Numerical Analysis
Convergence of isoparametric BGN methods for mean curvature flow of closed surfaces with high-order finite elements
Yifei Li
We prove the convergence of the isoparametric Barrett–Garcke–Nürnberg (BGN) method for mean
curvature flow of closed surfaces. For finite elements of degree $k\ge6$ and time steps $\tau=O(h^2)$, the numerical surfaces converge to the exact surfaces in the
manifold distance with order $h^2$. The proof relies on two ingredients. First, we construct approximation flows by an expansion in time, whose interpolations satisfy the BGN method with an improved defect. Second, we derive a refined estimate for the tangential error from an algebraic identity for tangential test functions.
Read the full abstract ▾
Robotics
HULK: Learning Whole-Body Forceful Loco-Manipulation for Humanoids
An Dang, Arturo Flores Alvarez, Yu-Ming Chen et al. (+5 more)
Humanoid loco-manipulation of large, heavy objects demands forceful interaction across the entire body. However, such payloads shift a
humanoid's center of mass and impose sustained loads across the upper body, challenging balance and command tracking. We present HULK, a whole-body control framework for forceful loco-manipulation. Using model predictive control (MPC) to guide reinforcement learning with predictions of the loaded dynamics, we train two teachers: one tracks arm motions under wrist forces, and the other locomotes while holding large objects against the body. A capture-point control barrier function augments the wrist-force teacher during training to improve balance under load. We distill both teachers into a single policy. Evaluation spans simulation and the Unitree G1. In simulation, the teacher with the barrier function achieves the lowest forward and lateral velocity tracking errors at 10 kg per arm among evaluated controllers and reduces aggregate divergent component of motion (DCM) excursion magnitude by 35.7% relative to MPC-guided reinforcement learning alone. Our wrist-force teacher withstands torso push disturbances of up to 130 N.
Read the full abstract ▾
All authors An Dang, Arturo Flores Alvarez, Yu-Ming Chen, Conor Mc Gartoll, Helen Sun, Aaron Ames, Nima Fazeli, Manikantan Nambi
Machine Learning
Work While They Sleep: Exploiting Evaluation Latency for Fully Bayesian Optimization
Gustavo Sutter, Alejandro Comas-Leon, David Holzmüller et al. (+4 more)
Black-box optimization problems are ubiquitous across science and engineering, often dealing with expensive objective functions. This objective latency has two consequences during optimization: (i) the objective evaluation dominates execution time, and (ii) sample-efficient
algorithms are crucial to accelerate development and avoid wasting resources.
Bayesian optimization (BO) methods are the de facto choice of planners for suggesting the next point to try. Standard BO fits the surrogate model's hyperparameters with a point estimate. Alternatively, a fully
Bayesian approach uses model averaging to account for uncertainty over the hyperparameters, leading to better uncertainty estimates—useful in the low-data regime that is pervasive in BO. However, it is often prohibitively expensive and thus rarely used. In this work, we propose ELF-BO, an
algorithm that uses the objective evaluation latency to headstart the computation of the next suggestion, allowing for fully
Bayesian optimization without incurring substantial decision-time costs. This is done by sampling from the hyperparameter posterior while the objective is being evaluated, only requiring reweighting of the samples once the objective value is observed. Across synthetic functions and real-world applications, we show that ELF-BO matches the performance of fully
Bayesian methods while only incurring decision latency on par with or better than standard BO. Thus, ELF-BO makes fully
Bayesian optimization practical in real-world use cases.
Read the full abstract ▾
All authors Gustavo Sutter, Alejandro Comas-Leon, David Holzmüller, Hao Wang, Luis Ricardez-Sandoval, Pascal Poupart, Agustinus Kristiadi
Game Theory · Cross-listed from econ.TH
Large player classes and approximately weighted voting
Ezra Einy and Ori Haimanko
It is known that completeness of the coalitional ($C$-)desirability relation in a simple game does not imply the existence of a weighted-majority representation. We show that weightedness is implied in an approximate sense. Call an $n$-player simple game $\varepsilon$-weighted if, for some weight vector in the $(n-1)$-simplex, any losing coalition weight exceeds any winning coalition weight by at most $\varepsilon$; let $\varepsilon^{\ast}$ be the smallest such $\varepsilon$. We establish an upper bound on $\varepsilon^{\ast}$ in a general $C$-complete game, and deduce that, if all classes of equally desirable players become large asymptotically but the number of classes is bounded, then $\varepsilon^{\ast}=O(1/n_{\min})$ for the minimal player class size $n_{\min}$, and the bound is tight. Importantly, the minimal attainable fraction $\delta$ of coalitions mislabeled by a weighting rule also tends to $0$ when $n_{\min}\rightarrow\infty$ and the number of player classes is bounded, with $\delta=O(1/\sqrt{n_{\min}})$.
Read the full abstract ▾
Artificial Intelligence
Socio-Foundation: A Model for Generalizable Individual Behavior Simulation via Hierarchical Capability Distillation
Liang Wang, Wenxuan Xie, Xinyi Mou et al. (+2 more)
Simulating individual behavior requires large
language models (LLMs) to preserve persona traits while adapting to dynamic social contexts. However, general-purpose LLMs often flatten distinct personas, while task-specific tuning suffers from fragmentation and generalization. To overcome these challenges, we organize individual simulation into the FONTS Taxonomy, comprising five complementary capability dimensions: persona fidelity (F), outcome realization (O), behavioral naturalness (N), trajectory coherence (T), and social grounding (S). Grounded in this taxonomy, we curate a standardized training corpus library of approximately 10 million instances across 14 representative
datasets and present Socio-Foundation. Socio-Foundation decouples specialization from integration via a three-stage pipeline: learning task experts via DAPO, consolidating them into capability experts via off-policy
distillation, and unifying them via multi-teacher on-policy
distillation (MOPD). We also establish IndiEval, consolidating 29 metrics across the FONTS dimensions. Experiments show that Socio-Foundation outperforms its Qwen3-8B base by 11.0 points and approaches frontier models such as GLM-5.2, with ablations and out-of-distribution evaluations further demonstrating the effectiveness and generalization of our model.
