The purpose of the RooFit toolkit is to facilitate the modeling of probability density models of arbitrary complexity. These probability models can be used to construct the likelihood function for statistical inference on any observed dataset. The concept of RooFit is that the individual mathematical elements of a probability function (observables, parameters, functions, integrals) are expressible as C++ objects and models are organically constructed from these components.
RooFit has had a large impact of experimental particle physics: since the introduction of persistable computable models, sharing and combination of probability models has resulted in many new state-of-the-art results from the Large Hadron Collider experiments. Such models combine hundreds of datasets and have thousands of parameters, and has greatly simplified the process of making these complex analyses in short time scales.
The goals of this project are: i) performance testing and tuning of new ‘parallel fit’ software on super-complex models, and ii) parallel uncertainty calculation based on parallel minimization infrastructure.
