Optimization and realization of boson sampling for true random number generation using the Xanadu X8
Joshua Ange, Mitchell Aaron Thornton · 2025
Random number generators (RNGs) are critical for problems involving security and cryptography. Much work has been done with true random number generators (TRNGs) that utilize quantum mechanical phenomena as sources of entropy to generate numbers according to some distribution without any underlying deterministic component. While qubit-based quantum computers can serve as high-quality sources of randomness, their effectiveness is limited by constraints due to their physical construction (e.g. not being feasible at room temperatures) and by the number of qubits that are necessary for fine-grained distributions (the number of bins scale as 2n for a n-qubit implementation). Continuous variable photonic quantum computers offer promising scalability, both from their experimental realization and from the use of linear optical components and photon number resolving detectors to generate varied distributions. We demonstrate that boson-sampling-based approaches realized with photonic quantum computers like the Xanadu X8 machine can be used to generate random numbers in diffuse, hard-to-predict distributions and they can be optimized for transformation to a uniform distribution. Specifically, boson sampling can act as a high-quality source of entropy and we introduce a strategy for scaling our TRNG with further experimental development (both in terms of the number of qumodes available and improvements for photonic components). We compare our results to the performance of qubit-based TRNGs and discuss experimental realizations.