GooStats, a multivariate spectrum fitting analysis package for particle physics accelerated by graphic processing units

2018-05-20T04:48:44Z (GMT) by Ding, Xuefeng

GooStats is an open source project for multivariate spectrum fitting analysis for particle physics accelerated by graphic processing units. It is a middle layer between GooFit, a minimization engine that can run on graphic processing units etc., and the module written by user customized for specific physics problem. The package as an analysis framework defines flexible data structures and interfaces allowing multivariate spectrum fitting. The parameter synchronization tool allows user to easily fit multiple datasets while synchronizing fit parameters between datasets at different granularity. As ROOT is popular in high energy physics community, the fit results, including all best fit parameters and fit errors, likelihood, chi-square, p-value etc. are saved in TTree format. Figures both in pdf and TFile formats are also produced, while the style of the plot can be customized easily. In this release, an example module dedicated for mass hierarchy determination through spectrum fitting in medium baseline reactor neutrino experiment is provided.