An excursion in finite-field-valued measures (Q1585842)

The aim of this paper is to extend classical finite probability theory to a theory over arbitrary finite fields. The axiomatic setup is as follows. Let \(\Omega\) be a finite set, let \({\mathcal B}\) be an algebra of subsets of \(\Omega\), and let \(\mathbb{F}\) be a finite field with addition \(\oplus\). By analogy with the word ``probability'', the term galoibility (after Galois) is used to designate a function \(Q:{\mathcal B}\to\mathbb{F}\) satisfying: (i) \(Q(\Omega)=1\) and (ii) \(Q(A\cup B)=Q(A) \oplus Q(B)\) whenever \(A,B\in{\mathcal B}\) and \(A\cap B= \emptyset\). A function \(x:\Omega\to\mathbb{F}\) is said to be a Galois random variable if \(x^{-1}(\{f\}) \in{\mathcal B}\) for any \(f\in\mathbb{F}\). Within this setup, the notions of conditioning and expectation are then taken up, and some initial ideas about Bernoulli trials and random walk are presented. A special attention is paid to the case \(\mathbb{F}= \text{GF}(p)\) for \(p\) prime.