{"id":"arxiv/2610.39537","url":"https://arxiv.org/abs/2610.39537","tags":["math.NA"],"score":0,"title":"Efficient variance reduction for randomized Kaczmarz methods via residual-tracking control variates","domain":"math.NA","source":"arxiv","authors":["Jiaxin Xie","Hou-Duo Qi","Deren Han"],"reviews":0,"subject":"math.NA","summary":"The randomized Kaczmarz (RK) method is a widely used and effective class of iterative algorithms for solving large-scale linear systems. For inconsistent systems, however, the persistent variance caused by randomization prevents the method from converging exactly to the least-squares solution. In this paper, we introduce a residual-tracking control variate framework for reducing this variance while retaining the row-action structure and low per-iteration cost of RK. The resulting variance-reduced RK method uses constant step sizes, accesses only sampled rows of the coefficient matrix, and…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-29","reactions":{}}