Skip to content

Running MNE Python on ARC Using Jupyter

Purpose

Set up a Jupyter server on ARC that you can access from your local computer's browser (Windows, macOS, or Linux). This allows you to run MNE Python interactively and use the mne_env kernel in Jupyter notebooks.

Prerequisites

  • A SSO (Single Sign-On) account at the University of Oxford
  • VPN access to the University of Oxford network (required for off-site access)
  • An account on the project psyc-neuosc at ARC
  • System environment configured; see Initialise Settings and Folders
  • MNE Python installed; see Setting Up MNE Python

Steps

1. Connect to ARC

Connect to the VPN, then access ARC via remote desktop or SSH. You will land on a gateway node.

SSH into the login node:

ssh -X htc-login

2. Start the Jupyter Server

Launch a Jupyter server that submits a job to ARC's compute cluster:

nog_jupyter_server

The command will take a few seconds to minutes to complete. It submits a job to run a Jupyter server and prints connection information including an SSH tunnel command.

Wait for the output. When complete, you will see an ssh command printed to your terminal. Copy this command—you'll need it in the next step.

3. Connect Your Local Machine to Jupyter

On your local machine (not on ARC), open a Terminal (Windows or MacOS) and paste the SSH tunnel command printed by nog_jupyter_server:

ssh -L 8888:htc-cXXX:YYYYY psycZZZZ@gateway.arc.ox.ac.uk

Here XXX:YYYY refer to name of htc-c node where the job was started. psycZZZZ is your SSO username.

Enter your Oxford password when prompted. This creates a secure tunnel from your local machine to the Jupyter server.

Keep this terminal window open while you use Jupyter.

4. Open Jupyter in Your Browser

On your local machine, open a web browser and navigate to:

http://localhost:8888/tree

You should see the Jupyter file browser. You're now connected to the Jupyter server running on ARC!

5. Select the MNE Kernel

When you create a new notebook or open an existing one:

  1. Click Kernel → Change kernel
  2. Select mne_env from the list
  3. The notebook will now use the MNE Python environment

You can now import and use MNE:

import mne
print(mne.__version__)

Advanced Options

Custom Server Settings

You can request specific resources (e.g., memory, CPU cores, walltime) using:

nog_jupyter_server -h

This shows all available options.

Monitor Jobs

View your submitted jobs and their job IDs:

squeue -u $USER

Stop the Server

To terminate the Jupyter server job:

scancel JOBID

Replace JOBID with the ID from squeue.

Note: Using Ctrl+C on your local machine will not terminate the ARC job. You must use scancel to stop it.

Troubleshooting

Can't connect to localhost:8888? - Ensure the SSH tunnel command is still running on your local machine - Check that the terminal running the tunnel hasn't closed

Kernel not available? - Ensure MNE is installed (see Setting Up MNE Python) - Restart Jupyter and check the kernel list again

Server timed out? - Jobs have time limits. Resubmit with nog_jupyter_server to start a new server