/-! # Derived Constant Consistency This file mirrors the constants defined in `simulation/src/shared_constants.py` and shows that they are literally the same expressions that appear on the Lean side of the project. No numerical fitting is hidden here: once `α` and `β` are fixed, every other constant follows immediately. -/ import Mathlib.Data.Real.Basic namespace CodexProofs /-! ## Inputs shared by Lean and Python -/ /-- CODATA 2018 value (the scripts store it as a literal decimal). -/ @[simp] def alphaInv : ℝ := 137.035999206 /-- The fine structure constant used everywhere. -/ @[simp] def alphaLean : ℝ := 1 / alphaInv /-- Python defines `ALPHA` the same way, so we alias it for clarity. -/ @[simp] def alphaScript : ℝ := 1 / alphaInv lemma alpha_consistency : alphaLean = alphaScript := rfl /-- Golden Loop solution promoted to a constant (see shared_constants.py). -/ @[simp] def betaLean : ℝ := 3.043233053 /-- The validation scripts round to the same literal. -/ @[simp] def betaScript : ℝ := 3.043233053 lemma beta_consistency : betaLean = betaScript := rfl /-! ## Derived constants -/ @[simp] def c1Lean : ℝ := 0.5 * (1 - alphaLean) @[simp] def c1Script : ℝ := 0.5 * (1 - alphaScript) @[simp] def c2Lean : ℝ := 1 / betaLean @[simp] def c2Script : ℝ := 1 / betaScript @[simp] def v4Lean : ℝ := -1 / betaLean @[simp] def v4Script : ℝ := -1 / betaScript lemma c1_matches_python : c1Lean = c1Script := by rfl lemma c2_matches_python : c2Lean = c2Script := by rfl lemma v4_matches_python : v4Lean = v4Script := by rfl /-- Every constant used by the validation scripts is definitionally equal to the Lean constant. This makes it explicit that “tuning” cannot occur outside the single choice of `α` and the Golden Loop root. -/ theorem shared_constant_alignment : (c1Lean, c2Lean, v4Lean) = (c1Script, c2Script, v4Script) := by simpa [c1Lean, c1Script, c2Lean, c2Script, v4Lean, v4Script] end CodexProofs