[Octave-bug-tracker] [bug #29487] Feature request: svd with 3 output variables should use dgesdd

[Jaroslav Hajek]

Update of bug #29487 (project octave):

                  Status:               Need Info => Fixed                  

    _______________________________________________________

Follow-up Comment #12:

I added svd_driver to Octave. Now I'm getting this:

octave:1> a = randn (1000);
octave:2> svd (1);
octave:3> tic; [u1, s1, v1] = svd (a); toc
Elapsed time is 10.1431 seconds.
octave:4> svd_driver ("gesdd")
ans = gesvd
octave:5> tic; [u2, s2, v2] = svd (a); toc
Elapsed time is 1.70709 seconds.
octave:6> norm (u1*s1*v1' - a, "fro")
ans =  9.7938e-12
octave:7> norm (u2*s2*v2' - a, "fro")
ans =  3.8047e-12

So it's true that gesdd can be much faster. Both drivers seem to be
comparably accurate on normal data, though the singular vectors differ
significantly due to lots of close singular values:

octave:8> norm (u1 - u2, "fro")
ans =  46.562

The default is still gesvd. Add svd_driver ("gesdd") in your .octaverc to
enjoy the speed-up. Beware of possible problems for highly ill-conditioned,
though I failed to produce a matrix that demonstrates this. 

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?29487>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/