# generate multi Diagonal matrices

3 views (last 30 days)
Seahawks on 10 Nov 2020
Edited: Setsuna Yuuki. on 10 Nov 2020
I need help. I have beable to generate 100 of random matrices H. then try to get S diagonal matrix but it is not succesful as both s11=0, s22=0, however s33 value is correct.
N = 100;
for j=1:length(N)
H = 0.5.*(randn(3,3, N) +(1i) * randn(3,3, N));
for k=1:length(H)
[S(3,3,k)]= svd(H(3,3,k));
%[U(3,3,k),S(3,3,k),V(3,3,k)]= svd(matrices(3,3,k));
end
end
H(:,:,92) =
-0.3382 + 0.3345i 0.1304 + 0.0842i -0.4230 + 0.3986i
0.0729 - 0.5495i -0.3290 - 0.2820i -0.2131 - 0.1437i
0.3096 + 0.0466i 0.0022 - 0.3723i -0.2904 - 0.0265i
S(:,:,92) =
0 0 0
0 0 0
0 0 0.2916
Thanks

Setsuna Yuuki. on 10 Nov 2020
In your code, you are only calculating the svd of the column 3 and row 3.
N = 100;
for j=1:length(N)
H = 0.5.*(randn(3,3, N) +(1i) * randn(3,3, N));
for k=1:length(H)
[S(1,1,k)]= svd(H(1,1,k)); % row 1 column 1
[S(2,2,k)]= svd(H(2,2,k)); % row 2 column 2
[S(3,3,k)]= svd(H(3,3,k)); % row 3 column 3
%[U(3,3,k),S(3,3,k),V(3,3,k)]= svd(matrices(3,3,k));
end
end
##### 2 CommentsShowHide 1 older comment
Setsuna Yuuki. on 10 Nov 2020
Edited: Setsuna Yuuki. on 10 Nov 2020 Christine Tobler on 10 Nov 2020
The SVD is usually computed for a matrix, but you're only passing in a scalar on each call.
Is it possible you meant to pass in a page H(:, :, k) (which would be a 3-by-3 matrix) in every iteration?