NAS provides several Python environments including the system default, /usr/bin/python3 (currently version 3.6.8), and modules such as python3/3.9.5, and so on. These environments provide a variety of Python packages and versions. However, if you need a Python package or version that is not available in these environments, we recommend that you install it in your own directory.

Note: NASA no longer allows Python 2 in agency environments. Only the Python 3 distribution is available on NAS systems. For more information about how we maintain software, see NAS Software Policy.

To install a Python package or version, we recommend using the virtualenv tool, which provides these benefits:

• It creates an isolated Python virtual environment. When activated, you can install and use packages through this environment, instead of through /usr/bin/python or the NAS-provided Python modules. 
• You can create separate virtual environments for separate projects that may need different versions of the same package.

Complete these steps to check whether a package is available on NAS systems, create a virtual environment, install a package from the Python Package Index (PyPI), and use the package in this virtual environment:

1. Load the latest Python module (currently python3/3.9.5), which already has virtualenv, pip and many other Python packages installed:

   % module load python3/3.9.5
   

2. View the packages and versions that are available under python3/3.9.5:

   % pip list

   In the output, look for a package with the version you want. If it is not available, or if you want to install your own private copy, continue with the steps below.

3. Create a virtual environment with a new directory name, such as ~/myenv. The virtualenv tool will create the directory for you.

   If you want this virtual environment to be able to access packages already available in the global site-packages (in this example, /nasa/pkgsrc/toss3/2021Q2/lib/python3.9/site-packages), so that you do not have to install everything from scratch, run:

   % virtualenv --system-site-packages ~/myenv

   Otherwise, run:

   % virtualenv ~/myenv

   At the end of this step, you should see the python (which is linked to python3.9), pip (which is same as pip3 or pip3.9), wheel, and multiple activate scripts under ~/myenv/bin.

4. Activate the virtual environment:

   % source ~/myenv/bin/activate (for bash)
   [myenv]%
   
   or
   
   % source ~/myenv/bin/activate.csh (for csh)
   [myenv]%

   Your prompt should now include the virtual environment [myenv].

5. Use the pip tool from this environment to install or upgrade a package (for example, SymPy) under ~/myenv/lib/python3.9/site-packages, and run the pip list command to check whether the installation is completed properly. (Note that SymPy depends on mpmath, so make sure that mpmath is installed before installing sympy.)

   [myenv]% which pip
   /u/username/myenv/bin/pip
   [myenv]% pip install --upgrade sympy
   [myenv]% pip list

6. When you finish using this isolated environment, run the deactivate command to exit. After you exit, your prompt should no longer include [myenv].

   [myenv]% deactivate  
   % 

To install another package or to use a package in this isolated virtual environment at a later time, you do not need to load the python/3.9.5 module again. Simply activate the virtual environment (step 4, above), then use the packages or install additional packages. For example:

% source ~/myenv/bin/activate (for bash)

or

% source ~/myenv/bin/activate.csh (for csh)

[myenv]% pip install --upgrade pendulum
[myenv]% pip list
[myenv]% which python
/u/username/myenv/bin/python
[myenv]% python
>>> import sympy as s
>>> import pendulum as p