Genetic algorithms

Genetic algorithm (GA) is a heuristic optimization method, which is aimed for finding a global optimum in a large search space. GAs belong to the group of evolutionary algorithms.
Genetic algorithms mimic following evolution mechanisms:
  • Evolution operates in chromosomes and the coding mechanism which forms new alternative s.
  • Natural selection forms the link between the chromosomes and the performance of the decoded structures.
  • Regeneration is based on crossover and mutation. Crossover is the core operator to evolve better solutions and mutation modifies chromosomes with random variations to prevent stopping in local optima.
  • Population includes the solution alternatives. It should be sufficiently wide to represent the differences of the alternatives. Too large a population might lead to slow convergence, The population changes during the calculations. Elitism is needed for keeping the best solutions within the population since the biological evolution does not have memory.
The solution mechanism does not know anything about the problem. The coding mechanism and the objective function connect the solution with the problem. In the building block hypothesis, the good parameter vectors represent feasible sub-solutions and even better global solutions are possible possible to reach by combining them. This methodology can solve very complicated problems in wide search spaces.
A set of good alternatives can be picked from the population to make a flexible menu.
Nutri-Flow Ltd
Villiperäntie 5, 90410 Oulu, FINLAND
Powered by:MetaVisual CMSMobile versionDesktop versionWebmasterLog in© 2020 - 2024 Nutri-Flow Oy