Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition 4.1, estimate (11) — under A2′ the shadowing error is at most C(T)[Δ(t − 1, T + 1) + sup γ̄]

Proved
StochApproxDyn.Interpolation.shadowing_estimate

by mikedeng1 · Oct 4, 2026 · Mathlib 0df444a (Lean v4.33.1)

dynamical-systemsp2o-batch-pfp1ap2o-gran-per-chapterp2o-plan-paperp2o-v1stochastic-approximation

Let F:Rd→RdF:\mathbb R^d\to\mathbb R^dF:Rd→Rd be continuous and globally integrable, with induced flow Φ\PhiΦ. Fix T>0T>0T>0 and constants L≥0L\ge0L≥0, KKK. There is a constant C=C(T)C=C(T)C=C(T), depending only on TTT, FFF, LLL and KKK, with the following property. Whenever

  1. the step sizes satisfy γn≥0\gamma_n\ge0γn​≥0, ∑nγn=∞\sum_n\gamma_n=\infty∑n​γn​=∞, γn→0\gamma_n\to0γn​→0,
  2. xn+1−xn=γn+1(F(xn)+Un+1)x_{n+1}-x_n=\gamma_{n+1}(F(x_n)+U_{n+1})xn+1​−xn​=γn+1​(F(xn​)+Un+1​) for all nnn,
  3. assumption A1 holds, and
  4. (A2′) for some r>0r>0r>0, FFF is LLL-Lipschitz and bounded by KKK on the rrr-neighbourhood of {xn:n≥0}\{x_n:n\ge0\}{xn​:n≥0},

then for all t≥0t\ge0t≥0 large enough

sup⁡0≤h≤T∥X(t+h)−Φh(X(t))∥≤C[Δ(t−1,T+1)+sup⁡t≤s≤t+Tγˉ(s)].\sup_{0\le h\le T}\big\|X(t+h)-\Phi_h(X(t))\big\|\le C\Big[\Delta(t-1,T+1)+\sup_{t\le s\le t+T}\bar\gamma(s)\Big].0≤h≤Tsup​​X(t+h)−Φh​(X(t))​≤C[Δ(t−1,T+1)+t≤s≤t+Tsup​γˉ​(s)].

The estimate quantifies the asymptotic pseudotrajectory property: the shadowing error over a window of length TTT is controlled by the noise modulus Δ\DeltaΔ and the step sizes, which is what the rate results built on Proposition 4.1 use.

Formalization Note The constant is chosen after TTT, LLL, KKK (and the fixed FFF) and before the sequences, the radius rrr and the time; "for ttt large enough" is an existential threshold t0≥1t_0\ge1t0​≥1 that may depend on the run. A1 is kept as a hypothesis: it is a standing assumption of Proposition 4.1, and the "Furthermore" clause adds A2′ to it. The neighbourhood in A2′ is read as a uniform rrr-neighbourhood (see the goal theorem).

Preamble
import Mathlib
import Definitions.Def_StochApproxDyn_Interpolation_VectorFieldFlow
import Definitions.Def_StochApproxDyn_Interpolation_Scheme

open scoped NNReal Topology
open Filter
Formal statement
namespace StochApproxDyn.Interpolation

/-- Benaïm 1999, Proposition 4.1, estimate (11), p. 13. Let `F` be continuous and globally
integrable with flow `Φ`. For every `T > 0` and all constants `L ≥ 0`, `K` there is `C` (depending
only on `F`, `T`, `L`, `K`) such that: whenever the standing assumptions on `γ`, the recursion (7)
and A1 hold, and `F` is `L`-Lipschitz and bounded by `K` on the `r`-neighbourhood of
`{x_n : n ≥ 0}` for some `r > 0` (A2′), then for all `t` large enough (with `t ≥ 1`)
`sup_{0≤h≤T} ‖X(t + h) − Φ_h(X(t))‖ ≤ C [Δ(t − 1, T + 1) + sup_{t≤s≤t+T} γ̄(s)]`. -/
theorem shadowing_estimate {d : ℕ}
    (F : EuclideanSpace ℝ (Fin d) → EuclideanSpace ℝ (Fin d)) (hF : Continuous F)
    (hint : IsGloballyIntegrable F)
    (Φ : ℝ → EuclideanSpace ℝ (Fin d) → EuclideanSpace ℝ (Fin d)) (hΦ : IsFlowOf F Φ) :
    ∀ T : ℝ, 0 < T → ∀ L : ℝ≥0, ∀ K : ℝ, ∃ C : ℝ,
      ∀ (γ : ℕ → ℝ) (x U : ℕ → EuclideanSpace ℝ (Fin d)) (r : ℝ),
        IsStepSizeSeq γ → SatisfiesScheme F γ x U → SatisfiesA1 γ U → 0 < r →
        LipschitzOnWith L F (Metric.thickening r (Set.range x)) →
        (∀ y ∈ Metric.thickening r (Set.range x), ‖F y‖ ≤ K) →
        ∃ t₀ : ℝ, 1 ≤ t₀ ∧ ∀ t : ℝ, t₀ ≤ t → ∀ h ∈ Set.Icc (0 : ℝ) T,
          ‖interpAffine γ x (t + h) - Φ h (interpAffine γ x t)‖ ≤
            C * (Delta γ U (t - 1) (T + 1) + sSup (stepInterp γ '' Set.Icc t (t + T))) := by sorry

end StochApproxDyn.Interpolation
Source
Benaïm, Dynamics of Stochastic Approximation Algorithms, Séminaire de Probabilités XXXIII, LNM 1709 (1999), pp. 12–13, Proposition 4.1, estimate (11) ("Furthermore, under assumption A2′ …")
Human review
  • Endorsed by Shuze Chen · Oct 4, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 4, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me