I also suspect it comes from some kind of rounding or precision issue, likely in MPFR since nearly all of the computations are done there. At least this is not super pressing since the Bober implementation is not the default implementation, but it is definitely something that should be fixed.
Best, Travis On Sunday, April 26, 2020 at 6:03:30 AM UTC+10, Sebastian Oehms wrote: > > see also #28549 <https://trac.sagemath.org/ticket/28549>. > > On Monday, April 20, 2020 at 9:23:52 AM UTC+2, Sebastian Oehms wrote: >> >> The existenz of the problem already appeared in Friedrich Wiemer's >> status report >> <https://groups.google.com/forum/#!searchin/sage-devel/Windows$20Subsystem/sage-devel/WoEWxYmwQnI/-LkPAc8VBAAJ> >> >> on Sage under Windows 10 using the Windows Subsystem for Linux (WSL). >> >> Now, that I have my office equipment (including a Windows 10 machine) at >> home, I took the occasion to install und play with WSL (using the Ubuntu >> App). I build Sage version 9.1.rc0 and run all doctests (like Friedrich >> did). The list of failing doctests (see the attached file) is not far from >> the corresponding list in Friedrich's log-file. Some issues disappeared >> while others remained and new ones have been added. But, the first issue in >> both lists is remarkable: >> >> *On WSL:* >> >> sage: number_of_partitions(241, algorithm='bober') >> 114540884553039 >> sage: number_of_partitions(241) >> 114540884553038 >> >> >> *whereas on proper Ubuntu (same Sage version) or Cygwin on the same >> machine (but stable 9.0):* >> >> sage: number_of_partitions(241, algorithm='bober') >> 114540884553038 >> sage: number_of_partitions(241) >> 114540884553038 >> >> I think, that is an issue which can damage trust even thought it probably >> is not caused by Sage. Shouldn't we trace that back? To me it looks like a >> rounding problem via MPFR. >> >> -- 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 sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/f1446d0f-bf01-40c2-8875-17f50e185e86%40googlegroups.com.
