Improving Quantum Approximate Optimization by Noise-Directed Adaptive Remapping

We present Noise-Directed Adaptive Remapping (NDAR), a heuristic meta-algorithm for approximately solving binary optimization problems by leveraging certain types of noise. We consider access to a noisy quantum processor with dynamics that features a global attractor state. In a standard setting, such noise can be detrimental to the quantum optimization performance. In NDAR, the algorithm bootstraps the attractor by iteratively gauge-transforming the cost-function Hamiltonian. In each iteration step, the gauge transformation effectively changes the attractor state into a higher-quality solution of the cost Hamiltonian based on the results of variational optimization in the previous step. The end result is that noise aids variational optimization, as opposed to hindering it. We demonstrate the effectiveness of our protocol in Quantum Approximate Optimization Algorithm experiments with subsystems of the newest generation of Rigetti Computing's superconducting device Ankaa-2. We obtain approximation ratios (of best-found solutions) 0. 9-0. 96 (spread across instances) for multiple instances of random, fully connected graphs (Sherrington-Kirkpatrick model) on n=82 qubits, using only depth p=1 (noisy) QAOA in conjunction with NDAR. This compares to 0. 34-0. 51 for vanilla p=1 QAOA with the same number of function calls.