Torus / Count Bridge For The Cn^3 Formalization #
This module contains the torus-coordinate model, the Fourier inversion bridge, and the normalized-count identification used throughout the formalization.
Readers primarily interested in the combinatorial-to-analytic translation should start with:
Equations
Equations
The edge-coordinate space has dimension n.choose 2.
Edge monomials attached to a Boolean sign vector.
Equations
- Cn3Torus.Z y e = Cn3Torus.spin (y (↑e).1) * Cn3Torus.spin (y (↑e).2)
Instances For
The edge-model phase.
Equations
- Cn3Torus.phase lam y = ∑ e : Cn3Torus.Edge n, lam e * Cn3Torus.Z y e
Instances For
Edge-model characteristic function.
Equations
- Cn3Torus.psi n lam = (∑ y : Fin n → Bool, Complex.exp (Complex.I * ↑(Cn3Torus.phase lam y))) / 2 ^ n
Instances For
Edge-model squared norm.
Equations
- Cn3Torus.sqNormEdge n lam = ∑ e : Cn3Torus.Edge n, lam e ^ 2
Instances For
Complex torus integral before taking real parts.
Equations
- Cn3Torus.torusIntegralC n m = ∫ (lam : Cn3Torus.Edge n → ℝ) in Cn3Torus.torusBox n, Cn3Torus.psi n lam ^ m
Instances For
Real quarter-scale torus integral.
Equations
- Cn3Torus.targetIntegral n t = ∫ (lam : Cn3Torus.Edge n → ℝ) in Cn3Torus.torusBox n, (Cn3Torus.psi n lam ^ (4 * t)).re
Instances For
Torus integral normalized by the torus volume (2π)^d.
Equations
- Cn3Torus.normalizedTargetIntegral n t = 1 / (2 * Real.pi) ^ Cn3Torus.d n * Cn3Torus.targetIntegral n t
Instances For
Edge-count as a real number.
Equations
- Cn3Torus.edgeCount n = ↑(Fintype.card (Cn3Torus.Edge n))
Instances For
Fixed local radius used in the torus decomposition package.
Equations
- Cn3Torus.delta = 3 * Real.pi / 16
Instances For
Local box around a lattice point in edge coordinates.
Equations
- Cn3Torus.localBox n = Set.univ.pi fun (x : Cn3Torus.Edge n) => Set.Icc (-Cn3Torus.delta) Cn3Torus.delta
Instances For
A single coordinate square is bounded by the total edge squared norm.
Local integral over the core box.
Equations
- Cn3Torus.localIntegral n t = ∫ (lam : Cn3Torus.Edge n → ℝ) in Cn3Torus.localBox n, (Cn3Torus.psi n lam ^ (4 * t)).re
Instances For
The tex prefactor converting local integrals to normalized torus scale.
Equations
- Cn3Torus.texPrefactor n = 2 ^ (2 * Cn3Torus.d n - n + 1) / (2 * Real.pi) ^ Cn3Torus.d n
Instances For
Fourier inversion on the quarter-scale torus model.
This identifies the real-valued target integral used in the paper with the real
part of the complex torus integral torusIntegralC.
The edge-model polynomial W, equal to phase.
Equations
- Cn3Torus.W mu y = ∑ e : Cn3Torus.Edge n, mu e * Cn3Torus.Z y e
Instances For
The edges incident to a given vertex.
Equations
- Cn3Torus.edgesIncident n i = {e : Cn3Torus.Edge n | (↑e).1 = i ∨ (↑e).2 = i}
Instances For
Row-degree of an edge monomial at vertex i.
Equations
- Cn3Torus.edgeNatRowSum b i = ∑ e ∈ Cn3Torus.edgesIncident n i, b e
Instances For
Edge monomial in the sign variables.
Equations
- Cn3Torus.edgeZMonomial b y = ∏ e : Cn3Torus.Edge n, Cn3Torus.Z y e ^ b e
Instances For
Edge multiplicities attached to an ordered 4-tuple of edges.
Equations
Instances For
Equations
Equations
Equations
Equations
Equations
- Cn3Torus.lambdaShifts n = Finset.image (fun (b : Cn3Torus.LambdaCode✝ n) => Cn3Torus.lambdaReal✝ b) (Cn3Torus.lambdaCodes✝ n)
Instances For
Edge Coordinate Transport #
The public transport maps edgeLam and matrixOfEdge identify matrix and edge
coordinates. The Boolean-encoding conversions below are kept internal so the
reader sees the transport results rather than the implementation vocabulary.
Pointwise transport of sign vectors to the Bool-valued edge model.
Equations
- signVecToBoolVec σ i = fin2ToBool✝ (σ i)
Instances For
Equivalence between the two encodings of sign vectors.
Equations
- signVecEquivBoolVec n = { toFun := signVecToBoolVec, invFun := boolVecToSignVec✝, left_inv := ⋯, right_inv := ⋯ }
Instances For
Edge-to-matrix transport is continuous coordinatewise.
Exact count-to-integral bridge for the torus model.
The torus integral equals the torus volume times the number of n × s partial
Hadamard matrices, divided by the ambient sign count 2^(ns). This is the
formal Fourier-inversion step connecting the combinatorial count to the
analytic integral.
Edge-coordinate transport of the Euclidean ball D_r.
Equations
- edgeEuclidBall n r = {mu : Cn3Torus.Edge n → ℝ | matrixOfEdge n mu ∈ euclidBall'✝ n r}
Instances For
Edge-coordinate transport of the core region D_t.
Equations
- edgeCoreRegion n t = {mu : Cn3Torus.Edge n → ℝ | matrixOfEdge n mu ∈ coreRegion n t}
Instances For
Edge-coordinate transport of the even far shell.
Equations
- edgeEvenFarShell n r = {mu : Cn3Torus.Edge n → ℝ | matrixOfEdge n mu ∈ evenFarShell✝ n r}
Instances For
Residual torus region left after removing all translated primary boxes.
Equations
- Cn3Torus.edgeResidualTorusRegion n delta = Cn3Torus.torusBox n \ Cn3Torus.edgePrimaryBoxUnion✝ n delta
Instances For
Fourier Inversion and Count Bridge #
This section packages the de Launey-Levin quarter-scale Fourier bridge in the form used by the rest of the active development.
Fact 2.3 [DL10]: quarter-scale Fourier inversion in the edge-coordinate torus model. This is the exact bridge from the combinatorial normalized count to the normalized torus integral.
Manuscript Fact 4.5(i): the degree-2 chaos W satisfies the L^6
hypercontractive bound with constant 5^6.
The actual primary-secondary torus decomposition from the manuscript, written in edge coordinates and normalized directly at the normalized-count scale.
The global normalized count is within the exponentially small Rdelta error
of the transported local-box contribution.