Open source software for building and using machine learning potentials based on E(3)-equivariant graph neural networks, which can be trained on output files of simulation codes that can be read by ASE. Molecular dynamics calculations with LAMMPS can be performed using the trained potentials.
An application for structure prediction based on the evolutionary algorithm. From an input of the atomic position in a unit cell and possible elements at each atomic position, this application predicts the stable structure and composition from the first-principles calculation and molecular dynamics in combination with the evolutionary algorithm. This application is written in Python, and uses Quantum ESPRESSO and GULP as an external program.
PHYSBO is a Python library for researchers mainly in the materials science field to perform fast and scalable Bayesian optimization based on COMBO (Common Bayesian Optimization). Users can search the candidate with the largest objective function value from candidates listed in advance by using machine learning prediction. PHYSBO can handle a larger amount of data compared with standard implementations such as scikit-learn.
2DMAT is a framework for applying a search algorithm to a direct problem solver to find the optimal solution. In version 1.0, for solving a direct problem, 2DMAT offers the wrapper of the solver for the total-reflection high-energy positron diffraction (TRHEPD) experiment. As algorithms, it offers the Nelder-Mead method, the grid search method, the Bayesian optimization method, and the replica exchange Monte Carlo method. Users can define original direct problem solvers or the search algorithms.
Python library for the design, simulation, and optimization of continuous-variable quantum optical circuits. It has high-level functions for solving problems including graph and network optimization, machine learning, and chemistry, and can perform training and optimization of quantum programs using the TensorFlow backend.
Software package that implements Behler-Parinello type neural network potential. The package provides tools for training and evaluating potentials based on given structure-energy data. It also provides an interface with LAMMPS for performing molecular dynamics calculations.
Software package to implement Behler-Parinello neural network potentials. Potentials can be trained from structure-energy/ interatomic forces/stress data, and molecular dynamics calculations using LAMMPS can also be performed using learned potentials. A prediction uncertainty measure can also be calculated simultaneously.
Python/C++ based software package that employs deep learning techniques for construction of interatomic potentials. It implements the Deep Potential, which defines atomic environment descriptors with respect to a local reference frame. The output of many first-principles and molecular dynamics applications can be used as training data, and the trained potentials can be used for molecular dynamics calculations using LAMMPS and path integral molecular dynamics calculations using i-PI.
A collection of software tools for molecular dynamics calculations. Various interatomic potentials and tight binding models are implemented, and numerous external applications can be invoked. It also supports training and evaluation of GAP (Gaussian Approximation Potential), which is a form of machine learning potential.
aenet is software for atomic interaction potentials using artificial neural networks. Users can construct neural network potentials using structures of target materials and their energies obtained from first principle calculations. The generated potentials can be used to molecular dynamics or Monte Carlo simulations.