Read the full abstract ▾
All authors Liang Wang, Wenxuan Xie, Xinyi Mou, Yixin Luo, Zhongyu Wei
Artificial Intelligence
Humanity's Sixth Sense: Benchmarking Intuitive Visual Reasoning in Multimodal Models
Xingang Guo, Jing Gu, Brian Jang et al. (+26 more)
Humans perceive far more in a scene than what is explicitly depicted: a single glance captures past causes and future trajectories; a quick peek determines if a vehicle can fit between two parked cars; a few seconds of
video reveals who holds authority in a room; and a fleeting clip highlights subtle abstract patterns like unwritten rules or hidden labels. This capacity reflects a form of humanity's sixth sense: an intuitive
reasoning mechanism that recovers implicit information beyond raw sensory perception. Crucially, this rapid, zero-shot visual intuition underpins everyday
navigation and social interaction, making it a vital capability for
Multimodal Large
Language Models (MLLMs) deployed alongside people. Existing visual
benchmarks, however, target either deliberate expert-level analysis in academic and mathematical domains or low-level perception, leaving the intuitive
reasoning that people perform largely untested. To bridge this gap, we introduce Humanity's Sixth Sense (HSS), a
benchmark for intuitive visual
reasoning. HSS spans diverse image and
video inputs, organizes items under a structured taxonomy, and pairs each with human-written
prompts probing the implicit temporal, spatial, social, and abstract structure that people infer at a glance. Frontier MLLMs fall short of human performance: participants reach 93.1% accuracy, while the strongest model, GPT-6-astra, reaches only 53.6% even at maximum
reasoning effort. Despite excelling in many complex tasks that require advanced perception and knowledge, current models still struggle significantly on these visual tasks that are intuitive for humans. We further explore agentic setup that apply dynamic visual manipulation to HSS, which narrows but does not close the gap. HSS establishes intuitive visual
reasoning as a measurable axis and directs
attention to a capability that scaling on current
benchmarks has so far left behind.
Read the full abstract ▾
All authors Xingang Guo, Jing Gu, Brian Jang, Renxiong Wang, Utkarsh Tyagi, Daniel Quigley, Steven Li, David Yan, Daniel Yue Zhang, Darvin Yi, Forrest Huang, HiJae Kim, Tianyi Zhang, Jared Lichtarge, Jihua Huang, Le Xue, Manan Tomar, Qiuyi Richard Zhang, Ruofei Yu, Seth Neel, Yaning Hu, Marcella Valentine, Xinzhe Jiang, Daniel Evans, Chenguang Wang, Dustin Tran, Tong Zhao, Yinfei Yang, Yunzhong He
Logic in Computer Science · Cross-listed from math.NT
More than 83.9% of the zeros of the Riemann zeta function are distinct and more than 67.35% are simple and on the critical line
Kristian Muri Knausgård
Let N(T) count the nontrivial zeros of the Riemann
zeta function up to height T with multiplicity, N_d(T) the distinct ones, and N_0^s(T) those that are simple and on the critical line. We prove liminf N_d(T)/N(T) >= 1645064/1960733 = 0.83900...; earlier work proves 0.83699..., and a report we have not verified claims 0.83805.... Hence more than 67.80% are simple. Also liminf N_0^s(T)/N(T) >= 0.67353...; earlier work proves 0.67250..., and reports we have not verified claim 0.673492. Two estimates of Lamzouri are improved: at least 88.93% of the zeros are simple or on the critical line, and the proportions of simple zeros and of zeros on the line average at least 83.67%. By the unconditional form of Montgomery's pair-correlation
theorem the energy, a sum of a test function over pairs of zeros, is known asymptotically. A zero of multiplicity d contributes d^2 to it and nearby zeros contribute too, so a lower bound for such pairs leaves less energy for multiple zeros. The proof has three steps. First, for zeros at least a fixed fraction of the mean spacing apart, a large sieve inequality lets their pairs be used in full, with multiplicities. Second, their contribution is bounded below by a computer-assisted inequality for seven or eight consecutive zeros which distinguishes simple and double zeros. Its correction terms
telescope, as increments of a storage function. Third, the test function is chosen to make this contribution large, at the cost of more energy. We also prove upper limits for what such inequalities can give with the two main test functions. The lower bounds and these limits are proved in Lean 4, analytic inputs included. The proofs use three axioms beyond those of Mathlib, one per inequality. Each records that a search program returned true; Lean's kernel does not check the run itself. This work is an experiment in AI-assisted mathematical research.
