Measures of inset information on open domains. II: Additive inset entropies with measurable sum property

Functional equations for functions with more general domains and/or ranges (39B52) Functional equations and inequalities (39B99) Measures of information, entropy (94A17)

Review text

In this (second) part of a series on inset entropies on open domains [part I is reviewed above (Zbl 0591.94012), part III below (Zbl 0591.94014)], we find explicit forms of those measures of inset information which are (3,2)-additive and have the measurable sum property. In the case that our ring S of sets is not an algebra, the entropy must have the form \[ I_ n\left( \begin{matrix} x_ 1,...,x_ n\\ p_ 1,...,p_ n\end{matrix} \right)=A\sum^{n}_{i=1}p_ i \log p_ i+B(1+n) \] for some constants A and B. (Here \((p_ 1,...,p_ n)\) is a complete distribution of positive probabilities, and \((x_ 1,...,x_ n)\) is a pairwise disjoint distribution of elements of S.) So the sets \(x_ i\) have no role in the measurement of the information content. The situation is much more interesting, however, if S is an algebra of subsets of some universal set \(\Omega\), and if \((x_ 1,...,x_ n)\) is required to be a partition of \(\Omega\) (which seems more natural for applications). Then it turns out (at least if S is non-atomic) that \[ I_ n\left( \begin{matrix} x_ 1,...,x_ n\\ p_ i,...,p_ n\end{matrix} \right)=\sum^{n}_{i=1}[Ap_ i+e(x_ i)\quad] \log p_ i+B(1+n), \] where e is a real-valued additive set function on S.

Recommendations

Cites work

Cited in

(10)

This page was built for publication: Measures of inset information on open domains. II: Additive inset entropies with measurable sum property

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1075290)