arXiv Science⌕ Search

arXiv subjects

Deng Pan

Publications and source records attributed to Deng Pan.

At least 19 recordsLinked to original sources
arXiv2026

Emergence and suppression of phonon vortices in two-dimensional crystals: Interplay of lattice symmetry, heavy impurities, and shear

Phonon vortices are vortex-like displacement fields that appear in the vibrational modes of two- dimensional materials. Here, we demonstrate that these vortices arise as symmetry-adapted linear combinations of degenerate planar phonon modes, with the superposition coefficients uniquely de- termined by the lattice point group. This symmetry principle establishes that vortices are intrinsic standing-wave solutions in perfect crystals, requiring neither impurities nor disorder. A heavy mass impurity favors vortex modes over planar modes through stronger resonance-induced frequency soft- ening, whereas shear deformation suppresses vortex modes by breaking rotational symmetry. The competition between these two effects gives rise to a precisely predictable strain threshold. The proposed framework, grounded in symmetry and energy-minimization, provides a useful basis for understanding vibrational topological defects induced by other types of impurities or defects. This study further suggests that defect and shear engineering constitutes an effective tuning strategy for controlling vibrational modes and the associated thermal and mechanical properties of crystals.

cond-mat.mtrl-sci↗
arXiv2026

Relaxation of quenched structural glasses: descent in a stiffening caging potential over inflection-point 'speed bumps'

The slow energy relaxation in quenched glasses is a ubiquitous yet poorly understood phenomenon. Despite extensive study, the microscopic origin of the observed power-law decay remains debated, with proposed mechanisms ranging from saddle-point slowdown and marginal stability to coarsening of localized excitations and phonon dynamics. Here, by simulating gradient descent in archetypal structural glass formers, we show that none of these scenarios can account for our data. Instead, the power-law behavior emerges from a remarkably simple caging effect: each particle experiences an effective stiffening potential that arises from many-body confinement and diverges at a characteristic cage size. This mechanism analytically yields the observed power-law decay and is quantitatively reproduced by a minimal single-particle cage model with fixed neighbours, demonstrating that collective relaxation modes are not essential. The dynamics is punctuated by fluctuations as the system rolls through inflection points on the energy landscape, which act as `speed bumps' but do not affect the overall power-law behaviour. In contrast to mean-field spin glass theory, we find no characteristic temperature that separates distinct dynamical regimes; state following within a given glass basin occurs universally for all initial temperatures whenever the system is sufficiently close to the inherent structure. Our results establish a complete physical picture of gradient descent dynamics in typical structural glasses.

cond-mat.dis-nn↗
arXiv2026

Predicting LLM Safety Before Release by Simulating Deployment

Pre-deployment safety evaluations aim to inform the downstream risks of releasing a new AI model. Yet most evaluations provide limited evidence about how often undesired model behavior will occur in deployment: they generally have insufficient coverage, are unrepresentative, and are generally recognizable as tests. To address these concerns, we study a simple way to simulate a model deployment: starting from de-identified conversations from a previous model deployment, we hold fixed the initial conversation prefix and regenerate the next response using a candidate model. The resulting responses can then both be audited for novel misalignments and used to estimate the prevalence of model misbehavior before deployment. We evaluate deployment simulation across four GPT-5-series deployments, using registered, outcome-blinded predictions for GPT-5.4 and retrospective analyses of three earlier releases. We find that deployment simulation produces informative estimates of post-deployment misbehavior rates and outperforms baselines based on adversarially selected production data; its evaluation-awareness point estimates were also much closer to production traffic than those from traditional evaluations. We also identify the realism of tool resampling as a central challenge for further improving predictions and share results suggesting that this challenge is surmountable even in complex tool-use settings. Finally, we show that deployment simulation can be seeded from public chat datasets and remain informative about production misbehavior rates, suggesting a path for external researchers to run deployment-grounded evaluations without access to private production logs. Overall, deployment simulation helps evaluators forecast how language models will behave in the real world and supports more quantitative assessment of deployment risk.

cs.LG↗
arXiv2026

CANN Bench: Benchmarking Agent Generated Kernels against Real NPU and Algorithmic Limits

