getting_started.md

Creating the recommended local environment

Run these commands from the repository root. This assumes that conda and mamba are already installed (e.g., conda install mamba -n base -c conda-forge).

mamba env create -f cssi_env.yml

conda activate cssi_evaluation

python -m ipykernel install --user --name=cssi_evaluation

The environment file installs the local cssi_evaluation package in editable mode with the [notebooks,dev] extras, so package imports, notebooks, and development tools are available in the same environment.

Creating the recommended environment on Verde

Follow these steps:

# 1. Go to a writable folder (Verde user home)
cd ~

# 2. Download and extract micromamba (Linux x86_64)
curl -L https://micromamba.snakepit.net/api/micromamba/linux-64/latest | tar -xvj bin/micromamba

# 3. Add micromamba to PATH temporarily (for this session)
export PATH=$PWD/bin:$PATH

# 4. Go to the cloned repository root
cd /home/<USER>/cssi_evaluation

# 5. Create the environment from the repository YAML
micromamba env create -f cssi_env.yml

# 6. Initialize micromamba for your shell
eval "$(micromamba shell hook --shell bash)"

# 7. Activate your environment
micromamba activate cssi_evaluation

# 8. Register the Jupyter kernel
python -m ipykernel install --user --name=cssi_evaluation

Notes

• pyproject.toml is the package metadata file used by pip install -e .. It contains core library deps.
• cssi_env.yml is the single Conda environment file intended for running examples and it contains heavy Conday/system deps.