Multicriteria Optimization in the DISO System

From the mathematical point of view multicriteria optimization (MCO) is a natural generalization of optimization problems. The need of decision making in contradictory situations makes MCO methods so interesting for us. MCO deals with one of the most sophisticated aspects of human activity which is to achieve several goals by the single act of decision making. MCO models and ordinary optimization are not very much di erent in task de nition, giving us hope to use the similar numerical methods. This paper gives the overview of the MCO package as one of the main parts of the DISO dialogue system for optimization problem solving which was developed in the Computing Center of the USSR Academy of Sciences. Two MCO methods are described in this paper. Both methods are based on the idea of non-uniform covering technique and inclusion function approach, which was initially developed for global extremum search [1] [5]. The complexity of MCO tasks makes it necessary to create e ective numerical methods to nd both a single point of Pareto set and an approximation of this set also. The paper describes two MCO algorithms which di er in the interpretation of the solution and as a consequence in the complexity of numerical calculations. The main features of the MCO package are described also.