Thank you very much for your answer and thank also to Robert Dodier for
his explanation.
I will study these topics and if I succeed, I put the result on my web
pages for public.
Robert
Richard Fateman wrote:
>
> First of all, let me point out that this is somewhat unnatural to do,
> with respect to sums and products... e.g. the matching program
> generally will succeed in matching a pattern p*q against an
> expression X by allowing p=X and q=1.
>
>
>
> For forms with a fixed number of arguments like sin, cos, etc. there is
> no problem.
>