Work out genetic algorithm population instantly with clear inputs, formula shown and shareable results.
Genetic algorithm population is conventionally scaled with chromosome length so the initial population samples every gene position several times; a multiplier of two to ten times the bit length is typical. Total cost is population times generations fitness evaluations, which is the number that matters when each evaluation is a simulation or a model training run. Mutation rate is usually set near 1/length so about one bit flips per individual.
GA sizing
population = length x multiplier; evaluations = population x generations; expected mutations = population x length x mutation rate; search space = 2^length
Around 1 divided by chromosome length, so roughly one bit changes per offspring. Much higher and the search becomes random; much lower and it stops escaping local optima.
Track best and mean fitness. When the gap closes and the best stops improving for many generations, the population has converged and further generations waste evaluations.