Code and Data

Research

OpenDiS Open Dislocation Simulator - Revolutionizing Dislocation Dynamics Simulations Through Open Source Collaboration. Updated version of ParaDiS.

Interface: Python3; Documentation: OpenDiS webpage.

Reference: Bertin et al., MSMSE, 2019, Arsenlis et al., MSMSE, 2007.

GNN FEM Polycrystal Plasticity Using graph neural networks to predict stress in polycrystal plasticity based on finite element meshes to accelerate simulations.

Dependence: torch, torch_geometric, Torch Scatter.

Reference: Zhai, Computational Mechanics, 2025.

Materials Discovery Benchmark This folder contains a series of optimization methods for materials design and discovery based on Materials Project database with targeted properties.

Dependence: pymatgen, MEGNet, MEALPY.

Reference: Zhai et al., APL Materials, 2024.

PyLAMDO Python-LAMMPS toolkit for Multiscale, Multiphysics, Materials Design & Optimization. PyLAMDO has been used for designing antibiofilm active surfaces and porous materials.

Dependence: LAMMPS 19.X, BayesianOptimization.

Reference: Zhai & Yeo, ACS Biomaterials Sci. & Eng., 2023, Zhai & Yeo, JMBBM, 2023.

BubbleNet BubbleNet is a deep learning framework designed to predict and obtain the physics of micro-bubble dynamics based on the general structure of PINNs.

Dependence: TensorFlow 1.X, COMSOL Multiphysics (data).

Reference: Zhai et al., AIP Advances, 2022.

Education

Density Functional Theory Computation of Materials This is the code & data archive for the course Computational Materials Science, containing Quantum ESPRESSO code for running DFT computation of materials electrical, chemical, mechanical, and various properties. The original code was running on the TACC Stampede2.