AI agents are now capable of writing, compiling, and iteratively optimizing low-level operator kernels on different hardware platforms. Existing benchmarks, however, focus almost exclusively on CUDA and Triton, leaving hardware ecosystems with less-exposed programming models without a common evaluation baseline. We present CANN Bench, an open benchmark for AI-generated operator code on Huawei's Ascend NPU. The current release covers 53 operators and 1060 test cases organized into four difficulty tiers -- from simple elementwise primitives to MoE dispatch and FlashAttention kernels -- spanning FP16, BF16, FP32, and INT8 precision formats. Evaluation adopts a \textbf{three-dimensional weighted composite score} that treats compilation, functional correctness, and performance as independent axes, providing a principled reward signal for kernel-generation agents. Performance is graded against an out-of-the-box PyTorch-on-Ascend baseline and an analytical per-case Hardware-Anchored Performance (HAP) limit on real NPU hardware, ensuring scores reflect genuine optimization headroom rather than measurement artifacts. The evaluation harness is designed to resist reward hacking from the ground up. CANN Bench is versioned within the official CANN repository and is designed for long-term community co-construction, providing the Ascend ecosystem with a quantitative, reproducible, and sustainably maintained yardstick for AI operator-authoring capability.

cs.AI↗
arXiv2026

Programming strain-stiffening in soft composites via structural memory near jamming

Soft composite solids, comprising discrete inclusions embedded within a compliant matrix, are emerging candidates for engineering synthetic tissues and soft robotic materials. Current strategies for controlling their nonlinear mechanics, such as strain-stiffening, have primarily relied on the nonlinear elasticity of polymer matrices. Although direct contacts between inclusions may enhance stiffening responses at high densities, the role of the non-equilibrium and history-dependent nature of disordered contact networks in composite mechanics remains unexplored. In this work, by applying a mechanical training protocol near a shear-jamming phase boundary, we demonstrate that the structural memory encoded in contact networks drives a crossover from granular-like to biopolymer-like strain stiffening. Simulations of a coarse-grained composite model reveal that this biopolymer-like mechanical response emerges from enhanced non-affine reconfigurations of nearly-jammed contact networks. Without relying on matrix nonlinearity, we establish a design strategy that leverages non-equilibrium memory effects intrinsic to granular systems to achieve highly programmable strain-stiffening in soft composites.

cond-mat.soft↗
arXiv2026

Epistemic Closure: Autonomous Mechanism Completion for Physically Consistent Simulation

The integration of Large Language Models (LLMs) into scientific discovery is currently hindered by the Implicit Context problem, where governing equations extracted from literature contain invisible thermodynamic assumptions (e.g., undrained conditions) that standard generative models fail to recognize. This leads to Physical Hallucination: the generation of syntactically correct solvers that faithfully execute physically invalid laws. Here, we introduce a Neuro-Symbolic Generative Agent that functions as a cognitive supervisor atop traditional numerical engines. By encapsulating physical laws into modular Constitutive Skills and leveraging latent intrinsic priors, the Agent employs a Chain-of-Thought reasoning workflow to autonomously validate, prune, and complete physical mechanisms. We demonstrate this capability on the challenge of thermal pressurization in low-permeability sandstone. While a standard literature-retrieval baseline erroneously predicts catastrophic material failure by blindly adopting a rigid "undrained" simplification, our Agent autonomously identifies the system as operating in a drained regime (Deborah number De << 1) via dimensionless scaling analysis. Consequently, it inductively completes the missing dissipation mechanism (Darcy flow) required to satisfy boundary constraints, predicting a stable stress path consistent with experimental reality. This work establishes a paradigm where AI agents transcend the role of coding assistants to act as epistemic partners, capable of reasoning about and correcting the theoretical assumptions embedded in scientific data.

cs.DB↗
arXiv2026

Engineering-Oriented Symbolic Regression: LLMs as Physics Agents for Discovery of Simulation-Ready Constitutive Laws

