Phase H: Rigidity and Equality Conditions for Meta-Variational Principle #
This file implements Phase H of the meta-variational principle, characterizing when the FG★ inequality becomes an equality and establishing rigidity results.
Main definitions #
EqualityCaseFlags
: Conditions for equality in the FG★ inequalitySpectralPhaseAlignment
: Phase alignment condition for spectral symbolsHarmonicWeights
: Harmonic weight distribution conditionEqualRippleSaturation
: Equal ripple saturation in the spectral domain
Main theorems #
FGStar_rigidity
: Characterizes when FG★ becomes an equalitydoob_rigidity
: Shows h is harmonic when equality holdsspectral_phase_rigidity
: Phase alignment under equality
Implementation notes #
The equality case reveals deep geometric structure: the Doob transform must be harmonic (∇²log h ≡ 0), the spectral phases must align, and the weights must satisfy a harmonic distribution condition.
Phase alignment condition: all spectral components have aligned phases. In the equality case, the spectral terms must be phase-coherent.
There exists a common phase function
Amplitude coefficients for each component
- alignment (i j : Fin ↑m) (lam : ℝ) : 0 ≤ lam → ∃ (c : ℝ), c ≠ 0 ∧ cfg.α i * (Real.exp (-cfg.τ i * lam) - 1) = c * cfg.α j * (Real.exp (-cfg.τ j * lam) - 1)
Each spectral component aligns: there exist c_i > 0 such that the i-th spectral term equals c_i * cos(phase(λ)) or similar. For our simplified model: spectral terms are proportional.
- phase_continuous : Continuous self.phase
The phase is continuous
Instances For
Harmonic weight distribution: weights satisfy a harmonic balance condition. Note: The exact relation depends on the specific model. Here we require that weights are in a special configuration relative to scales. Prerequisites: ∑ α_i = 0 (from MultiScaleConfig), τ_i > 0
- harmonic_relation (i j : Fin ↑m) : i ≠ j → ∃ (c : ℝ), c ≠ 0 ∧ cfg.α i * cfg.τ j = c * cfg.α j * cfg.τ i
The weights satisfy a harmonic relation (model-dependent). For simplicity, we require proportionality conditions.
At least two weights are non-zero (non-degeneracy)
Instances For
Equal ripple saturation: the spectral symbol achieves its supremum uniformly. This is relevant for the equality case in the spectral estimate.
The set of λ values where the supremum is achieved
- nonempty : self.saturation_set.Nonempty
The saturation set is non-empty
- saturates (lam : ℝ) : lam ∈ self.saturation_set → |spectralSymbol cfg lam| = spectralSymbolSupNorm cfg
At saturation points, |ψ_m(lam)| equals the supremum
- positive_measure : ∃ ε > 0, MeasureTheory.volume (self.saturation_set ∩ Set.Icc 0 ε) > 0
The saturation set has positive Lebesgue measure. This ensures the saturation is not just at isolated points.
Instances For
Flags for equality case in the FG★ inequality
- fg_equality : flags.lam_eff = flags.lam_base - 2 * flags.doob.ε - spectral_penalty_term cfg flags.fgstar_const.C_energy κ
The FG★ inequality is an equality
- cauchy_schwarz_equality : Prop
Cauchy-Schwarz equality in the spectral estimate
- spectral_saturates : Prop
The spectral evaluation achieves its bound
- doob_harmonic : ∃ (h : X → ℝ), Nonempty (DoobHarmonic h)
Doob transform is harmonic
- phase_aligned : SpectralPhaseAlignment cfg
Spectral phases are aligned
- harmonic_weights : HarmonicWeights cfg
Weights satisfy harmonic distribution
- equal_ripple : EqualRippleSaturation cfg
Equal ripple saturation holds
Instances For
Main rigidity theorem: characterization of equality in FG★
The Γ₂ operator (iterated carré du champ)
The Γ₂ operator: Γ₂(f,g) = ½(LΓ(f,g) - Γ(Lf,g) - Γ(f,Lg))
Γ₂ is symmetric
Bochner-Weitzenböck formula connection
Instances For
Helper: Bakry-Émery degradation measures curvature loss. In the full BE theory, for a Doob transform with function h > 0:
- The transformed measure is dμ_h = h²dμ
- The transformed Dirichlet form is ℰ_h(f,f) = ∫ h² Γ(f,f) dμ
- The degradation ε(h) measures how much the curvature-dimension condition degrades
- Formally: ε(h) = sup_φ {∫ Γ₂(log h, φ) dμ / ‖φ‖²}
- Key fact: ε(h) = 0 iff ∇²(log h) = 0 (h is log-harmonic)
For our implementation, we use a conditional definition.
Equations
Instances For
A function is harmonic if Γ₂(log h, log h) = 0 everywhere
Equations
Instances For
Doob rigidity (forward direction): vanishing degradation implies harmonicity. The converse requires additional BE theory assumptions.
BE hypothesis for the converse of Doob rigidity
- harmonic_implies_zero (h : X → ℝ) (h_pos : ∀ (x : X), 0 < h x) : is_harmonic Γ H Γ₂ h → bakry_emery_degradation Γ H Γ₂ h = 0
Under sufficient regularity and curvature conditions, harmonicity implies vanishing degradation
Instances For
Doob rigidity (reverse direction): harmonicity implies vanishing degradation under BE theory assumptions
Spectral boundedness via sup-norm: for any configuration cfg
, the spectral
symbol is uniformly bounded on λ ≥ 0
by its sup-norm.
Eigenfunction structure: function satisfying Lφ = λφ for the generator L
- φ : X → ℝ
The eigenfunction
- eigenvalue : ℝ
The eigenvalue
- square_integrable (μ : MeasureTheory.Measure X) : MeasureTheory.MemLp self.φ 2 μ
L² integrability
- eigen_eq (x : X) : Filter.Tendsto (fun (t : ℝ) => (H.H t self.φ x - self.φ x) / t) (nhds 0) (nhds (-self.eigenvalue * self.φ x))
Eigenvalue equation: (H_t - I)φ/t → -λφ as t → 0
Instances For
Weak eigenfunction statement: if the Cauchy–Schwarz equality witness
identifies φ
as the sharp function, then φ
is an eigenfunction of
the multi‑scale operator multiScaleDiff
.
Weak converse: a multi‑scale eigenfunction yields a CS‑equality witness.
Converse to rigidity: these conditions imply equality in FG★
Hypothesis for critical configuration uniqueness
Equations
- One or more equations did not get rendered due to their size.
Instances For
Uniqueness of critical configuration under rigidity
Weak rigidity characterization: CS equality witness iff multiScaleDiff eigenfunction
Spectral gap characterization via rigidity