[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-gsl] Re: Interpolation
From: |
Evgeny Kurbatov |
Subject: |
Re: [Bug-gsl] Re: Interpolation |
Date: |
Wed, 24 Nov 2010 19:18:59 +0300 |
Hi Rhys
On Wed, 24 Nov 2010 09:52:24 -0600
Rhys Ulerich <address@hidden> wrote:
> >> why the interpolating routines become extrapolating sometimes?
> >> Today I spent several hours to find the stupid bug in my program
> >> due to interpolation bounds violation. When an interpolation
> >> point becomes greater than right bound or lower than left then no
> >> error returns but the routine extrapolates a function value to
> >> that point giving very brave values.
>
> > That sounds like a good idea, to just hardcode an error for values
> > outside the range,
>
> Any chance that you'll break everyone who depended on this
> extrapolation "feature"? Such a check seems like an easy thing for
> the library user to do when/if it makes sense.
I think no chance. This check has already done in gsl_interp_eval_e
and related procs with _e suffix. It is strange not to do it in
others. In addition, it it necessary to clean situation with error
codes returning from interpolation procedures, when the code is
GSL_FAILURE in cspline_eval_deriv and deriv2, and the code
is GSL_EINVAL in linear_eval_deriv. I'll send the patch in a
coming days when rake an urgent tasks.
Regards,
Evgeny Kurbatov
>
> - Rhys
>