The discovery of constitutive laws for complex materials has historically faced a dichotomy between high-fidelity data-driven approaches, which demand prohibitive full-field experimental data, and traditional engineering fitting, which often yields numerically unstable models outside calibration regimes. In this work, we propose an Engineering-Oriented Symbolic Regression (EO-SR) framework that bridges this gap by leveraging Large Language Models (LLMs) as "Physics-Informed Agents." Unlike unconstrained symbolic regression, our framework utilizes an LLM Agent to zero-shot synthesize executable physical constraints -- specifically thermodynamic consistency and frame indifference -- transforming the search process from mathematical curve-fitting into a physics-governed discovery engine. We validate this approach on the hyperelastic modeling of rubber-like materials using standard Treloar datasets. The framework autonomously identifies a novel hybrid constitutive law that combines a Mooney-Rivlin linear base with a rational locking term. This discovered model not only achieves high predictive accuracy across multi-axial deformation modes (including zero-shot prediction of pure shear) but also guarantees unconditional convexity. Finite element validation demonstrates that while industry-standard models (e.g., Ogden N=3) fail due to numerical singularities under severe transverse compression, the EO-SR-discovered model maintains robust convergence. This study establishes a generalized, low-barrier pathway for discovering simulation-ready constitutive closures that satisfy both data accuracy and rigorous physical laws.

cs.CE↗
arXiv2026

Skill-Based Autonomous Agents for Material Creep Database Construction

The advancement of data-driven materials science is currently constrained by a fundamental bottleneck: the vast majority of historical experimental data remains locked within the unstructured text and rasterized figures of legacy scientific literature. Manual curation of this knowledge is prohibitively labor-intensive and prone to human error. To address this challenge, we introduce an autonomous, agent-based framework powered by Large Language Models (LLMs) designed to excavate high-fidelity datasets from scientific PDFs without human intervention. By deploying a modular "skill-based" architecture, the agent orchestrates complex cognitive tasks - including semantic filtering, multi-modal information extraction, and physics-informed validation. We demonstrate the efficacy of this framework by constructing a physically self-consistent database for material creep mechanics, a domain characterized by complex graphical trajectories and heterogeneous constitutive models. Applying the pipeline to 243 publications, the agent achieved a verified extraction success rate exceeding 90% for graphical data digitization. Crucially, we introduce a cross-modal verification protocol, demonstrating that the agent can autonomously align visually extracted data points with textually extracted constitutive parameters ($R^2 > 0.99$), ensuring the physical self-consistency of the database. This work not only provides a critical resource for investigating time-dependent deformation across diverse material systems but also establishes a scalable paradigm for autonomous knowledge acquisition, paving the way for the next generation of self-driving laboratories.

cs.DB↗
arXiv2025

Hierarchical and ultrametric barriers in the energy landscape of jammed granular matter

According to the mean-field glass theory, the (free) energy landscape of disordered systems is hierarchical and ultrametric if they belong to the full-replica-symmetry-breaking universality class. However, examining this theoretical picture in three-dimensional systems remains challenging, where the energy barriers become finite. Here, we numerically explore the energy landscape of granular models near the jamming transition using a saddle dynamics algorithm to locate both local energy minima and saddles. The multi-scale distances and energy barriers between minima are characterized by two metrics, both of which exhibit signatures of an ultrametric space. The scale-free distribution of energy barriers reveals that the landscape is hierarchical.

cond-mat.soft↗
arXiv2025

Generative AI in Depth: A Survey of Recent Advances, Model Variants, and Real-World Applications

In recent years, deep learning based generative models, particularly Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Diffusion Models (DMs), have been instrumental in in generating diverse, high-quality content across various domains, such as image and video synthesis. This capability has led to widespread adoption of these models and has captured strong public interest. As they continue to advance at a rapid pace, the growing volume of research, expanding application areas, and unresolved technical challenges make it increasingly difficult to stay current. To address this need, this survey introduces a comprehensive taxonomy that organizes the literature and provides a cohesive framework for understanding the development of GANs, VAEs, and DMs, including their many variants and combined approaches. We highlight key innovations that have improved the quality, diversity, and controllability of generated outputs, reflecting the expanding potential of generative artificial intelligence. In addition to summarizing technical progress, we examine rising ethical concerns, including the risks of misuse and the broader societal impact of synthetic media. Finally, we outline persistent challenges and propose future research directions, offering a structured and forward looking perspective for researchers in this fast evolving field.

