Identification of prompt and isolated muons using multivariate techniques at the CMS experiment in proton-proton collisions at $\sqrt{s}=13~\mathrm{TeV}$

Prompt and isolated muons as well as muons from heavy flavour decays represent a key object for many analyses at CMS either to select the signal final states or to reject the background events. In this note we present two multivariate techniques that have been developed to provide a highly efficient identification algorithm for muons with transverse momentum greater than $10~\mathrm{GeV}$. One has been trained as an alternative to the standard cut-based identification criteria but with higher efficiency working points, and offers a continuous variable which provides more flexibility to pick the desired working points. The second one aims to select prompt and isolated muons by using isolation requirements to reduce the contamination from nonprompt muons arising in heavy flavour hadron decays. Both algorithms are developed using $59.7~\mathrm{fb}^{-1}$ of data produced in proton-proton collisions at a center-of-mass energy of $\sqrt{s}=13~\mathrm{TeV}$ collected during 2018 with the CMS experiment at CERN LHC. Their performance has been assessed in both data and simulation. The measured efficiencies for the first MVA are similar or better than those achieved by the standard cut-based selection criteria. While the second MVA is key to reduce background contribution from nonprompt muons, which leads to an increase in sensitivity crucial both in precision standard model measurements as well as in beyond standard model searches.