Alternative Clustering Methods (ACM)

acm is a cosmological analysis pipeline built for the DESI collaboration. It provides alternative methods to the standard two-point correlation function analysis, including higher-order statistics and machine learning techniques. The package is designed to be modular and extensible, allowing users to easily add new estimators and analysis methods.

Warning

The ACM project is a research project in constant evolution. The content of this repository may change frequently, and some features may be experimental or under development. Users are advised to check for updates regularly and to use the software with caution.

Documentation

The complete documentation is available at acm.readthedocs.io.

Installation

Install with pip

To install acm at NERSC, the cosmodesi environment already has some required packages installed. You can install the package with:

pip install acm[sunbird] @ git+https://github.com/epaillas/acm

Install from source

If you want to install the package from source, you can clone the repository and install it with:

git clone
pip install -e .[sunbird]

Note

The -e flag is used to install the package in editable mode, which allows you to make changes to the code and have them reflected without reinstalling the package.

Requirements

Strict requirements are:

• numpy
• scipy
• pandas
• pyyaml
• matplotlib
• getdist

Warning

Make is required to compile some C files during the installation, so it should be available in your environment.

To run the emulators, you will need:

• sunbird

The package can be installed with the following dependencies:

• cosmodesi to install extra cosmodesi dependencies (already included in the cosmodesi environment at NERSC)
• docs to install the documentation building dependencies
• some estimators also have their own dependencies, which can be installed trough the estimator name.
• the estimator backends also have their own dependencies, which must be installed trough the backend name.

Tip

Add the dependency names separated by commas, e.g. pip install acm[sunbird,cosmodesi,estimator1,estimator2] to install several dependencies at once.

Cython building

The Cython files can be rebuilt (only in editable mode) with:

python setup.py build_ext --inplace 
python setup.py clean --all # Clean up build files

Note

The cython files are automatically built when installing the package, so you should not need to run this command unless you modify the Cython files.

Examples

You can find notebooks examples in the nb folder.

TODO (Github)

• Update the links to the doc