Interesting... the same parameter file you say?  That'd be odd.  As  
it turns out, there are *two* reasons for this error.  Either the  
number of trees being read from the file differs from the number of  
trees in the original individual, *or* the tree array in the original  
individual is null.  This second case is unlikely but you can test it  
by changing readGenotype in to be something like:

     /** Overridden for the GPIndividual genotype. */
     public void readGenotype(final EvolutionState state,
                              final DataInput dataInput) throws  
         int treelength = dataInput.readInt();
	if (trees == null)
	    state.output.fatal("Hmmm... trees are null in readGenotype 
(EvolutionState, DataInput).");
         if (treelength != trees.length) // wrong size!
             state.output.fatal("Number of trees differ in  
GPIndividual when reading from readGenotype(EvolutionState,  
DataInput).\n" +
	    "Original size: " + trees.length + " and read-size is " +  
         for(int x=0;x<trees.length;x++)

This would also tell us how many trees you have coming in and  
currently in the individual.

Some other oddities.  Though you're setting gp.individual.numtrees=1,  
this only gets checked (and used) if you don't have the local  
parameter set, namely pop.subpop.0.species.ind.numtrees=1  What do  
you have set in the local parameter?  Also, how many subpopulations  
do you have?  Do they have different species?  I'm grasping at straws  
here but won't be able to really look at it until after the GECCO  


On Jan 24, 2008, at 10:28 PM, Bradford Barr wrote:

> Hey List,
> I'm having an issue with an Island model. I keep getting this error:
> Number of trees differ in GPIndividual when reading from  
> readGenotype(EvolutionState, DataInput).
> But both Islands (I'm only using two so I can figure out what's  
> going on) are using the same parameter file, and I have a command  
> line parameter saying gp.individual.numtrees=1 for both runs just  
> to double check.
> I'm stumped. Can anyone offer any insight?
> Thanks in advance,
> Brad Barr