MATLAB kernel for Jupyter

[dougiesquire (Dougie Squire, ACCESS-NRI Ocean Modelling Team)]

MathWorks now officially support a MATLAB kernel for Jupyter. Details here:

Official MathWorks MATLAB kernel for Jupyter released

[dougiesquire (Dougie Squire, ACCESS-NRI Ocean Modelling Team)]

This could be a nice stepping-stone for people looking to transition from MATLAB to Python, for example - thanks MathWorks!

[Kim_Reid]

Awesome! Would we be able to get it on the Gadi VDI?

[dougiesquire (Dougie Squire, ACCESS-NRI Ocean Modelling Team)]

You could try creating your own conda environment containing the required packages and running on the NCI ARE JupyterLab App? I don’t have a Matlab license or any experience running Matlab on NCI so I can’t test if this works, but you could give it a try:

1. Follow the instructions under " Creating personal environments" here to create your own conda environment with the required packages. That is, create the ~/.condarc file as described and then run

   module use /g/data/hh5/public/modules
   conda create -n matlab -c conda-forge jupyterlab
   conda activate matlab
   pip install jupyter-matlab-proxy
   

2. Start a JupyterLab session via the ARE App. In addition to the standard fields, I presume you’ll have to select your Matlab software group under Software, e.g.:

   

   and to add Matlab modules in the Advanced Options under Modules, e.g.:

   

   You’ll also need to tell the ARE to use the conda environment you created (also in Advanced Options):

   

[Aidan (Aidan Heerdegen, ACCESS-NRI Release Team Lead)]

Given that NCI officially support MATLAB and they already provide a number of specialised conda environments, I’d say emailing help@nci.org.au would be a good option to see if they have plans to support this use case, or are willing to do so in the future.