Fitness Landscape: Adaptive Walk

This simulator renders Sewall Wright's classic "adaptive landscape" as an actual 3D terrain: a rugged, multi-peaked fitness surface that a population of genotypes climbs generation by generation. Each generation applies real Wright–Fisher-style dynamics — Gaussian mutation of each genotype, fitness read off the terrain height, and fitness-proportionate (softmax) selection of the next generation's parents. Tuning population size controls how strongly genetic drift's sampling noise can push the population downhill or across a valley toward a taller peak, mutation step size controls how far each generation can jump, and selection strength dials the surface from a random walk (β = 0) to near-deterministic hill-climbing. It's a hands-on way to see why real populations sometimes get stuck on a local fitness optimum, and how drift, not just selection, can occasionally free them.