Decide on values of K for K-landscape problems

NicMcPhee / XO-bias-study

Results and write-up of our genetic programming crossover bias study

MIT License

0 stars 0 forks source link

Decide on values of K for K-landscape problems #3

Closed NicMcPhee closed 9 years ago

NicMcPhee commented 9 years ago

If we're going to use the K-landscape problems (which I think we should), we'll have to decide what values to use.

NicMcPhee commented 9 years ago

The largest value "baked in" to ECJ is K=6, and that's actually fairly easy to solve. I don't really feel like implementing higher values just for this study, though, so I'm going to close this for now. If we find that the results look quite different on Much Harder Problems then we might come back and try to implement higher values of K, but I don't think that will be necessary.