Well my personal take on it is that the ATLAS build had a small issue. In the context of sage-on-gentoo i have seen particular version of Atlas or other cblas break only one doctest as an indication that there was a fault or a difference with the atlas sage usually ship. Early version of atlas 3.9 were breaking one test linked to iml. Volker has a problem with linbox in his early 3.9 spkg. I use openblas and it breaks a single doctest with an error of the order of 1e-15 (ok cython.py is also broken in two places but that's not numerics, it is a library list).
I am not quite what you could do to solve this. A good starting point would be accessing the machine where those atlas were produced. Francois On 26/01/2013, at 12:55, "Ethan Van Andel" <[email protected]<mailto:[email protected]>> wrote: I do not have any easy way to run that build. It certainly doesn't run on my machine (12.10 x86). Is there a straightforward way to run it that I'm unaware of? Also, can you clarify for me: does this build-related error appear on lots of machines (i.e. have other people encountered it), or only on your specific machine? One possible cause comes to mind. All of the riemann computations involve solving fairly large (100x100 to 1000x1000) complex-valued matrix equations. Are there any other packages in sage that do computations of this sort as part of the doctests? If there aren't many it's conceivable that the build error is related to that solve routine and only the riemann doctests happen to catch it. The riemann package really doesn't do anything particularly strange as it relates to ATLAS. It only makes one complex valued matrix solve call, and that is made through standard channels and the matrix shouldn't be particularly pathological. Ethan On Tuesday, January 22, 2013 11:35:18 AM UTC-8, Jeroen Demeyer wrote: On 2013-01-22 20:13, Ethan Van Andel wrote: > Has anyone else encountered the numerical errors? I'm happy to dig in > and try to figure out what's happening, but if the failed build errors > can't be duplicated (or even if I don't have a way to duplicate them), > there's not much I can do. Here is a binary of a bad build: http://boxen.math.washington.edu/home/buildbot/sage-5.6.rc1-bad-riemann.tar.bz2 This was made on sage.math (8.04.4 LTS x86_64). -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]<mailto:[email protected]>. To post to this group, send email to [email protected]<mailto:[email protected]>. Visit this group at http://groups.google.com/group/sage-devel?hl=en. This email may be confidential and subject to legal privilege, it may not reflect the views of the University of Canterbury, and it is not guaranteed to be virus free. If you are not an intended recipient, please notify the sender immediately and erase all copies of the message and any attachments. Please refer to http://www.canterbury.ac.nz/emaildisclaimer for more information. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/sage-devel?hl=en.
