量子计算机观测：超越经典稀疏基态求解器的准确性
量子物理
2026-03-05 v1
摘要
我们实验性地展示了混合量子-经典算法可超越纯粹经典、即插即用选取配置相互作用方法。首先，我们构建了一类具有稀疏基态的局部哈密顿问题，表明经典启发式方法在特定 49 位量子比特实例中未能找到基态。随后，我们表明在 IBM Heron R3 处理器上运行的基于样本的 Krylov 量子对角化算法在同一任务中成功。该算法使用来自时间演化量子态网格的量子样本，为稀疏基态问题提供可证收敛的保证。虽然该问题也可通过我们专为针对该哈密顿构造而设计的两个迭代求解器在经典上求解，但本工作解决了样本基量子对角化算法是否能超越标准选取配置相互作用启发式方法的先前未决问题。
引用
@article{arxiv.2603.03496,
title = {Observation of Improved Accuracy over Classical Sparse Ground-State Solvers using a Quantum Computer},
author = {William Kirby and Bibek Pokharel and Javier Robledo Moreno and Kevin C. Smith and Sergey Bravyi and Abhinav Deshpande and Constantinos Evangelinos and Bryce Fuller and James R. Garrison and Ben Jaderberg and Caleb Johnson and Petar Jurcevic and Su-un Lee and Simon Martiel and Mario Motta and Seetharami Seelam and Oles Shtanko and Kevin J. Sung and Minh Tran and Vinay Tripathi and Kazuhiro Seki and Kazuya Shinjo and Han Xu and Lukas Broers and Tomonori Shirakawa and Seiji Yunoki and Kunal Sharma and Antonio Mezzacapo},
journal= {arXiv preprint arXiv:2603.03496},
year = {2026}
}
备注
40 pages, 22 figures, comments appreciated!