Introduction

Introduction

Currently, there exist two $\tau$ paths in the HLT Upgrade menu. Both are DiTau paths, i.e., both target decays of particles into two $\tau$ leptons. Both of them make use of the "legacy" algorithm to reconstruct $\tau$ leptons, which is the so-called "Hadrons-plus-strips" algorithm. The ChargedIso path uses the "traditional" approach of applying cuts on predefined reconstructed variables to identify "real" $\tau$ leptons and to reject fake ones.

The DeepTau path is based on a Convolutional Neural Network (CNN) called DeepTau.

Cross-trigger paths, i.e., paths designed to trigger on one $\tau$ and one different lepton ($\mu$ or $e^-$) are currently in development.