arXiv subjects

Yan Gao

Publications and source records attributed to Yan Gao.

At least 19 recordsLinked to original sources

arXiv2026

Crystoria: A Large-Scale Crystal Dataset Linking Thermodynamic and Dynamical Stability for Materials AI

Large-scale crystal databases have transformed materials discovery by making millions of structures and thermodynamic properties accessible to materials informatics. Yet dynamical stability remains sparsely characterized at a comparable scale because its evaluation is substantially more computationally demanding than conventional structure and energy calculations. This mismatch creates a critical gap between the rapidly expanding crystal-structure space and the physical information available for materials research and machine learning. Here we introduce Crystoria, a large-scale crystal dataset designed to bring dynamical stability to the scale of modern materials data. We construct Crystoria by systematically curating low-energy and metastable inorganic crystals, standardizing their structural identities, and combining machine-learning-accelerated structural relaxation with phonon-based dynamical screening. The resulting dataset contains 2,649,640 crystal structures, spanning 89 elements, 219 of the 230 space groups, and primitive cells containing 1-48 atoms. Among them, 1,793,353 structures (67.68%) are classified as dynamically stable, revealing substantial variation in dynamical stability across chemical and crystallographic space. By bringing dynamical stability to the scale of millions of crystal structures, Crystoria provides a physically grounded foundation for materials artificial intelligence (AI) and enables systematic exploration of the dynamical crystal landscape.

cond-mat.mtrl-sci↗

arXiv2026

Continuity of biaccessibility dimension for polynomials

For a polynomial $f$ of degree $d\geq 2$, its biaccessible set $B(f)$ consists of angles whose external rays are smooth and land at a point receiving at least two smooth external rays. We prove that the function $f\mapsto {\rm dim}_H(B(f))$ is continuous on the entire space of degree-$d$ polynomials. Therefore $(\log d){\rm dim}_H (B(f))$ gives a continuous extension of classical core entropy from the postcritically finite case to all polynomials of degree $d$.

math.DS↗

arXiv2026

RM-OPD: On-Policy Distillation for Generative Reward Models

Reinforcement Learning on Mixed-domain preference data (Mix-RL) has become the mainstream approach for training general-purpose Generative Reward Models (GRMs). However, Mix-RL suffers from two critical bottlenecks: the reward sparsity induced by outcome rewards, and the cross-domain interference arising from jointly optimizing heterogeneous capabilities, which together give rise to imbalanced capability acquisition and pervasive reward hacking. Inspired by the success of On-Policy Distillation (OPD) in multi-domain capability integration, we introduce RM-OPD (Reward Modeling via On-Policy Distillation), a two-stage framework that separates specialist acquisition from capability integration. First, we cultivate domain-specialized teacher models via single-domain GRPO fine-tuning, allowing each expert to reach its performance ceiling in isolation; then we consolidate these teachers into one student model through (i) routed on-policy distillation, which provides dense signed token-level supervision on student-generated critiques; (ii) domain contribution reweighting, which reweights token guidance toward target domain shares, preventing specific domains from dominating optimization; and (iii) optional criteria context distillation, which conditions teachers on domain-specific evaluation criteria to sharpen token-level supervision. Across multiple RM benchmarks and two base models, RM-OPD consistently outperforms vanilla Mix-RL and other strong baselines, achieving overall superior performance and exhibiting an emergent "teacher-surpassing" effect. These results establish RM-OPD as a promising paradigm for building generalist GRMs.

cs.LG↗

arXiv2026

From Pilots to Production: Lessons in Cross-Institutional Federated Training and Artificial Intelligence for Science

Many of the most valuable scientific datasets cannot be centralized: they are proprietary, export-controlled, classified, or bound by data-sovereignty restrictions. This inverts the usual paradigm: the model must move to the data, making federated artificial intelligence (AI) core infrastructure for open science. We synthesize lessons from U.S. Department of Energy national laboratories, industry deployments, and the open-source community into a structured account of moving federated AI from pilot demonstrations to dependable, multi-site production. The lessons come from five concurrent efforts spanning leadership-class supercomputers and cloud infrastructure in regulated settings. We organize them around an adapted \textit{five-domain readiness frame} and a stratified view of the stack beneath a trained model: data architecture, privacy and security controls, trust and verification, governance and socio-technical factors, and operations, the layers that decide whether a pilot becomes dependable. The question shifts from ``can we train it?'' to ``can we operate it, audit it, and change it safely?'' We report systems lessons in memory efficiency, reliability, and synchronization; set out what privacy, security, and decentralized trust require in production, and what is not yet validated there; and identify cross-cutting open problems (asynchronous federation, leakage auditing, verification standards, and harmonized data contracts) that we argue warrant a dedicated, international, open-science working group.

