Optimized Top Quark Analysis with the Decision Tree

Abstract

We present an optimized and physically motivated method for separating top quark signal events from background events at the Tevatron. For the top quark signal $t\bar t \to e/\mu + 4$ jets, we show how to reject all but $25\%$ of the background in a data sample while retaining $80\%$ of the signal, without introducing bias into the subsequent mass measurement. The technique used is the Binary Decision Tree. Combining this highly efficient procedure for signal identification with a novel algorithm for top quark reconstruction, we propose a powerful new way to measure the top quark mass.