On Nov 3, 11:43 pm, "Georg S. Weber" <[EMAIL PROTECTED]>
wrote:
> On 4 Nov., 01:57, "Justin C. Walker" <[EMAIL PROTECTED]> wrote:
>
> > On Oct 31, 2008, at 5:13 PM, mabshoff wrote:
> > sage -t devel/sage/sage/calculus/calculus.py
> > this has been reported, but I just noticed that there is
> > "significant bit noise", not insignificant bit noise:
> > Expected:
> > 6.1232339957367663e-16
> > Got:
> > 6.1230317691118863e-16
>
> I looked into this.
> Short answer:
> The Python builtin "float(...)" sucks, the Sage builtin
> "RDF(...)" rules!
> Long answer:
> The command in question is (after "sage: f = 5 * sin(x)" IIRC)
> sage: float( f(pi) )
> The precision loss thus is due to the Python builtin "float()", and
> possibly only so on Mac.
> You'll get even on a Mac the awaited-for digits by doing instead:
> sage: RR( f(pi) )
> The mathematical correct result "0.0" is gotten by:
> sage: float( simplify( f(pi) ) )
> and of course also by
> sage: RR( simplify( f(pi) ) )
> But even without "simplify" you get the result "0.0" by:
> sage: RDF( f(pi) )
>
> It would be a good idea not only to "just fix" the numerical noise in
> this doctest,
> but also but some (or all) of the above into doctesting "calculus.py".
Nope, even though float sucks we still have to test it. The above
doesn't really solve the problem and so late in the game I don't want
to introduce additional numerical noise problems.
> @Michael:
> My own test did not yield more issues than already reported.
> If the ticket you wanted to open does not appear soon, I'll open an
> extra one
> just for this "calculus.py" stuff and apply a patch to it.
>
I got patches ready to go, so let me open the tickets.
> Cheers,
> gsw
>
Cheers,
Michael
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
