Searcharxiv⌕ Search

arXiv subjects

Feiyu Zhou

Publications and source records attributed to Feiyu Zhou.

6 recordsLinked to original sources

Distilling an atomistic foundation model is all you need for large-scale molecular dynamics of semiconductor interfaces

State-of-the-art atomistic foundation models provide broadly transferable energy and force predictions, but their direct deployment in large-scale molecular dynamics remains computationally costly. Here we examine whether their knowledge can be distilled into compact potentials while retaining the physical responses of heterogeneous semiconductor interfaces. From 128 pretrained models spanning 22 model families, we select a fixed EquiformerV3 teacher and train atomic cluster expansion (ACE) and neuroevolution potential (NEP) students without target-specific teacher fine-tuning, with both simulation speed and accessible system size improved by several orders of magnitude relative to the teacher model. Matched training configurations and architectures enable direct comparisons with students trained on density-functional-theory (DFT) reference labels. For 79,920-atom BAs/GaN interfaces, direct distillation preserves termination-dependent thermal conductance and dominant spectral features, with conductance differing by at most 6.6% from DFT-trained counterparts. In 29,792-atom Cu/diamond specimens loaded at 500 K, tensile strength and loading work differ by 2.8% and 1.5%, respectively. Exploratory SiC/a-C/SiC simulations retain the principal thermal bottleneck and the direction of the annealing response. Furthermore, a four-model committee retrieves 42-56% of the highest-error teacher configurations with a 10% selection budget and improves high-error force fitting, but does not consistently improve downstream properties. These findings demonstrate that direct distillation can preserve key interfacial thermal and mechanical responses in compact interatomic potentials, opening a path towards large-scale, high-throughput simulations of heterogeneous interfaces.

physics.comp-ph↗

A Transferable Machine-Learned Potential Enabling Large-Scale Copper-Carbon Simulations from Equilibrium to Extreme Conditions

Cu-C materials and interfaces are central to diamond thermal management, metallization, graphene growth and carbon-coating technologies, but atomistic simulations across these applications require a potential that describes metallic Cu, diverse carbon bonding, heterogeneous Cu-C interactions and close atomic encounters. Here we develop NEP-CuC, a transferable neuroevolution potential trained on a dataset of 2,727 uniformly relabelled configurations selected from a heterogeneous Cu-C candidate pool. Disagreement within an ensemble of five independently trained models provides an internal indicator of model sensitivity. A Ziegler-Biersack-Littmark (ZBL) correction describes short-range nuclear repulsion, while a runtime D3(BJ) correction supplements dispersion-sensitive calculations. Tests on source-derived held-out and independent external datasets, together with structural, elastic, defect, interfacial, vibrational and finite-temperature transport benchmarks, establish broad but property-dependent physical fidelity across elemental Cu, diverse carbon environments and heterogeneous Cu-C configurations. Large-scale simulations of amorphous-carbon formation and orientation-dependent shock compression of Cu further test transferability under strongly non-equilibrium conditions. For Cu/diamond interfaces, the same potential describes deposition, annealing and tensile response: preparation history leaves persistent differences in interphase structure, while the mean peak nominal tensile stress at 500 K decreases from $12.35\pm0.71$ to $6.65\pm0.98$ GPa as the incident Cu energy during the initial deposition stage increases from 1 to 50 eV. NEP-CuC provides an atomistic framework for linking processing, interfacial structure and physical response across Cu-C systems, with direct relevance to Cu/diamond thermal-management interfaces, diamond metallization and non-equilibrium Cu-C processing.

cond-mat.mtrl-sci↗

Transitions between bulk and interfacial fracture in diamond/$c$BN heterostructures

Whether an initially crack-free heterostructure fails at its interface or within an adjoining phase is controlled by the relative cohesion of competing atomic planes, but how interfacial chemistry, crystallographic orientation, and intermixing reshape this competition remains unclear. Here, we combine density functional theory (DFT) with a fine-tuned atomistic foundation model to resolve tensile fracture in coherent diamond/cubic boron nitride (cBN) heterostructures. The resulting potential reproduces independent DFT tensile responses, including an unseen (001) interface orientation. Interfacial termination, orientation, and diffusion-induced intermixing jointly determine fracture resistance and fracture-plane selection. C-N-bonded (111) and C-B-bonded (001) remain interface-controlled throughout the investigated intermixing range, whereas pristine C-B-bonded (111) fractures at a neighboring B-N plane inside cBN because the interface is more strongly bound. Increasing the diffusion fraction from 0 to 50.0% causes a nonlinear decrease in fracture strength from 50.3 to 15.8 GPa and drives a bulk-to-interface transition through three regimes: cBN fracture up to 4.86%, configuration-dependent competition between 7.6 and 10.1%, and interfacial fracture at 12.5% and above. Atom-resolved stress fields and DFT separation energetics show that this transition is associated with stress relocation and a reversal in the relative cohesion of competing planes, while electron localization analysis connects the cohesion hierarchy to termination- and orientation-dependent bonding. These results establish an atomistic framework for controlling fracture resistance and fracture pathways in strongly bonded heterostructures.

