A Pure Quantum Approximate Optimization Algorithm Based on CNR Operation

Abstract: By introducing the "comparison and replacement" (CNR) operation, we propose a general-purpose pure quantum approximate optimization algorithm and derive its core optimization mechanism quantitatively. The algorithm is constructed to a $p$-level divide-and-conquer structure based on the CNR operations. The quality of approximate optimization improves with the increase of $p$. For sufficiently general optimization problems, the algorithm can work and produce the near-optimal solutions as expected with considerably high probability. Moreover, we demonstrate that the algorithm is scalable to be applied to large size problems. Our algorithm is applied to two optimization problems with significantly different degeneracy, the Gaussian weighted 2-edge graph and MAX-2-XOR, and then we show the algorithm performance in detail when the required qubits number of the two optimization problems is 10.