Hallo Mark,

> Could you please be a bit more specific [...]
Maybe it would be easier (for me at least:-)) to tell you what I am looking
for:

It is a simple problem. Given a set of (x_i, y_i) - values for a funktion.
And another set (with another size) for a second one. Now I need the
convolution of both functiones and I want to reduce the number of
interpolating sections. Thats why I thought of (F)FT where I can ignore the
high frequencies. As I understand FFT there is probably no trivial way to
compute the transformation for a non-equidistant grid. How about wavelets?
(I have to confess, it is my first contact with them).

Regards
Ulf