cond-mat.mtrl-sci↗

YOLO with Kolmogorov-Arnold networks and vision-language foundation models for interpretable object detection with trustworthy multimodal AI in computer vision perception

The trustworthy object detection capabilities of a novel Kolmogorov-Arnold network framework are examined here. The approach addresses a key limitation in computer vision for vehicle detection perception, and beyond. These systems offer limited transparency regarding the reliability of their confidence scores in visually degraded or ambiguous scenes. To this end, a Kolmogorov-Arnold network is employed as an interpretable post-hoc surrogate to model the trustworthiness of the You Only Look Once (Yolov10) detections using seven geometric and semantic features. The additive spline-based structure of the Kolmogorov-Arnold network enables direct visualisation of each feature's influence. This produces smooth and transparent functional mappings that reveal when the model's confidence is well supported and when it is unreliable. Furthermore, a bootstrapped language-image (BLIP) foundation model generates descriptive captions of each scene. This tool enables a lightweight multimodal interface without affecting the interpretability layer. Experiments on both Common Objects in Context (COCO), and images from the University of Bath campus demonstrate that the framework accurately identifies low-trust predictions under blur, occlusion, or low texture. This provides actionable insights for acceptance, review, or downstream risk mitigation. The resulting system delivers interpretable object detection with trustworthy confidence estimates. It offers a powerful tool for transparent and practical perception component for autonomous and multimodal artificial intelligence applications.

cs.CV↗

Transformer self-attention encoder-decoder with multimodal deep learning for response time series forecasting and digital twin support in wind structural health monitoring

The wind-induced structural response forecasting capabilities of a novel transformer methodology are examined here. The model also provides a digital twin component for bridge structural health monitoring. Firstly, the approach uses the temporal characteristics of the system to train a forecasting model. Secondly, the vibration predictions are compared to the measured ones to detect large deviations. Finally, the identified cases are used as an early-warning indicator of structural change. The artificial intelligence-based model outperforms approaches for response forecasting as no assumption on wind stationarity or on structural normal vibration behavior is needed. Specifically, wind-excited dynamic behavior suffers from uncertainty related to obtaining poor predictions when the environmental or traffic conditions change. This results in a hard distinction of what constitutes normal vibration behavior. To this end, a framework is rigorously examined on real-world measurements from the Hardanger Bridge monitored by the Norwegian University of Science and Technology. The approach captures accurate structural behavior in realistic conditions, and with respect to the changes in the system excitation. The results, importantly, highlight the potential of transformer-based digital twin components to serve as next-generation tools for resilient infrastructure management, continuous learning, and adaptive monitoring over the system's lifecycle with respect to temporal characteristics.

cs.LG↗

An Information-Theoretic Method for Dynamic System Identification With Output-Only Damping Estimation

The system identification capabilities of a novel information-theoretic method are examined here. Specifically, this work uses information-theoretic metrics and vibration-based measurements to enhance damping estimation accuracy in mechanical systems. The method refers to a key limitation in system identification, signal processing, monitoring, and alert systems. These systems integrate various components, including sensors, data acquisition devices, and alert mechanisms. They are designed to operate in an environment to calculate key parameters such as peak accelerations and duration of high acceleration values. The current operational modal identification methods, though, suffer from limitations related to obtaining poor damping estimates due to their empirical nature. This has a significant impact on alert warning systems. This occurs when their duration is misestimated; specifically, when using the vibration amplitudes as an indicator of danger alerts for monitoring systems in damage or anomaly detection scenarios. To this end, approaches based on the Shannon entropy and the Kullback-Leibler divergence concept are proposed. The primary objective is to monitor the vibration levels in near real-time and provide immediate alerts when predefined thresholds are exceeded. In considering the proposed approach, both new real-world data from the multi-axis simulation table at the University of Bath, as well as the benchmark International Association for Structural Control-American Society of Civil Engineers (IASC-ASCE) structural health monitoring problem are considered. Importantly, the approach is shown to select the optimal model, which accurately captures the correct alert duration, providing a powerful tool for system identification and monitoring.

eess.SP↗