报告地点：行健楼学术活动室526
报告摘要：The Kaczmarz algorithm, proposed by the Polish mathematician Stefan Kaczmarz, is a simple iterative scheme with row projection for solving large linear systems. Since the second half of the 20th century, this method has been widely applied in many fields. In this report we present a block Kaczmarz method with oblique projection, which at each step, selects three rows of the coefficient matrix and projects the current iterate onto the solution space of the hyperplane formed by these target rows. Moreover, based on the greedy randomized Kaczmarz algorithm, we develop a greedy block randomized Kaczmarz method. Convergence analysis and numerical experiments show that these algorithms can be competitive with some state-of-the-art Kaczmarz-type methods in speed and efficiency.