Read the full abstract ▾
Machine Learning
On KL-Regularized Policy Optimization
Yifan Zhang
Asynchronous reinforcement learning (RL) for large
language model (LLM)
agents trains one policy on trajectories generated by another: rollouts come from stale checkpoints, and the inference engine's probabilities differ from the trainer's even at identical parameters. Standard remedies either clip importance ratios, which biases the update, or, as in GRPO, sample a group of responses per
prompt, which is costly when episodes are long. We propose KL-Regularized Policy Optimization (KLPO), a framework that anchors the KL regularizer at the sampler. The regularized improvement step then has a closed-form Gibbs solution, and KLPO fits its log-ratio optimality condition by least squares on the sampler's own trajectories, so the sampler probability enters through a log-ratio and no importance weights are needed. Profiling out the regression intercept replaces the intractable log-partition function with the signal's sampler mean plus a sampler-to-trainer KL divergence. For
token-level policy mirror descent targets, we show that the resulting gradient can be computed from terminal returns without a critic, via sampler-centered scores or a single trajectory residual, even under
stochastic tool outputs. We further prove that independent Monte Carlo estimates of the KL term keep these gradients unbiased, derive the exact KL gap of cheaper top-$K$ and binary approximations, and show that SPPO, GPO, REBEL, and BPO arise as special cases of KLPO. The result is a critic-free update that uses one rollout per
prompt and requires neither a learned normalizer nor a group of responses.
Read the full abstract ▾
Programming Languages
Automatically Detecting and Fixing Deadlocks in Go Code with GoDDaR
Gustavo Feio, Jorge Coelho, Miguel Flor et al. (+2 more)
The Go programming
language provides a lightweight abstraction for concurrent programming through goroutines, which are prone to deadlocks. Go includes a runtime detector that aborts execution when all threads are blocked (a global deadlock). However, due to nondeterministic thread scheduling, this runtime mechanism only detects global deadlocks that manifest during execution and cannot identify partial deadlocks, where a subset of goroutines is permanently blocked while at least one remains runnable. Statically detecting local deadlocks is essential for developing dependable concurrent software. While several tools statically detect deadlocks in concurrent programs, few assist developers in fixing them. Detecting and resolving partial deadlocks requires
reasoning about complex interleavings and communication patterns, an inherently challenging task. In this paper, we present GoDDaR, a tool that detects unobserved global or partial deadlocks in Go programs and suggests concrete fixes. GoDDaR translates Go source code into an intermediate representation reflecting message-passing communication events and composition patterns. Partial deadlocks are identified through symbolic analysis over this representation. To resolve detected deadlocks, our
algorithms transform the intermediate representation to eliminate problematic synchronization patterns, allowing GoDDaR to generate fix suggestions for the original Go source code as a diff. We evaluate GoDDaR on a
benchmark suite of representative deadlock examples from literature and git repositories to validate
algorithm correctness and compare performance against the state of the art. Results demonstrate that GoDDaR is competitive with top approaches in partial deadlock detection and advances the state of the art in automated deadlock fixing.
Read the full abstract ▾
All authors Gustavo Feio, Jorge Coelho, Miguel Flor, João M. Lourenço, António Ravara
Data Structures and Algorithms
Analysis of the two-for-one swap heuristic for approximating the maximum independent set in a k-polymatroid
Adrian Calinescu and Gruia Calinescu
Let f:2^N –> \cZ^+ be a polymatroid (an integer-valued non-decreasing submodular set function with f(emptyset) = 0). A k-polymatroid satisfies that f(e) <= k for all e in N. We call a subset S of N independent if f(S) equals the sum of f(e) over the elements e of S and f(e) > 0 for all e in S. Finding a maximum-size independent set in a 2-polymatroid has been studied and
polynomial-time
algorithms are known for linear polymatroids. For k >= 3, the problem is NP-hard, and an approximation
algorithm with ratio approaching 2/k is known and is obtained by swapping as long as possible a "large" subset from the current solution by a set with one more element. Here we give a simple analysis of the more particular two-for-one repeated swapping heuristic, obtaining a (weaker) 2/(k+1)-approximation.
Read the full abstract ▾
Machine Learning
Directed Temporal Representations for Offline Visual Control
Chenyang Yuan, Haoyu Wang, Zhuo Sun et al. (+1 more)
Predictive world models provide compact visual representations for control. Control requires a latent geometry aligned with temporal reachability rather than predictive similarity alone. We introduce Directed Temporal Representations for Control (DTRC), which learns such a geometry from offline visual trajectories on top of frozen LeWorldModel (LeWM) features. DTRC constructs a directed temporal quasimetric over the learned control representation. Short-range temporal offsets calibrate the distance scale. Bootstrapped targets extend temporal reachability across longer horizons. Action-conditioned consistency aligns the representation with local transition dynamics. The resulting distance estimates temporal reaching cost, and its change across a transition defines goal-relative temporal progress. We use this progress signal as a temporal critic for direct goal-conditioned policy learning. Model-assisted targets provide an additional training-time refinement under behavior-support and dynamics-agreement constraints. Across ten visual control tasks, DTRC achieves strong goal-conditioned control performance relative to planning and direct-policy baselines. Held-out diagnostics on the four LeWM tasks show consistent short-range temporal calibration, task-dependent long-range and directional structure, and positive transition-level progress. Temporal supervision improves the same flow-policy parameterization across all four LeWM tasks, while the resulting policy acts directly without iterative trajectory search at test time.
