Douglas Ashcroft; William Prescott. Synergizing Symbolic Logic and Reinforcement Learning for Provably Correct Reasoning in Large Language Model Decision Pipelines. CIS 2026, 4.