cs.CV↗
arXiv2025

Context Attribution with Multi-Armed Bandit Optimization

Understanding which parts of the retrieved context contribute to a large language model's generated answer is essential for building interpretable and trustworthy retrieval-augmented generation. We propose a novel framework that formulates context attribution as a combinatorial multi-armed bandit problem. We utilize Linear Thompson Sampling to efficiently identify the most influential context segments while minimizing the number of model queries. Our reward function leverages token log-probabilities to measure how well a subset of segments supports the original response, making it applicable to both open-source and black-box API-based models. Unlike SHAP and other perturbation-based methods that sample subsets uniformly, our approach adaptively prioritizes informative subsets based on posterior estimates of segment relevance, reducing computational costs. Experiments on multiple QA benchmarks demonstrate that our method achieves up to 30\% reduction in model queries while matching or exceeding the attribution quality of existing approaches. Our code is publicly available at https://github.com/pd90506/camab.

cs.AI↗
arXiv2025

Jamming as a topological satisfiability transition with contact number hyperuniformity and criticality

The jamming transition between flow and amorphous-solid states exhibits paradoxical properties characterized by hyperuniformity (suppressed spatial fluctuations) and criticality (hyperfluctuations), whose origin remains unclear. Here we model the jamming transition by a topological satisfiability transition in a minimum network model with simultaneously hyperuniform distributions of contacts, diverging length scales and scale-free clusters. We show that these phenomena stem from isostaticity and mechanical stability: the former imposes a global equality, and the latter local inequalities on arbitrary sub-systems. This dual constraint bounds contact number fluctuations from both above and below, limiting them to scale with the surface area. The hyperuniform and critical exponents of the network model align with those of frictionless jamming, suggesting a new universality class of non-equilibrium phase transitions. Our results provide a minimal, dynamics-independent framework for jamming criticality and hyperuniformity in disordered systems.

cond-mat.soft↗
arXiv2024

Universal activated aging and weak ergodicity breaking in spin and structural glasses

Glasses possess complex energy landscapes and exhibit non-equilibrium aging dynamics. Here, we propose a generalized trap model for activated aging based on a key static property of the energy landscape: the distribution of energy barriers. Our theory predicts that, upon cooling, weak ergodicity breaking (WEB) in quenching dynamics occurs prior to strong ergodicity breaking in equilibrium dynamics. Furthermore, the theory indicates that the characteristic size of activation clusters can be deduced from the logarithmic decay of the time-correlation function. We rigorously test the model's assumptions and predictions using the simplest spin glass model - the random energy model. The predicted aging behavior is also universally observed in paradigmatic structural glasses, including the Weeks-Chandler-Andersen (WCA) model and amorphous silica. Remarkably, applying our framework to the WCA model allows us to extract a static length from the non equilibrium dynamics, extending its observable growth range from a mere factor of 2-3 to a full order of magnitude and providing supportive evidence for the random first-order transition scenario. Finally, we propose a unified ergodic-WEB phase diagram for aging dynamics in general glassy systems.

cond-mat.dis-nn↗
arXiv2024

Gardner transition coincides with the emergence of jamming scalings in hard spheres and disks

The Gardner transition in structural glasses is characterized by full-replica symmetry breaking of the free-energy landscape and the onset of anomalous aging dynamics due to marginal stability. Here we show that this transition also has a structural signature in finite-dimensional glasses consisting of hard spheres and disks. By analyzing the distribution of inter-particle gaps in the simulated static configurations at different pressures, we find that the Gardner transition coincides with the emergence of two well-known jamming scalings in the gap distribution, which enables the extraction of a structural order parameter. The jamming scalings reflect a compressible effective force network formed by contact and quasi-contact gaps, while non-contact gaps that do not participate in the effective force network are incompressible. Our results suggest that the Gardner transition in hard-particle glasses is a precursor of the jamming transition. The proposed structural signature and order parameter provide a convenient approach to detecting the Gardner transition in future granular experiments.

