This paper introduces the Deterministic Hybrid Regulation and Semantic Reasoning (DHRSR) framework, a formal class of cyberphysical regulators designed to bridge the gap between probabilistic Large Language Model (LLM) reasoning and deterministiccontrol theory. We define DHRSR through five structural axioms: Bounded State Representation, Deterministic Action Kernel,Semantic Grounding, Temporal Causal Closure, and Homeostatic Stability. These axioms provide a necessary and sufficient formal basis to distinguish DHRSR-class systems from classical controllers and unstructured AI pipelines.
Martyna Benderyszyn (Fri,) studied this question.