Genetic Algorithm: Fitness Landscape Explorer

A genetic algorithm searches for good solutions the same way a population of organisms searches a real terrain: many individuals, each scored by a fitness function, compete and recombine generation after generation until the population converges on high ground. This simulator renders that search literally — a real 3D fitness landscape built from overlapping Gaussian peaks, with a population of genotypes rendered as instanced markers that climb it under selection, blend crossover and Gaussian mutation. Switch between a rugged six-peak landscape (where premature convergence to a local optimum is likely) and a smoother two-peak one, tune mutation rate, mutation strength and tournament size, and watch live readouts for generation count, best and mean fitness, and population diversity to see the exploration/exploitation trade-off play out in real time.