Gammapy Dependencies

The latest stable version of Gammapy is listed at https://gammapy.org

Gammapy works with Python 3.6 or later.

Linux and Mac OS are fully supported.

Gammapy itself, and most analyses, work on Windows. However, two optional dependencies don’t support Windows yet: Sherpa (an optional fitting backend) and healpy (needed to work with HEALPix maps, which is common for all-sky analyses).

Gammapy is a Python package built on Numpy and Astropy, as well as a few other required dependencies. For certain functionality, optional dependencies are used. The recommended way to install Gammapy is via a conda environment which includes all required and optional dependencies (see Installation).

Note that when you install Gammapy with conda (or actually any alternative distribution channel), you have a full package manager at your fingertips. You can conda or pip install any extra Python package you like (e.g. pip install pyjokes), upgrade or downgrade packages to other versions (very rarely needed) or uninstall any package you don’t like (almost never useful, unless you run out of disk space).

Required dependencies

Required dependencies are automatically installed when using e.g. conda install gammapy -c conda-forge or pip install gammapy.

  • numpy - array and math functions

  • scipy - numerical methods (interpolation, integration, convolution)

  • Astropy - core package for Astronomy in Python

  • regions - Astropy sky region package

  • click - used for the gammapy command line tool

  • PyYAML - support for YAML format (config and results files)

  • pydantic - support config file validation

Optional dependencies

The optional dependencies listed here are the packages listed in the conda environment specification (see Installation). This is a mix of packages that make it convenient to use Gammapy (e.g. ipython or jupyter), that add extra functionality (e.g. matplotlib to make plots, naima for physical SED modeling), and partly packages that aren’t used within Gammapy, only for example data download (parfive) or in one of the tutorials (sherpa).