Quantum computers are a big interest of mine, and in particular I think there’s a lot to be learned in using prototype devices available in the lab today. Unfortunately, a lot of the algorithms we talk about today are simply too complex to fit on even near-future quantum devices. However, by thinking about quantum computers as a specialized computation device, sort of a QPU analogous to GPUs we see today, and coupling it to a CPU, we can offload some of the more mundane parts of a calculation to our well developed CPUs and utilize our QPUs only where it counts. This was the idea when we developed the original VQE (Variational Quantum Eigensolver) algorithm and its implementation on a photonic quantum computer. Since that time, I’ve given a number of public talks on the topic, including two that can be found online here and here. The algorithm was then developed further including some special enhancements for ion traps and picked up the attention of some other theory groups who worked to optimize it further.
Given the excitement around this algorithm as a potential candidate for a first algorithm to become competitive with a classical computer at scale, we wanted to expand a bit more on our views of the theory and purpose of this algorithm, as well as offer some computational enhancements. We did just this in our new paper entitled “The theory of variational hybrid quantum-classical algorithms”, and we hope you enjoy reading it as much as we enjoyed writing it.