporal modeling. 2) We adopt cycle-consistent learning by introducing an auxiliary dance-to-music module. During training, discrepancies between the reconstructed and original music induce a stronger loss signal, effectively encouraging the consistency property between the music and dance motion. Extensive experimental results demonstrate that our proposed approach outperforms recent competitive methods on two benchmark datasets.
Read the full abstract

All authors Xiaojian Shen, Dahu Shi, Jianrong Zhang, Hai Li, Hongwei Zhao, Dawei Zhang, Yunzhi Zhuge, Zhiliang Wu, Guanghui Yue, Wei Zhou

Hardware Architecture

MEDEM: Multi-Engine DL Accelerator Design Methodology

Fareed Qararyah, Mohammad Ali Maleki and Pedro Trancoso

Multi-engine deep learning (DL) accelerators are becoming increasingly prevalent as they address the heterogeneity and growing complexity of modern DL workloads. To efficiently process diverse DL workloads, these accelerators must incorporate combinations of engines with complementary capabilities to match the distinct computational characteristics of these workloads' heterogeneous kernels. However, existing multi-engine DL accelerator design approaches lack a systematic methodology, leaving fundamental questions unresolved. These include how to co-design engines for workloads with diverse computational characteristics and which engine combinations minimize aggregate execution costs (such as time or energy) across such workloads. Addressing these questions requires efficient exploration of exponentially large design spaces. To address these questions systematically, this work proposes Multi-Engine DL Accelerator Design Methodology (MEDEM). MEDEM defines generic engine abstractions, co-designs candidate instances (engines), and selects a combination of co-designed engines to minimize aggregate execution cost given diverse DL workloads and a resource budget. MEDEM encompasses a set of design strategies that efficiently navigate the exponentially large design spaces of engine co-design and combination selection, identifying highly optimized multi-engine accelerators. A comprehensive evaluation demonstrates that MEDEM identifies accelerators that outperform state-of-the-art designs, delivering geometric-mean improvements of up to 4.84x in energy-delay product (EDP) and 1.59x in throughput. The improvements are achieved using different resource budgets, demonstrating MEDEM's scalability, and using 51 single- and multi-model DL workloads, demonstrating its generalizability.
Read the full abstract

Artificial Intelligence

Direct Experience World-Model Optimization: Learning the World Beyond Action Imitation

Xiangcheng Zhan, Zirui Chen, Yicheng Zhao et al. (+2 more)

World-Action Models (WAMs) couple action generation with predictions of how physical interactions unfold. However, current post-deployment learning paradigms typically improve behavior without requiring better world predictions. Especially in dexterous manipulation, small execution errors can compound in high-dimensional action spaces, hindering policy improvement and pushing interactions beyond the world model's training distribution. Motivated by this, we propose Direct Experience World-Model Optimization (DEWO), a post-deployment learning paradigm for WAMs that, alongside action imitation, refines world representations through visual experience to better condition action generation. Specifically, it identifies interaction turning points and learns from successful and failed futures to support classifier-free guidance. An additional value head estimates task progress from video representations and activates guidance when progress stalls during inference. Across five DexJoCo tasks, DEWO improves average success across all three WAM formulations. Ablations show that visual supervision from successful and failed continuations improves both prediction and control beyond action supervision alone. On four real-world tasks across Wuji and Sharpa, 3 x 3 grid evaluations show that two rounds of deployment learning increase success from 51.0% to 71.7% in cells with at least one initial success, a gain of 20.7 percentage points. These findings support continued predictive learning for improving control through deployment experience, making world modeling an active part of WAM adaptation.
Read the full abstract

All authors Xiangcheng Zhan, Zirui Chen, Yicheng Zhao, Ziteng Gao, Shuo Yang

Analysis of PDEs

Sharp stability and instability of stratified steady states for the incompressible porous media equation

Seyed Abdolhamid Banihashemi, Sepehr Mohammadkhani and Huy Q. Nguyen

We study the stability and instability of stratified steady states $\rho_s=\rho_s(y)$ for the two-dimensional incompressible porous media equation on $\mathbb{T}\times(-1,1)$ and $\mathbb{T}\times\mathbb{R}$. On the periodic channel, uniformly decreasing steady states satisfying natural boundary-compatibility conditions are nonlinearly stable under small $H^m$ perturbations, for every integer $m>2$. The solutions converge in $L^2$ to the measure-preserving stratification of the initial density at the rate $t^{-m/2}$. Conversely, every steady state with $\sup\rho_s'>0$ is nonlinearly unstable in $H^m$. On the infinite cylinder, we prove nonlinear instability under the additional gap condition \[ \sup_{\mathbb{R}}\rho_s' >\limsup_{|y|\to\infty}\rho_s'(y). \] In both settings, the instability is generated by positive eigenvalues of the linearized operator converging to $\sup\rho_s'$, which equals both its spectral bound and semigroup growth bound.
Read the full abstract