is maxima suitable for multiprocessor computations?



On 11/14/07, Stavros Macrakis <macrakis at alum.mit.edu> wrote:

> Just to clarify, I wasn't saying that parallelism couldn't be used in
> symbolic calculation, just that there is no straightforward way to take
> advantage of multiple processors in Maxima as it is currently built.

Agreed.

> I suppose some user calculations (numerical or not) would be an exception --
> when you are calculating f(X) for multiple values of X, you could simply
> split up the value space and delegate to multiple instances of Maxima, which
> the OS would take care of allocating to multiple processors.

Along these lines, I came across this blurb about MapReduce which
appears to be a system for automatically parallelizing programs.
http://labs.google.com/papers/mapreduce.html
If Google is on the bandwagon, it must be a good idea, right??

These days a big data set is often stuff someone pulled off the web.
Maxima is not usually applied to big piles of textual data,
but I see no reason to rule it out; text processing is just a variety
of symbolic computation anyway.

FWIW

Robert