Jupyter Notebook

Jupyter notebook is available via the Open OnDemand graphical console. If you use Conda to manage your environments you may add each of them to the Jupyter web interface with the following commands.

Adding Conda environments to OOD Jupyter Notebooks.

Change into each environment from the terminal, install ipykernel and register the kernel. Note the unique names given that match the conda environment names.

(base) [jflilley@login1 jupyter]$ conda env list

# conda environments:

base * /home/jflilley/miniconda3

my-new-env /home/jflilley/miniconda3/envs/my-new-env

conda activate base

conda install ipykernel

python -m ipykernel install --user --name base --display-name "Python (base)"

conda activate my-new-env

conda install ipykernel

python -m ipykernel install --user --name my-new-env --display-name "Python (my-new-env)"

Kill and relaunch your Jupyter via the Open OnDemand interface. Click "New" and you will be able to launch a new kernel based on the Conda environment of your choice.

Adding Julia language environments to OOD Jupyter Notebooks.

First load the Julia module and install the Julia Kernel

module load julia/1.4.2

julia> using Pkg

julia> Pkg.add("IJulia")

Then navigate to the OOD portal (VPN if off campus) > login > click Interactive Apps > Jupyter Notebook > select cores/time/gpus > Launch > Connect to Jupyter.

Once launched you should be able to select the Jupyter 1.4.2 kernel by clicking New > Julia 1.4.2

Secondary method

|Pkg.instantiate()|the project. |julia --project -e 'using IJulia;
installkernel("Julia VizClima", "--project=$(pwd())")'|

Changing default startup folder

You can change the default Jupyter startup folder by adding the following to the Jupyter config ~/.jupyter/jupyter_notebook_config.json

c.NotebookApp.notebook_dir="<directory>"

Another simple method that both retains the existing ~/ startup directory while allowing one to switch to directories outside of home is to create a symlink in ~/ that points to the group dir. At this point you will be able to click on the symlinked directory in Jupyter to navigate to the desired location.

For example:

ln -s /central/groups/<group> ~/my-group-dir