Read the full abstract ▾
All authors Chenyang Yuan, Haoyu Wang, Zhuo Sun, Xiaoyuan Cheng
Machine Learning
GraphOPD: Graph-Augmented On-Policy Distillation for LLM Agents
Bohan Lin, Liyi Chen, Zhuoning Guo et al. (+5 more)
On-policy
distillation post-trains large
language model
agents by supplying dense, step-level guidance from a teacher policy when the reinforcement-learning reward is sparse and arrives only once per trajectory. Existing instantiations allocate this guidance by the size of the teacher-student divergence at each step, on the single-turn intuition that a large disagreement marks a mistake worth correcting. Once decisions chain over many turns, that rule misfires, since an early drift enters every later context both policies condition on, leaving the teacher consistent with the drifted trajectory instead of flagging its cause, while interchangeable steps register large but outcome-irrelevant divergences. We demonstrate this on an agentic
benchmark, where distilling the highest-divergence steps brings no consistent benefit over random selection. To this end, we introduce GraphOPD, the first method to bring graph-based structural augmentation into on-policy
distillation for
agent capabilities. It reads which steps enabled which later ones from the environment's own record of state changes, immune to the drift that corrupts the teacher-student gap, organizes them into a dependency graph, scores each step by a random-walk stationary distribution over it, and fuses that structural credit with the divergence signal into a trajectory-relative mask concentrating supervision on each rollout's highest-aptitude steps. Across three model scales and eleven baselines on ALFWorld, WebShop, and SearchQA, GraphOPD shows competitive performance throughout, improving over the strongest baseline by up to +5.8 pp. An executed-replay audit further shows that this structural credit score tracks true causal impact far above chance, that both fused signals are independently necessary, and that the same signal transfers to out-of-domain tool-integrated
reasoning.
Read the full abstract ▾
All authors Bohan Lin, Liyi Chen, Zhuoning Guo, Muyang Li, Qimeng Wang, Yan Gao, Yao Hu, Yudong Zhang
Numerical Analysis
Numerical Bogoliubov approximation of bosonic many-body quantum dynamics
Yoann Le Hénaff, Christian Lubich and Peter Pickl
We propose an
algorithm for the numerical solution of the time-dependent Schrödinger equation for N >> 1
bosons interacting via a two-body interaction potential in the mean-field scaling regime. Unlike existing numerical
algorithms for
quantum dynamics, the proposed
algorithm has a computational complexity that is independent of the number N of particles. It is based on an ansatz for the wave function inspired by the Bogoliubov theory, which is an asymptotic theory for large N. In a first part, we follow Bogoliubov theory in evolving the condensate function by the Hartree equation, but we use the Dirac–Frenkel time-dependent variational principle to derive nonlinear equations of motion for the excitation functions in the Bogoliubov ansatz. By carefully exploiting the bosonic
symmetry, the so obtained approximate wave functions are shown to have derivatives bounded independently of N apart from a time-dependent global phase factor which rotates with a frequency proportional to N. In the second part, we numerically approximate the wave function (modulo the phase) by a composite
algorithm, which employs an appropriate splitting of the equations of motion and uses a dynamical low-rank approximation of the fundamental two-particle excitation function by linear combinations of products of single-particle functions. The low-rank excitation function is evolved in time by a basis update and Galerkin (BUG) integrator that is robust to small singular values of the coefficient
matrix and allows for rank adaptivity. Numerical results are provided to assess the performance of the proposed
algorithm.
Read the full abstract ▾
Computer Vision
RACER: Reflective Agent Coupling Query Interpretation and Tool-Based Retrieval for Frame Selection in Long Video Understanding
Yiyang Huang, Yitian Zhang, Yizhou Wang et al. (+6 more)
Video large
language models (Vid-LLMs) excel at diverse
video-
language tasks by
reasoning over selected frames. However, frame selection for long
videos remains challenging, as it requires retrieving relevant frames distributed across segments from a large candidate pool given complex queries. This paper investigates dominant approaches to long-
video frame selection from a task-decomposition perspective, identifying two key challenges: the Query Comprehension Gap in similarity-based methods and the Interpretation–Selection Gap in judgment-based methods. To address them, we propose RACER, a training-free reflective agentic framework that decomposes long-
video frame selection into query interpretation driven by a lightweight Vid-LLM and evidence localization supported by an
embedding model serving as a
retrieval tool. Specifically, the Vid-LLM is responsible solely for reformulating the complex query into sub-queries that make implicit information requirements explicit, mitigating the Query Comprehension Gap. Meanwhile, the
retrieval tool leverages these sub-queries to localize relevant evidence, relieving the Vid-LLM of direct frame selection and thus addressing the Interpretation–Selection Gap. Finally, the retrieved frames are fed back to the Vid-LLM for sub-query refinement, forming a reflection loop that iteratively improves query interpretation and frame selection. Experiments across multiple
benchmarks show that RACER consistently improves long
video understanding. Notably, RACER achieves effective frame selection even with limited-capability components, demonstrating that agentic integration enables these components to enhance more capable Vid-LLMs.
Read the full abstract ▾
All authors Yiyang Huang, Yitian Zhang, Yizhou Wang, Jianglin Lu, Qihua Dong, Hailing Wang, Huimin Zeng, Mingyuan Zhang, Yun Fu
Cryptography and Security
ASPIRE: Agentic Safety & Prompt Injection Red-teaming Engine
Pengfei He, Deep Mitra, Vishesh Sharma et al. (+4 more)
LLM
agents retrieve untrusted content and act through tools, creating indirect
prompt-injection risks that can cause unauthorized actions or persistent state changes. Existing automated red-teaming largely optimizes payloads for pre-specified scenarios, leaving latent vulnerabilities across the
agent's behavior space unexplored. We present ASPIRE, an Agentic Safety &
Prompt Injection Red-teaming Engine for open-ended, behavior-level vulnerability discovery. ASPIRE maintains an evolving
Agent Security Behavior Graph and uses complementary Explore and Exploit experts to discover, verify, and generalize consequence-centric tests. Trajectory evidence updates the graph and diagnoses partial or failed attempts, while cross-run
memory transfers useful red-team strategies. Experiments on various
benchmarks show that ASPIRE substantially expands coverage across consequences, injection methods, environments, and behavior paths while maintaining strong attack success.
