MyBigDFT

Code style: black

MyBigDFT provides a python wrapper over the BigDFT code. It can be considered as a sandbox for the PyBigDFT package. It might be used instead of the PyBigDFT package, even though both packages do not provide the same API and functionalities.

It is currently supported for the following python versions: 2.7, 3.6, 3.7.

Credit to mmoriniere for the original code.

Installation

BigDFT 1.8.2 must be installed in a given build directory (noted BUILDDIR). Then, run the following commands (where BUILDDIR is the appropriate build directory):

  • git clone this package

  • cd BUILDDIR/install/bin

  • source bigdftvars.sh’

  • unset PYTHONPATH

  • cd -

  • cd MyBigDFT

  • pip install .

This will copy the MyBigDFT sources, then set the environment variables needed by MyBigDFT before installing it.

To use pseudopotential files (optional)

To be able to use pseudopotential files in the calculations, the following line must be added to the ~/.bashrc file:

export PSEUDODIR="/path/to/the/directory/for/pseudos/files/"

The desired files must be named psppar. with the extension being the corresponding element atomic symbol.

To use SchnetPack models (optional)

To use SchnetPack models in calculations, the following line may be added to the ~/.bashrc file:

export MODELDIR="/path/to/the/directory/for/trained/models/"

The paths of models can then be expressed as relative paths to MODELDIRinstead of absolute paths.

Note: To use the machine learning related classes of this package, the linked fork of SchnetPack must be installed. The original repository will not work.

Extra steps you might need to follow

  • Make sure that the BIGDFT_ROOT environment variables does not lead to multiple directories. If so, choose only one of them (where BUILDDIR is the path to the actual BigDFT build directory you want to use):

  • export BIGDFT_ROOT=’BUILDDIR/install/bin’

  • If your default python version is python3, modify the first line of the bigdft-tool executable so that it forces the use of python2 (i.e., set it to ‘#!/usr/bin/env python2’). This executable will be found in the same directory as the bigdft one (i.e. BUILDDIR/install/bin)

Documentation

The documentation of MyBigDFT can be found here.

It can also be built locally by running the following commands (where MYBIGDFT_SOURCES is the directory where this README.md file is located):

  • cd MYBIGDFT_SOURCES

  • pip install -e .[doc]

  • cd doc

  • make html

The -e option is optional (use it if you wish to edit some source files locally), while the [doc] directive might install extra packages required for the doc to build correctly.

Tests

To be able to run all tests, make sure to install the package by running the following commands (where MYBIGDFT_SOURCES is the directory where this README.md file is located):

  • cd MYBIGDFT_SOURCES

  • pip install -e .[test]

  • pytest

Again, the -e option is optional, use it if you wish to edit the source files. The [test] directive might install extra packages required for the tests to run. The pytest command should run all the tests (including the doctests).

Notes

  • If all the python versions supported by MyBigDFT are installed, you can even run the ‘tox’ command to launch the tests for each of these versions instead of ‘pytest’.

Indices and tables