OAI/Probability/InvariantIsing/Arrays/NSpinTensorFrozen.lean
22 declarations, 0 contain sorry, 0 declared as axioms
A selected actual spectral/tree monomial as an independent Gaussian factor.
- abbrev
OAI.InvariantIsing.TensorGaussianKeyline 17 - def
OAI.InvariantIsing.tensorNamespaceTagline 22Source documentation
Namespace zero contains the linear field; namespace
j+1contains precisely thejth spectral tensor at all ancestor levels.def tensorNamespaceTag {N m k : ℕ} (I : Fin m → Finset (Fin N)) (degree : Fin k → Fin m → ℕ) (q : TensorGaussianKey I degree) : ℕ := …Mentions: OAI.InvariantIsing.TensorGaussianKey
- lemma
OAI.InvariantIsing.tensorNamespaceTag_injectiveline 28 - def
OAI.InvariantIsing.tensorNamespacedCoefficientsline 56 - theorem
OAI.InvariantIsing.tensorNamespacedFields_lawline 66Source documentation
Injective Gaussian reindexing identifies the entire countable spin/leaf field law with the flat tensor model.
theorem tensorNamespacedFields_law {N m k : ℕ} (U : Rotation N) (I : Fin m → Finset (Fin N)) (degree : Fin k → Fin m → ℕ) (amplitude : Fin k → ℝ) (n : ℕ) (v : Fin (n + 1) → SpinTensorIndex I degree → ℝ≥0) : gaussianCoordinates.map (fun g x => cylinderField (tensorLeafCoefficients U I degree amplitude n v x) g) = gaussianCoordinates.map (fun g x => cylinderField (tensorNamespacedCoefficients U I degree amplitude n v x) g) := by …Mentions: OAI.InvariantIsing.Rotation, OAI.InvariantIsing.SpinTensorIndex, OAI.InvariantIsing.tensorLeafCoefficients, OAI.InvariantIsing.tensorNamespacedCoefficients
- lemma
OAI.InvariantIsing.sqrt_varianceIncrementline 97 - def
OAI.InvariantIsing.tensorLinearCoefficientsline 102 - theorem
OAI.InvariantIsing.tensorNamespacedField_expandline 108Source documentation
The reindexed field is exactly the sum of the linear enrichment and the displayed independent spectral/tree Gaussian monomials.
theorem tensorNamespacedField_expand {N m k : ℕ} (U : Rotation N) (I : Fin m → Finset (Fin N)) (degree : Fin k → Fin m → ℕ) (amplitude : Fin k → ℝ) (n : ℕ) (treeDegree : Fin k → ℕ) (h : ℕ → ℝ) (hh : Monotone h) (h0 : 0 ≤ h 0) (x : Spin N × LabeledLeaf n) (g : ℕ → ℝ) : cylinderField (tensorNamespacedCoefficients U I degree amplitude n (fun i => tensorPathProfile I degree n treeDegree h i) x) g = cylinderField (tensorLinearCoefficients n h x) (fun i => g (Nat.pair 0 i)) + ∑ j, amplitude j * cylinderField (jointSpectralMonomialCoefficients U I (degree j) n (treeDegree j) x) (fun i => g (Nat.pair (j + 1) i)) := by …Mentions: OAI.InvariantIsing.Rotation, OAI.InvariantIsing.Spin, OAI.InvariantIsing.tensorNamespacedCoefficients, OAI.InvariantIsing.tensorPathProfile, OAI.InvariantIsing.tensorLinearCoefficients, OAI.InvariantIsing.jointSpectralMonomialCoefficients
- lemma
OAI.InvariantIsing.tensorNamespacedField_frozenline 140 - theorem
OAI.InvariantIsing.tensorNamespacedField_insertline 172Source documentation
Inserting the selected standard Gaussian family adds exactly its actual amplitude times
