Studying How the Configuration of a Genetic Algorithm Affects the Solution of a Problem

This paper discusses a genetic algorithm - an approach to solve complex optimization problems in a reasonable amount of time. It describes the principle of operation of the algorithm. The algorithm's efficiency is demonstrated by approximating the popular knapsack problem. The influence of each of the algorithm's parameters on the obtained solution is studied by freezing the other parameters and simulating an NP-complete problem with a known solution. It is concluded that the algorithm needs to be properly configured for the best results - its parameters individually don't have a decisive impact on the final solution, but in conjunction perform very well. It is also concluded that the genetic algorithm proves to be extremely effective in finding a close enough solution to the optimal one for NP-complete problems.