An interactive multiobjective procedure for selecting medium-term countermeasures after nuclear accidents

This paper presents a methodology for supporting medium-term decisions in the context of nuclear accidents. An interactive procedure based on a multiobjective linear model is introduced, allowing the set of feasible strategies to be explored. This procedure enables the users to express preferences on both criterion values and the structure of the strategies. Additional flexibility is provided by the possibility of integrating evolutive constraints during the decision process. © 1998 John Wiley & Sons, Ltd.     

Generating interior search directions for multiobjective linear programming

A new multiobjective linear programming (MOLP) algorithm is presented. The algorithm uses a variant of Karmarkar's interior-point algorithm known as the affine-scaling primal algorithm. Using this single-objective algorithm, interior search directions are generated and used to provide an approximation to the gradient of the (implicitly known) utility function. The approximation is guided by assessing locally relevant preference information for the various interior directions through interaction with a decision maker (DM). The resulting algorithm is an interactive approach that makes its progress towards the solution through the interior of the constraints polytope.     

Some results concerning the quality of vertex solutions found by a method for multiple-objective linear programming

The Zionts-Wallenius algorithm for multiple-objective linear programming terminates with an extreme point solution that is locally but not necessary globally optimal. To find the globally optimal solution, a search along the facets of the solution space polyhedron may be required. In this paper we report the results of an experiment to determine how close the local and global optima are. We discuss the concept of closeness and propose one measure. Computer simulation is used to determine, in general, the quality of the solution found by the Zionts-Wallenius method for two types of non-linear utility functions. The merits of a search procedure are discussed in the context of the results.