Robotics
Predictive Semantic Safety: From Visual Physical Reasoning to Safety-Critical Control
Taekyung Kim, Salem Fradi, Yanning Dai et al. (+2 more)
Physical interactions can create future hazards that are not apparent from the
robot's current
geometric surroundings. We present a framework termed Predictive Semantic Safety (PSS), which connects visual physical
reasoning to backup-based safety filtering. A
vision-
language model (VLM) predicts physical events and their timing or directly predicts object displacements. An explicit motion model converts event hypotheses into object trajectories. Split conformal prediction calibrates position errors jointly across specified objects, observation times, and future times;
geometric shape bounds convert the resulting position regions into predicted object occupancy. PSS evaluates a prescribed backup maneuver against this occupancy and derives input-affine constraints for minimally modifying the nominal input while preserving backup feasibility under the
robot dynamics and input limits. MuJoCo experiments with a Unitree Go1 consider falling fixtures, impact-driven support loss, and contact propagation. PSS achieves a safe episode rate of 99.3%, compared with 43.3% for a Backup Control Barrier Function baseline that only uses current obstacle geometry.
Read the full abstract ▾
All authors Taekyung Kim, Salem Fradi, Yanning Dai, Mateusz Ostaszewski, Jürgen Schmidhuber
Mesoscale and Nanoscale Physics
Observing Colossal and Tunable Near-field Thermal Radiation with A Highly Sensitive Annular Micro-thermocouple Junction
Ken Araki, Wei Han Won and Liping Wang
Near-field radiative heat transfer between two objects has been theoretically predicted and experimentally demonstrated to exceed far-field blackbody limit across nanoscale
vacuum gap distance enabled by evanescent surface waves coupling, while significant enhancement by more than 100 times usually requires sub-50-nm
vacuum gaps around room temperature. This is challenging for parallel-plate configuration with millimeter sample sizes due to intrinsic wafer bow and contaminant particles. Sphere-plate configuration with a microsphere attached to bimaterial cantilevers or micro-thermocouple tips has been used to experimentally demonstrate near-field radiative heat transfer down to 30-nm gaps, but it is much less developed because of the challenges in the sophisticated sensor fabrication, low sensitivity and weak signals. In this work, we overcome these challenges by an annular micro-thermocouple junction fabricated at the end of a glass fiber with straightforward thin-film deposition to achieve high measurement accuracy with large Seebeck coefficient 25 uV/K and thermal resistance 8.7e6 K/W. With a silica microsphere attached underneath the micro-thermocouple junction, we report experimental observation of colossal near-field radiation heat transfer over blackbody limit down to 10-nm gap up to 2600 times with quartz and 900 times with doped silicon. Upon phase transition of VO2 thin film emitter, tunable near-field heat transfer up to 430-fold enhancement is experimentally demonstrated at 15-nm gap with 64% reduction. Experimental data agrees well with rigorous modeling based on fluctuational electrodynamics and Derjaguin approximation, and underlying mechanism is understood by energy transmission calculations. The results will advance the experimental study and fundamental understanding of energy transport at nanoscale gaps.
Read the full abstract ▾
Machine Learning
FAST-Brain: A Flow-Aligned Spatio-Temporal Surrogate Brain Model
Shucheng Liu, Changchun Shi, Kai Zhang et al. (+1 more)
Modeling resting-state functional
magnetic resonance imaging (rs-fMRI) data is crucial for understanding
brain-wide neural activity. However, traditional methods struggle to capture complex temporal dynamics over long horizons, to account for the
brain's anatomical spatial structure, and to model high-dimensional ambient signals that lie on a low-dimensional intrinsic subspace. We propose FAST-
Brain, a unified flow-aligned spatio-temporal surrogate
brain model that addresses all three challenges. At its core is a flow-aligned generative framework that directly predicts the clean blood-oxygen-level-dependent (BOLD) signal, paired with a graph convolutional network that captures spatial structural constraints and a
Transformer that models long-range temporal dependencies. Theoretically, we show that under a low-dimensional subspace assumption, the approximation error of our model scales with the intrinsic dimension rather than the ambient dimension, which justifies our direct modeling of the BOLD signal. Extensive experiments on synthetic and Human Connectome Project
datasets demonstrate that FAST-
Brain achieves state-of-the-art performance in recovering functional connectivity, effective connectivity, and the implicit low-dimensional signal subspace.
Read the full abstract ▾
All authors Shucheng Liu, Changchun Shi, Kai Zhang, Hongtu Zhu
Artificial Intelligence
SemRD-V2X: Closure-Guided Communication with Bounded Inference for Cooperative Perception
Hu Xu, Chun Li, Siyuan Qiu et al. (+2 more)
Vehicle-to-Everything (V2X) cooperative perception improves 3-D detection by sharing intermediate features, but dense remote features may repeat context that the ego
agent can infer locally. Most communication-efficient designs optimize masks or codes empirically, leaving a more basic question open: which remote evidence is indispensable given the receiver's own observation? We introduce a closure-fidelity perspective on ego conditioned remote perception. Under a finite deductive abstraction and explicit conditions, its rate–distortion function decomposes over an irredundant core, and the exact zero-distortion rate becomes $P_A H(\pi_A)$. This analysis suggests a concrete design principle: transmit compact evidence and recover derivable context with bounded receiver-side inference. Guided by this principle, SemRD-V2X is an operational neural proxy that combines exact-budget BEV support selection, pointwise channel compression, and masked shared-weight reconstruction before standard fusion. Experiments on simulated V2XSet and real-world DAIR-V2X validate the resulting design. In a controlled five-run V2XSet comparison against a locally reproduced V2X-ViT-v1 baseline on one Tesla V100, SemRD-V2X reduces the analytical feature payload by $26.6\times$ while improving AP@0.5/AP@0.7 by 4.13/8.57 points, with 3.81% additional mean compute latency. These results position closure fidelity as both an analytical lens and an actionable design principle for communication-efficient cooperative perception.
