Search arXiv⌕ Search

arXiv subjects

Hang Yuan

Publications and source records attributed to Hang Yuan.

At least 19 recordsLinked to original sources
arXiv2026

Visualizing Three-Dimensional Quantum Fluctuations of Ground-State C$_3$H$_4$ via Deep Sub-Angstrom Coulomb Explosion Imaging

Coulomb explosion imaging (CEI) is a powerful tool for determining static geometries of isolated gas-phase molecules. In this work, taking allene and propyne as prototype, we demonstrate sub-angstrom resolution achieved with highly charged ion-induced CEI. Using 41.9 keV/u Xe$^{20+}$ beam, minute variations of CC single, double and triple bond lengths are clearly visualized, indicating that an ultrahigh resolution of 0.1 $\mathring{\mathrm{A}}$ is achieved. This unprecedented resolution not only makes it possible to distinguish CC bonds with different bond orders, but also lays a solid foundation for direct observation of quantum fluctuations inherent to 3D structure of a ground-state molecule. Notably, the coupled collective motion of the two CH$_2$ moieties spaced by a central carbon within allene is unambiguously visualized in our measurements.

physics.chem-ph↗
arXiv2026

PhysBrain 1.5: From Vision-Language Models to Physical Foundation Models

We present PhysBrain 1.5, a unified model for understanding physical environments, generating actions, and predicting future states. Motivated by the physical loop of observation, interaction, and environmental change, we bring these capabilities into a common learning framework. Starting from a general vision--language model, we encode language responses, end-effector motion, and dense visual targets as discrete sequences and jointly optimize them with autoregressive next-token prediction. Pre-training draws its embodied supervision entirely from human interaction videos, using task-centered episodes to pair semantic and spatial context with recovered motion and subsequent observations. We then adapt the model through supervised fine-tuning on a mixture of human demonstrations, robot trajectories, and simulated experience. Across 28 embodied understanding benchmarks, our 8B model achieves an average score of 72.5, setting a new open-source state of the art and performing on par with leading proprietary models such as GPT-6-Astra and Gemini 3.6 Flash. It achieves the best open-source results on 14 benchmarks while retaining general multimodal capabilities. Beyond these understanding evaluations, qualitative examples show the model's ability to produce end-effector trajectories and predict future scenes through spatially aligned RGB, depth, and robot-mask outputs.

cs.CV↗
arXiv2026

Hierarchical Codebook Design and Low-Overhead Beam Training for Near-Field Communications With Uniform Circular Arrays

Extremely large-scale multiple-input multiple-output (XL-MIMO) enables near-field location-specific beam focusing for sixth-generation (6G) communications. Uniform circular arrays (UCAs), with rotational symmetry and uniform azimuth coverage, have emerged as a key enabling architecture for near-field XL-MIMO systems. In this paper, we propose a resolution-aware hierarchical codebook for near-field UCA systems, along with an efficient two-stage beam training scheme to significantly reduce the training overhead. Specifically, we characterize the minimum resolvable distance of UCA systems in the near-field region based on a geometric spherical-wave propagation model, revealing their spatial resolution capability in the joint angle--distance domain. Guided by this result, we design a UCA-specific hierarchical codebook, where a power-efficient distance-robust beamforming (DRBF) codebook provides coarse azimuth localization and a full-precision (FP) codebook sampled according to the minimum resolvable distance enables refined angle--distance beam search. The regularized modal compensation suppresses weak-mode amplification and provides a controllable tradeoff between absolute amplitude gain under unit-norm transmission and distance robustness. Based on this codebook, we develop a hierarchical decoupled-architecture Bayesian regression (HDA-BAR) scheme for fast and accurate near-field beam training. For the considered array configuration, the resulting HDA-BAR training procedure requires $384$ probing slots, corresponding to an approximately \(99.66\%\) overhead reduction relative to the conventional near-field exhaustive-search benchmark.

