BLUSE kernels container

The files here can be used to create a Singularity container with software environments for analyzing Breakthrough Listen data in Jupyter notebooks running on JupyterHub.

Building the container

The easiest way to build the container is to run make. This will use singularity to build the container bluse-kernels-container.sif.

If you are already familiar with building Singlarity containers, the bluse-kernels-container.def file can be used directly with singularity:

singularity build bluse-kernels-container.sif bluse-kernels-container.def

Note that singularity requires root priviliges to build containers.

Using the container to provide Jupyter kernels

To provide the Jupyter kernels of the container to JupyterHub users, simply follow these steps:

1. Place the bluse-kernel-containers.sif file in a location that JupyterHub can access.

2. Edit the kernels/*/kernel.json files to match your setup: 2.1 Make sure the path to singularity is correct 2.2 Make sure to --bind any data directories so they will be visible from the notebooks 2.3 Make sure the path to the bluse-kernels-container.sif file is correct

3. Copy the kernels/* directory to the JupyterHub kernels directory, being sure to maintain the directory structure.

Done!

Users with existing JupyterHub servers running will have to restart them to see the new kernels.

Other container features

The container has two apps for command line users:

• ipython - runs ipython with the container's conda environement activted
• julia - runs julia with some preinstalled Julia packages on LOAD_PATH

To run these apps, use:

singularity run --app APP_NAME CONTAINER_NAME [ARGS]