Read the full abstract ▾
All authors Hu Xu, Chun Li, Siyuan Qiu, Zeyan Li, Jianfeng Xu
Statistical Mechanics
Second largest eigenvalue does not bound stationary entropy production
Naoto Shiraishi
We examine the spectral dissipation-coherence trade-off inequality conjectured by Oberreiter et al. [Phys. Rev. E 106, 014106 (2022)], stating that the coherent number, defined from the second largest
eigenvalue, provides a lower bound on the stationary
entropy production per unit oscillation. We disprove this
conjecture by explicitly constructing a counterexample, which accompanies finite coherent number with vanishing stationary
entropy production. This model also excludes a wide class of thermodynamic bound with the real and imaginary parts of the second largest
eigenvalue of the transition rate
matrix. Our result implies that the second largest
eigenvalue does not necessarily characterize the degree of coherent oscillation.
Read the full abstract ▾
Hardware Architecture
PolyCIM: Improving Data Reuse in Digital CIM Accelerators with Polyhedral-Based Compilation
Yingjie Qi, Cenlin Duan, Yiou Wang et al. (+4 more)
Digital Compute-in-
Memory (CIM) presents a promising solution for accelerating deep neural networks (DNNs) through the integration of computational logic directly within
memory arrays. However, mapping modern DNN operators to CIM accelerators often results in severe array underutilization, due to the strict data reuse constraints imposed by the rigid CIM array structure. We observe that data reuse in modern DNNs forms hyperplane structures often oriented along non-axial directions, rendering them invisible to conventional mapping methods that only exploit axis-aligned reuse. In this work, we propose PolyCIM, a polyhedral-based compilation framework for CIM architectures that systematically exposes and realigns these hyperplanes through affine transformations. PolyCIM provides a unified abstraction capable of efficiently representing both diverse DNN workloads and digital CIM architectures. Through data reuse exposure, computation mapping, and data movement optimization, PolyCIM generates mappings for CIM architectures that achieve superior array utilization and performance. Experimental results show that PolyCIM delivers up to $4\times$ improvement in macro utilization and $3.2\times$ speedup, effectively bridging the gap between modern DNN operators and CIM architectures.
Read the full abstract ▾
All authors Yingjie Qi, Cenlin Duan, Yiou Wang, Yikun Wang, Xiaolin He, Weisheng Zhao, Jianlei Yang
Plasma Physics
Electrons with Anomalous Energy Generated in Vacuum Diodes with Different Pulse Duration
Vasily Kozhevnikov, Victor Tarasenko, Maxim Vorobyov et al. (+2 more)
This paper presents results from an experimental study in which anomalous energy electrons (AEE) appear in
vacuum diodes under varying amplitude and pulse-duration conditions. AEE refer to high-energy discharge electrons with kinetic energy $T$ exceeding $eU$ values (where $e$ is the electron charge and $U$ is the amplitude gap voltage). Here, two experimental setups were used to generate electron beams with different pulse durations. The electron energy spectra were reconstructed from the beam attenuation curves using an AI-driven methodology for solving an ill-posed inverse problem for the Fredholm equation. This confirms that
vacuum diodes generate electron beams with a significant AEE fraction (up to 25~%) only when operating with nanosecond-long voltage pulses. It has been confirmed that AEE in
vacuum is generated using nanosecond- or sub-nanosecond voltage pulses. Experiments also indicate that photoelectric absorption of bremsstrahlung cannot produce significant AEE in
vacuum discharges.
Read the full abstract ▾
All authors Vasily Kozhevnikov, Victor Tarasenko, Maxim Vorobyov, Evgenii Baksht, Andrey Kozyrev
Artificial Intelligence
Fuzzy Distribution Modeling for Synthetic Tabular Data Generation with Causality Preservation
Michael Vasilakakis and Dimitris K. Iakovidis
Synthetic tabular data generation provides an effective alternative for the training of machine learning models when real-world data is limited or inaccessible. However, the heterogeneous, non-smooth, and incomplete nature of tabular data poses fundamental challenges to conventional probabilistic and deep generative models, where their interpretability remains limited. This paper proposes a novel fuzzy distribution modeling methodology for synthetic tabular data generation based on fuzzy sets theory. Feature distributions are represented using fuzzy sets and feature dependencies are modeled through Fuzzy Cognitive Maps, resulting in a low-parameter, and an interpretable data representation. Synthetic samples are generated by sampling fuzzy concepts rather than raw values, enabling native support for mixed data types, missing values, and domain constraints. The methodology further supports linguistic queries and IF-THEN
reasoning, facilitating transparent simulation of decision-making processes. Experimental results on
benchmark datasets demonstrate competitive performance with respect to utility, fidelity and
privacy compared to state-of-the-art methods, while offering substantially improved interpretability. These results establish fuzzy distribution modeling as a principled and effective approach for synthetic tabular data generation in fuzzy systems and decision support applications.
