一种用于求解高度超定线性系统的计数草图随机平均块Kaczmarz方法  

A Count Sketch Randomized Average Block Kaczmarz Method for Solving Highly Overdetermined Linear Systems

作　　者：姜昊辰 

机构地区：[1]广东工业大学数学与统计学院,广东 广州

出　　处：《应用数学进展》2025年第2期244-250,共7页Advances in Applied Mathematics

摘　　要：受计数草图对Kaczmarz方法加速效果的启发,文章将计数草图与随机平均块Kaczmarz方法相结合,得出了一种求解高度超定线性系统的新方法。为了验证新方法的可行性,进行了大量数值实验。数值实验表明,在相同精度下,新方法在计算时间上表现良好。Inspired by the acceleration effect of the count sketch on the Kaczmarz method, this paper combines the count sketch with the randomized average block Kaczmarz method to derive a new method for solving highly overdetermined linear systems. To verify the feasibility of the new method, a large number of numerical experiments were conducted. The numerical experiments show that, under the same precision, the new method performs well in terms of computing time.

关 键 词：计数草图 随机平均块Kaczmarz方法 高度超定线性系统 计算时间 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]