summertime hacking; making maxima 10X faster or more



-----maxima-admin at math.utexas.edu wrote: -----

>15 years ago I wrote a bunch of patches to the
>Maxima / Macsyma simplifier
>using hash coding and memorization, to make it
>much faster. 10X or more ??? replacing
> exponential-time routines with constant-time
>routines!

I think the time required to simplify some of
these sums could be doubled with some modest
changes to simp-add. My  evidence: (some time
ago RJF and I chatted about this same example)

(%i11) m1:sum(z[i],i,0,200)$
Evaluation took 0.45 seconds (0.45 elapsed)

(%i12) m1 : tree_reduce("+",makelist(z[i],i,0,200))$
Evaluation took 0.00 seconds (0.00 elapsed)  <-- average about 0.15 sec

As I remember, simp-add doesn't take advantage of
sorting as well as it might.

Barton