Title:Extension of a Pattern Recognition Validation Approach for Boson Sampling

Abstract:Boson sampling is one of the main quantum computation models to demonstrate the quantum computational advantage. However, this aim may be hard to realize considering noise sources such as photon distinguishability. Inspired by the Bayesian validation extended to evaluate whether photon distinguishability is too high to demonstrate this advantage, the pattern recognition validation is extended for boson sampling. Based on clusters constructed with the K means++ method, the distribution of characteristic values is nearly monotonically changed with the photon indistinguishability, especially when photons are close to be indistinguishable. We analyze the intrinsic data structure through calculating probability distributions and mean 2-norm distances of the sorted outputs. An approximation algorithm is also used to show the regular data structure changes with photon distinguishability.