How many trials do we need for reliable NISQ computing?

Gate-based quantum computing is an attractive candidate in the post-Moore era. Noisy intermediate-scale quantum (NISQ) computers are expected to be available in the next few years. It is required to repeatedly execute the target quantum application for reliable NISQ computing, e.g., users can set 1,024 as a repetition parameter in the IBM-Q machine, because NISQ computers output follows the probability distribution of execution trials. Since the distribution depends strongly on the effects of noise, it is difficult to determine a sufficient number of repetitions. This paper proposes a novel statistical approach for efficient NISQ computing. The key idea is to introduce a Bayesian credible interval model to obtain convergence of the probability distributions. We demonstrate that our execution method can detect all significant output values, that occur more often than the random situation (probability is 1/2n), using a NISQ simulator.