The Aumann functional equation for general weighting procedures (Q2354656)

Let \(I \subset \mathbb{R}\) be an interval. A function \(M: I \times I \to I\) is said to be a mean if \[ \min\{x,y\} \leq M(x,y) \leq \max\{x,y\}, \;x,y \in I. \] A function \(\overline{M}: [0,1] \to I\) is called a weighting of a mean \(M\) if {\parindent=10mm \begin{itemize}\item[(W1)] \(\overline{M}(w)\) is a mean defined on \(I\) for every \(w \in [0,1]\),\item [(W2)] \(\overline{M}(1/2) = M\),\item [(W3)] \(\overline{M}(0)(x,y) = x\) and \(\overline{M}(1)(x,y) = y\), for all \(x,y \in I\). \end{itemize}} Now, let \(\mathcal{M}(I)\) and \(\mathcal{N}(I)\), \(\mathcal{M}(I) \subset \mathcal{N}(I)\) be two classes of means defined on \(I\). A function \(M^w\) on two variables \(M\) and \(w\) defined for \((M,w) \in \mathcal{M}(I) \times [0,1]\) and with values in \(\mathcal{N}(I)\) is said to be a weighting procedure if the function \(w \mapsto M^w\) is a weighting of \(M\), for every \(M \in \mathcal{M}(I)\). Let \(M^w\) be a given weighting procedure and \(p, q \in [0,1]\). The functional equation \[ M(M^p(x,y), M^q(x,y)) = M(x,y), \,x,y \in I, \] (Aumann generalized equation with weights \(p, q\)), where \(M \in \mathcal{M}(I)\) unknown, is studied in this paper for different classes \(\mathcal{M}(I)\).