[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-gsl] On bug #14 - qagil
From: |
Brian Gough |
Subject: |
Re: [Help-gsl] On bug #14 - qagil |
Date: |
Thu, 27 Jan 2005 11:40:30 +0000 |
Maurice Leutenegger writes:
> I have written a function which uses qags to perform an integration.
> I am getting an error under certain conditions:
>
> gsl: qags.c:558: ERROR: roundoff error detected in the extrapolation table
> Default GSL error handler invoked.
>
> This looks to me like it may be related to Bug #14 in the BUGS.gz file
> accompanying the latest GSL release.
Hello,
That entry mainly concerns the sign of the extrapolated value when it
is close to zero. It is not an actual bug as such, just the way the
QUADPACK extrapolation algorithm works.
The error for roundoff is normal if you push the integration too far.
The solution is to use a larger tolerance or break up the integral in
some way. The reference in the manual gives more details.
--
best regards,
Brian Gough
Network Theory Ltd,
Commercial support for GSL --- http://www.network-theory.co.uk/gsl/