Read the full abstract ▾
All authors Pengfei He, Deep Mitra, Vishesh Sharma, Jiliang Tang, Vinay S Rao, Tomas Pfister, Long T. Le
Systems and Control
Quantifying Grid-Forming Requirements for System Strength: How Much and Where?
Ambuj Gupta, Balarko Chaudhuri and Mark O'Malley
As power systems transition toward inverter-based resource (IBR)-dominated grids, Grid-ForMing (GFM) IBRs are increasingly expected to provide the voltage-source characteristics required to maintain adequate system strength. However, how many GFM IBRs are required, and where they should be located in the grid, remains unclear. This paper investigates the impact of the number and location of GFM IBRs on small-signal system strength. The Jacobian Voltage Stiffness Metric (JVSM), which quantifies the voltage-magnitude and voltage-phase stiffness from frequency-domain Jacobian characteristics, is used as a system-strength measure. The analysis demonstrates that increasing the number of GFM IBRs generally improves system strength; however, configurations with the same number of GFM IBRs can exhibit significantly different bus-wise system-strength characteristics depending on their locations. Thus, this paper provides a quantitative basis for specifying GFM requirements in terms of the resulting system strength, rather than solely in terms of penetration level or the number of GFM IBRs. Case studies on a modified IEEE 39-bus system demonstrate how JVSM can be used to quantify and compare the system-strength impact of different numbers and locations of GFM IBRs, with the resulting strength assessments validated against small-signal oscillatory stability.
Read the full abstract ▾
Computational Engineering
MarketRank: ranking stocks by the stationary probabilities of a Markov chain over estimated money flows, with a topological landscape view
B. Kaan Karamete
Every trading day, money leaves some stocks and moves into others. MarketRank asks where that money concentrates and how the concentration shifts over time. It treats the
market as a Markov chain whose states are stocks. The chance of moving from one stock to another is the share of the first stock's outgoing dollars that goes to the second, and a stock's MarketRank score is its long run (stationary) probability in that chain, computed by damped power iteration on a sparse transition
matrix. PageRank ranks web pages with the same idea, using links instead of dollars. Who sold what to buy what is not public, so the flows are estimated from daily prices and volumes with a correlation tilted
gravity model, in a deterministic parallel C++ engine. On a universe of 10K US stocks and ETFs, the ranking is led by the large
semiconductor and technology names, and estimated flows stay within a sector more often than chance. The level of the score, however, does not predict next day returns: it shows where money sits, not where prices go next. A 3D landscape view, Fluxscape, turns the ranking into terrain: Louvain communities of the recent flow graph, reclustered on every bar, become territories, ordered by recursive spectral bisection along a generalized
Hilbert curve, interpolated by inverse distance weighting, smoothed by a centroidal Voronoi style relaxation with portfolio holdings pinned to their exact values, and rendered as a Gouraud shaded mesh. The surface is for reading trends; the exact ranked table remains the reference. The practical question is whether MarketRank can improve real portfolio moves, and we answer it with tests fixed in this study.
Read the full abstract ▾
Information Theory · Cross-listed from quant-ph
Efficiently computable bounds on the energy-constrained quantum reading capacity
Vishal Singh, Zixin Huang and Mark M. Wilde
In
quantum reading, classical messages are encoded in sequences of
quantum channels and recovered by probing the channels and processing their outputs. We study the reading capacity of a finite family of finite-dimensional channels under an average constraint on the probing energy, allowing arbitrary adaptive operations between channel uses. Using an input-dependent chain rule for the Belavkin-Staszewski relative
entropy, we derive a converse bound expressed as an optimization involving channel Choi operators and the operator relative
entropy. This formulation admits semidefinite approximations and reduces, without an energy constraint, to a channel information radius. To obtain achievable rates, we construct bilinear semidefinite lower approximations to a standard non-adaptive reading bound. Alternating optimization produces feasible probe states and encoding distributions, whose Holevo information gives a directly evaluable achievable rate. For jointly classical-
quantum channels, we derive a separate converse based on the Umegaki relative
entropy. Without an energy constraint, this converse matches a non-adaptive achievable rate, recovering a recent result of Pascual Abraldes and Winter: non-adaptive protocols suffice to achieve the unconstrained reading capacity of jointly classical-
quantum channels.
Read the full abstract ▾
Computer Vision
One-Slide Calibration of Pathology Foundation Models
Ming Ren Hou and Tianyi Huang
Scanner variation changes how pathology foundation models represent the same tissue. We introduce SlideRuler, which uses other regions within a slide as internal controls to predict a query region's scanner-induced shift. A transfer map learned from paired rescans enables calibration from a single scan at inference while keeping the foundation model fixed. Across two encoders and five SCORPION scanners, learned transfer reduces mean target-to-source
embedding distance by 16.3-38.5% relative to raw
embeddings. Replacing controls with unrelated same-scanner regions reveals a positive same-slide contribution in all four evaluation settings, including scanner holdout. A source-anchored variant reduces source-feature displacement by 47.7-83.6% relative to learned transfer while retaining most of its
alignment gain. By drawing calibration information from the slide itself, SlideRuler offers a path toward more consistent use of frozen pathology models across imaging systems.
