Installation

There are basically two ways of installing Pychemia, using pip or by cloning the Github repository.

Installing PyChemia with pip

This is probably the easiest way, pip will download the code, check and eventually install dependencies and installing the package on a system-wide place or the home directory. All you have to do is execute this command:

sudo pip install pychemia

or for python 3.x:

sudo pip3 install pychemia

If you are on a machine where you do not have superuser privileges you can install pychemia on your home directory by adding the command ‘–user’:

pip install pychemia --user

or for python 3.x:

pip3 install pychemia --user

This is a simple and usually will work for you. The only situation where it could give you problems is when pip tries to compile and install dependencies. In particular, pychemia requires the following packages and versions:

numpy  >= 1.11.0
scipy  >= 0.17.0
future >= 0.15.2
spglib >= 1.9.4

Now, most Linux distributions will probably come with older versions of numpy and scipy. Even more, Linux clusters usually have a very conservative approach related to packages and old versions of numpy and scipy are probably installed. numpy and scipy are python libraries that use the old and known BLAS and LAPACK libraries for all the usual linear algebra operations. If you encounter problems related with numpy and scipy, check if you have installed blas and lapack on your system.

The python library future is a small package that helps keeping compatibility with python 2 and python 3 on the same source code. Finally spglib is a C-library with python wrappers for computing space groups and related functionality.

We try to keep the dependencies of PyChemia to a very minimum. Some other libraries provides extra functionality that could be necessary for some tasks. Consider install pymongo, nose and matplotlib. You can do that using pip with the command:

sudo pip install pymongo nose matplotlib

Remember that you can use --user if you want to install on your home directory without special privileges. The package ‘pymongo’ offers connectivity with a Mongo database, a must if you plan to use ‘pychemia.population’ subpackage or doing global searches with the ‘pychemia.searcher’ subpackage. Matplotlib is the ‘standard de-facto’ for 2D plots on python. Many of the functionalities on ‘pychemia.visual’ subpackage depends on it. Nose is a python package for executing automatize tests for PyChemia. If you want to use it more information is below.

Installing PyChemia from github

The current stable repository for PyChemia is on Github, you can download the master branch with the command:

git clone https://github.com/MaterialsDiscovery/PyChemia.git

If you get a message such as:

$ git clone https://github.com/MaterialsDiscovery/PyChemia.git
-bash: git: command not found

You need to install git first. On machines from the Debian ‘lineage’ (Ubuntu, Mint, and many others) you can use the command:

sudo apt-get install git

On systems with yum you can use:

sudo yum install git

Now that you have ‘cloned’ the PyChemia repository you have two options. Install the package by using the set of commands:

cd PyChemia
python setup build
python setup install --user

Use python3 for the commands above, if you want to use python 3.x instead.

Another alternative is add path where you downloaded the repository to the variable $PYTHONPATH. You can do that by editing your .bashrc file. Supposing that you execute the git clone command directly on your home directory you can add the path for PyChemia adding this line to your .bashrc:

export PYTHONPATH=$HOME/PyChemia:$PYTHONPATH

If you want the changes on .bashrc take inmediate effect execute:

source $HOME/.bashrc

Importing the library

No matter how you installed PyChemia, you should be able to load the library. You can use the traditional python terminal, for example:

$ python3.5
Python 3.5.1 (default, Mar  2 2016, 03:38:02)
[GCC 4.2.1 Compatible Apple LLVM 7.0.2 (clang-700.1.81)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import pychemia
>>> pychemia.info()
PyChemia
--------

Version: 0.1.2
Path:    /Users/guilleaf/PyChemia/pychemia
Date:    May 13, 2016

Python version=3.5.1 (default, Mar  2 2016, 03:38:02)
[GCC 4.2.1 Compatible Apple LLVM 7.0.2 (clang-700.1.81)]

pymongo Not Found
     numpy     1.11.0   /opt/local/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/numpy
     scipy     0.17.1   /opt/local/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/scipy
    mayavi            Not Found
Scientific            Not Found
matplotlib      1.5.1   /opt/local/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/matplotlib
    future     0.15.2   /opt/local/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/future
      nose      1.3.7   /opt/local/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/nose
  coverage      4.0.3   /opt/local/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/coverage
    spglib      1.9.4   /Users/guilleaf/Library/Python/3.5/lib/python/site-packages/spglib
    pyhull            Not Found
  pymatgen            Not Found
      qmpy            Not Found
       ase            Not Found

The method pcyhemia.info() will inform about the several libraries that PyChemia uses, both mandatory and optional, their versions and path. That could be informative in case of something not working as expected.

Testing with nose

It is always important to test a library, not only from the developer point of view, but also for an user. Nose is a python package that offers a simple command to execute predefined test for a python package and report any errors or inconsistencies from the expected resuts.

Direct the terminal to the place where PyChemia is located. Lets suppose that you have pychemia on /Users/guilleaf/PyChemia, test PyChemia using the command:

cd /Users/guilleaf/PyChemia
nosetests -v

The name of the command could have small variations according to your distribution. On a MacOS using macports the name could be for example nosetests-2.7 or nosetests-3.5 for python 2 and 3 respectively.

nosetests will search for tests on the entire package and subpackages. If everything is fine (and you use -v for verbose output) you will get something like:

...
Example of a simple calc                                     : ... ok
Example of a multiple calc                                   : ... ok

----------------------------------------------------------------------
Ran 38 tests in 5.469s

OK

That is an indication that all tests were successful and eventually you are ready to use the library.