States on Bold Algebras: Categorical Aspects

• Published in: Journal of logic and computation 2011-06, Vol.21 (3), p.465-477
• Main Author: Fric, R.
• Format: Article
• Language: English
• Subjects: Algebra; Categories; Construction; Fuzzy; Fuzzy logic; Fuzzy set theory; Probability theory; Random variables
• ISSN: 0955-792X; 1465-363X
• DOI: 10.1093/logcom/exp014

We study bold algebras and states on bold algebras in the context of transition from classical probability theory to fuzzy probability theory. Our aim is to point out the role of bold algebras and states on bold algebras in a categorical approach to probability theory. In particular, we formulate several fundamental questions related to basic probability notions and constructions and provide possible answers in terms of bold algebras and states on bold algebras. We show that the category ID of D-posets of fuzzy sets and sequentially continuous difference homomorphisms can serve as a base category in which both classical and fuzzy probability theory can be developed and generalized. Classical and fuzzy random events such as fields of sets and measurable real-valued functions into the interval [0,1], considered as bold algebras, become special objects. Observables, considered as morphisms between objects, become dual to generalized random variables. States become morphisms into [0,1], considered as an object of ID. Properties of objects of ID follow from classical theorems of analysis such as the Lebesgue Dominated Convergence Theorem (states are sequentially continuous) and categorical constructions such as the product (the structure of a probability domain is completely determined by the states as the initial structure). We prove that each generated Lukasiewicz tribe is the epireflection of its underlying Butnariu-Klement sigma -field of sets. This helps to understand the transition from classical crisp random events to fuzzy random events. Indeed, the corresponding fuzzification is necessary to cover generalized random variables having a quantum character, i.e. fuzzy random variables in the Gudder-Bugajski sense sending a classical elementary event (point measure) to a non-trivial probability measure.