Authority-Constrained Decision Science (ACDS)

Authority-Constrained Decision Science (ACDS) is introduced as a formal, domain-agnostic framework for engineered decision systems operating under irreversibility, constraint intersection, and non-transferable authority. This paper establishes the axiomatic foundations of ACDS and derives theorem-level results including Authority Invariance, Feasible-Set Lock (hard-stop outside constraint manifolds), and Deterministic Replay Consistency under canonicalized inputs and fixed policy versions. The framework separates prediction from decision authority, models irreversibility as a time-asymmetric risk topology, and formalizes governance-constrained outputs (HOLD / NO-EXEC / LOCK) as valid decision outcomes in high-impact environments. A differential formulation for irreversibility dynamics is proposed, linking reversibility windows, exit costs, constraint pressure, and shock intensity to structurally stable decision gating. ACDS is positioned not as an AI model, but as a decision systems architecture applicable to capital-intensive, safety-critical, and regulated domains requiring audit-grade accountability and deterministic replay.