A Strongly Normalising System of Dependent Types for Transparent and Opaque Probabilistic Computation

We define an extension of lambda-calculus with dependents types that enables us to encode transparent and opaque probabilistic programs and prove a strong normalisation result for it by a reducibility technique. While transparent non-deterministic programs are formalised in a rather usual way, opaque non-deterministic programs are formalised by introducing in the syntax oracle constants, the behaviour of which is governed by oracular functions. The generality of these functions and the fact that their values are determined by the form of the whole term inside which the relative oracle occurs also enable us to simulate learning-like behaviours.