arXiv subjects

Youping Chen

Publications and source records attributed to Youping Chen.

14 recordsLinked to original sources

arXiv2025

Machine-Learned Interatomic Potential for Predictive Simulation of MoS2 Epitaxy

A machine-learned interatomic potential (MLIP) for multilayer MoS2 was developed using the ultra-fast force field (UF3) framework. The UF3 MLIP reproduces key properties in strong agreement with DFT including lattice constants, interlayer binding energies, and phase-stability. Furthermore, the potential reasonably captures the phonon spectra and the highly anisotropic elastic tensor across monolayer (1H) and bulk (2H, 3R) MoS2 phases. Critically, defect and edge formation energies are captured with high fidelity, exhibiting a strong correlation with DFT (R^2 = 0.91) across ten defective monolayers and reproducing the relative difference between the free energies of zigzag and armchair edges within 5% of DFT. Non-equilibrium molecular dynamics simulations reveal layered homoepitaxial growth consistent with experimental observations, demonstrating the formation of van der Waals gaps between successive epilayers and triangular domains bounded by zigzag edges. The robust UF3 MLIP, which is only ~2X slower than the fastest empirical potentials, enables large-scale atomistic simulations of MoS2 epitaxial growth.

cond-mat.mtrl-sci↗

arXiv2025

A Finite Element Method for Simulation of Coupled Dynamics of Dislocations and Fracture

This work presents a finite element method for simulating dynamic processes that involve the coupled evolution of dislocation motion and crack propagation. The method numerically solves the Concurrent Atomistic-Continuum (CAC) formulation of the conservation of linear momentum. A crystalline material is discretized at the unit-cell level using 6-node prism elements whose geometry allows dislocations and cracks to nucleate and propagate along element facets. Nanoscale simulations of single-crystal Cu, Fe, and Si demonstrate the initiation and propagation of dislocations and cracks, and these results are reproduced by the finite element method in excellent agreement with fully atomistic molecular dynamics simulations. Mesoscale simulations of single-crystal Cu further demonstrate the ability of the method to capture size-dependent brittle and ductile behavior. Under plane-strain conditions the Cu model fractures in a brittle manner, while a fully three-dimensional model exhibits curved and intersecting dislocations that blunt the crack tip and prevent crack propagation, resulting in ductile behavior. The accuracy, efficiency, and applicability of the method are discussed.

cond-mat.mtrl-sci↗

arXiv2025

Machine-learning interatomic potential for AlN for epitaxial simulation

A machine learned interatomic potential for AlN was developed using the ultra-fast force field (UF3) methodology. A strong agreement with density functional theory calculations in predicting key structural and mechanical properties, including lattice constants, elastic constants, cohesive energy, and surface energies has been demonstrated. The potential was also shown to accurately reproduce the experimentally observed atomic core structure of edge dislocations. Most significantly, it reproduced the experimentally observed wurtzite crystal structure in the overlayer during homoepitaxial growth of AlN on wurtzite AlN, something that prior potentials failed to achieve. Additionally, the potential reproduced the experimentally observed layer-by-layer growth mode in the epilayer. The combination of accuracy, transferability, and computational speed afforded by the UF3 framework thus makes large-scale, atomistic simulations of epitaxial growth of AlN feasible.

cond-mat.mtrl-sci↗

arXiv2025

Effect of Misfit and Threading Dislocations on Surface Energies of PbTe-PbSe Interfaces

This work quantifies the effect of misfit and threading dislocations on the surface energies of PbTe-PbSe interfaces, with the defect structures of the interfaces being obtained from atomistic and multiscale simulations of their manufacturing processes. Simulation results show that direct bonding produces semi-coherent interfaces with two-dimensional misfit dislocation networks, while heteroepitaxial processes produce complex three-dimensional dislocation structures with both misfit and threading dislocations. Surface energies at these interfaces were determined by computing the interaction energies across these interfaces. Compared with coherent interfaces, directly bonded interfaces exhibit up to ~23% lower surface energy, while the surface energies of epitaxially grown interfaces can be nearly 50% lower. The results demonstrate the significant effects of dislocations on interfacial energy.

cond-mat.mtrl-sci↗

arXiv2024

A Concise but High-performing Network for Image Guided Depth Completion in Autonomous Driving

Depth completion is a crucial task in autonomous driving, aiming to convert a sparse depth map into a dense depth prediction. Due to its potentially rich semantic information, RGB image is commonly fused to enhance the completion effect. Image-guided depth completion involves three key challenges: 1) how to effectively fuse the two modalities; 2) how to better recover depth information; and 3) how to achieve real-time prediction for practical autonomous driving. To solve the above problems, we propose a concise but effective network, named CENet, to achieve high-performance depth completion with a simple and elegant structure. Firstly, we use a fast guidance module to fuse the two sensor features, utilizing abundant auxiliary features extracted from the color space. Unlike other commonly used complicated guidance modules, our approach is intuitive and low-cost. In addition, we find and analyze the optimization inconsistency problem for observed and unobserved positions, and a decoupled depth prediction head is proposed to alleviate the issue. The proposed decoupled head can better output the depth of valid and invalid positions with very few extra inference time. Based on the simple structure of dual-encoder and single-decoder, our CENet can achieve superior balance between accuracy and efficiency. In the KITTI depth completion benchmark, our CENet attains competitive performance and inference speed compared with the state-of-the-art methods. To validate the generalization of our method, we also evaluate on indoor NYUv2 dataset, and our CENet still achieve impressive results. The code of this work will be available at https://github.com/lmomoy/CHNet.

