This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: periodic cspline first derivative discontinuity
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: David Necas (Yeti) <yeti at physics dot muni dot cz>
- Cc: gsl-discuss at sources dot redhat dot com
- Date: Mon, 15 Apr 2002 22:39:21 +0100 (BST)
- Subject: Re: periodic cspline first derivative discontinuity
- References: <20020406140517.A7885@physics.muni.cz>
David Necas (Yeti) writes:
> I've found a bug (or I think I've found a bug ;-) in periodic
> cubic spline implementation -- its first derivative is not
> continuous at the edges of the first segment.
> This is because length of next segment (h_ip1 in the patch below)
> is computed modulo even for the last segment and so it results as
> negative (well, not much clearly explained, perhaps the patch
> itself is clearer).
> Please note applying this patch actually fixes only first
> derivative discontinuity in right edge of the first segment, some
> dicontinuity may still appear at the left edge due to probems in
> cyclic tridiagonal linear system solver -- see my next bugreport.
Can you tell me how the patch to fix the cyclic solver affects this
result? Is any further change needed to the spline code?
Thanks
Brian