Title:Quantumness and Randomness of Quantum Random Number Generators via Photon Statistics

Abstract:Several quantum random number generator (QRNG) models have been proposed to produce random numbers, which, due to the quantum theory, are more secure than their classical counterparts. Many QRNG devices are commercially available as off-the-shelf black-box devices. Computationally, it is not possible to distinguish between a PRNG and a QRNG just by observing their outputs. The common practice for testing quantumness is a direct comparison between the mean and the variance of the experimental photon count. However, statistically, this is not a feasible solution either, because of finite sample size. A QRNG that uses single photons as the quantum source produces true quantum random numbers. Since single photons follow sub-Poissonian statistics, by determining the underlying distribution one can conclude whether a QRNG is truly quantum or not. In this work, we point out the limitations of existing methods of such a decision-making processes, and propose a more efficient two-fold statistical method, which can ensure whether an optical source is quantum or not up to a desired confidence level. Also, QRNGs can not produce true random numbers without deterministic classical post-processing. In this work, we also show that the two models of QRNGs, one producing random numbers from exponential distribution and the other from uniform distribution, become essentially similar under device noise. Detector outputs of both the above models can be tested to quantify the randomness coming from the quantum source of a QRNG, which in turn, dictates how much post-processing is required to produce good random numbers retaining quantumness. In this context, we also derive a relation when the underlying sampling distributions of the QRNGs will be $\epsilon$-random. Depending on this relation, a suitable post-processing algorithm can be chosen.