eess.SP↗
arXiv2026

IntentVLA: Short-Horizon Intent Modeling for Aliased Robot Manipulation

Robot imitation data are often multimodal: similar visual-language observations may be followed by different action chunks because human demonstrators act with different short-horizon intents, task phases, or recent context. Existing frame-conditioned VLA policies infer each chunk from the current observation and instruction alone, so under partial observability they may resample different intents across adjacent replanning steps, leading to inter-chunk conflict and unstable execution. We introduce IntentVLA, a history-conditioned VLA framework that encodes recent visual observations into a compact short-horizon intent representation and uses it to condition chunk generation. We further introduce AliasBench, a 12-task ambiguity-aware benchmark on RoboTwin2 with matched training data and evaluation environments that isolate short-horizon observation aliasing. Across AliasBench, SimplerEnv, LIBERO, and RoboCasa, IntentVLA improves rollout stability and outperforms strong VLA baselines

cs.RO↗
arXiv2026

Learning Human Health and Diseases from 24-hour Wrist Movement

Much of human health and function unfolds beyond the clinic, through the movements of everyday life. Wrist-worn accelerometers capture these movements continuously, yet their rich signals are often reduced to a small set of predefined behavioural summary measures. Here, we present Sensori, a self-supervised foundation model that learns general-purpose health representations directly from 24 hours of raw tri-axial wrist movement. We developed and evaluated the model across four population-based cohorts from the United Kingdom, China and the United States, comprising 122,640 participants contributing 683,617 person-days of free-living recordings. Sensori condensed each day of movement into a representation that captured diverse movement behaviours, demographic characteristics, health axes and physical function. Evaluation in independent cohorts showed that these representations generalised across populations and measurement settings without retraining. When added to common clinical covariates, Sensori significantly improved prevalent disease classification for 52 of 102 eligible conditions (median delta AUROC, 0.060; range, 0.012-0.242) and incident disease risk prediction for 26 of 87 eligible conditions (median delta Uno's C-index, 0.064; range, 0.025-0.172), with the largest gains for neurological and psychiatric disorders. These findings establish 24-hour wrist movement as a rich and scalable source of health information, with the potential to support passive health monitoring and disease prediction at population scale.

cs.LG↗
arXiv2026

Gene Expression-Informed Jointly Controlled Generative Modeling for Precision Molecular Design

Precision molecular design aims to discover personalized drug candidates through joint control of multiple conditions, such as biological relevance and molecular design strategies. Biological relevance reflects cellular functional states under disease or perturbation conditions, while molecular design strategies provide complementary guidance in terms of structural intentions and property optimization. In this study, we propose JoPMol, a jointly controlled precision molecular generative model that integrates biological states encoded by gene expression profiles with molecular structure information expressed in text, and chemical properties quantified by numerical values within a unified modeling framework. This formulation enables coordinated generation and optimization of candidate molecules under joint condition control. Experimental results show that JoPMol outperforms state-of-the-art methods across multiple evaluation metrics. Moreover, JoPMol demonstrates strong generalization ability in both transfer tasks and biologically grounded simulation scenarios, validating its effectiveness for precision molecular design. The source code is publicly available at https://github.com/hala-yh/JoPMol.

cs.LG↗
arXiv2026

Family Floer SYZ mirror algorithm for the Grassmannian $Gr(2,4)$

We give an explicit non-archimedean SYZ construction for the Landau-Ginzburg mirror of $Gr(2,4)$. This work is complementary to the approach of Hong-Kim-Lau \cite{hong2023immersed} to SYZ mirror symmetry for Grassmannians, while we focus on a more concrete fibration-level realization of the SYZ picture. Starting from a Lagrangian fibration on the A-side, we explicitly construct a non-archimedean analytic mirror fibration inside the Berkovich analytification of the Langlands dual Grassmannian on the B-side. We show that the two fibrations have identical smooth and singular loci and induce the same integral affine structure on the smooth locus. Moreover, the natural disk-counting Landau-Ginzburg superpotential agrees with the Marsh-Rietsch superpotential. While the construction is guided by the family Floer viewpoint, the proof proceeds mainly through explicit geometric constructions and does not rely on Floer-theoretic arguments. Thus, the Langlands-dual mirror and its superpotential are realized explicitly within a single framework, providing concrete geometric evidence for the SYZ principle.

