Pymatflow

Pymatflow is a workflow manager and highthroughput tool for DFT modeling of materials. It can automatically preparing input files for popular open source DFT programs like Quantum ESPRESSO, CP2K, Abinit, and SIESTA. And the development for the post processing is on the way.

Install

To install the current stable release, just use pip to install Pymatflow from PyPI:

$ pip3 install pymatflow

to install the current develpment version, you can clone the project and manually install it after chdir to the project root directory. Before that make sure you have scikit-build installed:

$ pip3 install scikit-build
$ git clone https://gitlab.com/deqitang/pymatflow.git
$ cd pymatflow
$ python setup.py install

or directly:

$ pip3 install scikit-build
$ pip3 install git+https://gitlab.com/deqitang/pymatflow.git

Usage:

There are two ways to use pymatflow:

use Pymatflow as a python library and build your own calculation.
use scripts provided by Pymatflow directly from command line.

The document for use Pymatflow as a python library is not ready at present! So we provide an example of using the script provided byPymatflow to show you how Pymatflow works. We take the scf calculation of one H2O molecule using Quantum ESPRESSO calculator as an example. First, preparing the following structure file h2o.xyz(xyz format, with the second line specifying the cell parameter):

3
cell: 10.0   0.0   0.0  |   0.0   10.0   0.0000000  |   0.0000000   0.0000000   10.0
H    0.000000    2.599887    5.074274
H    0.000000    3.543756    3.781971
O    0.000000    2.585050    4.077973

make sure you put the xyz structure file and the needed pseudopotential file for H and O in the same directory and in that dir, do:

$ mflow qe --xyz h2o.xyz -d h2o-scf --ecutwfc 60 --conv-thr 1.0e-5 --kpoints-option gamma --runopt=genrun --auto=0

The above command assumes you want to run the pw.x directly on your computer, and all the calculation happens inside directoryh2o-scf. Usually, we do DFT calculations on powerful cluster, and pymatflow can also automatically submit the job. However, only pbs job manager and modified version of slurm on TianHe II is supported now! For doing the calculation on the server, you should run the command on your server, and modifying the value of --runopt and --auto as follows:

$ mflow qe --xyz h2o.xyz -d h2o-scf --ecutwfc 60 --kpoints-option gamma --runopt=gen --auto=3 --servere=pbs

Which is actually the default setting for value of --runopt and --auto. This way pymatgen will automatically generate the input file and submit the job the pbs job manager. The support for other type of jobnamager is actually not hard to implement. If you have the need, you can get in touch with me through pymatflow@163.com.

Goals

powerful and stable calculator supports for popular Ab initio programs
fully featured results analysis system
user friendly structure builder
automated workflow for high throughput calculation[including database handling]
- automatical choice of appropriate parameters
- automatical correction of error during running
statistical analysis for properties, basic machine learning module
support for system visulization by third-party molecular visualizer like Avogadro, Vipster, paraview, vtk, vmd, xcrysden, vesta, blender
support for third-party structure generation tools, like mbuild, atomsk, msmbuilder,
using spgroup to get space group of crystals.

Documents & Resources

Official Document(ongoing!)

License