Read the full abstract ▾
Machine Learning
Commutator Memory: Sparse, Path-Local Reading and Steering in Language Models
John Sweeney
Gradient updates on different data generally do not commute: training a
language model on two data sources in opposite orders gives different weights, even with the same data and total exposure. Loss or
benchmark deltas show that the models differ, not where. We ask whether this path dependence leaves a parametric training-history
memory: a weight component that flips sign when the two sources are swapped, is localized in output space, changes the held-out loss gap between the two orders under targeted interventions, and reveals which trained model came from which order. For one small SGD step of size $\eta$ on each of sources $A$ and $B$, the weight difference $\theta_{AB}-\theta_{BA}$ is, to leading order, $\eta^2 b_{AB}$, where $b_{AB}=H_Bg_A-H_Ag_B$ is the Lie bracket of the two gradient fields at the base model. We define commutator
memory by projecting the bracket through the logits into one score per vocabulary
token; the scores sum to the bracket's prediction of the gap. The scores are localized: on three models, the same readout of the measured $\theta_{AB}-\theta_{BA}$, or of a bracket from disjoint batches, shares 82-99% of the original top-20
tokens, versus 35-49% for norm-matched random directions. They are causally actionable: in Qwen-3-4B SFT, downweighting the ten
tokens with the largest predicted share of the gap closes a median 32% of the measured gap, while frequency-matched
tokens with near-zero scores have almost no effect. The weights themselves carry the component: projecting the difference between the two trained models onto $b_{AB}$ identifies which came from which order in 92% of cases across four LLMs (chance 50%). Controlled tests also cover matched-batch DPO, a frozen-rollout GRPO-style objective, and an AdamW endpoint check. The
memory is defined per source pair, not per example, and its projection on $b_{AB}$ decays with further training.
Read the full abstract ▾
Sound
SAIL: Spatial Audio Intelligence with Large Language Models via Disentangled Acoustic-Spatial Encoding and Dual-Stream Q-Former
Zhengding Luo, Jinyang Wu, Haozhe Ma et al. (+3 more)
Spatial audio large
language models (LLMs) enable
embodied agents, wearable assistants, and immersive systems to recognize sound events, localize sources, and reason about their spatial relationships. However, existing spatial audio LLMs often rely on early fusion of acoustic and spatial features and source-agnostic
token representations. These designs make it difficult to preserve the correspondence between individual sound events and their spatial attributes, particularly in multi-source scenes. To address this limitation, we propose SAIL, a Spatial Audio Intelligence framework with LLMs that preserves acoustic-spatial structure and source-level correspondence from audio encoding to LLM
alignment. SAIL introduces a Disentangled Spatial Audio
Transformer that represents Mel-spectrogram and interaural phase difference features as separate acoustic and spatial streams. Source-discriminative task queries further learn event, direction, and distance information for each source. A Dual-Stream Q-Former then aligns the two streams with the LLM using acoustic and spatial queries organized by source slots. Compared with the early-fusion baseline, SAIL achieves consistent improvements in dual-source sound event detection, direction and distance estimation, and spatial
reasoning. These results demonstrate the importance of structured, source-discriminative audio representations for multi-source spatial understanding and
reasoning.
Read the full abstract ▾
All authors Zhengding Luo, Jinyang Wu, Haozhe Ma, Yanghao Zhou, Woon-Seng Gan, Wenwu Wang
Computer Vision
E-WAVE: Event-based Continuous Optical Flow via Warping-Aligned Visual Encoding
Jiale Wu, Xiaoyang Bai, Haoming Yu et al. (+3 more)
Temporally dense optical flow is essential for dynamic perception in immersive VR/AR systems, where rapid head, hand, and object motion must be continuously captured and tracked. Existing frame-based optical flow estimation methods are constrained by the tradeoff between temporal resolution and computational cost; while event cameras, with their high temporal resolution and energy efficiency, serve as a natural solution to the dilemma. However, event-based approaches commonly rely on correlation volumes to capture pairwise voxel correspondences, which incur substantial
memory and computation overhead. We present E-WAVE, a correlation-free framework for high-temporal-resolution (HTR) optical flow estimation from event streams. Instead of constructing all-pairs correlation volumes, E-WAVE employs global
attention mechanism to model long-range feature dependencies and performs trajectory guided feature warping using Bézier curve. Through iterative updates, it predicts trajectories that allow for querying at arbitrary timestamps without repeated inference. Experiments on MultiFlow and DSEC-Flow demonstrate a 25% lower trajectory error and comparable endpoint flow estimation accuracy relative to state-of-the art baselines. Additional evaluations on self-captured data using a head-mounted prototype validate that E-WAVE remains robust under challenging real-world conditions.
Read the full abstract ▾
All authors Jiale Wu, Xiaoyang Bai, Haoming Yu, Yiwei Chen, Yifan Peng, Weiwei Xu
Computation and Language
CRISP: Cultural Reward Modeling for Implicit Situated Propriety
Zekun Yuan, Yangfan Ye, Baohang Li et al. (+6 more)
As large
language models (LLMs) are increasingly deployed across countries and regions, the ability to recognize and respond appropriately to diverse cultural contexts becomes increasingly important. However, existing research has largely focused on cultural knowledge or tasks with predefined response spaces, while open-ended culturally situated behavior remains comparatively underexplored. In this work, we introduce CRISP-RM, a culturally situated reward model that assigns rewards according to cultural appropriateness in open-ended social scenarios. During policy optimization, we further introduce Norm Grounding Supervision (NGS), providing guidance that enhances the policy's sensitivity to relevant cultural norms. To construct culturally situated data, we employ a collaborative multi-
agent framework that instantiates implicit cultural norms into diverse social scenarios and further curate NormCompass as a dedicated testbed. We conduct comprehensive experiments to evaluate the effectiveness of CRISP-RM in both reward modeling and policy optimization. Best-of-\(N\) experiments show that CRISP-RM consistently outperforms strong general reward models. During GRPO policy optimization, CRISP-RM generally improves culturally situated behavior, while incorporating NGS yields further gains. Further analyses demonstrate the advantages of CRISP-RM in distinguishing culturally appropriate behavior beyond superficial fluency and politeness, while NGS provides complementary gains during policy optimization by improving norm grounding.
