Re: extrapolate forbidden with interp1, spline

[Eric Chassande-Mottin]

as a first impression, I am against. the results given by those procedures
are likely to be crazy when extropolating away from the data support.
the use of "extrap" ensures that the user knows that what he/she is
doing. the warning I was suggesting in my first message would
help the user to understand that interp1 works slightly differently
in Octave and would point him/her to a simple fix.

eric.

On Nov 27, 2007 5:40 PM, David Bateman <address@hidden> wrote:
>
> John W. Eaton wrote:
> > On 27-Nov-2007, David Bateman wrote:
> >
> > | This is a bug in matlab, as extrapolation should only be done if the
> > | "extrap" flag is given..
> >
> > I just checked the Matlab docs and I think it is more subtle than
> > this.  It seems that methods other than nearest, linear, and v5cubic
> > do extrapolate for out of range values even without the extra "extrap"
> > argument.
> >
> > jwe
> >
> >
> Then this is inconsistent behavior.. Do we want it have it for
> compatibility?
>
>
> D.
>
>
> --
> David Bateman                                address@hidden
> Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph)
> Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob)
> 91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax)
>
> The information contained in this communication has been classified as:
>
> [x] General Business Information
> [ ] Motorola Internal Use Only
> [ ] Motorola Confidential Proprietary
>
>