Analyzing CMS open collider data through topic modeling

Abstract

We conduct an investigation of jet substructure at the Large Hadron Collider (LHC) on the CMS Open Data. We analyze a sample of jets detected by the CMS experiment in 2011 with the companion simulated (both pre- and post-detector) jet datasets to demonstrate the ability of our analysis framework to accurately quantify jet substructure observables as well as properly describe the effects of the CMS detector. We move to a specific analysis of jet classification using the unsupervised algorithm of jet topics to provide a new way of defining the categories of quark and gluon jets through their observable properties. We finally outline a means of accessing our sample of 883,431 high quality CMS-detected jets so that other scientists may conduct their own analyses on this sample.