[Octave-bug-tracker] [bug #54383] accuracy issues with betainc

[Rik]

Update of bug #54383 (project octave):

                  Status:                    None => In Progress            

    _______________________________________________________

Follow-up Comment #1:

The tolerances were chosen based on the most restrictive values that still
passed on Michele Ginesi's computer.  Some of the tolerances were relaxed in
this cset (https://hg.savannah.gnu.org/hgweb/octave/rev/f80e68529bbf) in order
to have the test suite pass on other i386 platforms.  Mike was the author of
that cset.  I think it would probably be fine to just increase the tolerance
for this single test.  The two tests are


%!assert <54383> (betainc (0.99, 20, 30, "upper"), 1.5671643161872703e-47,
-5e-14);
%!assert (betainc (0.999, 20, 30, "upper"), 1.850806276141535e-77, -7e-14);


and the second one already has a larger tolerance.


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?54383>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/