Dortmund 2021 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

T: Fachverband Teilchenphysik

T 40: Experimental methods II

T 40.6: Vortrag

Dienstag, 16. März 2021, 17:15–17:30, To

Estimating the fake lepton contribution in the framework of Bayesian statistics — Johannes Erdmann, Cornelius Grunwald, Kevin Kröninger, Salvatore La Cagnina, and •Lars Röhrig — TU Dortmund, Experimentelle Physik IV

In a particle detector, the signature of prompt leptons like electrons and muons can be faked by other objects, such as jets or non-prompt leptons from hadron decays. Huge amounts of simulated events would be needed to estimate the fake lepton background contribution due to the small fake probability, thus data-driven techniques like the matrix method are commonly used instead. The method is based on the definition of two identification criteria, referred to as loose and tight and the probabilities of a real or fake lepton to pass either one of these criteria.

Known limitations of the classical matrix method are the possible prediction of negative fake rates and the approximation of Poisson distributed counting rates as Gaussian distributions. A Bayesian approach is presented, facing these restrictions and leading to a stable fake rate estimation in cases, in which the classical matrix method has limited validity. Several studies and comparisons to alternative methods are presented.