The Optimization of ATLAS Track Reconstruction in Dense Environments

This note presents recent changes in the ATLAS track reconstruction chain derived from detailed studies of track reconstruction in dense environments. The cores of high $p_{T}$ jets and $\tau$-leptons are characterized by charged particle distances comparable to the inner detector sensor dimensions. The ambiguity processor stage of the reconstruction chain was over-halled including an improvement of the usage of a NN based approach to identify clusters created by multiple charge particles. Single particle samples are used to demonstrate the alteration in a simple environment. The impact of these changes on tracks in high $p_{T}$ jets are shown to result in more pixel hits on track, a more meaningful split hit definition, and improved track parameter estimation. A 10% increase in b-jet identification for an equal fake rate has been shown.