Research Symposium

25th annual Undergraduate Research Symposium, April 1, 2025

Jake Serwe Poster Session 2: 10:45 am - 11:45 am/ Poster #158


jake_headshot.jpg

BIO


Jake Serwe is an undergraduate studying Computer Science at Florida State University. He is interested in computational solutions to scientific problems, particularly in physics and political science. He was born and raised in rural Germany and moved to Virginia in 2017.

Identifying Overlapping Particle Reactions in GlueX Data

Authors: Jake Serwe, Dr. Edmundo Barriga
Student Major: Computer Science
Mentor: Dr. Edmundo Barriga
Mentor's Department: Physics
Mentor's College: Arts & Sciences
Co-Presenters:

Abstract


The GlueX experiment at the Thomas Jefferson National Accelerator Facility studies particle reactions using a photon beam scattered off of a liquid hydrogen target. The GlueX spectrometer measures neutral and charged particles produced as the decay products of a myriad of particle reactions. Many of these reactions have similar final states. This introduces backgrounds to datasets that overlap in their final state events. The GlueX collaboration aims to aid in the identification of a reactionโ€™s signal through the use of a kinematic fitter. The fitter assigns ๐œ’2 values to each event indicating how well it satisfies physics conservation laws. Here, we propose a tool that compares the ๐œ’2 results for particle reactions and identifies which reaction hypothesis is favored for the events that overlap. By comparing the ๐œ’2 values of identical events but different hypotheses, we can determine which events form part of the desired signal, and which ones have been misidentified. The tool was tested using three particle reactions with a final state 4๐›พ ฯ€+ ฯ€โˆ’. Preliminary results show significant cross-contamination between these processes, including ฯ€0 being misidentified as ฮท. This overlap is demonstrated between the 2ฯ€0 ฯ€+ ฯ€โˆ’ and ฯ€0ฯ€+ฯ€โˆ’ฮท final states. The tool is written in C++ and demonstrates excellent performance by processing two ~300,000 event datasets in ~30 seconds using FSU Physics' computing machines.

Bildschirmfoto 2025-03-17 um 4.59.05 PM.png

Keywords: GlueX, Physics