English

Fast and Practical Quantum-Inspired Classical Algorithms for Solving Linear Systems

Data Structures and Algorithms 2023-12-01 v2 Computational Complexity Numerical Analysis Numerical Analysis Quantum Physics

Abstract

We propose fast and practical quantum-inspired classical algorithms for solving linear systems. Specifically, given sampling and query access to a matrix ARm×nA\in\mathbb{R}^{m\times n} and a vector bRmb\in\mathbb{R}^m, we propose classical algorithms that produce a data structure for the solution xRnx\in\mathbb{R}^{n} of the linear system Ax=bAx=b with the ability to sample and query its entries. The resulting xx satisfies xA+bϵA+b\|x-A^{+}b\|\leq\epsilon\|A^{+}b\|, where \|\cdot\| is the spectral norm and A+A^+ is the Moore-Penrose inverse of AA. Our algorithm has time complexity O~(κF4/κϵ2)\widetilde{O}(\kappa_F^4/\kappa\epsilon^2) in the general case, where κF=AFA+\kappa_{F} =\|A\|_F\|A^+\| and κ=AA+\kappa=\|A\|\|A^+\| are condition numbers. Compared to the prior state-of-the-art result [Shao and Montanaro, arXiv:2103.10309v2], our algorithm achieves a polynomial speedup in condition numbers. When AA is ss-sparse, our algorithm has complexity O~(sκlog(1/ϵ))\widetilde{O}(s \kappa\log(1/\epsilon)), matching the quantum lower bound for solving linear systems in κ\kappa and 1/ϵ1/\epsilon up to poly-logarithmic factors [Harrow and Kothari]. When AA is ss-sparse and symmetric positive-definite, our algorithm has complexity O~(sκlog(1/ϵ))\widetilde{O}(s\sqrt{\kappa}\log(1/\epsilon)). Technically, our main contribution is the application of the heavy ball momentum method to quantum-inspired classical algorithms for solving linear systems, where we propose two new methods with speedups: quantum-inspired Kaczmarz method with momentum and quantum-inspired coordinate descent method with momentum. Their analysis exploits careful decomposition of the momentum transition matrix and the application of novel spectral norm concentration bounds for independent random matrices. Finally, we also conduct numerical experiments for our algorithms on both synthetic and real-world datasets, and the experimental results support our theoretical claims.

Keywords

Cite

@article{arxiv.2307.06627,
  title  = {Fast and Practical Quantum-Inspired Classical Algorithms for Solving Linear Systems},
  author = {Qian Zuo and Tongyang Li},
  journal= {arXiv preprint arXiv:2307.06627},
  year   = {2023}
}

Comments

Theorem 3 and Theorem 5 are incorrect, and more efforts are needed to fix existing issues

R2 v1 2026-06-28T11:29:13.027Z