cs.DC↗

arXiv2026

GraphOPD: Graph-Augmented On-Policy Distillation for LLM Agents

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.

cs.LG↗

arXiv2026

Symmetry-constrained coexistence of charge-three and charge-one Weyl points in nonmagnetic crystals

High-charge Weyl semimetals (WSMs), particularly those hosting charge-three Weyl point (C-3 WPs), have attracted considerable interest owing to their unconventional topological properties and enhanced boundary phenomena. However, ideal C-3 WSMs remain rare, and the symmetry-constrained coexistence of charge-three and charge-one Weyl point (C-1 WPs) in nonmagnetic crystals remains largely unexplored. Here, we investigate a representative eight-node Weyl configuration, consisting of two C-3 WPs and six C-1 WPs with opposite chirality, corresponding to two identical $\{3,-1,-1,-1\}$ irreducible Weyl molecules. By combining topological charge conservation with space-group symmetry analysis, we systematically classify all 230 space groups and identify 12 and 23 candidate groups without and with spin--orbit coupling (SOC), respectively. First-principles calculations without SOC reveal that the boron allotropes TBR-B$_{48}$ and chiral OBD-B$_{48}$ realize this configuration, hosting hourglass-type and non-hourglass-type C-3 WPs, respectively. Surface-state calculations reveal six Fermi-arc branches at the shared projection of the charge-three pair and three at each individual projection. Reversing the structural chirality of OBD-B$_{48}$ reverses the signs of all Weyl charges, establishing a direct correspondence between crystal and momentum-space chirality. Furthermore, breaking the protecting $C_6$ symmetry in TBR-B$_{48}$ splits each C-3 WP into three C-1 WPs and reconstructs the surface Fermi arcs. Our work establishes symmetry-based guidelines for designing and searching for nonmagnetic WSMs hosting charge-three and charge-one Weyl-point coexistence.

cond-mat.mtrl-sci↗

arXiv2026

Rigidity of Simple Parabolic Maps

We prove that two simple parabolic rational maps with conformally conjugate completely invariant parabolic Fatou domains are globally conformally conjugate. The proof combines quasiconformal rigidity with the absence of invariant line fields for simple parabolic maps. Together with our previous existence theorem, the rigidity result gives existence and uniqueness of simple parabolic models for infinitely-connected fixed parabolic Fatou domains. We further obtain a rational-map analogue of the Goldberg--Milnor perturbation conjecture.

math.DS↗

arXiv2026

Population Scaling or Data Dilution? Dynamics of Local Topology Evolution in Decentralized Learning

Scaling decentralized learning changes not only the number of clients $N$, but also the dynamics of information propagation and consensus. We argue that the effect of increasing $N$ cannot be understood in isolation, because data allocation, topology-dependent mixing, and communication capacity may change simultaneously. We study these coupled effects on CIFAR-10 with $N\in\{10,50,100,200\}$, comparing a degree-two Ring, a Static Random graph, and Local-First Heuristic Evolution (LFHE), a locally adaptive topology process based on friend-of-friend discovery. The Ring provides an analytically transparent failure mode: its Metropolis spectral gap decays as $Θ(N^{-2})$, implying progressively slower contraction of model disagreement as the population grows. Experiments show that holding the nominal local dataset size fixed substantially reduces the apparent population penalty observed when a fixed total dataset is divided among more clients. The remaining degradation depends strongly on communication structure: Ring enters a high-disagreement regime, whereas Static Random and LFHE remain close to consensus. Increasing LFHE's degree threshold further improves accuracy and consensus, but at a substantially higher model-transmission cost. These results show that decentralized scaling is governed by coupled learning and communication dynamics, rather than by the number of clients alone.

