On 6/29/06, Karl Glazebrook <[EMAIL PROTECTED]> wrote:
Karl,
I created a patch to the matrixops.t test to do precisely what you are saying. That patch is attached to the bug report. I only test one matrix.
If someone gives me cvs write privs, I'll check it in. I can also come up with more examples.
Craig,
My suggestion was to put a test of symmetry in front of the call to eigens, and if it is symmetric, use the old eigens function. That way we don't loose any functionality and possibly gain some working non-symmetric eigens.
-Bill
Bill could you write a test script to check in to the tests
directory? They are quite straight forward and it would be good to
give MatrixOps a work out. You already had something like that in
your report.. Also assign the bug to Doug on sourceforge.
Karl,
I created a patch to the matrixops.t test to do precisely what you are saying. That patch is attached to the bug report. I only test one matrix.
If someone gives me cvs write privs, I'll check it in. I can also come up with more examples.
Craig,
My suggestion was to put a test of symmetry in front of the call to eigens, and if it is symmetric, use the old eigens function. That way we don't loose any functionality and possibly gain some working non-symmetric eigens.
_______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
