Genetic Algorithm: Evolving a Route

A genetic algorithm searches for good solutions by evolving a whole population of candidates at once rather than following a single gradient. This simulator applies that idea to a small traveling-salesman problem: each individual is a full tour through a set of scattered cities, and its fitness is simply how short that tour is. Generation after generation, the fittest tours survive, pairs of parents recombine through order crossover to produce valid child tours, and occasional swap mutations keep the population exploring new orderings instead of settling on the first decent route it finds. Watch the best, second-best and third-best routes redraw themselves live, and track the population's improvement on the fitness-convergence curve in the corner. Tune population size, mutation rate, crossover rate and evolution speed to see how each reshapes the search — and to feel directly why a genetic algorithm trades the certainty of gradient descent for the ability to search spaces no derivative could touch.