HEP advanced tracking algorithms at the exascale (Project Exa.TrkX)

Reconstructing the trajectories of thousands of charged particles from a collision event as they fly through a HEP detector is a combinatorially hard pattern recognition problem. Exa.TrkX, a DOE CompHEP project and a collaboration of data scientists and computational physicists from the ATLAS, CMS, and DUNE experiments, is developing Graph Neural Network models aimed at reconstructing millions of particle trajectories per second from Petabytes of raw data produced by the next generation of detectors at the Energy and Intensity Frontiers. Exa.TrkX is also exploring the scaling of distributed training of GNN models on DOE pre-exascale systems and the deployment of GNN models with microsecond latencies on FPGA-based real-time processing systems.

Performance of a track finding Graph Neural Network as a function of the message-passing iteration. Best results are found after eight iterations.


Exa.TrkX is a follow-up to the the HEP.TrkX pilot project. It relies on the aCTS toolkit to simulate a generic HL-LHC detector, and more recently to benchmark the performance of its models. Exa.TrkX is collaborating with the FastML Lab to deploy GNN models on FPGA systems. Exa.TrkX is also collaborating with the NERSC Big Data center, and the Exalearn co-design center to demonstrate distributed training and model hyperparameter optimization at scale on HPC systems.
collaborators, partners & toolkits

meetings & workshops

Collaboration tools

reports, publications & presentations

project co-PI's

powered by