Improved Qubit Routing for QAOA Circuits

TL;DR

This paper introduces a polynomial-time qubit routing algorithm for QAOA circuits that optimizes circuit depth and SWAP gate count, outperforming existing methods on various graph types and sizes.

Contribution

It presents a novel, efficient qubit routing algorithm based on edge coloring and greedy SWAP application, improving QAOA circuit compilation.

Findings

Outperforms existing routing algorithms on $k$-regular and Erdős-Rényi graphs.

Effective for problem sizes up to N=400.

Balances circuit depth and SWAP gate count.

Abstract

We develop a qubit routing algorithm with polynomial classical run time for the Quantum Approximate Optimization Algorithm (QAOA). The algorithm follows a two step process. First, it obtains a near-optimal solution, based on Vizing's theorem for the edge coloring problem, consisting of subsets of the interaction gates that can be executed in parallel on a fully parallelized all-to-all connected QPU. Second, it proceeds with greedy application of SWAP gates based on their net effect on the distance of remaining interaction gates on a specific hardware connectivity graph. Our algorithm strikes a balance between optimizing for both the circuit depth and total SWAP gate count. We show that it improves upon existing state-of-the-art routing algorithms for QAOA circuits defined on $k$-regular as well as Erd\"os-Renyi problem graphs of sizes up to $N \leq 400$.