Re: AW: [Gretl-devel] bessel K function
On Fri, 19 Jun 2009, Klein, Christoph wrote:
thank you *so* much!!! this function is much more accurate than
my version.
I can't claim credit for the revised version; it comes from the
netlib "specfun" repository.
Additionaly, I propose to dump my bessel_I function, too. With
the attached patch bessel_I for negative orders is calculated
using the new besselK function. bessel_I for negative orders
can't be calculated with the cephes function, because cephes
assumes symmetry around 0 for the order (see code), which is
false.
Thanks. I've applied your patch. I've also added (minimal)
docmentation for the bessel function. One little refinement: you
don't need to quote the first argument. For example,
x = bessel(J, 2, 5.6)
will do OK.
Allin.