gauss2dfit is a package for defining and evaluating 2D Gaussian mixture models on imaging data. It is being developed primarily for use in astronomy - specifically, by Vera C. Rubin Observatory Data Management for the Legacy Survey of Space and Time, and the MultiProFit source modelling package - but it can be used for any kind of image or domain.
gauss2dfit is provided in two parts; a C++ shared library (libgauss2dfit) and optional Python bindings. Both depend on the base package gauss2d <https://github.com/lsst-dm/gauss2d/>, as well as the parameters <https://github.com/lsst-dm/parameters/> package for defining parameters with limits and transformations.
gauss2dfit is built with Meson and uses pkg-config to store package information. You will likely want to configure it to install in a local directory. For example, if you are using a conda environment:
PKG_CONFIG_PATH=$CONDA_PREFIX/.local/lib64/pkgconfig meson --prefix=$CONDA_PREFIX/.local build
You will also need to run meson build once, followed by meson compile -C build && meson install -C build to (re)compile. This must be done in the base directory and then the python subfolder if bindings are desired.
A full example setup script is provided in setup_conda-release.sh. This defaults to using $CONDA_PREFIX, but can be configured to output elsewhere (e.g. ~/.local):
CONDA_PREFIX=~ sh setup-conda-release.sh
Once the build command is run once to create the build directories, subsequent rebuilds can use the provided build*.sh scripts.
One can also build a configuration for debugging purposes: see setup-debug.sh for an example. Note that meson defaults to a build directory named builddir, to disambiguate the common meson build command. The example scripts configure separate build-debug and build-release directories. If you are using an IDE that expects to find a build, simply create a symbolic link (e.g. ln -s build-debug build).
Note: You may need to change include and/or LD paths to debug the Python bindings with build-debug; this has not been tested yet.