The genetic algorithm for 0-1 programming with linear constraints

This paper suggests a new method for solving 0-1 programming problem with linear constraints by genetic algorithm. The existing way of handling such problem allows constraint violation at the expense of some penalty and meets some difficulties of choosing penalty parameter which affects directly numerical result and possible premature convergence. In order to improve this situation, a genetic algorithm for 0-1 programming with linear constraints is proposed. The basic line of this algorithm is that the constraints are to be satisfied by every trial point in the genetic algorithm. By using nonbinary code, the genetic operators are different from classical genetic algorithm. Computational results are compared with those of conventional method and possibility of parallel processing is discussed