Evidence for WZ/WW->l nu + Heavy Flavors Vector Boson Production in 7.5 fb^-1 of CDF Data

TL;DR

This paper reports a search for WZ and WW production with heavy flavor decays using 7.5 fb^-1 of CDF data, observing a signal consistent with Standard Model predictions and employing advanced multivariate techniques.

Contribution

It introduces a novel analysis method combining heavy flavor tagging and support vector machine discrimination to identify WZ/WW events in collider data.

Findings

Observed a 3.03 sigma signal significance.

Measured cross section consistent with Standard Model.

Demonstrated effective background suppression with multivariate analysis.

Abstract

This contribution describes a search for the associated production of Standard Model vector bosons WZ where the W boson decays leptonically (W->l nu) and the Z boson decays to a heavy flavor quark pair (Z->b b-bar,c c-bar). At least one identified ("tagged") heavy flavor jet in the final state is required. Given the small di-jet invariant mass separation between the W and the Z resonances, the production of WW events where one W decays leptonically and the second W decays into an heavy flavor jet (e.i. W^+ -> c s-bar) contributes to our the signal. This search uses data collected with the CDF II detector at the Tevatron Collider at Fermilab, and corresponding to an integrated luminosity of approximately 7.5 fb^-1. Events consistent with the signature of a charged lepton (electron or muon), large missing transverse energy and exactly two jets, of which at least one is required to contain a secondary vertex displaced from the jet origin, are selected. A multivariate discriminant based on the Support Vector Machine algorithm is used to reduce greatly the multi-jet background contamination. We observe a signal of significance of 3.03 sigma over the background only hypothesis. A cross section of 1.085 +0.26/-0.40 times the expected Standard Model value for the combined WZ/WW production and decay into heavy flavors is measured, consistent with the Standard Model prediction.