Print

Print


Hi,

Thanks for the reply Shane.  The below is for the initialisation of the 
population, but I'm actually concerned about how large the trees can 
become within the run itself.  Is there some upper limit?  Sorry, I 
don't think I phrased my question very well!

Cheers

David

Shane wrote:
> Take a look at this in koza.params:
>
> # The tree uses an ec.gp.koza.HalfBuilder to create
> # itself initially.
> # HalfBuilder will pick GROW half the time and FULL
> # the other half, with a ramp from 2 to 6 inclusive.
> # By ramp we mean that it first picks a random number between
> # 2 and 6 inclusive.  This then becomes the *maximum* tree size
> # (for the FULL approach, it's the tree size of the tree, for
> # GROW, the tree can get no bigger than this)
>
> gp.tc.0.init = ec.gp.koza.HalfBuilder
>
> # We set the default for HalfBuilder to be a ramp of 2--6,
> # with a grow probability of 0.5
> gp.koza.half.min-depth = 2
> gp.koza.half.max-depth = 6
> gp.koza.half.growp = 0.5
>
>
>
> On Thu, Jul 24, 2008 at 12:08 PM, David Robert White 
> <[log in to unmask] <mailto:[log in to unmask]>> wrote:
>
>     Hi,
>
>     Please could someone help me with this query:
>
>     Is there a limit on the tree sizes that ECJ can create?  Is there
>     any restriction on tree size that I should know about?  I'm asking
>     because I am experiencing bloat in a run up to a point, and no
>     further increase in size after that point.  Perhaps some internal
>     limit has been reached?
>
>     Thanks,
>
>     David
>
>     -- 
>     David R White
>
>     Research Student
>     Department of Computer Science
>     University of York
>     York YO10 5DD
>     United Kingdom
>
>

-- 
David R White

Research Student
Department of Computer Science
University of York
York YO10 5DD
United Kingdom

Phone: +44 (0)1904 434756