Read the full abstract ▾
All authors Zekun Yuan, Yangfan Ye, Baohang Li, Shuaibo Zhao, Zekun Zhou, Ziming Li, Qichen Hong, Kun Chen, Xiaocheng Feng
Machine Learning
Learning to Steer, Steering to See: Unveiling the Geometry of RLVR in Large Language Models via Trainable Vectors
Yuchen Cai, Ding Cao, Qixiang Yin et al. (+10 more)
Reinforcement learning (RL) has become a key paradigm for enhancing the
reasoning of large
language models, yet the high dimensionality of parameter updates makes its training dynamics hard to analyze. We study reinforcement learning with verifiable rewards (RLVR) and use vector steering to identify a low-dimensional effective
manifold in activation space associated with RL-induced gains. We uncover two
geometric properties. (1) Effective
Manifold Capacity: the capacity needed to reproduce RL gains can be very small but is not infinitely compressible; at extremely low capacity, intervention dimensionality and input-dependent expressiveness become key constraints, and this requirement varies with injection depth. (2) Control
Manifold Separation: effective control directions lie mainly in the low-variance complement of the activation principal subspace. Within a task and base model, the learned geometry stays largely consistent across training configurations, and across tasks
geometric alignment correlates with capability transfer. Experiments on 5 LLMs and 6 verifiable-reward tasks support these findings. We then propose Alpha-Stabler, a plug-and-play framework with a Predictor that monitors principal-subspace intrusion for early collapse warnings, and a Controller that removes the principal-subspace component of activation gradients during backpropagation while preserving the orthogonal complement. Alpha-Stabler stabilizes training for 2,000 steps and consistently improves RL gains, offering practical insights for robust post-training. Code: https://github.com/caiyuchen-ustc/On_Policy_Vector_Training
Read the full abstract ▾
All authors Yuchen Cai, Ding Cao, Qixiang Yin, Xin Xu, Kai Yang, Siye Wu, Pengyuan Wang, Jiaxuan Wang, Weijie Liu, Saiyong Yang, Guangzhong Sun, Guiquan Liu, Junfeng Fang
Machine Learning
Guang Li, Takahiro Ogawa and Miki Haseyama
Dataset distillation compresses a training set into a small synthetic set, usually evaluated one at a time. In
federated and data-governance settings, several parties distill their own data and a user trains on their union. We ask whether the union of separately distilled sets reproduces training on the union of the real data composability and show that it can fail even when every source is distilled exactly and the total budget admits an exact joint distillate. Compressing a training trajectory into fewer steps transforms the source statistics nonlinearly, so averaging compressed sources differs from compressing their average. For quadratic objectives we derive the exact composition error for two-to-one step compression in terms of the source-Hessian variance and the linear terms of the losses, and on a smooth network at small step sizes this prediction captures the local endpoint discrepancy in magnitude and direction. For learned synthetic sets, however, the composed error decomposes exactly into this local discrepancy and an aggregate source residual. Under endpoint matching the residual exceeds the structural term by more than an order of magnitude, and under distribution matching the two terms partly cancel. Joint
distillation also retains an accuracy advantage when both sets are distilled from the same
dataset, where the local discrepancy is exactly zero. Training fidelity and downstream accuracy are therefore distinct requirements, neither established by evaluating each set on its own.
Read the full abstract ▾
Artificial Intelligence
SAGE: Structured Strategic Reasoning for Efficient LLM Game Playing
Zhiwei Chen, Tianchun Wang, Zhongtao Rao et al. (+3 more)
A strong LLM strategic
agent should reason prospectively over uncertain futures, adapt its strategy to opponents' behavioral tendencies, and continuously recalibrate its decision process from interaction experience. However, incorporating these sources in free-form
reasoning could lead to unsupported strategic assumptions, inconsistent opponent estimates, and harmful interference from irrelevant historical interactions. To address these issues, we propose SAGE, a training-free inference-time framework that structures LLM strategic
reasoning around three coordinated operations: anchor, adapt, and recalibrate. SAGE first anchors
reasoning to an
equilibrium policy that provides a strategically valid prior. It then conditions deviations from this anchor on a soft belief over opponent behavioral tendencies, enabling opponent-specific exploitation. Finally, SAGE distills strategically related interactions into counterfactual hypotheses about previously missing considerations, allowing past experience to recalibrate the model's
reasoning. We evaluate SAGE on three repeated imperfect-information games: Leduc Hold'em, Liar's Dice, and Goofspiel, against various opponent types in each game. Compared with
reasoning-intensive LLM
agents, including Suspicion-
Agent, ReTA,
Agent-Pro, EMO, and Hypothetical Minds, SAGE achieves up to a 127.6% payoff improvement in Liar's Dice while reducing input and output
token usage by up to 80% and 90%, respectively. In direct match-up play, it attains non-negative mean payoff against 5/10, 8/10, and 8/10 evaluated opponents in Leduc Hold'em, Liar's Dice, and Goofspiel, respectively, while using relatively fewer
tokens. Code is available at https://github.com/chenzhwsysu57/SAGE.
