Sage 4.4.4 numerical noise in mwrank.pyx on 32-bit cicero

[williamstein]

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

--
Regards
Minh Van Nguyen

Component: numerical

Author: William Stein

Reviewer: John Cremona

Merged: sage-4.5.alpha1

Issue created by migration from https://trac.sagemath.org/ticket/9360

[williamstein]

comment:1

Attachment: trac_9360.patch.gz

[JohnCremona]

comment:2

Looks good.

For the pedants out there, this number is a no-way strict upper bound on something and it makes no sense to worry about the low order bits.

[89c6e537-b2e3-45e6-882d-d4957b74ffe5]

Merged: sage-4.5.alpha1

[89c6e537-b2e3-45e6-882d-d4957b74ffe5]

Author: William Stein

[89c6e537-b2e3-45e6-882d-d4957b74ffe5]

Reviewer: John Cremona