報告題目:Randomized Block Kaczmarz Methods for Inner Inverses of a Matrix
報告專家:李維國教授,中國石油大學(華東)
報告地點:騰訊會議(會議號328-171-9712;密碼314159)
報告時間:2023年12月27日, 19:30 — 21:30
報告摘要:In this talk, some randomized block Kaczmarz algorithms to compute inner inverses of any rectangle matrix are presented, where A is full rank or rank deficient. Convergence results are provided to guarantee the convergence of the proposed methods theoretically. Numerical examples are given to illustrate the effectiveness. Since the proposed algorithms only require one row of A at each iteration without matrix-matrix product, it is suitable for the scenarios where the matrix A is too large to fit in memory or matrix multiplication is considerably expensive. In addition, if the MII-RABK method is implemented in parallel, the running time will be greatly reduced. Therefore, in some practical applications, the MII-RABK method is more feasible when the Moore-Penrose generalized inverse is unknown or the calculation is too expensive.
報告人簡介:李維國,博士,教授,博士生導師。中國石油大學(華東)校學術委員會委員。1997年博士畢業于南京大學數學系,2000年被評為中國石油大學(華東)教授,至今在中國石油大學(華東)理學院工作。發表科研論文100余篇,主持、參與國家自然科學基金6項,教育部和山東省基金3項。主要從事圖像處理及分析中的數值計算問題及其應用、廣義逆計算、微分方程周期解的數值計算、線性和非線性方程組的數值解等方面的研究工作。講授本科生、研究生課程10余門,出版教材3部。
作者:吳念慈;編輯;劉鹍;審核:郭暉;發布:郭敏。