NISQ computing for decision making under uncertainty

Abstract

Noisy Intermediate-Scale Quantum computers are expected to be available this year. It is proposed to exploit such a device for decision making under uncertainty. The probabilistic character of quantum mechanics reflects this uncertainty. Concomitantly, the noise may add to it. The approach is standard in the sense that Bayes decision rule is used to decide on the basis of maximum expected reward. The novelty is to model the various action profiles and the development of ‘nature’ as unitary transformations on a set of qubits. Measurement eventually yields samples of classical binary random variables in which the reward function has to be expressed. In order to achieve sufficiently low variances for reliable decision making more runs of such a quantum algorithm are necessary. Some simple examples have been worked out to elucidate the idea. Here the calculations are still analytically feasible. Presently lacking an operating quantum device, the QX simulator of Quantum Inspire has been used to generate the necessary samples for comparison and demonstration. First obtained results are promising and point at a possible useful application for noisy intermediate-scale quantum computers.