1
.TH im_inv 1 "April 1993" "Scilab Group" "Scilab Function"
7
[X,dim]=im_inv(A,B [,tol])
8
[X,dim,Y]=im_inv(A,B, [,tol])
13
: two real or complex matrices with equal number of columns
16
: orthogonal or unitary square matrix of order equal to the
17
number of columns of \fVA\fR
20
: integer (dimension of subspace)
23
: orthogonal matrix of order equal to the number of rows
24
of \fVA\fR and \fVB\fR.
26
.Vb [X,dim]=im_inv(A,B)
32
i.e vectors whose image through \fVA\fR are in range(\fVB\fR)
34
The \fVdim\fR first columns of \fVX\fR span
38
.LA $ A^{-1}(\cal B).$
40
\fVtol\fR is a threshold used to test if subspace inclusion;
41
default value is \fVtol = 100*%eps\fR.
42
If \fVY\fR is returned, then \fV[Y*A*X,Y*B]\fR is partitioned as follows:
44
[A11,A12;0,A22],[B1;0]
46
where \fVB1\fR has full row rank (equals \fVrank(B)\fR) and \fVA22\fR has
47
full column rank and has \fVdim\fR columns.
50
A=[rand(2,5);[zeros(3,4),rand(3,1)]];B=[[1,1;1,1];zeros(3,2)];
51
W=rand(5,5);A=W*A;B=W*B;
53
svd([A*X(:,1:dim),B]) //vectors A*X(:,1:dim) belong to range(B)
54
[X,dim,Y]=im_inv(A,B);[Y*A*X,Y*B]
57
rowcomp, spaninter, spanplus, linsolve