This is just a quick “how-to” use virtualenv. Full documentation available here.

You can also check out the how-to written by Andrej Krevl.

#cd to a folder where you want to store the new virtualenv (in this example we are using python2.7)
virtualenv -p /usr/bin/python2.7 newenv

#or if you want to use python3.7
virtualenv -p /usr/local/bin/python3.7 newenv

#activate virtualenv enviroment that we created before
source newenv/bin/activate

#just a test that you see which pip executalbe is in use
(newenv) <csid>@madmax:/dfs/scratch0/<csid>$ which pip
/dfs/scratch0/<csid>/newenv/bin/pip
#install a new package
(newenv) <csid>@madmax:/dfs/scratch0/<csid>$ pip install numpy

#check which python is in use while using virtualenv
(newenv) <csid>@madmax:/dfs/scratch0/<csid>$ which python
/dfs/scratch0/<csid>/newenv/bin/python

#check where is the package that we installed before located
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy
>>> numpy.__file__
'/dfs/scratch0/<csid>/newenv/lib/python3.5/site-packages/numpy/__init__.py'

#deactivate virtualenv enviroment
deactivate
#check wich python is in use when the virtualenv is deactivated
<csid>@madmax:/dfs/scratch0/<csid>$ which python
/usr/bin/python

To use Jupyter Notebook with a specific virutalenv you need to do the following:

#active the virtualenv you want to use
source newenv/bin/activate

# install ipykernel
pip install ipykernel

# create a new kernel
python -m ipykernel install --user --name=newenv
# output: Installed kernelspec newenv in /afs/cs.stanford.edu/u/user/.local/share/jupyter/kernels/newenv

Listing all the available kernels:

jupyter kernelspec list

#sample output
#Available kernels:
#  newenv.         /afs/cs.stanford.edu/u/user/.local/share/jupyter/kernels/newenv
#  python2         /usr/local/share/jupyter/kernels/python2
#  python3         /usr/local/share/jupyter/kernels/python3

If you remove the virtualenv and you don't need the kernel anymore you can remove it with:

jupyter kernelspec uninstall newenv