cs.CV↗

arXiv2023

Unifying temperature definition in atomistic and field representations of conservation laws

This work presents a formalism to derive field quantities and conservation laws from the atomistic using the theory of distributions as the mathematical tool. By defining temperature as a derived quantity as that in molecular kinetic theory and atomistic simulations, a field representation of the conservation law of linear momentum is derived and expressed in terms of temperature field, leading to a unified atomistic and continuum description of temperature and a new conservation equation of linear momentum that, supplemented by an interatomic potential, completely governs thermal and mechanical processes across scales from the atomic to the continuum. The conservation equation can be used to solve atomistic trajectories for systems at finite temperatures, as well as the evolution of field quantities in space and time, with atomic or multiscale resolution. Four sets of numerical examples are presented to demonstrate the efficacy of the formulation in capturing the effect of temperature or thermal fluctuations, including phonon density of states, thermally activated dislocation motion, dislocation formation during epitaxial processes, and attenuation of longitudinal acoustic waves as a result of their interaction with thermal phonons.

cond-mat.stat-mech↗

arXiv2022

Self-supervised Sequential Information Bottleneck for Robust Exploration in Deep Reinforcement Learning

Effective exploration is critical for reinforcement learning agents in environments with sparse rewards or high-dimensional state-action spaces. Recent works based on state-visitation counts, curiosity and entropy-maximization generate intrinsic reward signals to motivate the agent to visit novel states for exploration. However, the agent can get distracted by perturbations to sensor inputs that contain novel but task-irrelevant information, e.g. due to sensor noise or changing background. In this work, we introduce the sequential information bottleneck objective for learning compressed and temporally coherent representations by modelling and compressing sequential predictive information in time-series observations. For efficient exploration in noisy environments, we further construct intrinsic rewards that capture task-relevant state novelty based on the learned representations. We derive a variational upper bound of our sequential information bottleneck objective for practical optimization and provide an information-theoretic interpretation of the derived upper bound. Our experiments on a set of challenging image-based simulated control tasks show that our method achieves better sample efficiency, and robustness to both white noise and natural video backgrounds compared to state-of-art methods based on curiosity, entropy maximization and information-gain.

cs.LG↗

arXiv2022

Integrating Contrastive Learning with Dynamic Models for Reinforcement Learning from Images

Recent methods for reinforcement learning from images use auxiliary tasks to learn image features that are used by the agent's policy or Q-function. In particular, methods based on contrastive learning that induce linearity of the latent dynamics or invariance to data augmentation have been shown to greatly improve the sample efficiency of the reinforcement learning algorithm and the generalizability of the learned embedding. We further argue, that explicitly improving Markovianity of the learned embedding is desirable and propose a self-supervised representation learning method which integrates contrastive learning with dynamic models to synergistically combine these three objectives: (1) We maximize the InfoNCE bound on the mutual information between the state- and action-embedding and the embedding of the next state to induce a linearly predictive embedding without explicitly learning a linear transition model, (2) we further improve Markovianity of the learned embedding by explicitly learning a non-linear transition model using regression, and (3) we maximize the mutual information between the two nonlinear predictions of the next embeddings based on the current action and two independent augmentations of the current state, which naturally induces transformation invariance not only for the state embedding, but also for the nonlinear transition model. Experimental evaluation on the Deepmind control suite shows that our proposed method achieves higher sample efficiency and better generalization than state-of-art methods based on contrastive learning or reconstruction.

cs.LG↗

arXiv2021

LF-YOLO: A Lighter and Faster YOLO for Weld Defect Detection of X-ray Image

X-ray image plays an important role in manufacturing industry for quality assurance, because it can reflect the internal condition of weld region. However, the shape and scale of different defect types vary greatly, which makes it challenging for model to detect weld defects. In this paper, we propose a weld defect detection method based on convolution neural network, namely Lighter and Faster YOLO (LF-YOLO). In particularly, a reinforced multiscale feature (RMF) module is designed to implement both parameter-based and parameter-free multi-scale information extracting operation. RMF enables the extracted feature map capable to represent more plentiful information, which is achieved by superior hierarchical fusion structure. To improve the performance of detection network, we propose an efficient feature extraction (EFE) module. EFE processes input data with extremely low consumption, and improves the practicability of whole network in actual industry. Experimental results show that our weld defect detection network achieves satisfactory balance between performance and consumption, and reaches 92.9 mean average precision mAP50 with 61.5 frames per second (FPS). To further prove the ability of our method, we test it on public dataset MS COCO, and the results show that our LF-YOLO has a outstanding versatility detection performance. The code is available at https://github.com/lmomoy/LF-YOLO.

