[SciPy-Dev] optimize ncg approx_fhess_p bug?
Thu Apr 8 09:16:34 CDT 2010
On Thu, Apr 8, 2010 at 10:12 AM, <firstname.lastname@example.org> wrote:
> On Thu, Apr 8, 2010 at 7:17 AM, Ralf Gommers
> <email@example.com> wrote:
>> On Tue, Apr 6, 2010 at 10:56 PM, <firstname.lastname@example.org> wrote:
>>> On Tue, Apr 6, 2010 at 9:00 AM, <email@example.com> wrote:
>>> there is still something strange:
>>> I get the convergence failure when I use nosetests directly, but not
>>> when I run the testfile (which uses nosetests through numpy.testing)
>> I can't reproduce this on either XP or OS X. Alan was on Vista, what are you
> I'm also on XP.
> I wouldn't bother about this now if it cannot be consistently reproduced.
> I put a print statement into the function of the test and sometimes
> the arguments turn into nans, and in this case the optimization breaks
> down. I have no idea why this happens sometimes but not always, or why
> there should be an random element to the result.
> I think the default epsilon is too small for practical purposes, but
> that's a different issue, and the test seems to work correctly in
> general. However, I never get a failure with smaller epsilon.
with a larger epsilon (not smaller)
>> SciPy-Dev mailing list
More information about the SciPy-Dev