Read the full abstract ▾
Human-Computer Interaction
"I'm Very Happy for It to Start Hallucinating a Little Bit": Using ClayFlect to Negotiate Multimodal AI Representations in Material Meaning-Making
Kellie Yu Hui Sim, Quoc-Nam Nguyen, Shuenn Yuen Han et al. (+1 more)
As AI enters reflection and emotional support, understanding how it can participate in personal meaning-making while preserving users' authority over interpretation is increasingly important. We present ClayFlect, a novel MLLM-powered system integrating tactile clay-making with conversational and visual generative AI, and report a mixed-methods study with 50 participants. Reflection developed across material, conversational, and generated forms rather than through AI interaction alone. Participants treated AI representations as provisional: they compared, redirected, reinterpreted, selectively incorporated, or left them aside as their artefacts and meanings evolved. Clay provided a directly manipulable space in which participants could continue developing meaning independently of the AI, while generated representations externalised possibilities beyond what they could readily make or visualise. We show how generative AI can participate through representations that remain negotiable, and derive implications for supporting movement across representations, preserving parallel sites of control, and allowing AI support to recede or deepen as reflection unfolds.
Read the full abstract ▾
All authors Kellie Yu Hui Sim, Quoc-Nam Nguyen, Shuenn Yuen Han, Kenny Tsu Wei Choo
Computer Vision
SPW-Nav: A Streaming Panoramic World Model for Language-Guided Navigation
Yunheng Liu, Ziqi Cai, Siqi Yang et al. (+7 more)
Language-guided panoramic
video generation benefits various downstream applications, such as interactive 3D scene exploration, virtual reality experiences, and
embodied agent training. Existing panoramic generators follow predefined trajectories, and interactive world models act through low-level actions in perspective views. We propose SPW-Nav, a streaming panoramic world model that understands movement instructions and streams one minute of 2K 360-degree
video in real time from a single panorama. SPW-Nav interprets each instruction in the previously generated panorama as camera motion. Spherical rotation decoupling applies rotation exactly on the sphere, pose-aligned conditioning keeps translation inputs bounded over long streams, and a multi-term
memory with a few-step generator continues the scene as instructions change. We also build SPW-NavSet, panoramic
videos with camera trajectories and verified instructions. Driven by
language, SPW-Nav outperforms prior panoramic generators in camera-following accuracy and
video quality, and supports on-the-fly instruction switching.
Read the full abstract ▾
All authors Yunheng Liu, Ziqi Cai, Siqi Yang, Yimu Wang, Minggui Teng, Jiaming Tan, Shuchen Weng, Erwin Wu, Kaipeng Zhang, Boxin Shi
Artificial Intelligence · Cross-listed from q-bio.BM
A Shortcut to Structure in AlphaFold 3
Jonathan Feldman and Jeffrey Skolnick
AlphaFold 3 predicts
protein structures with remarkable accuracy, yet how structural information emerges within the model remains poorly understood. Here, through causal interventions on internal representations and direct probing of every Pairformer block, we trace the formation of global
protein geometry and identify the multiple sequence
alignment (MSA) as a structural shortcut to the fold. Removing the MSA largely preserves local secondary structure while disrupting the long-range relationships that define global
topology. Restoring the MSA-enriched pair representation at only forty residues recovers most of this lost organization, including at pairs never directly modified. This contribution depends on the detailed direction of the MSA module's output rather than its magnitude. The Pairformer rapidly converts this signal into global geometry: the final fold becomes recoverable by approximately block 9 of 48 for a majority of
proteins, roughly twenty-seven blocks before the model's decoder can render it, whereas without the MSA it remains inaccessible for most
proteins throughout the pass. Which homologs are supplied shapes this trajectory more strongly than which query is supplied; it persists for a designed query that never evolved but collapses for a shuffled sequence. Most importantly, an
alignment built for a different
protein that shares the fold, supplied only at the structurally corresponding columns, raises the median TM-score against experiment from 0.44 to 0.72, while the same
alignment shifted a few residues along the chain performs worse than supplying no
alignment at all. What AlphaFold 3 reads from an
alignment is therefore a description of the fold itself, transferable between
proteins that share one, rather than the query's own evolutionary history. This explains both its accuracy and the limits of what it has solved.
Read the full abstract ▾
Computer Vision
VCR-Bench: A Modular Open-Source Benchmark for Video Classification Robustness
Maksim Plinskiy, Aleksandr Gushchin, Sergey Lavrushkin et al. (+2 more)
Robustness of image classification has several
benchmarks, but their
video counterparts are absent. In
video classification temporal dimension introduces additional degrees of freedom for
adversarial attacks, defenses, and preprocessing. Temporal sampling, perturbation budgets, and metric aggregation also interact in ways with no direct analogue in the image setting. Therefore, robustness for
video classifiers is studied across scattered, incompatible implementations, making reported numbers hard to reproduce and analyze. We introduce VCR-Bench, a modular open-source
benchmark framework that standardizes
video loading, wrappers for classifiers,
adversarial attacks and defenses, perceptual metrics, configuration presets, and result logging. VCR-Bench currently integrates 30
video classification models, 14
adversarial attacks, and 10 defense wrappers under a common evaluation protocol. We evaluate representative
video classifiers, attacks, and defenses on Kinetics-400 subset, reporting clean accuracy, attack success rate, perceptual quality, runtime, and
memory usage. VCR-Bench is released with documented installation, reproducible run presets, component-extension interfaces, and scripts for reproducing the reported results at https://github.com/msu-
video-group/vcr-bench.
