Stochastic Lexical Calculus for Language Model Probabilities

This framework defines formal conditions for transforming prompt-dependent language model probabilities into stable, sequential state representations on a probability simplex. The method provides mathematical bounds on irreducible nonclosure and accumulated error without requiring internal model architecture changes.