Sampling electronic structure QUBOs with Ocean and Mukai solvers

TL;DR

This paper compares two heuristic software tools, qbsolv and Mukai, for solving electronic structure problems on quantum annealers, demonstrating Mukai's superior performance in classical mode.

Contribution

It introduces a comparison between Ocean's qbsolv and QCI's Mukai solvers for electronic structure calculations, highlighting Mukai's improved efficiency.

Findings

Mukai outperforms qbsolv in all tested cases.

Both ground and excited state calculations favor Mukai.

Results encourage development of software for quantum annealer utilization.

Abstract

The most advanced D-Wave Advantage quantum annealer has 5000+ qubits, however, every qubit is connected to a small number of neighbors. As such, implementation of a fully-connected graph results in an order of magnitude reduction in qubit count. To compensate for the reduced number of qubits, one has to rely on special heuristic software such as qbsolv, the purpose of which is to decompose a large problem into smaller pieces that fit onto a quantum annealer. In this work, we compare the performance of two implementations of such software: the original open-source qbsolv which is a part of the D-Wave Ocean tools and a new Mukai QUBO solver from Quantum Computing Inc. (QCI). The comparison is done for solving the electronic structure problem and is implemented in a classical mode (Tabu search techniques). The Quantum Annealer Eigensolver is used to map the electronic structure eigenvalue-eigenvector equation to a type of problem solvable on modern quantum annealers. We find that the Mukai QUBO solver outperforms the Ocean qbsolv for all calculations done in the present work, both the ground and excited state calculations. This work stimulates the development of software to assist in the utilization of modern quantum annealers.