cs.LG↗

arXiv2026

LFHE: Local-First Heuristic Evolution for Bounded Local Topology Search in Decentralized Learning with Non-IID Data

Decentralized learning is highly sensitive to communication topology under non-IID data. Adaptive peer-selection methods can exploit local model information, but broader peer discovery may require increasingly large control state, whereas direct spectral optimization typically relies on graph-wide information. We study the intermediate setting of bounded local topology search and propose Local-First Heuristic Evolution (LFHE), a representation-driven rewiring framework whose candidate discovery and scoring use only ego-neighborhood and friend-of-a-friend (FoF) information. The structural score admits an exact interpretation through graph Dirichlet energy: its sum across clients equals twice the representation Dirichlet energy, which under standard linear consensus dynamics governs the instantaneous dissipation of representation disagreement. LFHE combines this state-dependent structural signal with early exploration and degree control, while algebraic connectivity remains an offline graph diagnostic. Under bounded sparse degree, its FoF candidate state remains local rather than expanding toward population-wide peer tracking. Across four image, speech, and text benchmarks, LFHE achieves competitive decentralized learning performance. Matched-protocol controls identify the structural term as the principal empirical topology-selection signal, while comparison with broader peer discovery exposes a trade-off between predictive performance and discovery-state locality. Together, these results motivate state-aware bounded local topology search between pairwise peer selection and globally informed topology optimization.

cs.AI↗

arXiv2026

RankBuffer: Efficient Ranking-Based Rewards for Open-Ended Generation

Open-ended generation lacks canonical answers, making pointwise rewards difficult to calibrate for group-based reinforcement learning. Directly ranking same-query rollouts provides a more suitable relative reward signal, but existing ranking-based reward methods can incur substantial judging cost. We introduce RankBuffer, which maintains an ordered, query-specific buffer of previously judged responses as a reusable quality scale. Each rollout is first inserted into an anchor interval through an independent coarse judgment, after which only rollouts assigned to the same interval undergo local fine ranking. The resulting complete order is converted into bounded rank rewards, while boundary expansion, local refinement, and inactive-anchor pruning adapt the buffer as the policy evolves. Across four open-ended benchmarks, RankBuffer consistently outperforms all pointwise baselines. It also achieves nearly on-par performance with the strongest ranking-based reward baseline while substantially reducing judging cost. Ablations demonstrate the importance of both local fine ranking and anchor response content, while buffer analyses show that rollout-derived anchors progressively extend and refine the covered quality scale. These results establish response reuse as an effective approach to efficient relative reward construction.

cs.AI↗

arXiv2026

Irreducible Weyl Semimetals

A key objective in Weyl-semimetal (WSM) research is to identify phases built from minimal configurations of Weyl points (WPs), which provide the simplest settings for investigating the intrinsic physics of chiral topological charges. Pang et al. [Phys. Rev. Res. 8, 033304 (2026)] recently established that crystallographically realizable WSMs can be decomposed into linear combinations of 16 minimal "irreducible Weyl molecules" (IWMs). A fundamental question remains unresolved: which of these non-decomposable configurations can themselves form standalone crystalline phases? Here we introduce irreducible Weyl semimetals (IWSMs), in which two bands crossing near the Fermi level form an elementary symmetry-compatible complex whose complete Weyl-node configuration is charge neutral and crystallographically indivisible. Combining all 1651 magnetic space groups with Weyl-orbit multiplicities and band-compatibility relations, we classify IWSMs formed by crystalline-symmetry-protected twofold WPs. Only ten of the sixteen primitive inventories admit one-copy realizations, comprising four Pair, four Split, and two Mixed charge-node architectures. Remarkably, this classification uncovers five unconventional and previously unrecognized topological phases: the charge-three Pair IWSM {3, -3}; the Split IWSMs {3, -1, -1, -1}, and {4, -1, -1, -1, -1}; and the Mixed IWSMs {3, 1, -2, -2} and {3, 3, -2, -2, -2}. For all ten classes, symmetry-constrained lattice models verify the complete node inventories, Chern charges, and surface chiral-flow incidence. Our results establish crystalline irreducibility as a phase-level organizing principle beyond individual Weyl nodes and provide a symmetry-resolved framework for identifying minimal Weyl complexes in electronic, phononic, and photonic systems.

