First,an explicit representation A(2)T,S=(GA+E)^-1G of the outer invers A(2)T,S for a matrix A∈Cm×n with the prescribed range T and null space S is derived,which is simpler than A(2)T,S=(GA+E)^-1G-V(UV)-2UG prop...First,an explicit representation A(2)T,S=(GA+E)^-1G of the outer invers A(2)T,S for a matrix A∈Cm×n with the prescribed range T and null space S is derived,which is simpler than A(2)T,S=(GA+E)^-1G-V(UV)-2UG proposed by Ji in 2005.Next,a new algorithm for computing the outer inverse A(2)T,S based on the improved representation A(2)T,S=(GA+E)^-1G through elementary operations on an appropriate partitioned matrix GAInIn0 is proposed and investigated.Then,the computational complexity of the introduced algorithm is also analyzed in detail.Finally,two numerical examples are shown to illustrate that this method is correct.展开更多
基金The National Natural Science Foundation of China(No.11771076).
文摘First,an explicit representation A(2)T,S=(GA+E)^-1G of the outer invers A(2)T,S for a matrix A∈Cm×n with the prescribed range T and null space S is derived,which is simpler than A(2)T,S=(GA+E)^-1G-V(UV)-2UG proposed by Ji in 2005.Next,a new algorithm for computing the outer inverse A(2)T,S based on the improved representation A(2)T,S=(GA+E)^-1G through elementary operations on an appropriate partitioned matrix GAInIn0 is proposed and investigated.Then,the computational complexity of the introduced algorithm is also analyzed in detail.Finally,two numerical examples are shown to illustrate that this method is correct.
