============
Installation
============

Download
--------

The package can be downloaded from `Github repository
<https://github.com/Gao-Group/atomdnn>`_.


Enviromental variable
---------------------

Once you have a local copy of atomdnn, include the top directory to your
$PYTHONPATH enviromental variable:

.. code-block:: sh

   export PYTHONPATH=path_to_pacakge/atomdnn:$PYTHONPATH


Required packages
-----------------


- `ASE <https://wiki.fysik.dtu.dk/ase/index.html>`_ is used for file and data manipulation.

- `Tensorflow 2 <https://www.tensorflow.org/install/pip>`_ is used
  to build and train the potential. We tested both Version 2.5 and 2.6.

- `Jupyter Notebook <https://jupyter.org/>`_ is needed for interactive
  running the code.