Re: [Help-gsl] looking for gsl_integration_long_double

[Tobias Munk]

The problem I have to solve is a summation of integrals of Mathieu function 
(some special
periodic functions, in gsl since v 1.9). The integral solver for oscillatory 
functions
does not work in this case, since the integrand is known only numerically, thus 
one
cannot factor out a trigonometric term. In any case, a long double integration 
alone
would not be sufficient to increase accuracy for this problem, since the Mathieu
functions must be available at the same precision. Currently, they are only 
present as
double, too.

Mathematica is able to approximate the series much better than the code I wrote 
using
gsl, presumably since Mathematica uses arbitrary precision arithmetics. I am 
not sure
which approach is more promising, implementing everything in long double, or 
using
arbitrary precision arithmetics, like gmp (http://gmplib.org).

Tobias

--- Evgeny Kurbatov <address@hidden> schrieb:

> Please tell, what the task you solve with such requirement to precision?
> 
> Tobias Munk wrote:
> > I am looking for a integration routine with "long double" type, and cannot 
> > find
> anything
> > in the gsl. Has anyone already thought of implementing something like this?
> > 
> > Cheers, Tobias
> 




      __________________________________  Yahoo! Clever: Sie haben Fragen? 
Yahoo! Nutzer antworten Ihnen. www.yahoo.de/clever