A large fraction of statistical models within HEP can be expressed through a flexible, declarative template, HistFactory, for binned densities of observables.
Originally implemented only for ROOT and RooStats, HistFactory has recently been re-implemented based on the scientific python packages in order to make use of auto-differentiation and hardware acceleration via an integration with Machine Learning tensor libraries such as TensorFlow and PyTorch.
The Go scientific ecosystem is lacking such a facility.
During this GSoC project, you will work on implementing a similar library and class of models in Go using Machine Learning libraries such as Gorgonia.
We propose the following steps:
einsum) in Go ML libraries,A package that creates statistical models for multi-bin histogram-based analyses, written in Go.