Re: [sage-release] Sage 4.4.4 numerical noise on 32-bit cicero

[leif]

Minh Nguyen schrieb:
> Hi folks,
>
> Doctesting Sage 4.4.4 on cicero.skynet, a 32-bit Pentium 4 machine, I
> got the following numerical noise:
>
> sage -t -long "devel/sage/sage/libs/mwrank/mwrank.pyx"
> **********************************************************************
> File "/tmp/mvngu/sage-4.4.4-9338-pycrypto/devel/sage/sage/libs/mwrank/mwrank.pyx",
> line 340:
> sage: E.silverman_bound()
> Expected:
> 6.5222617951910102
> Got:
> 6.5222617951910111
> **********************************************************************
> File "/tmp/mvngu/sage-4.4.4-9338-pycrypto/devel/sage/sage/libs/mwrank/mwrank.pyx",
> line 372:
> sage: E.silverman_bound()
> Expected:
> 6.5222617951910102
> Got:
> 6.5222617951910111
>

On a Pentium 4 (Prescott), Ubuntu 9.04, gcc 4.3.3*, all doctests passed
(ptestlong).

-Leif

____
* with CFLAGS and CXXFLAGS="-march=native -fno-strict-aliasing",
CPPFLAGS="-march=native"

[John Cremona]

This number comes from an eclib function which returns a C double,
which becomes a Python float. Now eclib has not changed since
previous release. Has Python?

This bound is something which is in no way tight, so it would be
absolutely fine to do the usual thing with numerical fuzz, namely
replace the last digits with ... . But it would still be nice to work
out why it has changed at all.

John

> --
> You received this message because you are subscribed to the Google Groups "sage-release" group.
> To post to this group, send email to sage-r...@googlegroups.com.
> To unsubscribe from this group, send email to sage-release...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/sage-release?hl=en.
>
>

[William Stein]

Patch up at

http://trac.sagemath.org/sage_trac/ticket/9360

Please review.

William

--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org

[Dr. David Kirkby]

On 06/26/10 04:30 PM, John Cremona wrote:
> This number comes from an eclib function which returns a C double,
> which becomes a Python float. Now eclib has not changed since
> previous release. Has Python?

I did update the python .spkg for sage-4.5.rc0, but I very much doubt the change
had anything to do with numerical results you see. It was related to making sure
the _socket module built on Solaris.

[John Cremona]

The patch has a positive review now.

John

[William Stein]

On Mon, Jun 28, 2010 at 11:21 AM, John Cremona <john.c...@gmail.com> wrote:
> The patch has a positive review now.
>
> John

Thanks. Here's a new bugfix release of sage that has just this patch
in it (plus fixing that Makefile issue):

http://sage.math.washington.edu/home/wstein/build/release/4.4.4.1/sage-4.4.4.1.tar

I'm not going to build binaries of this.

--