Running Jupyter Remotely from OLCF

To run Jupyter remotely at OLCF, you can use the OLCF JupyterHub, as described in the OLCF Jupyter documentation. In order to use extra python libraries (e.g. yt), it’s best to first install these as part of a conda environment, which can then be loaded when you create your Jupyter notebook.

Creating a conda environment

If you wish to install extra python libraries (e.g. yt), then they can be installed using a conda environment. To do this, make sure you have the correct python module loaded on Summit (e.g. module load python/3.7.0-anaconda3-5.3.0), and create a new conda environment with the modules you require installed. At this point, it’s also a good idea to make sure this environment has the libraries ipykernel and nb_conda_kernels:

conda create -n my_env -y ipykernel nb_conda_kernels
conda install -n my_env -c conda-forge yt
conda install -n my_env jupyterlab

Note

To use the conda environment for python subsequently, you do not need to do module load python/3.7.0-anaconda3-5.3.0

Note

yt is distributed via conda-forge, you can add this channel to your conda search via:

conda config --add channels conda-forge