On: Stochastic Superoptimization

[jacques_chester]

Well to pick between GA and GP, the question is: are you trying to evolve a list of parameters, or create a novel function/program?

Parameters into function: use GA.

Function/program: use GP.

(speaking very roughly, that's the historical distinction between them)

[coherentpony → jacques_chester]

From a mathematical perspective, it looks as though your statement could imply that GA may potentially be viewed as a finite dimensional analogue of GP. Interesting.