On 23 May 2011 05:51, Rob Beezer <[email protected]> wrote:
> On May 22, 7:26 pm, François <[email protected]> wrote:
>> If I am not mistaken about the meaning of the distribution_function method
>> gsl may have actually corrected a bug and +oo is actually the value you are
>> supposed to get, see this 
>> graph:http://en.wikipedia.org/wiki/File:Chi-square_distributionPDF.svg
>> Would statisticians on the list care to comment?
>
> That sounds correct to me.  GSL's "nu" is Wikipedia's "k", according
> to the GSL documentation.  The pdf has an x^{k/2-1} in it, and nothing
> to counteract this term growing large for x close to zero when k=1, so
> a value of +oo looks right.
>
> Rob

This is another worrying example of where someone wrote a doctest
based on the output from Sage, without actually verifying the result
is mathematically correct. Neither did the reviewer do so - assuming
this test was written after the review process was introduced into
Sage.

The fact GSL had a bug (now corrected) is not so bad, as all
non-trivial software has bugs. But for us to actually test for the
buggy result seems pretty bad to me.

Dave

-- 
To post to this group, send an email to [email protected]
To unsubscribe from this group, send an email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Reply via email to