Duality and ordinality in fuzzy measure theory (Q2572273): Difference between revisions

For a finite fuzzy measure \(m\) (i.e., a non-decreasing set function defined on a sigma-algebra \({\mathcal A}\) of a universe \(X\) vanishing at the empty set), its dual \(m^{\text{d}}:{\mathcal A}\to[0,\infty[\;\) is given by \(m^{\text{d}}(A)=m (X)-m(A^c)\). Several properties of finite fuzzy measures have their counterpart for the corresponding dual fuzzy measures. For example, the submodularity is related to the supermodularity, the continuity from below to the continuity from above, Choquet integrability of a measurable function \(f\) with respect to \(m\) with the Choquet integrability of \(f\) with respect to \(m^{\text{d}}\), etc. This type of duality in the fuzzy measure theory is called cardinal and was discussed, e.g., by \textit{X. Liu} [J. Fuzzy Math. 2, No. 3, 555--563 (1994; Zbl 0816.28013)]. In this paper, a new type of ordinal duality is introduced, applicable to all fuzzy measures. Ordinal properties of fuzzy measures are based on increasing bijections between the domain \([0,m(X)]\) and the range \([0,\infty]\). Then the ordinal duality of properties \(P\) and \(Q\) of fuzzy measures means that both \(P\) and \(Q\) are ordinal properties, and for finite fuzzy measures they are cardinally dual. Several examples of ordinally dual properties are included. For example, the maxitivity of the Sugeno integral with respect to a maxitive fuzzy measure is ordinally dual to the minitivity of the Sugeno integral with respect to minitive fuzzy measures.