Quantum-Enhanced Optimization by Warm Starts

Abstract

We present an approach, which we term quantum-enhanced optimization, to accelerate classical optimization algorithms by leveraging quantum sampling. Our method uses quantum-generated samples as warm starts to classical heuristics for solving challenging combinatorial problems like Max-Cut and Maximum Independent Set (MIS). To implement the method efficiently, we introduce novel parameter-setting strategies for the Quantum Approximate Optimization Algorithm (QAOA), qubit mapping and routing techniques to reduce gate counts, and error-mitigation techniques. Experimental results, including on quantum hardware, showcase runtime improvements compared with the original classical algorithms.