Description
Quantum random numbers based on quantum origins possess unpredictability and infinite periodicity, achieving information-theoretic absolute security. In our work, we employ a DOPO system based on a warm atomic vapor cell, where the randomness originates from the quantum collapse of polarization states. Through polarization self-rotation and cavity mode selection, the system exhibits robustness against fluctuations in laser intensity and frequency detuning. Traditional methods utilizing photon vacuum fluctuations or laser phase fluctuations suffer from poor entropy source randomness and require extensive post-processing, while fiber-based DOPO systems are limited by cavity linewidth and long photon lifetime, resulting in low generation rates. In contrast, atomic motion-induced coherence in our system helps narrow the linewidth and can enhance the random number generation rate. With the aid of machine learning, long-term stable unbiased randomness is achieved, passing all NIST tests with only weak post-processing. Furthermore, the sensitivity of randomness to magnetic fields near the unbiased point holds promise for applications in precision measurement and multi-parameter sensing. Looking forward, the programmable flexibility of spatial light modulators (SLM) can be exploited to flexibly tailor optical patterns, enabling parallel quantum random number sources that are highly promising for all-optical computing and parallel acceleration architectures.
