Automatic Quantum Circuit Generation for Finding Ground States

Quantum computing could advance basic science if we had software that automatically generates quantum circuits, circuits that compute the ground state of a given Hamiltonian. Our free and open-source software development [*] already does exactly that, but it not efficient yet. This talk will discuss the roadmap to make it efficient, and the roadblocks that we might encounter in the way. I will claim that tensor networks is the only known approach that can be both efficient and accurate. For this approach to work, we need to penalize or discard deep layered circuits, so that I will discuss the resulting impact to convergence. I will address also the need to use high performance (classical) computing, and the benefits of hybrid computing, where quantum circuit fitness can be evaluated on quantum hardware. Condensed matter, computational chemistry, high energy physics, all should benefit from this approach, as our generator would provide insightful, non-intuitive quantum circuits that could not have been designed manually.

(*) https://code.ornl.gov/gonzalo_3/evendim