%0 Journal Article
%T Cholesky分解的单Pass随机算法<br>Single-Pass Randomized Algorithm for Cholesky Decomposition
%A 刘雪翠
%J Pure Mathematics
%P 87-93
%@ 2160-7605
%D 2024
%I Hans Publishing
%R 10.12677/PM.2024.141010
%X 对于大规模数据矩阵，数据的读取成本远高于算法本身的成本；对存储在磁盘外部的流型数据，往往只有一次读取数据的机会。而以往的Cholesky分解的随机算法都至少需要读取输入数据两次，难以满足实际应用中的低成本需求。本文基于矩阵分解的单pass随机算法研究，提出了Cholesky分解的单pass随机算法，并给出了该算法的误差上界，最后通过数值实验验证了该算法的可行性以及有效性。<br />
For large-scale data matrices, the cost of data reading is much higher than the cost of algorithms. For streaming data stored outside the disk, there is only one chance to read the data. In the past, the randomized algorithm of Cholesky decomposition needed to read the original data at least twice, which was difficult to meet the low-cost requirements in practical applications. Based on the research of single-pass randomized algorithm with matrix decomposition, a single-pass randomized algorithm based on Cholesky decomposition is proposed, and the error upper bound of the algorithm is given, and finally the feasibility and effectiveness of the algorithm are verified by numerical experiments.
%K Cholesky分解，矩阵分解，单Pass随机算法<br>Cholesky Decomposition
%K Matrix Factorization
%K Single-Pass Randomized Algorithm
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=79465