计算加权Moore Penrose逆A_(M ,N)^+的一个改进的并行算法(英文)

An Improved Parallel Algorithm for Computing the Weighted Moore-Penrose Inverse A_(M ,N)^+

利用PREPARTAFP和SARWATEDV[6] 的一些结果 ,给出一个计算加权Moore Penrose逆A+MN 的改进的并行算法 ,改善了文献 [8]中提出的算法。在与PREPARTAFP和SARWATEDV文 [6 ]中相同的假设下证明了改进的并行算法的时间复杂性和处理机台数分别为 T =0 ((logn) 2 ) , P =max m/n 2 nα/logn ,2r1 / 2 nα(logrlogn)时空积 (成本最优性 ) T× P小于T×P(T和P分别为 [8]中原有并行算法的时间复杂性和处理机台数 )。

HT5”SS] We give an improved parallel algorithm for computing the weighted Moore Penrose inverse A + M,N  by using some results of PREPARTATA F P and SARWATE D V [6] and modifying the algorithm of WANG Guo rong and LU Sen quan [8] . We show that with the same assumption as that given by PPEPARATA F P and SARWATE D V, the time complexity and the number of processors following the improved parallel algorithm are. =0(( log n) 2) and = max ?m/n? 2n α/ log n, 2r 1/2 n α log r log n respectively. The cost optimality * is less than P*T (T and P are the time compleity and the number of processors accompanying the origimal parallel algorithm).