Read the full abstract ▾
All authors Zhiwei Chen, Tianchun Wang, Zhongtao Rao, Haiming Zhu, Ding Cao, Tianxiang Zhao
Computational Engineering
A Comparative Study on Robust Topology Optimization of Design-Dependent Pressure-Actuated Compliant Mechanisms with Quadrilateral Elements
Swagatam Islam Sarkar and Prabhat Kumar
This paper presents a comparative study of compliant mechanisms generated using a robust
topology optimization technique involving design-dependent pressure loads. Design domains are parameterized using standard and higher-order quadrilateral elements. Both eroded and blueprint configurations are considered. A min-max optimization model combined with an output-spring method is employed to extremize the mechanisms' output displacements. A volume and a strain energy constraint are applied to the blueprint and the eroded designs, respectively. The optimization process is executed using the method of moving asymptotes. Numerical experiments are performed to optimize the pressure-actuated inverter and gripper mechanisms using Q4, Q8, and Q9 elements, and the results are compared. The research highlights how quadrilateral element selection influences both the resulting
topologies and performance characteristics.
Read the full abstract ▾
Distributed Computing
VarioPath: Workload-Aware All-to-All Communication for PCIe GPU Clusters
Yao Fei, Jin Fang, Size Zheng et al. (+4 more)
AlltoAllv communication is a critical primitive in distributed large-model inference, particularly for mixture-of-experts (MoE) models. The growing adoption of PCIe GPU systems for cost-efficient inference makes AlltoAllv performance on these systems increasingly important. Without a dedicated scale-up interconnect (e.g., NVLink or Infinity Fabric), PCIe GPU systems carry both intra-node and inter-node traffic through the PCIe hierarchy, where concurrent transfers can contend for PCIe link bandwidth. This link contention, compounded by skewed traffic distributions and dynamic traffic demand, makes efficient AlltoAllv scheduling challenging. Existing approaches are either poorly suited to PCIe GPU systems or incur substantial schedule synthesis overhead that reduces their practicality in real-world deployments. We present VarioPath, an efficient AlltoAllv scheduling framework for PCIe GPU systems. It combines an offline
topology-aware analyzer with an online demand-aware scheduler. The analyzer records contention-free transfer patterns as AlltoAllv channels and exploits
topology symmetry to build a compact catalog for efficient search. The online scheduler decomposes each AlltoAllv invocation's demand across a sequence of channels, adapting to rapidly changing and skewed traffic while incurring low planning overhead. Evaluation on four platforms (up to 256 GPUs) shows average AlltoAllv speedups of 5.88x over FAST and 1.72x over DeepEP. End-to-end experiments show that VarioPath reduces Qwen3 inference latency by up to 27.2% and Wan2.1 generation latency by 6.1%.
Read the full abstract ▾
All authors Yao Fei, Jin Fang, Size Zheng, Gongming Zhao, Hongli Xu, Jiacheng Zhu, Zhijing Xin
Data Structures and Algorithms
The directed temporal exploration problem
Marcelo Garlet Milani, Lucas Picasarri-Arrieta, Chaoliang Tang et al. (+1 more)
We study the temporal exploration problem on temporal digraphs. We prove that a lifetime of $O(n^2)$ suffices to guarantee the existence of a temporal exploration on always-unilateral temporal digraphs. We complement this with a $\Omega(n^2)$ lower bound, even in the case where each snapshot has maximum undirected degree 2; for always-strong temporal digraphs, the lower bound still holds even if the maximum undirected degree is 3. This stands in stark contrast with the undirected setting. For the large minimum degree setting, we show that a lifetime of $4n/3 - 1$ is sufficient and necessary for guaranteeing the existence of a temporal exploration on temporal digraphs where each snapshot is semicomplete. For always-strong temporal digraphs where each snapshot has minimum undirected degree at least $n - c - 1$, we prove that a lifetime of $O(cn)$ guarantees the existence of a temporal exploration, and we also prove that this is asymptotically tight. From a computational perspective, our results for temporal semicomplete digraphs also yield a
polynomial-time, factor-$4/3$
algorithm for deciding if a temporal semicomplete digraph admits a temporal exploration within the first $\ell$ snapshots. We complement this showing that no
polynomial-time, factor-$(4/3 - \epsilon)$ approximation
algorithm exists, even if every snapshot is a tournament, unless P$=$NP.
Read the full abstract ▾
All authors Marcelo Garlet Milani, Lucas Picasarri-Arrieta, Chaoliang Tang, Hehui Wu
Nuclear Theory
Neutron star cooling with medium-modified Urca rates
B. X. Zhou, J. B. Wei, Z. H. Li et al. (+2 more)
We study the effects of in-medium modification of the elementary cooling processes on observable properties of isolated
neutron stars. We deduce the
neutron star mass distributions compatible with the cooling analysis and compare with current theoretical models. We conclude that current cooling data require fast direct Urca cooling, moderated by
proton superfluidity, to be active in most
neutron stars, and that the DU onset threshold must lie below canonical masses. In that case medium modifications of modified Urca rates are practically insignificant, but the nn Bremsstrahlung rate plays a dominant role.
