含噪声量子计算机上的可扩展核壳模型计算
Scalable nuclear shell model calculations on noisy quantum computers
浏览论文内容
中文总结 AI 辅助
本研究将SQD框架首次应用于核壳模型，结合云连接的NISQ硬件与经典HPC集群，可在内存和执行时间上优于传统超级计算机，实现大规模核壳模型计算。
中文摘要 AI 辅助
核壳模型的精确对角化呈指数级缩放，在经典高性能计算（HPC）中会导致严重的内存瓶颈。尽管变分量子本征求解器（VQE）等混合量子算法旨在克服这些限制，但其深度量子电路和迭代反馈回路易受当前含噪声中等规模量子（NISQ）硬件固有噪声的显著影响。尽管有复杂的噪声缓解技术，这种噪声仍使VQE等多种算法无法用于大规模计算。作为一种能容忍这些问题的实用方法，我们首次将基于采样的量子对角化（SQD）框架应用于核壳模型。以$^{38}\text{Ar}$为基准验证数值精度后，我们将SQD扩展到$^{32}\text{Mg}$，求解了在给定HPC系统中无法用传统经典方法直接对角化其底层希尔伯特空间的核壳模型哈密顿量。我们对SQD与标准变分量子方案及精确经典求解器进行了系统比较。通过利用经云连接到经典HPC集群的NISQ硬件，基于SQD的方案在内存缩放和总执行时间方面可优于传统超级计算机，从而实现更严格的大规模壳模型计算。
英文摘要
The exact diagonalization of the nuclear shell model scales exponentially, leading to severe memory bottlenecks in classical high-performance computing (HPC). While hybrid quantum algorithms like the Variational Quantum Eigensolver (VQE) aim to overcome these limits, their deep quantum circuits and iterative feedback loops are susceptible to substantial noise inherent in the current Noisy Intermediate-Scale Quantum (NISQ) hardware. This noise renders several algorithms, such as the VQE, impractical for large-scale calculations despite sophisticated noise-mitigation techniques. As a pragmatic approach tolerant to these issues, we apply the Sample-based Quantum Diagonalization (SQD) framework to nuclear shell models for the first time. Using $^{38}\text{Ar}$ as a benchmark to confirm the numerical accuracy, we extend SQD to $^{32}\text{Mg}$, solving a nuclear shell-model Hamiltonian whose underlying Hilbert space cannot be directly diagonalized using conventional classical methods in a given HPC system. We present a systematic comparison of SQD with standard variational quantum schemes and exact classical solvers. By leveraging NISQ hardware connected via the cloud to classical HPC clusters, the SQD-based scheme could outperform conventional supercomputers in memory scaling and total execution time, enabling more rigorous large-scale shell model calculations.
