On 7/4/06, Stavros Macrakis <macrakis at gmail.com> wrote:
> Is there any (good) reason that nusum and GosperSum don't handle
> sum(...,i,inf)? No reason for the core logic to understand what's going on,
> but why not perform sum(...,i,gensym) then if it's a nounform, subst
> gensym=inf, else return limit( ... , gensym, inf).
I am inclined to leave alone nusum and GosperSum,
and to put the gensym / limit logic into simpsum instead.
(I don't see why simpsum shouldn't invoke the Gosper
and/or Zeilberger stuff; so far as I know it's not very
time or memory consuming.)
So what I propose is that if simpsum can't make progress,
it calls Gosper and/or Zeilberger, after substituting a gensym
into the expression if need be, and calls the limit stuff
afterward.
Comments?
Robert