>
> Hi Dmitry,
>
> DB> Hi, anyone here heard of lagrange
> DB> interpolation?
>
> Definition:
> http://forum.swarthmore.edu/dr.math/problems/wang8.8.97.html
>
> Some code (???):
> http://www.netlib.org/c++/linalg.tgz
>
> Java applet (did NOT work on my IE 5.01):
> http://www.students.cs.ruu.nl/~jcgronde/java/LA.html
>
> Lots of info (including frequency response etc.):
> http://www-ccrma.stanford.edu/~jos/Interpolation/Lagrange_Interpolation_Coef.html
>
> DB> It is supposed to be really good method for
> DB> interpolation. Better than cubic/quadratic
> DB> (for down/up sampling)
>
It is just high order polynomial interpolation. Using
Lagrange polynomials just gives a simple formula for the
interpolant in terms of the original data. (cubic/quadratic
interpolation is the same as 3rd/2nd order Lagrange interpolation)
But it is only stable and accurate if the original data points are
clustered at the end points. When the original data is equally
spaced, it will be wildly unstable as you increase the polynomial
degree.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