Read the full abstract ▾
All authors Maksim Plinskiy, Aleksandr Gushchin, Sergey Lavrushkin, Dmitriy S. Vatolin, Anastasia Antsiferova
Robotics
Toward Evidence-Driven Human-Agent-Robot Teaming for Earth-Independent Anomaly Triage
Ignacio G Lopez-Francos, Alexis Gallagher and Samira Shalal
Deep-space crews cannot rely on real-time ground support for urgent off-nominal events. Initial alerts may underdetermine cause, while discriminating evidence may reside in crew observations or at locations that are unsafe, costly, or unavailable for crew inspection. We present an evidence-driven architecture for human-
agent-
robot teaming in Earth-independent anomaly triage. Agentic AI is treated as a stateful coordinator over bounded, inspectable services rather than as a fully autonomous vehicle controller. A triage state manager maintains hypotheses, evidence provenance, uncertainty, operational context, and tool status; a crew-facing
embodied agent elicits observations and explains assessment changes; and a mobile
robot acquires targeted, localized evidence. Typed interfaces separate dialogue and orchestration from monitoring,
robot command, context
retrieval, and safety-critical control. Two scenarios illustrate the architecture: a crewed deep-space mission based on an actual ISS ammonia false alarm, where suspected contamination restricts crew access, and a power-interface anomaly at a crewed lunar base, where robotic inspection distinguishes a local connector fault from other causes ambiguous in remote telemetry. Our main contribution is an authority-bounded closed evidence-loop architecture, exercised in a hardware-in-the-loop integration prototype using Reachy Mini and an Innate MARS mobile
robot.
Read the full abstract ▾
Information Theory
A Unified Framework for Characterizing General MIMO Channels
Zeyan Zhuang, Anzheng Tang, Xin Zhang et al. (+2 more)
Modern
MIMO systems are evolving toward higher-dimensional and more flexible architectures, such as distributed and
holographic MIMO, offering substantial benefits while giving rise to increasingly complex channel correlation structures. This growing architectural diversity makes it difficult to characterize the fundamental limits of different
MIMO systems on a case-by-case basis, motivating a unified analytical framework applicable across a broad range of architectures. This paper addresses this research gap by investigating a generally correlated Rayleigh channel whose vectorized form follows a general Gaussian distribution with an arbitrary covariance
matrix. For that purpose, we first prove that the spectral norm of the channel
matrix is bounded in the asymptotic regime where the numbers of transmit and receive
antennas grow proportionally. We then derive a deterministic approximation for the ergodic mutual information, with an explicit convergence rate governed by the structure of the channel correlation. The approximation is characterized by a pair of
matrix-valued self-consistent equations, for which we establish the existence and uniqueness of the solution and propose an iterative numerical
algorithm. Furthermore, we develop a framework based on positive linear maps to analyze the stability of these equations, which can facilitate the spectral analysis of broader classes of random
matrices. The proposed characterization unifies many existing results for structured
MIMO channels as special cases while remaining applicable to a broad class of general
MIMO architectures that are difficult to evaluate using existing analytical frameworks. To demonstrate its utility, we apply the developed theory to two representative systems, namely downlink distributed
MIMO and uplink heterogeneous
MIMO. Numerical results confirm the accuracy of the derived deterministic approximations.
Read the full abstract ▾
All authors Zeyan Zhuang, Anzheng Tang, Xin Zhang, Dongfang Xu, Shenghui Song
Distributed Computing
Part of the Strassen algorithm can speed up matrix multiplication in a parallel pebbling game
Aaron Dayton and Thomas E. Baker
We present a novel variant of the Strassen
algorithm called the Partial Strassen
algorithm, which uses a fraction of the Strassen steps to perform
matrix multiplication in less time than traditional implementations. The
memory footprint required to implement this
algorithm is provably small whether used with a single thread or in a multi-threaded context. Data comparing the Partial Strassen
algorithm at depths one, two, and three with BLAS
matrix multiplication show that the three-level Partial Strassen
algorithm is able to perform
matrix multiplication in $80\%$ the time of BLAS with $2.25$ times the
memory requirement, with a theoretical improvement of $75\%$ for sufficiently large
matrices. The Partial Strassen
algorithm is also compared against an implementation of the traditional Strassen
algorithm, and is shown to be more performant with increasing
matrix size due to decreased
memory consumption. An open source implementation of the Partial Strassen
algorithm for arbitrary depth and rectangular
matrix multiplication is also included.
Read the full abstract ▾
Cryptography and Security
Multi-Aspect Runtime Verification for Simulation-Based V&V of LLM-Enabled Autonomous Agents
Nikolaos Kekatos, Dimitrios Nikou, Anastasios Temperekidis et al. (+4 more)
LLM-based
agents are entering decision-support roles in defence staff work, where the obligations they must respect are already written down and binding, and where retraining is not available as a control because models arrive as procured components. What can be placed under engineering control is the interface between the
agent and the systems it acts on. Those obligations are at once spatial, temporal and text-semantic, and a violation typically lives in the composition of a multi-step interaction, which is why per-event guardrails miss sequential tool-attack chains. We present a multi-aspect runtime-verification framework that decomposes a natural-
language policy clause into a typed spatial/temporal/semantic triple over one canonical event stream, checks each aspect with its own monitoring specification, and fuses the verdicts through a four-valued
algebra that carries provenance. The spatial aspect is interpreted over a weighted two-sorted location graph in which mission geometry and information-release
topology are one object; we show that these spatial obligations are not in general subsumed by a first-order temporal specification. The past-time aspect runs on the unmodified MonPoly engine, which agrees with our reference monitor at every time point. Across two mission domains, casualty evacuation and contested sustainment, and one civil domain, composition under the precautionary blocking policy drives attack success to zero with no observed false positives and microsecond-scale per-event cost, while every single aspect and every pair leaves a substantial share of attacks succeeding. In a closed-loop experiment a policy-naive planner reaches a violating state in most unshielded missions and in none when shielded, and four refused episodes in five still recover to a compliant outcome.
