I'm working with MTMs "svd" command. I encountered the following error message
Error, (in LinearAlgebra:-SingularValues) 8 off-diagonals failed to converge
The program consists of
with(MTM,svd);
Digits := 450;
M:=myFunctionToGenerateMatrix();
U,S,V:=svd(M);
It works fine for a precision of 400 Digits. I previously wrote the same program in Matlab and it worked fine. Depending on the matrix, the lowest precision at which the error appears varies, but I haven't found a matrix where it worked for any precision. I tried singular and non-singular matrices. This is surprising to me since the SVD-algorithm should be stable. Any ideas are welcome.
I'm particularly looking for answers regarding: Why does this error appear (I haven't found it anywhere, neither in the maple documentation, nor in the web)? Are there any alternatives, either to Digits or to MTMs svd? Are there any obvious mistakes?