math.AG↗
arXiv2026

Existence of pseudo-holomorphic disks via non-archimedean disk potentials

We show that if a graded monotone Lagrangian $L_0$ has a non-vanishing disk potential, then for every smooth isotopy $\{L_s\}_{s\in[0,1]}$ of Lagrangians starting from it and for every tame almost complex structure $J$, each $L_s$ bounds a $J$-holomorphic disk of Maslov index two. The main input is a non-archimedean analytic potential function, defined as an invariant up to analytic isomorphisms, generalizing the classical disk potential of a monotone Lagrangian. The techniques are inspired by recent developments in the Strominger-Yau-Zaslow mirror construction via family Floer theory and non-archimedean geometry. We also discuss applications such as recovering a simple case of Audin's conjecture.

math.SG↗
arXiv2026

Family Floer SYZ conjecture for $A_n$ singularity

We resolve a mathematically precise SYZ conjecture for $A_n$ singularity by building a quantum-corrected T-duality between two singular torus fibrations related to the Kähler geometry of the $A_n$-smoothing and the Berkovich geometry of the $A_n$-resolution, respectively. Our approach involves novel computations that embody a non-archimedean version of the partition of unity, and it confirms the strategy that patching verified local singularity models brings global SYZ conjecture solutions (like K3 surfaces) within reach. There is also explicit extra evidence concerning the collision of singular fibers and braid group actions. On one hand, we address the central challenge of matching SYZ singular loci identified by Joyce. In reality, we construct not merely an isolated SYZ mirror fibration partner, but a parameter-dependent one that always keeps the matching singular loci plus integral affine structure, even when the collision of singular fibers occurs. On the other hand, our SYZ result displays a visible tie, regardless of the parameter choice, between the $(A_n)$-configuration of Lagrangian spheres occurred as vanishing cycles in the $A_n$-smoothing and the exceptional locus of rational $(-2)$-curves in the $A_n$-resolution, which aligns with the celebrated works of Khovanov, Seidel, and Thomas.

math.AG↗
arXiv2026

Factorization-Error-Free Discrete Diffusion Language Model via Speculative Decoding

Discrete diffusion language models improve generation efficiency through parallel token prediction, but standard $X_0$ prediction methods introduce factorization errors by approximating the clean token posterior with independent token-wise distributions. This paper proposes Factorization-Error-Free Discrete Diffusion Language Modeling (FeF-DLLM), which replaces independent clean-token prediction with an exact prefix-conditioned factorization of the clean posterior to better preserve token dependencies. To reduce the sequential cost introduced by prefix conditioning, FeF-DLLM further incorporates speculative decoding within diffusion denoising, accelerating inference while maintaining the parallel prediction and re-masking properties of DLLMs. Theoretically, we prove that FeF-DLLM generates from the true joint distribution and derive its expected acceleration ratio. Experiments on GSM8K, MATH, HumanEval, and MBPP demonstrate that our method improves accuracy by an average of 5.04 percentage points while achieving an average inference speedup of $3.86\times$.

cs.CL↗
arXiv2026

PhysBrain 1.0 Technical Report

Vision-language-action models have advanced rapidly, but robot trajectories alone provide limited coverage for learning broad physical understanding. PhysBrain 1.0 studies a complementary route: converting large-scale human egocentric video into structured physical commonsense supervision before robot adaptation. Our data engine extracts scene elements, spatial dynamics, action execution, and depth-aware relations, then turns them into question-answer supervision for training PhysBrain VLMs. The resulting physical priors are further transferred to VLA policies through a capability-preserving and language-sensitive adaptation design. Across multimodal QA benchmarks and embodied control benchmarks, including ERQA, PhysBench, SimplerEnv-WidowX, LIBERO, and RoboCasa, PhysBrain 1.0 achieves SOTA results and shows especially strong out-of-domain performance on SimplerEnv. These results suggest that scaling physical commonsense from human interaction video can provide an effective bridge from multimodal understanding to robot action.