Read the full abstract ▾
All authors Nikolaos Kekatos, Dimitrios Nikou, Anastasios Temperekidis, Alexios Lekidis, Nikolaos Kolokotronis, Panagiotis Katsaros, Stylianos Basagiannis
Artificial Intelligence
Can AI Agents Make Open-Ended Scientific Discovery? Evidence from Station
Wenyu Du and Stephen Chung
Recent AI systems have made rapid progress in scientific discovery when given well-defined metrics, but whether they can autonomously undertake open-ended scientific discovery remains unclear. We investigate AI's ability to tackle open-ended tasks in Station, an open-world environment in which multiple
agents simulate a scientific ecosystem. To tackle challenges specific to open-ended tasks, we propose augmenting Station with two mechanisms: a Supervisor mechanism and periodic Meta Reflection, which encourage persistent exploration even when intermediate metrics are lacking. We construct open-ended tasks from three recent oral papers presented at ICLR. We give
agents the main research question studied in each paper while withholding the paper's results and disabling web access. We then measure how many of the original findings-partitioned into individual criteria-
agents rediscover. We find that Station rediscovers 62.7% of the criteria on average, compared with 15.4% for Codex Multiagent-v2 and 14.4-20.6% for AI Scientist-v2. Ablation and behavioral analyses indicate that adding the two mechanisms together improves research coverage and continuity. We further evaluate Station on two open-ended tasks without oracle papers and find that some of the discoveries made by the
agents closely match discoveries reported by researchers after the knowledge cutoff date. Together, these results indicate that a suitable environment can enable
agents to autonomously make meaningful progress in open-ended scientific discovery.
Read the full abstract ▾
Artificial Intelligence
AdaGuard: Enhancing Safety and Policy Compliance with Reasoning-Enabled LLM-As-A-Judge Guardrails
Melissa Kazemi Rad, Sihui Dai, Isha Slavin et al. (+6 more)
Enterprise generative AI applications require robust safety mechanisms that can accommodate diverse risk postures, evolving policies, and varying latency constraints. Current guardrail solutions often suffer from
rigidity, relying on fixed policy sets and offering limited transparency or
reasoning flexibility. We present Adaguard, an adaptive LLM-as-a-Judge framework designed to address these challenges through dynamic policy enforcement and adaptive
reasoning-budget allocation. Built using supervised fine-tuning (SFT) and reinforcement learning (GRPO), AdaGuard generalizes to user-defined safety and compliance policies at runtime without requiring frequent model updates. A core innovation of our approach is the ability to dynamically infer the complexity of input-policy pairs, allowing the model to switch between high-speed black-box inference and explainable,
reasoning-enabled moderation. This flexibility enables developers to balance stringent latency requirements with the need for actionable transparency. This adaptive capability allows AdaGuard to rival other guardrail and frontier models several times its size, while its auto-
reasoning mode recovers the accuracy of always-on
reasoning at a fraction of the latency
Read the full abstract ▾
All authors Melissa Kazemi Rad, Sihui Dai, Isha Slavin, Kushal Chawla, Mann Patel, Jian Ni, William M. Campbell, Stephen Rawls, Sambit Sahu
Computation and Language
CARE: Certifying Acceleration for Vision-Language-Action Inference
Rui Liu, Tong Zheng, Jindong Gu et al. (+1 more)
While
vision-
language-action (VLA) models have advanced rapidly, running them at every control step remains expensive. Prior work accelerates VLA inference using techniques like action chunking and visual-
token pruning, typically evaluating based on latency and average task success. However, acceleration may discard information and break tasks the original policy would solve, a risk hidden by average metrics. Measuring these failures is challenging because action deviations compound over closed-loop trajectories, meaning task failure is only observable across full episodes. We therefore define an acceleration-induced failure via paired rollouts from identical initial conditions, tracking when the reference succeeds but the accelerated policy fails. To manage this, we introduce CARE, an approach for certified accelerator selection. CARE uses paired rollouts on a calibration set to provide finite-sample guarantees that acceleration-induced failure risk stays below a user-specified budget. It deploys the fastest certified candidate, falling back to the reference if none qualify. By relying only on terminal outcomes and measured compute, CARE applies unchanged across diverse acceleration mechanisms, while sequential testing and failure-triggered reference rollouts keep certification affordable. On four LIBERO suites with OpenVLA-OFT, CARE certifies $9.0$–$10.8\times$ speedups while guaranteeing (at $95\%$ confidence) that at least $85.8\%$ of reference-solved episodes are preserved. Under tight budgets, selectors without guarantees exceed the budget in up to $75\%$ of trials, whereas CARE stays within budget and its sequential form uses $78.9\%$ fewer rollouts than exhaustive evaluation. CARE further generalizes to flow-step reduction for $\pi_{0.5}$, and to Qwen3.5-9B and Llama-3.1-8B
agents in Crafter.
Read the full abstract ▾
All authors Rui Liu, Tong Zheng, Jindong Gu, Zhipeng Wang