cond-mat.mtrl-sci↗

arXiv2026

Enantiomer-specific pumping of chiral molecules in high-$J$ rotational states

Enantiomer-specific state transfer (ESST) based on electric-dipole transitions has attracted considerable interest in the studies of manipulating molecules with electromagnetic fields. It also holds promise as a powerful tool for enantiomer detection and separation. To date, experimental demonstrations of ESST have focused mainly on molecules in the lowest rotational angular-momentum states, namely, $J=0$ and $J=1$. Realistic molecular samples, however, may contain appreciable populations in higher-$J$ rotational states, making it desirable to develop ESST schemes applicable to such states. Nevertheless, one cannot derive such schemes via a simple and straightforward extension of existing low-$J$ schemes, because higher-$J$ manifolds contain more magnetic sub-levels and involve more complicated networks of electric-dipole transitions. Here, we propose an ESST scheme for chiral molecules involving the $J=1$ and $J=2$ rotational states. Specifically, our scheme is based on enantiomer-specific pumping driven by three linearly polarized microwave fields and one laser beam. Under this scheme, one enantiomer is pumped into an enantiomer-specific dark state within a sub manifold with $J=1,2$, whereas the other enantiomer is dissipatively pumped out of this manifold. Our strategy can be extended to systems involving rotational states with even higher angular momenta.

quant-ph↗

arXiv2026

STRUCTUREDAGENT: Planning with AND/OR Trees for Long-Horizon Web Tasks

Existing LLM-based web agents struggle on complex, long-horizon tasks due to limited in-context memory, weak planning abilities, and greedy behaviors that lead to premature termination. To address these challenges, we propose \SA{}, a hierarchical planning framework that interleaves planning and execution via dynamic $\ANDOR$ trees. The framework separates structural planning from LLM-based reasoning, enabling principled error recovery through node repair, systematic exploration of alternatives via OR nodes, and modular plans that can facilitate human intervention. On WebArena (630 tasks), \SA{} achieves a $\sim$53\% success rate vs.\ $\sim$46\% for AgentOccam, and on complex multi-constraint Amazon shopping tasks, gains reach 10\% over the strongest baseline.

cs.AI↗

arXiv2026

Symmetry-protected triplet Weyl complexes

The Nielsen-Ninomiya theorem dictates that Weyl nodes must appear in pairs of opposite chirality to preserve global charge neutrality. However, in crystals, specific crystalline symmetries can stabilize multi-Weyl nodes, circumventing this pairwise constraint and enabling compensated Weyl complexes with mixed chiral charges. The minimal configuration of this type is a triplet Weyl complex (TWC), comprising exactly three Weyl nodes. Here, we systematically investigate the symmetry conditions required to realize TWCs. By screening all 1651 magnetic space groups (MSGs) in both spinless and spinful systems, we establish that: (i) Only TWCs with charge magnitudes of $\{1,1,2\}$ and $\{1,2,3\}$ are permitted; (ii) the $\{1,1,2\}$ configuration can be realized in 166 spinless MSGs and 70 spinful MSGs; and (iii) the $\{1,2,3\}$-TWCs, which has not been reported before, can occur in 10 MSGs for both spinless and spinful cases. We explicitly demonstrate the existence of $\{1,2,3\}$-TWC in a tight-binding model. Furthermore, we present the first electronic realization of $\{1,1,2\}$-TWC topological semimetal state in the chiral carbon allotrope DZQH-C$_{36}$, in which the three Weyl nodes form a collinear configuration, leading to a characteristic ``S''-shaped surface Fermi arc pattern. Our findings uncover novel topological states featuring mixed chiral charges and provide guidance for exploring their physics in concrete material systems.

