NK Fitness Landscape — Epistasis & Adaptive Walk Trapping

N (genome length): 10 K (epistasis): 3 Walks: 15 Build Landscape Run Adaptive Walks Clear Walks

The NK model (Kauffman 1993) creates tunable ruggedness: N binary genes, each interacting with K others. K=0 gives a single peak (Mount Fuji); K=N-1 gives a maximally random landscape with ~2^N/N+1 local optima. Most adaptive walks get trapped in local optima — more so at higher K. The density of local optima scales as 1/N+1 of all genotypes for K=N-1.