Skip to content

Recipes

Recipes are self-contained, task-oriented guides. Each one walks you through a specific workflow on Unity, from setting up a Python environment to running multi-node MPI jobs. Unlike the Fundamentals section, which builds concepts from the ground up, recipes assume you already have the basics down and need to get something done.

  • Python

    Manage environments with uv or Conda, parallelize CPU-bound work across multiple cores, run PyTorch and Hugging Face Transformers workloads, and scale training across multiple GPUs.

    Python recipes

  • R

    Load R, install packages with renv, and submit R scripts as batch jobs.

    R recipes

  • MATLAB

    Run MATLAB scripts as non-interactive batch jobs on the cluster.

    MATLAB recipes

  • Containers

    Learn when containers make sense on HPC and how to build and run them with Apptainer.

    Container recipes

  • Open OnDemand

    Access the cluster through a web browser: launch Jupyter notebooks, RStudio, and other interactive apps without SSH.

    Open OnDemand recipes

  • Slurm Patterns

    Interactive sessions, job arrays, and intermediate submission patterns for more complex workflows.

    Slurm recipes

  • MPI

    Compile and run MPI programs across multiple nodes, including a Python example with mpi4py.

    MPI recipes

New to the cluster?

If you haven't submitted a job yet, start with Getting Started instead. These recipes assume you're comfortable connecting to Unity, loading modules, and writing basic Slurm scripts.