Thanks for your answer, first of all.
I will try with the custom tree builder approach. Surely I will post
another few questions as I continue with my research.
Thanks again for your time!!
On Tue, Mar 9, 2010 at 12:53 AM, Sean Luke <[log in to unmask]> wrote:
> Beto, I think you're going to need to make a custom tree builder. I would
> suggest restricting the minimum and/or maximum height of nodes in your logic
> area with a straightforward GROW algorithm.
> You could also modify PTC2 as follows: instead of populating the remaining
> horizon spots with terminals, PTC2_MODIFIED might populate them all with
> comparators, then pass through and populate again with terminals and random
> Ideally you could use the UNIFORM algorithm, which should suit you almost
> perfectly. But unfortunately I think it's suffering from some bit-rot, or
> so I'm told. And the typed version (which I built for ECJ based on the
> untyped version in the literature) is quite complex -- and complex to debug.