[bug-gsl] Re: Re: rk8pd (2008-04-10)
|
From: |
Brian Gough <bjg@network-theory.co.uk> |
|
Subject: |
Re: Re: rk8pd |
|
Date: |
Thu, 10 Apr 2008 17:01:39 +0100 |
|
To: |
"Andrew W. Steiner" <awsteiner@.........> |
|
Cc: |
Bug-gsl@gnu.org |
At Mon, 7 Apr 2008 10:16:16 -0400,
Andrew W. Steiner wrote:
>
> Ok. I checked a couple other functions and they're fine.
> It seems I just picked a case where the higher-order
> stepper does worse. There's probably some deep reason why the Bessel
> function is
> difficult at higher order. I have a couple guesses, but if someone
> wants to enlighten
> me that'd be great too.
You could calculate the error growth analytically by putting the
formulas and bessel recurrence into Maxima and expanding them.
--
Brian Gough
