Re: [Help-gsl] Incomplete elliptic integral (Legendre) (quasi-)periodicity issue

[Brian Gough]

At Fri, 27 Oct 2006 12:43:05 +0100,
Lionel Barnett wrote:
> It appears that the function gsl_sf_ellint_E(phi,m) is periodic with
> period 2\pi. However, E(\phi|m), even under the definition at:
> 
> http://www.gnu.org/software/gsl/manual/html_node/Definition-of-Legendre-Forms.html#Definition-of-Legendre-Forms
> 
> is actually *quasi*-periodic, satisfying the relation:
> 
> E(\phi+n\pi,m) = 2n E(m) + E(\phi,m)
> 
> Now I can use this relationship to calculate the correct value of
> E(\phi,m) for larger \phi ... except for \pi/2 < \phi < \pi
> 

Just to let you know this has now been fixed in CVS for the next
release (http://sources.redhat.com/gsl/devel.html) - the values should
be correct for all values of phi now.

-- 
Brian Gough
(GSL Maintainer)

Network Theory Ltd,
Publishing the GSL Manual - http://www.network-theory.co.uk/gsl/manual/