cs.CV↗

arXiv2021

A Lightweight and Accurate Recognition Framework for Signs of X-ray Weld Images

X-ray images are commonly used to ensure the security of devices in quality inspection industry. The recognition of signs printed on X-ray weld images plays an essential role in digital traceability system of manufacturing industry. However, the scales of objects vary different greatly in weld images, and it hinders us to achieve satisfactory recognition. In this paper, we propose a signs recognition framework based on convolutional neural networks (CNNs) for weld images. The proposed framework firstly contains a shallow classification network for correcting the pose of images. Moreover, we present a novel spatial and channel enhancement (SCE) module to address the above scale problem. This module can integrate multi-scale features and adaptively assign weights for each feature source. Based on SCE module, a narrow network is designed for final weld information recognition. To enhance the practicability of our framework, we carefully design the architecture of framework with a few parameters and computations. Experimental results show that our framework achieves 99.7% accuracy with 1.1 giga floating-point of operations (GFLOPs) on classification stage, and 90.0 mean average precision (mAP) with 176.1 frames per second (FPS) on recognition stage.

cs.CV↗

arXiv2017

Recent Progress in the Concurrent Atomistic-Continuum (CAC) Method and its Application in Phonon Transport

This work presents recent the progress in the development of the Concurrent Atomistic-Continuum (CAC) method for coarse-grained space- and time-resolved atomistic simulations of phonon transport. Application examples, including heat pulses propagating across grain boundaries and phase interfaces, as well as the interactions between phonons and moving dislocations, are provided to demonstrate the capabilities of CAC. The simulation results provide visual evidence and reveal the underlying physics of a variety of phenomena including: phonon focusing, wave interference, dislocation drag, interfacial Kapitza resistance caused by quasi-ballistic phonon transport, etc. A new method to quantify fluxes in transient transport processes is also introduced.

cond-mat.mes-hall↗

arXiv2017

Ballistic-diffusive Phonon Heat Transport across Grain Boundaries

The propagation of a heat pulse in a single crystal and across grain boundaries (GBs) is simulated using a concurrent atomistic-continuum method furnished with a coherent phonon pulse model. With a heat pulse constructed based on a Bose-Einstein distribution of phonons, this work has reproduced the phenomenon of phonon focusing in single and polycrystalline materials. Simulation results provide visual evidence that the propagation of a heat pulse in crystalline solids with or without GBs is partially ballistic and partially diffusive, i.e., there is a co-existence of ballistic and diffusive thermal transport, with the long-wavelength phonons traveling ballistically while the short-wavelength phonons scatter with each other and travel diffusively. To gain a quantitative understanding of GB thermal resistance, the kinetic energy transmitted across GBs is monitored on the fly and the time-dependent energy transmission for each specimen is measured; the contributions of coherent and incoherent phonon transport to the energy transmission are estimated. Simulation results reveal that the presence of GBs modifies the nature of thermal transport, with the coherent long-wavelength phonons dominating the heat conduction in materials with GBs. In addition, it is found the phonon-GB interaction can result in the reconstruction of the GBs.

cond-mat.mes-hall↗

arXiv2017

Passing Waves from Atomistic to Continuum

Progress in the development of coupled atomistic-continuum methods for simulations of critical dynamic material behavior has been hampered by a spurious wave reflection problem at the atomistic-continuum interface. This problem is mainly caused by the difference in material descriptions between the atomistic and continuum models, which results in a mismatch in phonon dispersion relations. In this work, we introduce a new method based on atomistic dynamics of lattice coupled with a concurrent atomistic-continuum method to enable a full phonon representation in the continuum description. This then permits the passage of short-wavelength, high-frequency phonon waves from the atomistic to continuum regions. The benchmark examples presented in this work demonstrate that the new scheme enables the passage of all allowable phonons through the atomistic-continuum interface; it also preserves the wave coherency and energy conservation after phonons transport across multiple atomistic-continuum interfaces. This work is the first step towards developing a concurrent atomistic-continuum simulation tool for non-equilibrium phonon-mediated thermal transport in materials with microstructural complexity.

cond-mat.mtrl-sci↗

arXiv2016

The origin of the distinction between microscopic formulas for stress and Cauchy stress

Stress is calculated routinely in atomistic simulations. The widely used microscopic stress formulas derived from classical or quantum mechanics, however, are distinct from the concept of Cauchy stress, i.e., the true mechanical tress. This work examines various atomistic stress formulations and their inconsistencies. Using standard mathematic theorems and the law of mechanics, we show that Cauchy stress results unambiguously from the definition of internal force density, thereby removing the long-standing confusion about the atomistic basis of the fundamental property of Cauchy stress, and leading to a new atomistic formula for stress that has clear physical meaning and well-defined values, satisfies conservation law, and is fully consistent with the concept of Cauchy stress.

cond-mat.stat-mech↗