Genetic Algorithm

genetic_algorithm() evolves a population of solutions with selection, crossover and mutation. Use it when good solutions can be combined.

from pymetaheuristics.genetic_algorithm import genetic_algorithm

ga = genetic_algorithm(
    knapsack,
    stop=max_iterations(30),
    rng=1,
    population_size=20,
    mutation=bit_flip_neighbor,
    repair=drop_heaviest,
)
print(ga.best_solution, ga.best_value)

Each generation:

1. Breeds population_size children. selection picks two parents and crossover breeds them, until the population is full.
2. Mutates every child. For an infeasible mutant, mutation is retried up to max_tries times, and if every try is infeasible the child is kept unmutated.
3. Passes children that are still infeasible through repair (if given). Any that remain infeasible are replaced by a fresh feasible genome.
4. Evaluates the children. Elitism then puts the best genome so far in place of the worst child, so the best is never lost.

Keyword	Default
population_size	10
selection	random_weighted_selection
crossover	single_point_crossover
mutation	inter_mutation
repair	None
max_tries	1000

To set an operator's knob, bind it with functools.partial, for example mutation=partial(inter_mutation, num_swaps=3).

Warning

Each generation costs population_size evaluations. stop is checked once per generation, so a max_evaluations budget can be exceeded by up to one generation.

Recap

• Pass operators as keywords; defaults work for bit lists.
• Elitism keeps the best genome, and repair handles infeasible children.

Next: Simulated Annealing.