摘要
A genetic algorithm to solve the set covering problem proposed in the literature had some improvements which gave better solutions, i.e., better chromosomes in the first starting population, taking full account of domain specific knowledge with sound programming skill. We have further investigated the input data dependency of their genetic algorithm, i.e., the dependency on costs and density. We have found that for input problem data sets with densities greater than or equal to 3%, our genetic algorithm is still practical both in computing time and approximation ratio.
A genetic algorithm to solve the set covering problem proposed in the literature had some improvements which gave better solutions, i.e., better chromosomes in the first starting population, taking full account of domain specific knowledge with sound programming skill. We have further investigated the input data dependency of their genetic algorithm, i.e., the dependency on costs and density. We have found that for input problem data sets with densities greater than or equal to 3%, our genetic algorithm is still practical both in computing time and approximation ratio.
