A QAOA approach with fake devices: A case study for the maximum cut in ring graphs

Abstract

The quantum approximate optimization algorithm (QAOA) can require considerable processing time for developers to test and debug their codes on expensive quantum devices. One avenue to circumvent this difficulty is to use the error maps of quantum devices, where a local simulator can be automatically configured to mimic an actual device backend. In our work, we evaluated some error maps of quantum devices, known as fake devices, that are freely available in the cloud. The QAOA and the problem of maximum cut in 2-regular connected graphs, known as ring of disagrees, were used as tools for the noise analysis. The approximation ratio, the expectation energy and the probability of success for this problem have been evaluated in two scenarios. First, the quantities were studied through noisy simulations using fake devices. Second, error mitigation methods such as optimization levels and translation (connectivity mapping) of the original problem were applied. These results were then compared with the analytical solution of the ring graph. The study shows that error mitigation methods were crucial in obtaining better results for the expectation value of the energy, the approximation ratio, and the probability of success for the ring graphs.