cond-mat.soft↗
arXiv2024

Long-range angular correlations of particle displacements at a plastic-to-elastic transition in jammed amorphous solids

Understanding how a flow turns into an amorphous solid is a fundamental challenge in statistical physics, during which no apparent structural ordering appears. In the athermal limit, the two states are connected by a well-defined jamming transition, near which the solid is marginally stable. A recent mechanical response screening theory proposes an additional transition above jamming, called a plastic-to-elastic transition here, separating anomalous and quasi-elastic mechanical behavior. Through numerical inflation simulations in two dimensions, we show that the onsets of long-range radial and angular correlations of particle displacements decouple, occurring respectively at the jamming and plastic-to-elastic transitions. The latter is characterized by a power-law diverging correlation angle and a power-law spectrum of the displacements along a circle. This work establishes two-step transitions on the mechanical properties during ``decompression melting'' of an athermal over-jammed amorphous solid, reminiscent of the two-step structural melting of a crystal in two dimensions. In contradistinction with the latter, the plastic-to-elastic transition exists also in three dimensions.

cond-mat.soft↗
arXiv2024

Fast Explanations via Policy Gradient-Optimized Explainer

The challenge of delivering efficient explanations is a critical barrier that prevents the adoption of model explanations in real-world applications. Existing approaches often depend on extensive model queries for sample-level explanations or rely on expert's knowledge of specific model structures that trade general applicability for efficiency. To address these limitations, this paper introduces a novel framework Fast Explanation (FEX) that represents attribution-based explanations via probability distributions, which are optimized by leveraging the policy gradient method. The proposed framework offers a robust, scalable solution for real-time, large-scale model explanations, bridging the gap between efficiency and applicability. We validate our framework on image and text classification tasks and the experiments demonstrate that our method reduces inference time by over 97% and memory usage by 70% compared to traditional model-agnostic approaches while maintaining high-quality explanations and broad applicability.

cs.LG↗
arXiv2024

Jamming is a first-order transition with quenched disorder in amorphous materials sheared by cyclic quasistatic deformations

Jamming is an athermal transition between flowing and rigid states in amorphous systems such as granular matter, colloidal suspensions, complex fluids and cells. The jamming transition seems to display mixed aspects of a first-order transition, evidenced by a discontinuity in the coordination number, and a second-order transition, indicated by power-law scalings and diverging lengths. Here we demonstrate that jamming is a first-order transition with quenched disorder in cyclically sheared systems with quasistatic deformations, in two and three dimensions. Based on scaling analyses, we show that fluctuations of the jamming density in finite-sized systems have important consequences on the finite-size effects of various quantities, resulting in a square relationship between disconnected and connected susceptibilities, a key signature of the first-order transition with quenched disorder. This study puts the jamming transition into the category of a broad class of transitions in disordered systems where sample-to-sample fluctuations dominate over thermal fluctuations, suggesting that the nature and behavior of the jamming transition might be better understood within the developed theoretical framework of the athermally driven random-field Ising model.

cond-mat.soft↗
arXiv2024

Criticality in the Luria-Delbr\"uck model with an arbitrary mutation rate

The Luria-Delbr\"uck model is a classic model of population dynamics with random mutations, that has been used historically to prove that random mutations drive evolution. In typical scenarios, the relevant mutation rate is exceedingly small, and mutants are counted only at the final time point. Here, inspired by recent experiments on DNA repair, we study a mathematical model that is formally equivalent to the Luria-Delbr\"uck setup, with the repair rate $p$ playing the role of mutation rate, albeit taking on large values, of order unity per cell division. We find that although at large times the fraction of repaired cells approaches one, the variance of the number of repaired cells undergoes a phase transition: when $p>1/2$ the variance decreases with time, but, intriguingly, for $p<1/2$ even though the fraction of repaired cells approaches 1, the variance in number of repaired cells increases with time. Analyzing DNA-repair experiments, we find that in order to explain the data the model should also take into account the probability of a successful repair process once it is initiated. Taken together, our work shows how the study of variability can lead to surprising phase-transitions as well as provide biological insights into the process of DNA-repair.

physics.bio-ph↗