Classification of concatenation measurement structures according to scale type (Q1066844): Difference between revisions

A concatenation structure consists of a triple \(<X,\preccurlyeq,\circ >\) where X is a non-empty set, \(\preccurlyeq\) is a total order and \(\circ\) is a binary relation; in addition certain properties relating the binary relation and the total ordering hold. Concatenation structures abstract the notion of measurement to a situation in which the entities being measured can be combined in some way. Thus a ''standard'' concatenation structure would be the real numbers with their natural ordering and with addition as the operation. This paper is concerned to classify concatenation structures through the scale types that they induce on the relational structure \(<X,\preccurlyeq >\). To any relational structure one can associate a scale type (M,N) where M is the largest degree of homogeneity of the structure and N is the largest degree of uniqueness. Homogeneity measures the length of the longest chain \(x_ 1\preccurlyeq...\preccurlyeq x_ m\) such that any two chains of that length can be mapped to one another by an order preserving map of X. Uniqueness is measured in terms of the longest chain on which one can specify the values of an automorphism of X. It is shown that for concatenation structures with \(M>0\) and \(N<\infty\) the only scale types are (1,1), (1,2) and (2,2). Furthermore, it is shown that for \(M>0\) concatenation structures are all isomorphic to numerical ones for which the operation can be written \(x\circ y=yf(x/y).\) In order to obtain representations of concatenation structures the authors introduce the notion of a conjoint structure. A conjoint structure is a special binary relation on a set \(A\times P\). Among the conditions that the relation must satisfy for conjointness are requirements that it induces total ordering on A and P. As an example of a conjoint structure one can let A stand for a collection of substances (gold, silver, lead, etc.) and let P be the positive reals. The partial order then enables one to compare a quantity of one substance with a different quantity of another substance. Given a concatenation structure \(<X,\preccurlyeq,\circ >\) one gets an associated conjoint structure on \(X\times X\) defined by uv\(\geq xy\) iff \(u\circ v\geq x\circ y\). If one has a conjoint structure on \(A\times P\) a notion of distributivity of the binary operation of A in the conjoint structure is given. This ''distributive interlock'' provides the means to produce representations of concatenation structures. In particular the following result is shown: Theorem. Suppose that \({\mathcal X}=<X,\geq,\circ >\) is a concatenation structure that is totally ordered, closed, dense and Archimedean. Then \({\mathcal X}\) is either 1- or 2- point unique and is isomorphic to a real closed structure \(<R,\geq,\otimes >\) where \(R\subseteq Rc^+.\) In the final section of the paper the results are applied to utility theory, in particular the utility of gambles.