Samplinglib
Lean gate not recorded for this source state main · 0e31a3cda412
ASTIS mathematical exposition

A Jensen gap yields nonnegative Γ through right-generator limits

AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.CarreDuChamp.carreDuChamp_nonneg_of_markov_jensen_rightGenerator · theorem · Teaching coverage

Statement

Let X be any type, L a real-linear operator on real functions, P_h a family of arbitrary operators indexed by h∈ℝ≥0, f:X→ℝ, and x∈X fixed. Assume (P_h f(x))²≤P_h(f²)(x) for every h>0; as h↓0 through positive nonnegative-real times, the difference quotients for f and f² converge respectively to Lf(x) and L(f²)(x); and P_h f(x)→f(x). Then ΓL(f,f)(x)≥0.

\[\begin{gathered}(P_hf(x))^2\le P_h(f^2)(x)\quad(h>0),\\\frac{P_hf(x)-f(x)}h\to Lf(x),\qquad\frac{P_h(f^2)(x)-f(x)^2}h\to L(f^2)(x),\qquad P_hf(x)\to f(x)\\\Longrightarrow\quad0\le\Gamma_L(f,f)(x)=\tfrac12\big[L(f^2)(x)-2f(x)Lf(x)\big].\end{gathered}\]

All objects and hypotheses

  • X is an arbitrary type, with no measurable-space, topological-space, normed-space, or finiteness hypothesis.
  • L:(X→ℝ)→ₗ[ℝ](X→ℝ) is a real-linear operator on all real-valued functions on X. No Markov semigroup or invariant law is supplied.
  • P:ℝ≥0→(X→ℝ)→X→ℝ is an arbitrary family; no linearity, semigroup law, positivity-preserving property, or value at h=0 is assumed of P.
  • f:X→ℝ and x∈X are fixed. hjensen holds for every positive h∈ℝ≥0, for this f and this x.
  • hf is convergence of (P_hf(x)−f(x))/(h:ℝ) to Lf(x); hf2 is convergence of (P_h(f*f)(x)−f(x)²)/(h:ℝ) to L(f*f)(x). Both filters are nhdsWithin 0 (Ioi 0) on ℝ≥0.
  • hcontinuous asserts only P_hf(x)→f(x) along that same right-hand filter. It is not operator strong continuity or continuity at every state.

Mathematical proof

1. Define the nonnegative scaled Jensen gap

For h>0 the numerator is nonnegative by hjensen and the denominator is positive, so the gap is nonnegative on the right-hand filter.

\[G(h):=\frac{P_h(f^2)(x)-(P_hf(x))^2}{2h}\ge0\quad(h>0).\]
Corresponding Lean step

gap; hgap_nonneg; self_mem_nhdsWithin; div_nonneg

2. Factor the difference of squares

Subtract and add f(x)², then factor (P_hf(x))²−f(x)². This algebra is used only for h>0, where cancellation is valid.

\[G(h)=\tfrac12\frac{P_h(f^2)(x)-f(x)^2}{h}-\tfrac12\frac{P_hf(x)-f(x)}h\big(P_hf(x)+f(x)\big).\]
Corresponding Lean step

hrewrite; field_simp; ring

3. Take the sum and product limits

The supplied right-continuity gives P_hf(x)+f(x)→2f(x). Combine this with hf to identify the product's limit.

\[P_hf(x)+f(x)\to2f(x),\qquad\frac{P_hf(x)-f(x)}h\big(P_hf(x)+f(x)\big)\to2f(x)Lf(x).\]
Corresponding Lean step

tendsto_const_nhds; Filter.Tendsto.add; Filter.Tendsto.mul

4. Identify the gap limit with Γ

Use hf2 and the preceding product limit, multiply by one-half, subtract, and unfold the diagonal Γ formula. The eventually valid factorization transfers the limit back to G.

\[G(h)\to\tfrac12L(f^2)(x)-\tfrac12Lf(x)(2f(x))=\Gamma_L(f,f)(x).\]
Corresponding Lean step

