[Maxima] speeding up programs (was slowness in repeated Maxima evaluations) from Sage
fateman at cs.berkeley.edu
Mon Dec 8 00:17:10 CST 2008
Right, I understand the problem as indicated by the Sage people.
Nevertheless, if the program was somehow
automatically generated by Sage, it was a rather poor program for the
reasons given. I'm not sure that the
one I posted is actually equivalent entirely, but it should be about the
same time. And time is often of concern.
For example, I believe that one of the goals of the Sage project is that
in Maxima should be rewritten in python because Maxima is too slow
..... (Another seems to be
that the rewriting of such programs should be done by high school
students during the summer.)
My point is that if you write truly bad programs that are 30X slower
you might as well rewrite in Maxima! It is also possible to compile
fsum(), but the
problem is too small to justify the expense of compiling, even though
fsum runs several
then the Robert Dodier wrote:
> On 12/7/08, Richard Fateman <fateman at cs.berkeley.edu> wrote:
>> There are a number of peculiarities in the way this computation was set up,
>> perhaps some because the original author was not concerned with efficiency,
>> or aware of the consequences of programming in certain ways. or perhaps Sage
>> is itself producing bad stuff from good input. I don't know which.
> Actually the problem as it was described on the Sage mailing list is
> not that the program was too slow, but rather that it got slower and
> slower (by a substantial factor each time) if it was executed
> repeatedly from Sage.
> I wrote that program as an approximation to the description of the
> troublesome program (which wasn't posted). When I run it repeatedly
> in Maxima, the run time remains constant (ignoring noise).
> The evidence such as it is suggests the problem is in Sage, not Maxima.
> Robert Dodier
> Maxima mailing list
> Maxima at math.utexas.edu
More information about the Maxima