Jupyter notebook on Vorna

IT4sci Jupyter notebook interface

University of Helsinki IT4SCI has introduced a (beta-version) Jupyter notebook interface. It can be found at https://hub.cs.helsinki.fi and can be accessed only from the internal network or using a VPN.

1. Log in with your UH credentials
2. Launch a compute task (e.g. 1 day Vorna - make sure you select a task with enough memory). Launching and contacting the server can take some time, so be patient.
3. Using the file interface button on the left-hand-side, navigate to your home directory (or to another directory where you have write access to)
4. Click on New Launcher - Notebook - Python3
5. Now you should have a Python notebook with a title something like UntitledXX.ipynb with a command line. Type import pytools as pt and start using Analysator!

NB: A command (cell) is submitted with shift+Enter

No LaTeX support

The Jupyter environment does not currently include a LaTeX installation, so you should either include export PTNOLATEX=1 in your Vorna ~/.bashrc or before importing pytools, write

import os
os.environ['PTNOLATEX'] = '1'

Creating Python virtual environment in jupyterhub

Start your jupyterhub and go to file-> New-> terminal and go to the proj/$USER directory and create your virtual environment "virtualEnvName" by typing

python -m venv virtualEnvName

Activate the virtuelEnvName and start installing required packages and/or packages you want, for example

source /proj/$USER/virtualEnvName/bin/activate
pip install virtualenvwrapper ipykernel matplotlib==3.2.1 scipy 

To add analysator in your python virtual environment, type the following in the jupyterhub terminal (virtualenvwrapper is required here)

source /proj/$USER/virtualEnvName/bin/virtualenvwrapper.sh
add2virtualenv /directory/where/analysator/is/located/analysator

To use the created virtual environment in Jupyterhub, go to the directory where your virtualEnvName is located and type the following in the jupyterhub terminal, a python environment name will appear "MyPythonEnv" in your jupyterhub interface and ready to use

python -m ipykernel install --user --name=virtualEnvName --display-name MyPythonEnv

When you want to use the kernel "MyPythonEnv", you can just click the MyPythonEnv at jupyterhub interface to create jupyter notebook like UntitledXX.ipynb or change .ipynb files you already have by using the kernel menu or by clicking Python 3 on the top right corner on the jupyterhub interface.

If somehow you want to remove the python environment completely, first get all the kernels by typing jupyter kernelspec list and remove the "kernel-to-remove" kernel by typing jupyter kernelspec uninstall kernel-to-remove. And finally remove the "virtuelEnvName" directory

Optional: Changing the notebook backend

By default, the Python3 notebook uses an inline backend which creates simple static plots. Alternatively, before importing pytools, you can activate more interactive backends such as %matplotlib widget or %matplotlib ipympl (for re-scalable and rotatable 3D plots)