A speech about generative datalog and non-measurable sets

Generative Datalog is the first component of PPDL (short for Probabilistic-Programming Datalog), a recently proposed probabilistic programming language. Specifically, generative Datalog provides constructs to refer to parameterized probability distribution, and is used for the specification of stochastic processes. Possible outcomes of such a stochastic process are possibly filtered according to logical constraints, which constitute the second component of PPDL. This speech is about generative Datalog, and hints on the possibility to represent non-measurable sets by combining generative Datalog constructs with addition over real numbers and a single, atomic, ground constraint.