Read the full abstract ▾
All authors B. X. Zhou, J. B. Wei, Z. H. Li, G. F. Burgio, H.-J. Schulze
Audio and Speech Processing
Audio Tokens as a Budgeted Resource: Marginal-Utility Allocation for Scalable Audio Representations
Mingyu Zhao, Jinchao Zhang and Zhiyong Wu
Discrete audio
tokens are widely used as a representation interface, yet fixed-depth RVQ tokenizers allocate equal capacity to every frame despite varying refinement value. We introduce UniAdapt, which learns marginal utility of RVQ refinements on a frozen codec and allocates them under exact serialized-bit budgets. A rate-independent causal controller predicts acoustic utility, while an optional semantic head supports
speech-only utterance-level allocation; measured acoustic and semantic marginal gains on
speech have a correlation of 0.42. For causal allocation, a primal-dual allocator selects prefix-valid depths, while an exact guard constrains each sequence prefix to its matched fixed-depth serialized budget. Under utterance-level allocation, UniAdapt reduces Log-STFT distortion by 1.07-4.39 percent across
speech, music, and environmental audio without larger budgets. Causally, it improves three of four
speech rates with zero violations across 800 utterance-rate evaluations and runs faster than real time. A 20-listener utterance-level MUSHRA study shows a significant 3.52-point
speech improvement, with no significant differences on music or environmental audio. These results support separating utility prediction from budget enforcement for scalable, budget-conditioned audio representations.
Read the full abstract ▾
Astrophysics of Galaxies
Toward Resolution Convergence for Star Formation and Galactic Winds in Galaxy Simulations
Xue-Fu Li, Weishan Zhu, Tian-Rui Wang et al. (+2 more)
The predictive power of hydrodynamic simulations of
galaxies is often hampered by the resolution dependence of sub-grid models for
star formation (SF) and
stellar feedback. We present three-dimensional hydrodynamic simulations of an M82-like low-mass starburst
galaxy at resolutions of 4, 8, 16, 32 pc and develop optimized sub-grid prescriptions to mitigate these numerical inconsistencies. For
star formation, we adopt a $\sim20$ pc physical-scale control volume, corresponding to the typical size of
giant molecular clouds (GMCs), for gas
accretion onto sink particles, reducing its dependence on numerical resolution. We further incorporate an empirical, resolution- and density-dependent
star formation scheme that accounts for variations in gas surface density across spatial scales, as suggested by observations. Furthermore, we refine the
supernova feedback coupling by anchoring the injection volume to the physical cooling radius of the
supernova remnant ($R_{\rm cool}$) and utilizing a resolution-dependent thermal-to-kinetic energy partition. These improvements substantially enhance convergence across 4 to 32 pc: variations in the total
stellar mass formed and kinetic energy are reduced to $\sim20\%$, while relative errors in the cool-phase mass
outflow rate and integrated X-ray luminosity decrease to below $\sim40\%$. The warm and hot
outflow phases remain less well converged, with variations of $\sim50\%$ and $\sim75\%$, respectively. The upgraded model also improves the convergence of normalized wind properties, with relative errors below $\sim50\%$, $\sim80\%$, and $\sim20\%$ for the time-averaged mass loading factor, energy loading factor, and X-ray emission efficiency, respectively. Our results provide a practical framework for achieving more robust simulations of starburst
galaxies.
Read the full abstract ▾
All authors Xue-Fu Li, Weishan Zhu, Tian-Rui Wang, Huan Chen, Long-Long Feng
Computer Vision
SkillPE: Creativity-Oriented Cinematic Skill Evolution for Text-to-Video Prompt Engineering
Yanwei Huang, Mingxuan Zhu, Shujie Li et al. (+3 more)
Achieving high-quality, cinematic results in text-to-
video generation remains challenging for non-experts, whose
prompts often lack professional narrative and creative design. We propose SkillPE, a
prompt engineering (PE) framework that evolves reusable cinematic skills from expert-authored seeds. SkillPE represents shot logic, composition, lighting, sound design, and other filmmaking cues in a fine-grained format, and retrieves movie references categorized as resonators (good matches), dissonants (weak matches), and divergents (creatively useful near-misses). The first two refine when and how a skill should be applied, while divergents inspire alternative cinematic realizations at different degrees of modification while preserving the user intent. Candidate skills are assessed through generated
videos along
prompt fidelity, cinematic quality, narrative appeal, and creativity to construct the final skill libraries. Experiments on StoryEval and VBench show improvements of up to 1.40 points over the strongest external baseline and 0.51 points over seed skills on 7-point four-dimensional evaluation, while remaining competitive on
benchmark-native metrics. Overall, SkillPE offers a practical approach to balancing fidelity and creativity in cinematic text-to-
video generation. Code is available at https://github.com/Ais0n/SkillPE .
Read the full abstract ▾
All authors Yanwei Huang, Mingxuan Zhu, Shujie Li, Shiyuan Liu, Yuanxing Zhang, Arpit Narechania
Networking
HOCCL: Offloading Collective Communication from GPU Cores to Accelerate Distributed Training
Yao Fei, Gongming Zhao, Hongli Xu et al. (+5 more)
Large
language model training involves massive computation on GPU streaming multiprocessors (SMs), the primary compute units of GPUs. Since SMs host specialized accelerators such as Tensor Cores, their efficient utilization is critical to training efficiency. Unfortunately, existing collective communication systems compete with computation for SMs, as they consume SMs for communication-related data movement and synchronization operations. We observe that communication can, in principle, be driven by DMA engines, thereby eliminating SM involvement in communication. Based on this insight, we propose HOCCL, a zero-SM collective communication framework consisting of three components: a stream manager, a point-to-point (P2P) executor, and a collective scheduler. The stream manager preserves operator-level temporal ordering with other GPU kernels. The P2P executor enables zero-SM point-to-point communication, while the collective scheduler orchestrates P2P transfers to maximize bandwidth. Experiments show that HOCCL preserves near-peak communication performance, achieving within 3% of the state of the art on average, while eliminating communication occupancy on nearly 10% of total GPU SMs. By freeing SM resources for computation, HOCCL improves end-to-end training throughput by up to 5%.
