Parallel circuit implementation of variational quantum algorithms

TL;DR

This paper introduces a parallel circuit method for variational quantum algorithms that enables larger problem solving and efficient training by exploiting problem structure and circuit slicing, with promising results for optimization tasks.

Contribution

It proposes a novel circuit slicing technique for VQAs, allowing parallel execution and training, and demonstrates its effectiveness on combinatorial optimization problems.

Findings

Enables solving larger problems with fewer qubits.

Allows full VQA training using only one circuit slice.

Redundancies in quantum circuits can be exploited for efficiency.

Abstract

We present a method to split quantum circuits of variational quantum algorithms (VQAs) to allow for parallel training and execution, that maximally exploits the limited number of qubits in hardware to solve large problem instances. We apply this specifically to combinatorial optimization problems, where inherent structures from the problem can be identified, thus directly informing how to create these parallelized quantum circuits, which we call slices. We test our method by creating a parallelized version of the Quantum Approximate Optimization Algorithm, which we call pQAOA, and explain how our methods apply to other quantum algorithms like the Variational Quantum Eigensolver and quantum annealing. We show that not only can our method address larger problems, but that it is also possible to run full VQA models while training parameters using only one slice. These results show that the loss of information induced by splitting does not necessarily affect the training of parameters in quantum circuits for optimization. This implies that combinatorial optimization problems are encoded with redundant information in quantum circuits of current VQAs. Therefore, to attain quantum advantage for combinatorial optimization, future quantum algorithms should be designed to incorporate information that is free of such redundancies.