A fuzzy preference relation in the vector maximum problem (Q1089249)

The authors propose a formulation of the fuzzy multicriteria optimization problem. Such a formulation is based on the possibility of specifying a fuzzy preference relation that expresses the degree of preference for each pair of alternatives in the solution space of a standard multicriteria optimization problem. The solution of the ensuing optimization problem maximizes the membership function of the set of non- dominated (in terms of preference) elements of the alternative space. Under the assumption that the original utility vector function can be associated with a weight vector, a special case of fuzzy preference relation is considered. The corresponding optimization problem is studied, giving an optimality condition, and is connected with the concept of Pareto optimality. In this latter respect it is essentially shown that, allowing the weights to vary, one recovers just the efficient solutions of the original multicriteria optimization problem. A more general case is then introduced. The upshot of the analysis is here the derivation of a sufficient existence condition for the solution in terms of saddle points of a function, that is immediately defined in terms of the fuzzy preferences relation. A more loose connection with efficient solutions of the original multicriteria optimization problem is given in this case too.