Read the full abstract ▾
All authors Yao Fei, Gongming Zhao, Hongli Xu, Jin Fang, Jiacheng Zhu, Shuo Xu, Kun Huang, Zhuolong Yu
Signal Processing
Stacked Intelligent Metasurface-Diffractive Deep Neural Networks for Onboard Terrain Classification from SAR Level-0 Raw Data
Mengbing Liu, Xin Li, Jiancheng An et al. (+1 more)
Real-time terrain classification directly from Level-0 raw Synthetic Aperture Radar (SAR) data remains restricted by traditional digital-centric paradigms, where the processing of noisy, high-dimensional patches is hindered by computationally intensive processors and significant downlink latency. To address these fundamental limitations, this work establishes a new research paradigm for autonomous on-board sensing by proposing a Stacked Intelligent Metasurface-Diffractive Deep Neural Network (SIM-D$^2$NN). This architecture leverages the physical wave-propagation medium to offload inference tasks from digital processors to a physical device. By executing in-wave feature mapping, the SIM-D$^2$NN facilitates a move toward an integrated `compute-while-transmitting' framework, providing an alternative to the traditional `digitize-then-process' sequence. The multi-layer metasurface is positioned at the forefront of the satellite communication module. The initial layer modulates the raw SAR data through both amplitude and phase adjustments, where a 90$^\circ$ phase rotation is introduced as a lightweight but effective augmentation strategy to enhance robustness against noise and Doppler distortions. Subsequent layers learn variable phase shifts, enabling advanced feature mapping for the classification task. The classification results at the terrestrial station can be directly obtained based on the signal amplitude received at each
antenna. This design reduces reliance on downlink bandwidth and high-power terrestrial computing, achieving performance around 90% in the binary task directly from real raw SAR data in terms of accuracy, precision, recall, and F1 Score. Therefore, our method helps bridge the gap between next-generation remote sensing tasks and in-orbit processing needs, paving the way for computationally efficient remote sensing applications.
Read the full abstract ▾
All authors Mengbing Liu, Xin Li, Jiancheng An, Chau Yuen
Quantum Physics
Observation of a topological edge state among localized bulk states in the anisotropic quantum Rabi model
Sungjoo Lim, Chanyang Im, Christopher G. Yale et al. (+7 more)
Topological phases are governed by discrete
symmetries that protect boundary modes against local perturbations. When translational periodicity is absent, the bulk states also become localized, so that a topological edge state can no longer be distinguished from them by spatial localization alone. Here, we investigate the topological edge state (TES) and bulk eigenstates of the anisotropic
quantum Rabi model (AQRM) in a trapped-ion
quantum simulator. The AQRM hosts a topological phase in a one-dimensional synthetic
lattice, whose translational
symmetry is broken by the non-uniform couplings scaling with the site index. While both the TES and bulk states show localized distributions, we find that the TES exhibits well-defined chirality and near-complete
spin–
boson separability as signatures of the topological phase, in contrast to the bulk states. Phase-space
tomography further reveals that the bosonic component of the TES is a squeezed
vacuum state, with squeezing up to 6.45 dB. These results identify the TES through its intrinsic topological signatures and establish eigenstate-level characterization as a route to probing topological phenomena.
Read the full abstract ▾
All authors Sungjoo Lim, Chanyang Im, Christopher G. Yale, Brian K. McFarland, Edward C. Tortorici, Daniel S. Lobser, Melissa C. Revelle, Susan M. Clark, Mahn-Soo Choi, Junki Kim
Functional Analysis
Riesz transform on eventually Gaussian local trees
Fabrice Baudoin, Aobo Chen and Li Chen
We study the Riesz transform $\mathcal{R}=\partial(-\Delta)^{-\frac{1}{2}}$ on uniform local trees, metric measure spaces that are locally real trees and whose canonical Dirichlet form is built from weak derivatives along the skeleton. The reference measure $m$ may be singular with respect to the length measure $\nu$, so boundedness of $\mathcal R$ is understood from $L^{p}(m)$ to $L^{p}(\nu)$. In contrast with fractal-like
manifolds and cable systems, the
diffusion is sub-Gaussian at small scales and Gaussian at large scales. Under uniform volume growth, two-sided heat kernel estimates and a pointwise gradient estimate for the heat kernel, we prove that a local Dini condition on the scale function implies boundedness of $\mathcal{R}$ on $L^{p}$ for every $p\in[2,\infty)$, and hence the reverse Riesz inequality for every $p\in(1,2]$. Conversely, boundedness of $\mathcal{R}$ for some $p<2$, or a reverse Riesz inequality for some $p>2$, forces the space to be one-dimensional at small scales. We show that a reverse Hölder inequality for harmonic functions yields the gradient estimate, and verify all hypotheses for spaces carrying a
geometric group action whose generators have bounded displacement. As an application, for the alternating Vicsek fractafold in $\mathbb Z^{d}$ we determine the exact ranges of $p$ for which the Riesz and reverse Riesz inequalities hold.