cs.RO↗
arXiv2026

FrameSkip: Learning from Fewer but More Informative Frames in VLA Training

Vision-Language-Action (VLA) policies are commonly trained from dense robot demonstration trajectories, often collected through teleoperation, by sampling every recorded frame as if it provided equally useful supervision. We argue that this convention creates a temporal supervision imbalance: long low-change segments dominate the training stream, while manipulation-critical transitions such as alignment, contact, grasping, and release appear only sparsely. We introduce FrameSkip, a data-layer frame selection framework that scores trajectory frames using action variation, visual-action coherence, task-progress priors, and gripper-transition preservation, then remaps training samples toward high-importance frames under a target retention ratio. Because FrameSkip operates only in the dataloader, it leaves the VLA architecture, action head, training objective, and inference procedure unchanged. Across RoboCasa-GR1, SimplerEnv, and LIBERO, FrameSkip improves the success-retention trade-off over full-frame training and simpler frame selection variants, achieving a macro-average success rate of 76.15% across the three benchmarks compared with 66.50% for full-frame training while using a compressed trajectory view that retains 20% of unique frames in the main setting.

cs.RO↗
arXiv2026

DanceCrafter: Fine-Grained Text-Driven Controllable Dance Generation via Choreographic Syntax

Text-driven controllable dance generation remains under-explored, primarily due to the severe scarcity of high-quality datasets and the inherent difficulty of articulating complex choreographies. Characterizing dance is particularly challenging owing to its intricate spatial dynamics, strong directionality, and the highly decoupled movements of distinct body parts. To overcome these bottlenecks, we bridge principles from dance studies, human anatomy, and biomechanics to propose \textit{Choreographic Syntax}, a novel theoretical framework with a tailored annotation system. Grounded in this syntax, we combine professional dance archives with high-fidelity motion capture data to construct \textbf{DanceFlow}, the most fine-grained dance dataset to date. It encompasses 41 hours of high-quality motions paired with 6.34 million words of detailed descriptions. At the model level, we introduce \textbf{DanceCrafter}, a tailored motion transformer built upon the Momentum Human Rig. To circumvent optimization instabilities, we construct a continuous manifold motion representation paired with a hybrid normalization strategy. Furthermore, we design an anatomy-aware loss to explicitly regulate the decoupled nature of body parts. Together, these adaptations empower DanceCrafter to achieve the high-fidelity and stable generation of complex dance sequences. Extensive evaluations and user studies demonstrate our state-of-the-art performance in motion quality, fine-grained controllability, and generation naturalness.

cs.CV↗
arXiv2026

3D-Mix for VLA: A Plug-and-Play Module for Integrating VGGT-based 3D Information into Vision-Language-Action Models

Vision-Language-Action (VLA) models leverage Multimodal Large Language Models (MLLMs) for robotic control, but recent studies reveal that MLLMs exhibit limited spatial intelligence due to training predominantly on 2D data, resulting in inadequate 3D perception for manipulation tasks. While recent approaches incorporate specialized 3D vision models such as VGGT to enhance spatial understanding, they employ diverse integration mechanisms without systematic investigation, leaving the optimal fusion strategy unclear. We conduct a comprehensive pilot study comparing nine VGGT integration schemes on standardized benchmarks and find that semantic-conditioned gated fusion, which adaptively balances 2D semantic and 3D geometric features based on task context, achieved the strongest performance among all nine evaluated fusion schemes in our pilot study. We present 3D-Mix, a plug-and-play module that integrates into diverse VLA architectures (GR00T-style and $π$-style) without modifying existing MLLM or action expert components. Experiments across six MLLM series (nine model variants, 2B--8B parameters) on SIMPLER and LIBERO show that 3D-Mix delivers consistent performance gains, averaging +7.0% on the out-of-domain (OOD) SIMPLER benchmark across all nine GR00T-style variants, establishing a principled approach for enhancing spatial intelligence in VLA systems.

cs.RO↗
arXiv2026

Higher operad structure for Fukaya categories

Operads often arise from geometry. The standard $A_\infty$ operad can be derived from the cellular chains on the Stasheff associahedra, and an $A_\infty$ algebra is an algebra over this operad. The notion of an $\mathbf{fc}$-multicategory, also called a virtual double category, is a two-dimensional generalization of operads and multicategories. Here $\mathbf{fc}$ stands for the free category monad. We establish a natural $\mathbf{fc}$-multicategory structure on the collection of moduli spaces of pseudo-holomorphic polygons with boundary on sequences of Lagrangian submanifolds in a symplectic manifold. These moduli spaces are known to underlie the construction of Fukaya categories. Based on this, we develop the theory of differential graded (dg) variants of $\mathbf{fc}$-multicategories and show that a broad range of $A_\infty$-type structures, such as $A_\infty$ algebras, $A_\infty$ (bi)modules, and $A_\infty$ categories (possibly curved), admit a uniform operadic formulation as algebras over dg $\mathbf{fc}$-multicategories.

