Particle ID required

Identifying the specific types of particles in the DZero detector is absolutely crucial for performing measurements and searches. Different particles have characteristic signatures when interacting with the detector material, which allow them to be differentiated, as described in two recent DZero papers.

Disponible en español

What does an electron look like in the DZero detector? Or what about a bottom quark? Or even a photon? Answering such questions is a crucial part of any particle physics measurement. For a given process of interest, there is a distinct set of decay products. For example, single top quark production may be characterized by two bottom quarks, an electron and a neutrino. To select such events, we need to identify these various final-state particles in the detector. The performance of such “particle ID” feeds directly into the sensitivity to rare processes and the precision when measuring fundamental parameters.

The DZero collaboration recently submitted two publications describing different types of particle ID. The first paper covers the identification of photons, electrons and positrons (anti-electrons), or EM-ID for short, since all these particles interact through the electromagnetic (EM) force in the calorimeter detector. The second paper describes the identification of bottom quarks (b-ID), in particular the tools used to distinguish them from lighter quarks. While the specific techniques are quite different in the two papers, the overall strategies are very similar.

The first step is to design an algorithm that differentiates between the particle of interest and the relevant backgrounds. The algorithm relies on the fact that different particles interact with the detector in different ways, leading to distinct experimental signatures. For example, both photons and electrons tend to deposit their energy in the inner layers of the calorimeter, as distinct from quark jets, which penetrate farther. Similarly, jets from bottom quarks tend to originate farther from the original proton-antiproton collision than light-quark jets, due to the relatively long bottom quark lifetime. For both papers, scientists use multivariate techniques to combine several discriminating variables, giving a single output number that expresses the confidence in the identification.

The next step is to quantify the performance of the ID algorithm. Specifically, for a given value of the output, we need to know the efficiency of identifying the desired particle, as well as the rate of false positives (or “fakes”). The usual strategy is to use control samples with a well-known composition. For instance, a very pure electron sample can be collected from Z boson decays into an electron-positron pair, allowing the efficiency of electron ID to be extracted as a function of various properties of the particle. Similarly, a sample dominated by light quark jets allows the fake rate of the b-ID algorithm to be extracted.

The two papers represent the culmination of years of gradual improvements, testimony to the innovation and hard work of many DZero members. The resulting gains in ID performance feed directly into the analysis program, giving more precise measurements and more sensitive searches. As a particular (and timely!) example, the observation of single top quark production in the s-channel, one of the rarest of Standard Model processes, is now within reach at the Tevatron, something that would not have been possible without these state-of-the-art particle ID tools. The combination of the Tevatron experiments’ results on s-channel single top quark production will be presented at the wine-and-cheese seminar tomorrow at 4 p.m., so please come along to learn more.

Mark Williams

These DZero members all made significant contributions to the b quark identification paper.
These DZero members all made significant contributions to the EM-ID paper. We recognize the major contributions of Rajendran Raja (1948-2014) over the course of both Run I and Run II to DZero electron and photon identification, including contributions to the NIM paper submitted on Dec. 30, 2013.