Performance of the bwHPC cluster in the production of μ -> t embedded events used for the prediction of background for H -> tt analyses

Abstract

In high energy physics, a main challenge is the accurate prediction of background events at a particle detector. These events are usually estimated by simulation. As an alternative, data-driven methods use observed events to derive a background prediction and are often less computationally expensive than simulation. The lepton embedding method presents a data-driven method to estimate the background from Z ! events for Higgs boson analyses in the same final state. Z ! μμ events recorded by the CMS experiment are selected, the muons are removed from the event and replaced with simulated leptons with the same kinematic properties as the removed muons. The resulting hybrid event provides an improved description of pile-up and the underlying event compared to the simulation of the full proton-proton collision. In this paper the production of these hybrid events used by the CMS collaboration is described. The production relies on the resources made available by the bwHPC project. The data used for this purpose correspond to 65 million di-muon events collected in 2017 by CMS.