Re: Interfacing GALib to Swarm

[J.J. Merelo Guervos]

>>>>> "Nelson" == Nelson Minar <address@hidden> writes:

Hi,

    Nelson> Are there any good GA libraries written in plain C? That
    Nelson> would be your easiest course.

There are lots of them, in C, Lisp, you name it. Check them out at
your nearest ENCORE Web site, like http://krypton.ugr.es/~encore or
http://alife.santafe.edu/~joke/encore/. Most of them allow binary
representation, or floating point (usually called Evolution Strategies, or 
Evolutionary Programming) 

    Nelson> This is a nice idea, something we always wanted to do with
    Nelson> Swarm. I recommend making the actual communication layer
    Nelson> between the two processes be as abstract as possible. Or
    Nelson> at least, avoid named pipes.  They aren't reliable across
    Nelson> Unices, and I don't think they're supported in Windows
    Nelson> NT. It's probably better to use a TCP/IP socket - that
    Nelson> way, you could even run the GAlib part on another
    Nelson> CPU. Another option is to use shared memory, but again
    Nelson> that's not very portable.

Or RPC. Or ActiveX (for the WinWorld). But you'll get much more
mileage by implementing a new ObjC floating-point representation GA
library. If you want to include all features of GAlib (or any other
library, for that matter), you'll have to do a lot of programming
anyways. And take into account that GAlib is written in C++, which
limits a lot its run-time functionality, indeed, you won't be able to
define new classes "on the run".

It all boils down to
1. It's probably better to write a new floating point GA library
using, if possible, current protocols (that way, all classes will be
interchangeable).
2. Program to program communication is a good idea, but is slow, you
won't be able to take advantage of all the characteristics of both
languages, and you'll have to do lots of programming anyways.

                        JJ

-- 
JJ Merelo                         | http://kal-el.ugr.es/htbin/jj-plex
Grupo Geneura ---- Univ. Granada  | http://kal-el.ugr.es/



                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================