Argonne’s Computational Science Division is developing code and algorithms for simulations of Quantum Field Theory (QFT) to help study the fundamental building blocks of nature. QFT combines relativity and quantum mechanics to describe the many-body interactions of subatomic particles. Our main focus is on simulating strongly interacting theories such as Quantum Chromodynamics (QCD), which describes how quarks and gluons combine to form nuclei, and other models of interest to the DOE high energy and nuclear physics programs.

Field theories can be discretized on a four-dimensional space-time lattice for simulation on a supercomputer.  We are researching ways to improve the efficiency of these lattice field theory calculations on upcoming exascale hardware. This involves finding new or improved algorithms for large sparse linear algebra (including eigenvector deflation and multigrid), Monte Carlo techniques (such as Hybrid Monte Carlo), and incorporating machine learning methods.

We are also investigating methods for simulating QFT on emergent technologies such as quantum computers. As part of this work, we are using tensor network representations of field theories to help find improved mappings of the QFT degrees of freedom onto qubits, and to perform direct simulations of the theories. We collaborate with other researchers from universities and national labs to develop improved code and algorithms and to help perform leadership class calculations related to advances in high energy and nuclear physics.