Installation#

Prerequisites#

ANDALUS requires Python 3.10 or later. The following dependencies are installed automatically:

Package	Purpose
`pandas`	Core data structures for sensitivities and covariances
`numpy`	Numerical operations
`serpentTools`	Reading Serpent Monte Carlo output files
`sandy`	Reading ERRORR covariance files and exporting to ACE format
`pyyaml`	Parsing YAML configuration files
`tables`	HDF5 storage via `pandas.HDFStore`

Stable release#

Install the latest published version from PyPI:

pip install andalus

If you prefer uv:

uv add andalus

From source#

Clone the repository and install in editable mode for development:

git clone https://github.com/daan1392/andalus
cd andalus
pip install -e .

With uv (recommended — also installs dev dependencies from pyproject.toml):

git clone https://github.com/daan1392/andalus
cd andalus
uv sync

uv sync creates an isolated virtual environment under .venv/. Activate it before running commands:

# Linux / macOS
source .venv/bin/activate

# Windows (PowerShell)
.venv\Scripts\Activate.ps1

Verifying the installation#

import andalus

print(andalus.__version__)

Optional: NJOY / SANDY for ACE export#

The AssimilationSuite.to_ace() method requires a working NJOY installation accessible on your PATH, in addition to the sandy Python package. This is only needed when you want to export adjusted nuclear data to ACE format for use in a Monte Carlo transport code.

NJOY is not required if the adjustment is instead applied directly to an existing ACE library. This approach is preferred, as it skips the processing step entirely. Consistency is strongest when the adjustments are applied to the same ACE library that was used to calculate the response of interest and the sensitivity coefficients. An added benefit is that we do not need to reprocess the ACE files, which saves a lot of time.