On 7/17/06, Barton Willis <willisb at unk.edu> wrote:
> To evaluate sum(a[i],i,1,n), Maxima calls
> great n^3 / 6 - n / 6 times (experimentally determined).
> Surely even a simple-minded algorithm would be O(n^2).
I've committed a revision of src/asum.lisp such that the time
to evaluate sum(a[i], i, 1, n) scales like n^2 instead of n^3.
Before, the code (in DOSUM iirc) was constructing an
expression ((MPLUS) ...) and simplifying it. Now the code
calls ADD for each term instead.
HTH
Robert