Filter.Tendsto.const_mul; Filter.Tendsto.sub; hvalue; Filter.Tendsto.congr'

5. Pass nonnegativity through the limit

The half-line [0,∞) is closed, and G is eventually in it along a nontrivial right-hand time filter. Therefore its limit lies there too.

\[G(h)\ge0\ \text{eventually},\quad G(h)\to\Gamma_L(f,f)(x)\Longrightarrow\Gamma_L(f,f)(x)\ge0.\]
Corresponding Lean step

isClosed_Ici.mem_of_tendsto hlimit hgap_nonneg

Lean statement · carreDuChamp_nonneg_of_markov_jensen_rightGenerator

The three convergence assumptions are scalar assertions at one selected state. No topology or measurable structure on X is needed. The theorem name describes the motivating Markov route, but the formal inputs already supply its Jensen and generator-limit consequences.

Braces mark parameters Lean can infer; square brackets request structures such as a measurable space or probability measure. Named hypotheses are mathematical premises, not facts established by this declaration. Section parameters are described in the mathematical hypotheses above; the module link retains their exact source context.

theorem carreDuChamp_nonneg_of_markov_jensen_rightGenerator
    (P : ℝ≥0 → (X → ℝ) → X → ℝ)
    (generator : (X → ℝ) →ₗ[ℝ] (X → ℝ))
    (f : X → ℝ) (x : X)
    (hjensen : ∀ h : ℝ≥0, 0 < h → (P h f x) ^ 2 ≤ P h (f * f) x)
    (hf : Tendsto
      (fun h : ℝ≥0 => (P h f x - f x) / (h : ℝ))
      (𝓝[>] 0) (𝓝 (generator f x)))
    (hf2 : Tendsto
      (fun h : ℝ≥0 => (P h (f * f) x - (f x) ^ 2) / (h : ℝ))
      (𝓝[>] 0) (𝓝 (generator (f * f) x)))
    (hcontinuous : Tendsto (fun h : ℝ≥0 => P h f x)
      (𝓝[>] 0) (𝓝 (f x))) :
    0 ≤ carreDuChamp generator f f x

Exact module and namespace context

Lean proof · carreDuChamp_nonneg_of_markov_jensen_rightGenerator

The proof constructs a nonnegative gap, factors a difference of squares, applies the supplied three limits, and uses closedness of the nonnegative real half-line. It never constructs P, establishes a semigroup law, or derives the difference-quotient hypotheses.

Braces mark parameters Lean can infer; square brackets request structures such as a measurable space or probability measure. Named hypotheses are mathematical premises, not facts established by this declaration. Section parameters are described in the mathematical hypotheses above; the module link retains their exact source context.

