search

NicMcPhee / XO-bias-study

Results and write-up of our genetic programming crossover bias study
MIT License
0 stars 0 forks source link

Do K Landscapes runs #8

Closed NicMcPhee closed 9 years ago

NicMcPhee commented 9 years ago

We'll have to decide on some reasonable values of N, K to run these on.

NicMcPhee commented 9 years ago

klandscapes_k6_bias_results

These are results with K=6. For K=2, almost every run solves the problem almost every time, and even here solutions are awfully easy to come by. K=6 seems to be the highest value "baked in" to ECJ, though, and I'm not sure I care enough to implement a harder version just for this paper.