Use of charged-track information to subtract neutral pileup

Abstract

The use of charged pileup tracks in a jet to predict the neutral pileup component in that same jet could potentially lead to improved pileup removal techniques, provided there is a strong local correlation between charged and neutral pileup. In a Monte Carlo simulation we find that the correlation is however moderate, a feature that we attribute to characteristics of the underlying nonperturbative dynamics. Consequently, pure "neutral-proportional-to-charge" (NpC) pileup mitigation approaches do not outperform existing, area-based pileup removal methods. This finding contrasts with the arguments made in favor of a new method, "jet cleansing," in part based on the NpC approach. We identify the critical differences between the performances of linear cleansing and trimmed NpC as being due to the former's rejection of subjets that have no charged tracks from the leading vertex, a procedure that we name "zeroing." Zeroing, an extreme version of the "charged-track trimming" proposed by ATLAS, can be combined with a range of pileup mitigation methods, and appears to have both benefits and drawbacks. We show how the latter can be straightforwardly alleviated. We also discuss the limited potential for improvement that can be obtained by linear combinations of the NpC and area-subtraction methods.