Foundations of global genetic optimization

Mehr zum Buch

Genetic algorithms are effective global optimization methods used to tackle complex real-life problems in science and technology. Despite their computational complexity, they can explore vast data sets and address challenging cases where objective functions are irregular and multimodal, and where extrema location information is inaccessible through other means. These algorithms belong to the class of iterative stochastic optimization strategies, generating and evaluating a set of admissible points, known as the random sample or population, during each step. Unlike Monte Carlo strategies that sample uniformly across the search domain, genetic algorithms adjust the probability distribution at each step. Drawing from biological mechanisms, they utilize processes such as genetic code mutation, crossover, and selection among individuals. These methods have been validated by solving multimodal problems in nature, evidenced by various species adapted to different ecological niches. They enhance the search process's efficiency compared to completely random approaches, maintaining the exploration capability of the entire admissible set, which is crucial for global optimization. Consequently, genetic algorithms represent a novel class of artificial intelligence methods that incorporate heuristics proven effective in other fields into the traditional framework of stochastic global search.