On the existence of a possibility distribution function (Q703392): Difference between revisions

A possibility space \((X,{\mathcal B},\Pi)\) consists of a set \(X\) together with a \(\sigma\)-algebra \({\mathcal B}\) of subsets and a possibility measure \(\Pi:{\mathcal B}\to [0,1]\) subject to the conditions \(\Pi(\emptyset)=0,\Pi(X)=1\) and \(\Pi(A\cup B)=\max\{\Pi(A),\Pi(B)\}\). A possibility distribution for such a possibility space is a function \(\pi:X\to [0,1]\) satisfying \(\sup_{x\in X}\pi(x)=1\) and \(\Pi(A)=\sup_{x\in A}\pi(x)\) for each \(A\subseteq X\). And such a possibility space (or possibility measure) is continuous iff one has \(\Pi(\bigcap_{i\to\infty}B_i)=\lim_{i\to\infty}\Pi(B_i)\) for each non-increasing sequence \((B_i)_{i\geq 0}\) of subsets of \(X\). For infinite sets \(X\) the existence of a possibility distribution is a non-trivial problem. The authors study the so-called pdf-property of a set \(X\), which means that each particular continuous possibility space over \(X\) for which the \(\sigma\)-algebra \( {\mathcal B}\) is the full power set of \(X\) has a possibility distribution function. It is shown that this pdf-property depends only on the cardinality of \(X\). And the main result is that a set \(X\) has this pdf-property if and only if the cardinal of \(X\) is non-measurable.