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.
An open-source application for the first-principles calculation by the all-electron calculation method based on plane wave bases. This application can perform electronic state calculation by the density functional theory (DFT). This appication also supports the LDA+U method, treatment of spin-orbit interaction and noncolinear magnetism, the GW approtimation, and downfolding by the constraint RPA method.
Automatic generation tool for codes of tensor contraction. This tool can automatically generate codes of an optimal computing sequence for construction of a single tensor from a tensor network composed of tensors. Netcon algorithm proposed by Pfeifer et al. is used, and it is possible to search optimal solution quickly. Generated codes are available in Numpy and mptensor in Python.
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.
An open-source application for molecular dynamics simulation of biomolecules, especially designed for massively parallel computing. This package enables us to perform efficient parallel calculation on parallel computers ranging from 100 to 20,000 cores. For preparation of calculation and analysis of orbits, it uses visualization software VMD. It supports file formats compatible with other applications such as AMBER and CHARMM, and can be used on various computing environments.
A collection of C++ interfaces for simulation of mesoscale properties based on grid data. By using provided header files, one can easily construct programs for simulation of various phenomena such as solidification, crystal growth, and spinodal decomposition, based on a Monte Carlo method, cellar automaton, and a phase-field method. This interface supports parallel computing by MPI, and also provides converters of output files for visualization software such as ParaView.
An application for first-principles calculation by the joint-DFT method based on a plane-wave basis. By implementation of the joint-DFT method, this application realizes a good convergence for electronic state calculation of molecules in liquid, particular for charged systems. This application is written by C++11, and supports GPU calculation by CUDA. This application also supports diffusive Monte Carlo simulation in cooperation with CASINO.
An open-source numerical library for machine learning. Using other machine learning numerical libraries (TensorFlow, CNTK, Theano, etc.), users can construct neural networks by relatively short codes. Since a number of methods in machine learning and deep learning are implemented, users can try state-of-the-art methods easily. This package is written by Python.
An open-source application for simulation based on the density-matrix renormalization group (DMRG). This application can perform high-speed calculation of low-dimensional quantum systems with high accuracy. It implements generic programming techniques in the C++ language, and can easily extend simulation to new models and geometries. It is developed putting emphasis on user-friendly interfaces and low dependences on environments.
ComDMFT is a massively parallel computational package to study the electronic structure of correlated-electron systems. Users can perform a parameter-free method based on ab initio linearized quasiparticle self-consistent GW (LQSGW) and dynamical mean field theory (DMFT).