Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 4.2 — strong ergodic convergence to the minimum-norm minimizer when ∫ ε(t)/t dt = +∞

Proved
TikhonovHDD.Ergodic.theorem_4_2

by mikedeng1 · Oct 8, 2026 · Mathlib 0df444a (Lean v4.33.1)

convex-optimizationdynamical-systemsp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1tikhonov-regularization

Let H\mathcal HH be a real Hilbert space, t0>0t_0>0t0​>0, β≥0\beta\ge0β≥0, and let ggg and ϵ\epsilonϵ satisfy the General assumption: g:H→Rg:\mathcal H\to\mathbb Rg:H→R is convex, twice Fréchet differentiable, with gradient Lipschitz continuous on bounded sets and argmin⁡g≠∅\operatorname{argmin} g\ne\emptysetargming=∅; ϵ:[t0,+∞)→[0,+∞)\epsilon:[t_0,+\infty)\to[0,+\infty)ϵ:[t0​,+∞)→[0,+∞) is nonincreasing, of class C1C^1C1, with lim⁡t→+∞ϵ(t)=0\lim_{t\to+\infty}\epsilon(t)=0limt→+∞​ϵ(t)=0. Let xxx be a global C2C^2C2-solution of

x¨(t)+αtx˙(t)+β∇2g(x(t))x˙(t)+∇g(x(t))+ϵ(t)x(t)=0,t≥t0,x(t0)=u0, x˙(t0)=v0.\ddot x(t)+\frac{\alpha}{t}\dot x(t)+\beta\nabla^2 g(x(t))\dot x(t)+\nabla g(x(t))+\epsilon(t)x(t)=0,\qquad t\ge t_0,\quad x(t_0)=u_0,\ \dot x(t_0)=v_0 .x¨(t)+tα​x˙(t)+β∇2g(x(t))x˙(t)+∇g(x(t))+ϵ(t)x(t)=0,t≥t0​,x(t0​)=u0​, x˙(t0​)=v0​.

Assume

∫t0+∞ϵ(t)t dt=+∞,\int_{t_0}^{+\infty}\frac{\epsilon(t)}{t}\,dt=+\infty,∫t0​+∞​tϵ(t)​dt=+∞,

and let x∗=argmin⁡{∥x∥:x∈argmin⁡g}x^*=\operatorname{argmin}\{\|x\|:x\in\operatorname{argmin} g\}x∗=argmin{∥x∥:x∈argming} be the element of minimum norm of the nonempty closed convex set argmin⁡g\operatorname{argmin} gargming. If α>0\alpha>0α>0, then

lim⁡t→+∞1∫t0tϵ(s)s ds∫t0tϵ(s)s∥x(s)−x∗∥2 ds=0andlim inf⁡t→+∞∥x(t)−x∗∥=0.\lim_{t\to+\infty}\frac{1}{\int_{t_0}^{t}\frac{\epsilon(s)}{s}\,ds}\int_{t_0}^{t}\frac{\epsilon(s)}{s}\|x(s)-x^*\|^2\,ds=0 \qquad\text{and}\qquad \liminf_{t\to+\infty}\|x(t)-x^*\|=0 .t→+∞lim​∫t0​t​sϵ(s)​ds1​∫t0​t​sϵ(s)​∥x(s)−x∗∥2ds=0andt→+∞liminf​∥x(t)−x∗∥=0.

When the Tikhonov parameter decays slowly enough that ∫ϵ(t)/t dt\int\epsilon(t)/t\,dt∫ϵ(t)/tdt diverges, the trajectory converges strongly in this weighted ergodic sense to the minimum-norm minimizer, extending the corresponding result for the system without Hessian damping.

Formalization Note The statement is made for every global C2C^2C2-solution; existence and uniqueness are Theorem 2.1 of the paper. The hypothesis α>0\alpha>0α>0 is the theorem's own and replaces the standing α≥3\alpha\ge3α≥3 of (5). Divergence of the integral is ∫t0Tϵ(t)t dt→+∞\int_{t_0}^{T}\frac{\epsilon(t)}{t}\,dt\to+\infty∫t0​T​tϵ(t)​dt→+∞ as T→+∞T\to+\inftyT→+∞. The ergodic average is written with the reciprocal of ∫t0tϵ(s)s ds\int_{t_0}^t\frac{\epsilon(s)}{s}\,ds∫t0​t​sϵ(s)​ds, which vanishes for t≥t0t\ge t_0t≥t0​ only at t=t0t=t_0t=t0​ (the divergence hypothesis forces ϵ>0\epsilon>0ϵ>0 on [t0,+∞)[t_0,+\infty)[t0​,+∞)); the value there does not affect the limit. "lim inf⁡∥x(t)−x∗∥=0\liminf\|x(t)-x^*\|=0liminf∥x(t)−x∗∥=0" is stated as: for every δ>0\delta>0δ>0 there are arbitrarily large ttt with ∥x(t)−x∗∥<δ\|x(t)-x^*\|<\delta∥x(t)−x∗∥<δ.

Preamble
import Mathlib
import Definitions.Def_TikhonovHDD_Ergodic_TikhonovCurve
import Definitions.Def_TikhonovHDD_Ergodic_Setting

open Set Filter Topology MeasureTheory
Formal statement
namespace TikhonovHDD.Ergodic

/-- Theorem 4.2 (p. 18): if `∫_{t₀}^{+∞} ε(t)/t dt = +∞` and `α > 0`, every global
`C²`-solution of (5) converges strongly in the ergodic sense to the minimum-norm element `x⋆`
of `argmin g`:
`(∫_{t₀}^{t} ε(s)/s ds)⁻¹ ∫_{t₀}^{t} ε(s)/s ‖x(s) − x⋆‖² ds → 0`, and
`lim inf_{t → +∞} ‖x(t) − x⋆‖ = 0` (stated as: for every `δ > 0`, `‖x(t) − x⋆‖ < δ` for
arbitrarily large `t`). -/
theorem theorem_4_2 {H : Type*} [NormedAddCommGroup H] [InnerProductSpace ℝ H]
    [CompleteSpace H]
    (g : H → ℝ) (α β t₀ : ℝ) (ε ε' : ℝ → ℝ) (u₀ v₀ : H) (x xd xdd : ℝ → H) (xstar : H)
    (ht₀ : 0 < t₀) (hα : 0 < α) (hβ : 0 ≤ β)
    (hg : GeneralAssumptionG g) (hε : GeneralAssumptionEps t₀ ε ε')
    (hx : IsSolution g α β ε t₀ u₀ v₀ x xd xdd)
    (hdiv : Tendsto (fun T => ∫ t in t₀..T, ε t / t) atTop atTop)
    (hxstar : TikhonovHDD.Strong.IsMinNormMinimizer g xstar) :
    Tendsto (fun t => (∫ s in t₀..t, ε s / s)⁻¹ * ∫ s in t₀..t, ε s / s * ‖x s - xstar‖ ^ 2)
        atTop (𝓝 0) ∧
      ∀ δ > 0, ∃ᶠ t in atTop, ‖x t - xstar‖ < δ := by sorry

end TikhonovHDD.Ergodic
Source
Boţ, Csetnek, László, Tikhonov regularization of a second order dynamical system with Hessian driven damping, arXiv:1911.12845v2, p. 18, Theorem 4.2
Human review
  • Endorsed by Shuze Chen · Oct 8, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 8, 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