Arai Shunta, Kadowaki Tadashi
Institute of Science Tokyo, Ookayama, 152-8550, Tokyo, Japan.
Global R&D Center for Business by Quantum-AI Technology, National Institute of Advanced Industrial Science and Technology, Ibaraki, Japan.
Sci Rep. 2025 Jul 1;15(1):21427. doi: 10.1038/s41598-025-07293-y.
In this study, we propose quantum annealing-enhanced Markov Chain Monte Carlo (QAEMCMC), where QA is integrated into the MCMC subroutine. QA efficiently explores low-energy configurations and overcomes local minima, enabling the generation of proposal states with a high acceptance probability. We benchmark QAEMCMC for the Sherrington-Kirkpatrick model and demonstrate its superior performance over the classical MCMC method. Our results reveal larger spectral gaps, faster convergence of energy observables, and reduced total variation distance between the empirical and target distributions. QAEMCMC accelerates MCMC and provides an efficient method for complex systems, paving the way for scalable quantum-assisted sampling strategies.
在本研究中，我们提出了量子退火增强的马尔可夫链蒙特卡罗方法（QAEMCMC），即将量子退火（QA）集成到马尔可夫链蒙特卡罗（MCMC）子例程中。量子退火能够有效地探索低能量构型并克服局部最小值，从而能够生成具有高接受概率的提议状态。我们对Sherrington-Kirkpatrick模型的QAEMCMC进行了基准测试，并证明了它相对于经典MCMC方法的优越性能。我们的结果显示出更大的谱隙、能量可观测量更快的收敛速度以及经验分布与目标分布之间总变差距离的减小。QAEMCMC加速了MCMC，并为复杂系统提供了一种有效的方法，为可扩展的量子辅助采样策略铺平了道路。