math.AT↗
arXiv2026

Motion Capture is Not the Target Domain: Scaling Synthetic Data for Learning Motion Representations

Synthetic data offers a compelling path to scalable pretraining when real-world data is scarce, but models pretrained on synthetic data often fail to transfer reliably to deployment settings. We study this problem in full-body human motion, where large-scale data collection is infeasible but essential for wearable-based Human Activity Recognition (HAR), and where synthetic motion can be generated from motion-capture-derived representations. We pretrain motion time-series models using such synthetic data and evaluate their transfer across diverse downstream HAR tasks. Our results show that synthetic pretraining improves generalisation when mixed with real data or scaled sufficiently. We also demonstrate that large-scale motion-capture pretraining yields only marginal gains due to domain mismatch with wearable signals, clarifying key sim-to-real challenges and the limits and opportunities of synthetic motion data for transferable HAR representations.

cs.LG↗
arXiv2026

A$^2$-LLM: An End-to-end Conversational Audio Avatar Large Language Model

Developing expressive and responsive conversational digital humans is a cornerstone of next-generation human-computer interaction. While large language models (LLMs) have significantly enhanced dialogue capabilities, most current systems still rely on cascaded architectures that connect independent modules. These pipelines are often plagued by accumulated errors, high latency, and poor real-time performance. Lacking access to the underlying conversational context, these pipelines inherently prioritize rigid lip-sync over emotional depth. To address these challenges, we propose A$^2$-LLM, an end-to-end conversational audio avatar large language model that jointly reasons about language, audio prosody, and 3D facial motion within a unified framework. To facilitate training, we introduce FLAME-QA, a high-quality multimodal dataset designed to align semantic intent with expressive facial dynamics within a QA format. By leveraging deep semantic understanding, A$^2$-LLM generates emotionally rich facial movements beyond simple lip-synchronization. Experimental results demonstrate that our system achieves superior emotional expressiveness while maintaining real-time efficiency (500 ms latency, 0.7 RTF).

cs.LG↗
arXiv2025

Euclid's Gift: Enhancing Spatial Perception and Reasoning in Vision-Language Models via Geometric Surrogate Tasks

Spatial intelligence spans a rich suite of abilities, including visualising and transforming shapes, mentally rotating objects, judging relational positions and containment, and estimating numerosity. However, it still remains a critical unresolved challenge for Multimodal Large Language Models (MLLMs). To fill this gap, we propose to treat Euclidean geometry problem-solving as a surrogate task. Specifically, we meticulously constructed a curated multimodal dataset, called Euclid30K, comprising approximately 30K plane and solid geometry problems. Furthermore, to enable the model to learn and apply Euclidean principles from these geometry problems, we fine-tuned seven model variants (spanning 3--72B parameters) from the Qwen2.5VL, Qwen3VL, and RoboBrain2.0 families using Group Relative Policy Optimization (GRPO), inspiring the models to identify shapes, count, and relate entities, and perform multi-step deductive reasoning using Euclidean principles. Our experiments demonstrate that the resulting models achieve substantial zero-shot gains across four spatial reasoning benchmarks (Super-CLEVR, Omni3DBench, VSI-Bench, and MindCube) without any task-specific adaptations. Notably, after training on the Euclid30K, the mean VSI-Bench accuracy rose from 36.6\% to 41.8\% (+5.2\%), and the mean MindCube accuracy rose from 31.4\% to 38.1\% (+6.7\%). To our knowledge, this is the first systematic study showing that geometry-centric fine-tuning can confer vision-language models with broadly transferable spatial skills. Code and Euclid30K dataset can be found in \href{https://zgca-ai4edu.github.io/Euclids_Gift}{this}.

cs.CV↗