A representation theorem for maxitive measures (Q1042752)

Following [\textit{N.\,Shilkret}, Nederl.\ Akad.\ Wet., Proc., Ser.\,A 74, 109--116 (1971; Zbl 0218.28005)], a nonnegative set function \(\eta\) on a \(\sigma\)-algebra \(\Sigma\) such that \(\eta(\emptyset)=0\) is said to be a maxitive measure if \(\eta(\bigcup_iA_i)=\sup_i\eta(A_i)\) for every countable disjoint family \((A_i)\) in \(\Sigma\). In the paper, the following representation theorem for maxitive measures is established: Let \((S,\Sigma,\mu)\) be a \(\sigma\)-finite positive measure space and \(\eta\) be a maxitive measure on \(\Sigma\) vanishing on all \(\mu\)-null sets. Then there exists a unique, up to equality \(\mu\)-a.e., measurable function \(f:S\to \overline{\mathbb{R}}_+\) such that \(\eta(A)=\mu-\sup_Af\) for every \(A\in \Sigma\). This theorem is applied to represent a Köthe function \(M\)-space as \(L_{\infty}\)-space.