Approximating the noninferior set in linear biobjective programs using multiparametric decomposition

An algorithm is developed to generate an approximate representation of the noninferior set in the objective space for linear biobjective programs. A sharp measure of geometrical error based on multiparametric decomposition is used to obtain a subset of the noninferior objective vectors distributed over the entire noninferior set. The deviation of the approximate representation from the exact noninferior set in the objective space can be controlled by specifying the maximum possible deviation. The algorithm attempts to select each additional noninferior objective vector to be included in the approximation to reduce the deviation as quickly as possible.