Exact Positive-Cone Propagation
Abstract
The positive-cone crossing sandwich lowers the exact winding phase by two.
Theorem 1.1 (The crossing sandwich lowers the winding phase by two).
Proof. Machine-checked in Lean as D5/S3/PrimeForms/Crossing/ExactPropagation.exact_propagation_positive_cone (✓ std3). ∎
Source. Repository-derived.
Commentary.
Let T, b, and c be natural numbers with 2 <= T, b <= c, and b^2 + c^2 + 1 = T^2 + bc. Define gamma = [[T+c-b,b],[c,T+b-c]] and M = [[3,1],[2,1]]. The cone equation forces every coefficient used by the phase formula to be positive; in particular T+b-c > 0.
Direct multiplication gives the complete lower-left chain c, 2, c+2T+2b, and 8T+7c. Completing the square in the cone equation gives c-b <= sqrt(T^2-1), while sqrt(T^2-1) < T. These two bounds prove the displayed inequality and make the endpoint sign corrections positive.
The phase proof uses the repository’s finite rational Dedekind sum and its proved reciprocity theorem. A residue-permutation argument first proves invariance under inverse numerators. Two reciprocity calculations then establish the right and left multiplication corrections separately, each with correction -3; no cocycle law is assumed. Since the fixed matrix has Phi(M)=2, the winding phase of M gamma M is exactly Psi(gamma)-2.
Local searches in D5 and pinned Mathlib found no ready-made Rademacher phase or cocycle theorem. The exact imported hits are the finite Dedekind sum, its residue-permutation lemma, and Dedekind reciprocity. Loogle returned Unknown identifier Rademacher; the grep.app query returned HTTP 503, and the attempted LeanSearch endpoint returned HTTP 404.
References
- Truth anchor:
D5/S3/PrimeForms/Crossing/ExactPropagation.exact_propagation_positive_cone - Dependency: D5/S1/Phase/Interference/DedekindReciprocity