cond-mat.mtrl-sci↗

arXiv2026

Multi-Image Visual Token Pruning in Large Visual Language Models

With the growing demand for processing multiple image sequences in real-world applications, various visual token pruning methods have emerged to mitigate the computational and context length constraints faced by Large Vision Language Models (LVLMs). However, most existing pruning approaches rely on static strategies that struggle to adapt across different architectural LVLMs and multi-image scenarios, and are additionally constrained by their dependence on attention computations that are incompatible with efficient techniques like FlashAttention. To address these limitations, we propose a training-free, Adaptive Visual Token Pruning (AVTP) framework, applicable to diverse LVLM architectures. We strategically determine pruning layers based on empirical analysis of visual attention distributions across various LVLMs, and implement adaptive pruning ratios in multi-image contexts where images of higher importance retain proportionally more tokens. We conduct extensive experiments across different LVLMs to demonstrate the effectiveness and robustness of AVTP. Specifically, Qwen3VL-8B achieves 2 times inference speedup while maintaining 96.1\% of its original accuracy on multiple multi-image benchmarks, InternVL3.5-8B retains 94.1\% accuracy, and LLaVA-OV-7B even exceeds its original baseline performance. Our code is available at \href{https://github.com/zry13/AVTP}{this link}.

cs.CV↗

arXiv2026

Tournament-GRPO: Group-Wise Tournament Rewards for Reinforcement Learning in Open-Ended Long-Form Generation

Reinforcement learning in open-ended long-form generation is challenging because reliable reference answers and automatic metrics are often unavailable. Existing rubric-based methods typically rely on pointwise LLM-as-a-judge scoring, but absolute scores are difficult to calibrate across complex responses, may provide weak discrimination among same-query rollouts, and can become saturated during optimization. We propose Tournament-GRPO, a group-wise reward framework that converts rubric-guided LLM judgments into relative rewards through repeated multi-round tournaments among same-query rollouts. Tournament-GRPO compares candidates within groups, accumulates tournament outcomes, and normalizes them into group-wise rewards for GRPO training. Experiments on Deep Research Bench show that Tournament-GRPO consistently outperforms existing reward-design baselines, achieving a 4.52-point overall-score improvement over the strongest baseline. Further analyses show that tournament rewards provide a favorable effectiveness--efficiency trade-off and that tournament design affects training dynamics. These results suggest that rubric-guided tournament comparison provides an effective reward signal for reinforcement learning in open-ended long-form generation.

cs.CL↗

arXiv2026

Flower Hub: A Reproducible Benchmarking Platform for Federated Learning in Simulation and Deployment

Federated learning (FL) has emerged as a key approach for training models across decentralized data, yet benchmarking in FL remains difficult to reproduce, compare, and extend. Existing evaluations are often tied to custom infrastructure, released as incomplete research code, and conducted primarily in simulation, which limits portability and practical relevance. We present Flower Hub, a platform for publishing, discovering, and executing decentralized and federated applications. We show how it enables reproducible benchmarking by packaging benchmarks as executable, versioned applications with standardized metadata, pinned dependencies, and explicit evaluation workflows. We instantiate this approach with a multi-domain benchmark suite spanning cross-silo and cross-device settings, and including tasks in medical imaging, financial tabular learning, legal instruction tuning, phishing URL detection, and audio tagging. We further demonstrate that the same benchmarking application can run across both simulation and deployment runtimes without changing the application code, enabling unified evaluation across varying learning environments. Beyond model quality, our benchmark design supports system-aware reporting, including runtime and communication metrics. This work advances benchmarking in FL settings from ad hoc code artifacts towards portable, executable, and reusable benchmark applications.

cs.LG↗

arXiv2026

SIGMA-Lane: Scale-pyramId Gated MAmba for Temporally Consistent Video Lane Detection

Video lane detection requires predictions that remain stable across frames, yet severe vehicle occlusions can break temporal cues. In streaming recurrent models, corrupted observations may enter the hidden state and produce errors that persist into later frames. Existing occlusion-aware refinements usually provide obstacle masks as auxiliary inputs, so the state-update path is only indirectly protected. We propose SIGMA-Lane, which treats this failure mode as state contamination in State Space Model (SSM)-based temporal modeling. SIGMA-Lane places occlusion-aware gates on the SSM write and residual-fusion paths, controlling how current observations enter temporal memory and are fused back after temporal propagation. After coordinate-consistent affine alignment, the model combines two complementary paths: SSM-consistent dual-gating for temporal filtering and Structural Spatial Retrieval (SSR) for recovering missing lane structure from aligned historical priors. Experiments on VIL-100 and OpenLane-V show improved temporal stability under heavy occlusion, with competitive F1 and mIoU scores.

cs.CV↗