The aim of virtualenv is to provide your own instance of Python as maintaining different versions of packages is really tricky. This tutorial shows how to have a tailored Python environement for your program : 

You can create a directory for your program "mon_programme" :

  nicolas.greneche@magi1:~$ mkdir mon_programme
  nicolas.greneche@magi1:~$ cd mon_programme

Then, you can create a virtualenv based mon Magi's Python in a subdirectory called "venv" :

  nicolas.greneche@magi1:~$ virtualenv -p /usr/bin/python3.9 venv

You may activate the brand new environement :

  nicolas.greneche@magi1:~$ source venv/bin/activate

You can check that your default Python is from the virtualenv :

  (venv) nicolas.greneche@magi1:~/mon_programme$ which python
  /nfs/homes/nicolas.greneche/mon_programme/venv/bin/python

Finally, you can install your own packages in the virtualenv :

  (venv) nicolas.greneche@magi1:~/mon_programme$ pip install numpy