Read the full abstract ▾
Computer Vision
MiCo: Mutual Information Coverage Optimization through Semantic Erasure Modeling for Efficient MLLM Inference
Tinghao Wang, Yichen Guo, Qizhe Zhang et al. (+11 more)
Multimodal large
language models (MLLMs) have demonstrated impressive performance in
multimodal understanding, but processing large numbers of visual
tokens results in high computational costs. While many methods have been proposed to reduce the number of visual
tokens, most of them rely on heuristics and are prone to discarding substantial visual information during
pruning, leading to degradation in model performance. In this work, by using a semantic erasure model, we derive a general mutual information coverage objective from task log-loss and propose MiCo, a training-free two-stage
pruning method. MiCo first uses visual signals to select a representative candidate pool before visual
tokens enter the
language model, then performs task-aware subset selection within it. At each stage, suitable observable proxies instantiate the derived objective as a monotone submodular coverage function, which MiCo greedily optimizes under the
token budget. MiCo is evaluated on diverse MLLMs ranging from 7B to 13B parameters across a broad range of image and
video benchmarks spanning general visual
reasoning, fine-grained OCR and grounding,
hallucination detection, and long-
video understanding. MiCo consistently achieves the best performance across nearly all evaluated models under all
pruning ratios. On LLaVA-NEXT-13B, MiCo uses only 5.6% visual
tokens, retains 97.5% of baseline performance, and achieves a 3.8-fold inference speedup. Our experiments demonstrate the effectiveness of MiCo and our mutual information coverage objective for visual
token pruning. Our code is available at https://github.com/MAX-7237/MiCo.
Read the full abstract ▾
All authors Tinghao Wang, Yichen Guo, Qizhe Zhang, Yuan Zhang, Weimin Ouyang, Rui Huang, Jiajun Cao, Sixiang Chen, Hao Jiang, Jixian Wu, Zheng Lu, Bofan Zhu, Renyuan Li, Shanghang Zhang
Astrophysics of Galaxies
A Study on the Evolution of NGC6822. I. Investigating the Star Formation History through Evolved Stellar Populations
Fatemeh Khatamsaz, Mahdi Abdollahi, Atefeh Javadi et al. (+4 more)
NGC6822 is an isolated
dwarf irregular
galaxy in the Local Group at a distance of $\sim$ 490 kpc . In this paper, we derived the
star formation history (SFH) of NGC6822 employing a method based on evolved asymptotic
giant branch (AGB)
stars, known as long-period variable (LPV)
stars. We utilized a
dataset of 329 LPVs in JHK$_s$ bands to estimate the
star formation rate (SFR) over time and reconstructed the SFH of the
galaxy. In addition to obtaining the SFH assuming constant
metallicity values within the range of 0.0001 $<$ Z $<$ 0.012, we also adopted three distinct age-
metallicity relations (AMRs) to account for variations in the chemical content of the
galaxy throughout its lifetime. The SFH has been investigated in two regions. The bar region encloses a central area of 189 arcmin$^{2}$, and the outer region encircles a field beyond the bar region, extending to the radial distance of 3 kpc. In the bar region, we identified three episodes of
star formation peaking at $t \sim$ 4.7 Gyr (log $t = 9.67$), $t \sim$ 1.6 Gyr (log $t = 9.21$), and $t \sim$ 48 Myr (log $t = 7.63$), where $t$ is the look-back time. In the outer region, we found one episode of
star formation occurring at $t \sim$ 4.6 Gyr. The significant peak observed $\sim$ 4-5 Gyr ago further supports the hypothesis suggested by the previous research, indicating that NGC6822 passed through the virial radius of the Milky Way in the past.
Read the full abstract ▾
All authors Fatemeh Khatamsaz, Mahdi Abdollahi, Atefeh Javadi, Hamidreza Mahani, Jacco Th. van Loon, Hedieh Abdollahi, Sepideh Ghaziasgar
General Topology
From samples to shape: a finite approach to topological invariants
Pedro J. Chocano
Given a metric space $Y$ and a compact subspace $X$, we present a method for constructing an inverse sequence of finite topological spaces by sampling points from neighborhoods of $X$ in $Y$. We show that these inverse sequences can be employed to define shape theory and to approximate shape invariants, such as Čech homology groups.
Read the full abstract ▾
Artificial Intelligence
Knowing When Thinking Is Not Enough: Teaching Small Reasoning Models to Reason Beyond Their Parametric Knowledge
Chanuk Lee, Minki Kang, Sangwoo Park et al. (+3 more)
Scaling test-time computation is a powerful way to improve
language-model
reasoning, and is particularly appealing for small
reasoning models (sRMs) that are cheap to serve. However, is additional thinking always the right operation? By intervening at intermediate
reasoning states across two model families and multiple scales, we find that self-refinement largely consolidates probability mass onto solutions already reachable from the current state, rather than making new ones reachable. These interventions reveal two failure regimes: execution bottlenecks, where the correct path is reachable and reflection can recover it, and knowledge bottlenecks, where relevant external information makes it reachable. Motivated by this distinction, we introduce FlyBy, a selective querying framework, and train 4B and 8B variants to reason first, diagnose what remains unresolved, and, at a knowledge bottleneck, query stronger models whose parametric knowledge extends beyond its own. Supervised fine-tuning bootstraps a multi-depth query action, and cost-aware reinforcement learning calibrates whether to query, what to ask, and how much to spend. On 1,158 hard problems across six
benchmarks, FlyBy-4B achieves 45.96% pass@8, surpassing Qwen3-14B (41.64%) at 2.7 times lower serving cost, while also exceeding Qwen3-8B in pass@1 (16.85% vs. 15.31%). Scaling to FlyBy-8B further improves pass@8 to 51.81%.
Read the full abstract ▾
All authors Chanuk Lee, Minki Kang, Sangwoo Park, Woongyeong Yeo, Jinheon Baek, Sung Ju Hwang