light-roast-test

Example scripts for running on the UChicago AF.

How to run these examples

To run the eventloop example, use the python 3.6 kernel.

To run the columnar example, you'll need a more recent version of python, and you probably want to work in a virtual environment. Here are a few options for that:

Using pip

The simplest way to get a new environment up and running is with pip:

bash # in case you're not already in a bash shell
new_env = "my_new_virtual_env"
python -m venv ${new_env}
source ${new_env}/bin/activate
pip install ipykernel numpy dask_jobqueue parse
python -m ipykernel install --user --name=${new_env}

Then restart your jupyter server and you should see the new kernel available for use. To add a module later, just open a terminal in jupyter:

bash # in case you're not already in a bash shell
new_env = "my_new_virtual_env"
source ${new_env}/bin/activate
pip install scikit-learn

No need to restart the jupyter kernel if you're just adding packages.

Using pixi

In a Jupyter shell, set up a new kernel to use:

curl -fsSL https://pixi.sh/install.sh | bash
export PATH=$PATH:.pixi/bin/
pixi init
pixi add python=3.12
pixi add atlas-schema
pixi add ipykernel
pixi add pixi-kernel
pixi add numpy                  # this syntax should work in most cases, but maybe not all packages support this
pixi add pip                    # if you need pip
pip install dask_jobqueue parse # this should also work
python -m ipykernel install --user --name=light-roast-kernel

Then restart your Jupyter server. On restart, you should see an option to use the light-roast-kernel, which is the one you want. You may need to add some additional packages, which you can do from within a shell:

pixi shell
pip install parse               # this should also work

You may need to restart your Jupyter server (not just the kernel) again if you do that.

Using ALRB

If you prefer to use ALRB, you can try something like the following, but it hasn't worked smoothly in the past:

export ATLAS_LOCAL_ROOT_BASE="/cvmfs/atlas.cern.ch/repo/ATLASLocalRootBase"
alias setupATLAS="source ${ATLAS_LOCAL_ROOT_BASE}/user/atlasLocalSetup.sh"
setupATLAS
lsetup "python 3.11.9-x86_64-el9"
python3 -m venv lr-kernel
source lr-kernel/bin/activate
pip3 install --upgrade pip
pip3 install atlas-schema dask_jobqueue parse ipykernel
python3 -m ipykernel install --user --name=lr-kernel

The tool referenced here may help with getting the environment right, but I'm not sure that's the problem.