Has GA ever been used for code optimization?

Compudolt Compudolt at aol.com
Sat Jan 24 05:39:41 EST 1998


  This would be a pretty tricky problem to implement, because the best
solution would be to actually run the test code (so that it is specific to the
platform being run on) and base the correctness of the code on the results and
on the timings.
  However, I assume this type of approach is crash prone.

  Also, what is the largest (ie most complex) problem that GA has handled?  It
would be interesting to see a genetic algorithm handle a problem like creating
a better genetic algorithm than itself.

Thanks for your time,
Max.






More information about the galib mailing list