jointSpectralMonomialCoefficients.theorem tensorNamespacedField_insert {N m k : ℕ} (U : Rotation N) (I : Fin m → Finset (Fin N)) (degree : Fin k → Fin m → ℕ) (amplitude : Fin k → ℝ) (n : ℕ) (treeDegree : Fin k → ℕ) (h : ℕ → ℝ) (hh : Monotone h) (h0 : 0 ≤ h 0) (j : Fin k) (x : Spin N × LabeledLeaf n) (z : ℕ → ℝ) (g : OtherNamespaces (j + 1)) : cylinderField (tensorNamespacedCoefficients U I degree amplitude n (fun i => tensorPathProfile I degree n treeDegree h i) x) (gaussianNamespaceJoin (j + 1) (z, g)) = cylinderField (tensorNamespacedCoefficients U I degree (Function.update amplitude j 0) n (fun i => tensorPathProfile I degree n treeDegree h i) x) (gaussianNamespaceJoin (j + 1) (0, g)) + amplitude j * cylinderField (jointSpectralMonomialCoefficients U I (degree j) n (treeDegree j) x) z := by …Mentions: OAI.InvariantIsing.Rotation, OAI.InvariantIsing.Spin, OAI.InvariantIsing.tensorNamespacedCoefficients, OAI.InvariantIsing.tensorPathProfile, OAI.InvariantIsing.jointSpectralMonomialCoefficients
- lemma
OAI.InvariantIsing.measurable_tensorNamespacedFields_jointline 204 - theorem
OAI.InvariantIsing.tensorNamespacedGibbs_lawline 218Source documentation
The full finite-volume Gibbs-law distribution is preserved by the Gaussian namespace reindexing, for each actual rotation and cascade prior.
theorem tensorNamespacedGibbs_law {N m k : ℕ} (eig : Fin N → ℝ) (U : Rotation N) (c : Fin N → ℝ) (I : Fin m → Finset (Fin N)) (degree : Fin k → Fin m → ℕ) (amplitude : Fin k → ℝ) (n : ℕ) (v : Fin (n + 1) → SpinTensorIndex I degree → ℝ≥0) (T : LabeledTree n) : let ν := labeledSpinReference n (uniformSpinPrior N : Measure (Spin N)) T …Mentions: OAI.InvariantIsing.Rotation, OAI.InvariantIsing.SpinTensorIndex, OAI.InvariantIsing.uniformSpinPrior, OAI.InvariantIsing.Spin
- abbrev
OAI.InvariantIsing.TensorFrozenDataline 248 - def
OAI.InvariantIsing.tensorFrozenLawline 250 - instance
OAI.InvariantIsing.tensorFrozenLaw_probabilityline 254 - def
OAI.InvariantIsing.tensorFrozenBaseEnergyline 259 - def
OAI.InvariantIsing.tensorFrozenReferenceline 268 - instance
OAI.InvariantIsing.tensorFrozenReference_probabilityline 275 - lemma
OAI.InvariantIsing.measurable_tensorFrozenBaseEnergyline 281 - lemma
OAI.InvariantIsing.measurable_tensorFrozenReferenceline 315 - theorem
OAI.InvariantIsing.tensorFrozenEnergy_insertline 329 - theorem
OAI.InvariantIsing.tensorFrozenInsertion_measurePreservingline 345Source documentation
The selected field is a genuinely independent Gaussian product factor; all remaining marks, the rotation, and the tree form external disorder.
theorem tensorFrozenInsertion_measurePreserving {N : ℕ} (μ : Measure (SpecialOrthogonal N)) [IsProbabilityMeasure μ] (n : ℕ) (b : ℕ → ℝ) (j : ℕ) : MeasurePreserving (fun p : TensorFrozenData N n j × (ℕ → ℝ) => (p.1.1, gaussianNamespaceJoin (j + 1) (p.2, p.1.2))) ((tensorFrozenLaw μ n b j).prod gaussianCoordinates) ((μ.prod (labeledCascadeLaw n b : Measure (LabeledTree n))).prod gaussianCoordinates) := …Mentions: OAI.InvariantIsing.SpecialOrthogonal, OAI.InvariantIsing.TensorFrozenData, OAI.InvariantIsing.tensorFrozenLaw