wrong symbolic sum



On 7/26/09, Barton Willis <willisb at unk.edu> wrote:

> From the perspective of consistency with these (well-known)
> simplifications, I would say that the default should be for sum and
> product to use the sumhack / prodhack identities.

Yeah. OK, I hereby drop my opposition to sumhack.
If you want to restore it, I won't stop you.

> Of course, lsum cannot sumhack. If lsum were extended somewhat and we
> had a simple way to represent subsets of the integers, we could make
> lsum the non-sumhack sum. Something like lsum(1/k,k, 1 .. 10^7), where
> 1 .. 10^7 = set(1,2,3,..., 10^7, or 1 .. -1 = empty set.

Apart from summation, I'd like to see symbolic or lazy subsets of integers.

best

Robert Dodier