Samplinglib
Lean gate not recorded for this source state main · 9bb7a7394ab0
Samplinglib mathematical lesson

Dissipation with forcing converges to a floor

AutoSamplingTheory.TechnicalLemmas.FunctionalInequalities.SemigroupDecay.forced_exponential_decay_of_scaled_dissipation_from · theorem · Teaching coverage

Read this first

Mathematical statement

Let an energy curve have the exact right derivative E'(u)=-qD(u). If r>0 and rE(u)≤qD(u)+b at every time, then between any s≤t the energy is bounded by the exponential interpolation between E(s) and the forcing floor b/r.

\[E(t)\le E(s)e^{-r(t-s)}+\frac{b}{r}\bigl(1-e^{-r(t-s)}\bigr)\]
Lean statement · forced_exponential_decay_of_scaled_dissipation_from

Lean keeps the dissipation scale q, decay rate r, forcing b, curve, and both endpoint times explicit.

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 forced_exponential_decay_of_scaled_dissipation_from
    {scale rate forcing : ℝ} (hrate : 0 < rate)
    (curve : DissipationCurve scale)
    (hcoercive : ∀ u : ℝ,
      rate * curve.energy u ≤ scale * curve.dissipation u + forcing)
    {s t : ℝ} (hst : s ≤ t) :
    curve.energy t ≤
      curve.energy s * Real.exp (-rate * (t - s)) +
        (forcing / rate) * (1 - Real.exp (-rate * (t - s)))

Exact module and namespace context

Proof roadmap

  1. 1Convert coercivity into a differential inequality

    Negating the forced coercivity inequality and using the derivative identity gives E'(u)≤-rE(u)+b.

  2. 2Apply one-sided Grönwall

    Mathlib's right-slope Grönwall theorem applies directly because the curve is continuous and the exact right derivative controls every right slope.

  3. 3Expose the forcing floor

    Since r is nonzero, unfold the closed Grönwall expression and rewrite b/(-r)(e^{-rΔ}-1) as (b/r)(1-e^{-rΔ}).

Full assumptions and notation

All objects and hypotheses

  • The energy curve is continuous on the real line.
  • At each time u, its right derivative within [u,∞) is exactly -qD(u).
  • The decay rate r is strictly positive.
  • The forced coercivity inequality rE(u)≤qD(u)+b holds for every real u.
  • The terminal time satisfies s≤t.

Mathematical proof

1. Convert coercivity into a differential inequality

Negating the forced coercivity inequality and using the derivative identity gives E'(u)≤-rE(u)+b.

\[E'(u)=-qD(u)\le-rE(u)+b\]
Corresponding Lean step

The proof combines neg_le_neg (hcoercive u) with linear arithmetic.

2. Apply one-sided Grönwall

Mathlib's right-slope Grönwall theorem applies directly because the curve is continuous and the exact right derivative controls every right slope.

\[E(t)\le\operatorname{gronwallBound}(E(s),-r,b,t-s)\]
Corresponding Lean step

Use le_gronwallBound_of_liminf_deriv_right_le with K=-r and ε=b.

3. Expose the forcing floor

Since r is nonzero, unfold the closed Grönwall expression and rewrite b/(-r)(e^{-rΔ}-1) as (b/r)(1-e^{-rΔ}).

\[\frac{b}{-r}(e^{-r\Delta}-1)=\frac{b}{r}(1-e^{-r\Delta})\]
Corresponding Lean step

Unfold gronwallBound; field_simp and ring verify the scalar identity.

Lean verification

The exact proof begins folded immediately after the mathematical argument; the statement is folded beside the claim above.

Lean proof · forced_exponential_decay_of_scaled_dissipation_from

The proof is a direct specialization of Mathlib's one-sided Grönwall theorem, followed by exact field algebra.

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.

Download module (.lean)
theorem forced_exponential_decay_of_scaled_dissipation_from
    {scale rate forcing : ℝ} (hrate : 0 < rate)
    (curve : DissipationCurve scale)
    (hcoercive : ∀ u : ℝ,
      rate * curve.energy u ≤ scale * curve.dissipation u + forcing)
    {s t : ℝ} (hst : s ≤ t) :
    curve.energy t ≤
      curve.energy s * Real.exp (-rate * (t - s)) +
        (forcing / rate) * (1 - Real.exp (-rate * (t - s))) := by
  have hbound : ∀ x ∈ Ico s t,
      -scale * curve.dissipation x ≤
        (-rate) * curve.energy x + forcing := by
    intro x _hx
    have hneg := neg_le_neg (hcoercive x)
    linarith
  have hgronwall :=
    le_gronwallBound_of_liminf_deriv_right_le
      (f := curve.energy)
      (f' := fun x => -scale * curve.dissipation x)
      (δ := curve.energy s)
      (K := -rate)
      (ε := forcing)
      (a := s)
      (b := t)
      curve.energy_continuous.continuousOn
      (fun x _hx r hr => by
        simpa [slope] using
          (curve.energy_hasDerivWithinAt x).liminf_right_slope_le hr)
      le_rfl
      hbound
      t
      ⟨hst, le_rfl⟩
  rw [gronwallBound] at hgronwall
  simp only [if_neg (neg_ne_zero.mpr hrate.ne')] at hgronwall
  calc
    curve.energy t ≤
        curve.energy s * Real.exp (-rate * (t - s)) +
          forcing / -rate *
            (Real.exp (-rate * (t - s)) - 1) := hgronwall
    _ = curve.energy s * Real.exp (-rate * (t - s)) +
          (forcing / rate) *
            (1 - Real.exp (-rate * (t - s))) := by
      field_simp [hrate.ne']
      ring

end

end SemigroupDecay
end FunctionalInequalities
end TechnicalLemmas
end AutoSamplingTheory

Exact module and namespace context

Scope and omitted-condition boundaries
  • The theorem does not construct a semigroup or prove a Bakry–Émery gradient identity. It consumes an exact right-derivative dissipation law and a forced coercivity inequality.
Sources, dependencies, and reuse

Source and reuse

Samplinglib parents called

Mathlib API called (external library)

  • le_gronwallBound_of_liminf_deriv_right_le
  • HasDerivWithinAt.liminf_right_slope_le
  • gronwallBound

Mathematical sources

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