Simultaneous Monte Carlo analysis of parton distributions: a new paradigm in global QCD studies

We report the results of the first simultaneous extraction of unpolarized and polarized parton distributions and parton-to-hadron fragmentation functions from a Monte Carlo analysis of global high energy lepton and hadron scattering data. We use data resampling techniques to thoroughly explore the Bayesian posterior distribution of the extracted functions, and use $k$-means clustering on the parameter samples to identify the configurations that give the best description across all reactions. Inclusion of semi-inclusive deep-inelastic scattering data reveals a strong suppression of the strange quark distribution in the proton, in contrast with recent results from $W^\pm$ and $Z$ production at the LHC. The analysis is the first to include jet production data in both unpolarized and polarized proton-proton collisions, which provides the best constraints to date on the momentum and spin carried by gluons in the proton.