At Tue, 12 Sep 2006 13:30:41 -0600, Patrick Alken wrote: > I believe there typos in eigen/test.c. On line 247 (in the function > test_eigen_symm()) it sorts the eigenvalues with VAL_DESC a second > time - I think this should be ABS_DESC instead. The same thing > occurs in test_eigen_herm().
Thanks you're right, I will fix that. I guess we also need to extend the tests to have some negative eigenvalues so that they cover the different cases. -- Brian Gough
