Collaborative Research: Elements: Lattice QCD software for nuclear physics on heterogeneous architectures

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

Nucleons (protons and neutrons) make up atomic nuclei and are a common building block of all matter in the universe. To better understand atomic nuclei, as well as to assist current experiments to find new particle physics and to understand particles known as neutrinos, it is important to carry out calculations of nucleons interacting with not only other nucleons but other particles in nature, such as electrons, muons, and mesons. The development of optimized software to carry out such computations on modern GPU-accelerated supercomputer systems is pursued. The physics of hadron-hadron interactions can be studied using Monte Carlo estimates of path integrals involving quark and gluon fields on a space-time lattice. Baryon-meson and baryon-baryon scattering phase shifts can be computed, yielding important information on hadron structure. Form factors involving the nucleon and transitions through the Delta baryon are particularly important since they are crucial to accelerator-based neutrino experiments, such as the Deep Underground Neutrino Experiment (DUNE). New computational techniques, such as the stochastic LapH method, have made possible such computations in lattice quantum chromodynamics (LQCD). A key, but computationally intensive, ingredient in such computations is the evaluation of individual baryon sources and sinks. One goal of this work is the development of highly-optimized software to evaluate baryon sources/sinks on modern GPU-accelerated architectures. A plethora of tensor contractions are needed in the end stages of such scattering calculations, which are best performed in separate program executions requiring different wall times and numbers of computing processors. Effectively bundling such numerous runs together into a handful of batch jobs on supercomputer systems is crucial in the work flow of these computations. This work will also develop the software to efficiently carry out this bundling. This award by the Office of Advanced Cyberinfrastructure is jointly supported by the Physics at the Information Frontier program in the Division of Physics within the Directorate for Mathematical and Physical Sciences. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.