theorem carreDuChamp_nonneg_of_markov_jensen_rightGenerator
    (P : ℝ≥0 → (X → ℝ) → X → ℝ)
    (generator : (X → ℝ) →ₗ[ℝ] (X → ℝ))
    (f : X → ℝ) (x : X)
    (hjensen : ∀ h : ℝ≥0, 0 < h → (P h f x) ^ 2 ≤ P h (f * f) x)
    (hf : Tendsto
      (fun h : ℝ≥0 => (P h f x - f x) / (h : ℝ))
      (𝓝[>] 0) (𝓝 (generator f x)))
    (hf2 : Tendsto
      (fun h : ℝ≥0 => (P h (f * f) x - (f x) ^ 2) / (h : ℝ))
      (𝓝[>] 0) (𝓝 (generator (f * f) x)))
    (hcontinuous : Tendsto (fun h : ℝ≥0 => P h f x)
      (𝓝[>] 0) (𝓝 (f x))) :
    0 ≤ carreDuChamp generator f f x := by
  let gap : ℝ≥0 → ℝ := fun h =>
    (P h (f * f) x - (P h f x) ^ 2) / (2 * (h : ℝ))
  have hgap_nonneg : ∀ᶠ h in 𝓝[>] (0 : ℝ≥0), 0 ≤ gap h := by
    filter_upwards [self_mem_nhdsWithin] with h hh
    have hh0 : 0 < h := by simpa only [mem_Ioi] using hh
    exact div_nonneg (sub_nonneg.mpr (hjensen h hh0)) (by positivity)
  have hrewrite : ∀ᶠ h in 𝓝[>] (0 : ℝ≥0),
      gap h =
        (2 : ℝ)⁻¹ * ((P h (f * f) x - (f x) ^ 2) / (h : ℝ)) -
        (2 : ℝ)⁻¹ * (((P h f x - f x) / (h : ℝ)) *
          (P h f x + f x)) := by
    filter_upwards [self_mem_nhdsWithin] with h hh
    have hh0 : (h : ℝ) ≠ 0 := by
      have : 0 < h := by simpa only [mem_Ioi] using hh
      exact_mod_cast this.ne'
    dsimp [gap]
    field_simp
    ring
  have hlimit : Tendsto gap (𝓝[>] (0 : ℝ≥0))
      (𝓝 (carreDuChamp generator f f x)) := by
    have hconst : Tendsto (fun _ : ℝ≥0 => f x)
        (nhdsWithin 0 (Set.Ioi 0)) (nhds (f x)) :=
      tendsto_const_nhds
    have hsum : Tendsto (fun h : ℝ≥0 => P h f x + f x)
        (nhdsWithin 0 (Set.Ioi 0)) (nhds (f x + f x)) :=
      hcontinuous.add hconst
    have hprod := hf.mul hsum
    have hcombined := (hf2.const_mul (2 : ℝ)⁻¹).sub
      (hprod.const_mul (2 : ℝ)⁻¹)
    have hvalue :
        (2 : ℝ)⁻¹ * generator (f * f) x -
            (2 : ℝ)⁻¹ * (generator f x * (f x + f x)) =
          carreDuChamp generator f f x := by
      simp only [carreDuChamp]
      ring
    have htarget : Tendsto
        (fun h : ℝ≥0 =>
          (2 : ℝ)⁻¹ * ((P h (f * f) x - (f x) ^ 2) / (h : ℝ)) -
          (2 : ℝ)⁻¹ * (((P h f x - f x) / (h : ℝ)) *
            (P h f x + f x)))
        (𝓝[>] 0) (𝓝 (carreDuChamp generator f f x)) := by
      rw [← hvalue]
      exact hcombined
    exact htarget.congr' (hrewrite.mono fun h hh => hh.symm)
  exact isClosed_Ici.mem_of_tendsto hlimit hgap_nonneg

/-- Chewi Theorem 1.2.14: stationarity and generator symmetry imply the
fundamental integration-by-parts identity between the Dirichlet form and the
integrated carre du champ.

The three integrability hypotheses are the exact terms expanded from Gamma;
they prevent the totalized Bochner integral from hiding a domain failure. -/

Exact module and namespace context

Scope and omitted-condition boundaries

  • No global invariant-law, stationarity for every observable, reversibility, semigroup construction, or concrete diffusion identity is inferred from this declaration.
  • The conclusion is only at the supplied f and x; a global statement requires those inputs for every relevant observable and state.
  • The limit excludes h=0; no division-by-zero argument or two-sided derivative is used.
  • No diffusion chain rule or invariant-law theorem is produced.

Source and reuse

ASTIS parents called

Mathlib API called (external library)

  • self_mem_nhdsWithin
  • div_nonneg
  • sub_nonneg
  • tendsto_const_nhds
  • Filter.Tendsto.add
  • Filter.Tendsto.mul
  • Filter.Tendsto.const_mul
  • Filter.Tendsto.sub
  • Filter.Tendsto.congr'
  • isClosed_Ici
  • IsClosed.mem_of_tendsto

Mathematical sources

ASTIS prose is not a quotation or a source-equivalence certificate. Definitions and aliases are explained as constructions, not counted as new mathematical proofs.