IRIS-HEP Fellow: Caitlin Patterson
Fellowship dates: Jan - Jun 2021
Home Institution: The Ohio State University
Project: Scaling up implementations of GNNs with FPGA co-processors for charged particle track reconstructionCharged particle tracking is an important part of characterizing particles produced in colliders. Tracking algorithms are very computationally expensive and scale poorly with the number of hits. The upcoming run of high-luminosity the LHC will feature an increase in collision rate, leading to more data production than ever before. It is therefore advantageous to implement these algorithms using FPGAs, which have lower latency and higher energy efficiency than CPUs. This project proposes building on the work of Aneesh Heintz by scaling up graph neural network implementations co-processed on FPGAs.
More information: My project proposal
Isobel Ojalvo (Princeton University)
Savannah Thais (Princeton University)