Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 3.1, pp. 40–41, closing displays — studentized θ̂ᵈ equals minus the studentized score plus o_P(1)

Proved
LagrangeRank.Debias.asymptotic_linearity

by mikedeng1 · Oct 22, 2026 · Mathlib 0df444a (Lean v4.33.1)

asymptotic-normalityconstrained-inferencedebiased-estimatorp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Work on one probability space (Ω,F,P)(\Omega,\mathcal F,\mathbb P)(Ω,F,P). For each sample size nnn let dnd_ndn​ be the dimension, Ln(ω,⋅)L_n(\omega,\cdot)Ln​(ω,⋅) the random loss, fnf_nfn​ the constraint, θn∗\theta^*_nθn∗​ the true parameter, θ^n\widehat\theta_nθn​ the initial estimator, Ω^n\widehat\Omega_nΩn​ the inverse-Hessian estimator and Ωn∗\Omega^*_nΩn∗​ its target, and let in≠jni_n\neq j_nin​=jn​ be coordinates. Assume Assumptions 3.1–3.4 with the standing assumptions (the structure Assumptions, with rates r1,…,r5r_1,\dots,r_5r1​,…,r5​, constants K,c1,c2,c3,L1,L2K,c_1,c_2,c_3,L_1,L_2K,c1​,c2​,c3​,L1​,L2​ and events EnE_nEn​), r1→0r_1\to0r1​→0, r4→0r_4\to0r4​→0, the scaling condition

n(r12r5+(r1+r4)(r12+r2)+r1r3)→0,\sqrt n\big(r_1^2r_5+(r_1+r_4)(r_1^2+r_2)+r_1r_3\big)\to0,n​(r12​r5​+(r1​+r4​)(r12​+r2​)+r1​r3​)→0,

and the variance lower bound of Assumption 3.5: for some C>0C>0C>0 and all nnn, vn=[Θ11∗Σ11∗Θ11∗⊤]jnjn≥Cv_n=[\Theta^*_{11}\Sigma^*_{11}\Theta^{*\top}_{11}]_{j_nj_n}\ge Cvn​=[Θ11∗​Σ11∗​Θ11∗⊤​]jn​jn​​≥C and wn=(ein−ejn)⊤Θ11∗Σ11∗Θ11∗⊤(ein−ejn)≥Cw_n=(e_{i_n}-e_{j_n})^\top\Theta^*_{11}\Sigma^*_{11}\Theta^{*\top}_{11}(e_{i_n}-e_{j_n})\ge Cwn​=(ein​​−ejn​​)⊤Θ11∗​Σ11∗​Θ11∗⊤​(ein​​−ejn​​)≥C. Then, in probability,

n θ^jd−θj∗vn+n [Θ11∗∇L(θ∗)]jvn→0,\sqrt n\,\frac{\widehat\theta^d_{j}-\theta^*_{j}}{\sqrt{v_n}}+\sqrt n\,\frac{[\Theta^*_{11}\nabla L(\theta^*)]_{j}}{\sqrt{v_n}}\to0,n​vn​​θjd​−θj∗​​+n​vn​​[Θ11∗​∇L(θ∗)]j​​→0, n (θ^id−θi∗)−(θ^jd−θj∗)wn+n [Θ11∗∇L(θ∗)]i−[Θ11∗∇L(θ∗)]jwn→0,\sqrt n\,\frac{(\widehat\theta^d_{i}-\theta^*_{i})-(\widehat\theta^d_{j}-\theta^*_{j})}{\sqrt{w_n}}+\sqrt n\,\frac{[\Theta^*_{11}\nabla L(\theta^*)]_{i}-[\Theta^*_{11}\nabla L(\theta^*)]_{j}}{\sqrt{w_n}}\to0,n​wn​​(θid​−θi∗​)−(θjd​−θj∗​)​+n​wn​​[Θ11∗​∇L(θ∗)]i​−[Θ11∗​∇L(θ∗)]j​​→0,

with i=ini=i_ni=in​, j=jnj=j_nj=jn​, θ^d\widehat\theta^dθd the debiased estimator (3.9) and Θ∗\Theta^*Θ∗ the block-inverse matrix of p. 16 at (Ωn∗,∇fn(θn∗))(\Omega^*_n,\nabla f_n(\theta^*_n))(Ωn∗​,∇fn​(θn∗​)).

This is the "=−n[Θ11∗∇L(θ∗)]j/⋯+oP(1)=-\sqrt n[\Theta^*_{11}\nabla L(\theta^*)]_j/\sqrt{\cdots}+o_{\mathbb P}(1)=−n​[Θ11∗​∇L(θ∗)]j​/⋯​+oP​(1)" step of the proof of Theorem 3.1; with Assumption 3.5 and Slutsky's lemma it gives the theorem.

Formalization Note. The page prints the normalization as [Θ11∗Σ11∗Θ11∗]jj[\Theta^*_{11}\Sigma^*_{11}\Theta^*_{11}]_{jj}[Θ11∗​Σ11∗​Θ11∗​]jj​, without the transpose that Theorem 3.1 and Assumption 3.5 carry; the theorem's normalization is used. oP(1)o_{\mathbb P}(1)oP​(1) is TendstoInMeasure to 000. The variance lower bound is the condition under which Assumption 3.5 is stated; it is needed to divide the remainder by vn\sqrt{v_n}vn​​.

Preamble
import Mathlib
import Definitions.Def_LagrangeRank_Debias_Setting
Formal statement
namespace LagrangeRank.Debias
open MeasureTheory ProbabilityTheory Matrix Filter Topology
theorem asymptotic_linearity {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
    [IsProbabilityMeasure P]
    (d : ℕ → ℕ) (L : (n : ℕ) → Ω → (Fin (d n) → ℝ) → ℝ) (f : (n : ℕ) → (Fin (d n) → ℝ) → ℝ)
    (thetaStar : (n : ℕ) → Fin (d n) → ℝ) (thetaHat : (n : ℕ) → Ω → Fin (d n) → ℝ)
    (OmegaHat : (n : ℕ) → Ω → Matrix (Fin (d n)) (Fin (d n)) ℝ)
    (OmegaStar : (n : ℕ) → Matrix (Fin (d n)) (Fin (d n)) ℝ)
    (r1 r2 r3 r4 r5 : ℕ → ℝ) (K c1 c2 c3 L1 L2 : ℝ) (E : ℕ → Set Ω)
    (i j : (n : ℕ) → Fin (d n)) (hij : ∀ n, i n ≠ j n)
    (hA : Assumptions P d L f thetaStar thetaHat OmegaHat OmegaStar
      r1 r2 r3 r4 r5 K c1 c2 c3 L1 L2 E)
    (hr1 : Tendsto r1 atTop (𝓝 0)) (hr4 : Tendsto r4 atTop (𝓝 0))
    (hscale : Tendsto (fun n : ℕ => Real.sqrt (n : ℝ) *
      (r1 n ^ 2 * r5 n + (r1 n + r4 n) * (r1 n ^ 2 + r2 n) + r1 n * r3 n)) atTop (𝓝 0))
    (hvar : ∃ C : ℝ, 0 < C ∧ ∀ n,
      C ≤ popVar P (L n) (f n) (OmegaStar n) (thetaStar n) (j n) ∧
      C ≤ popVarPair P (L n) (f n) (OmegaStar n) (thetaStar n) (i n) (j n)) :
    TendstoInMeasure P (fun (n : ℕ) (ω : Ω) =>
        Real.sqrt (n : ℝ) *
          (estThetaD (L n ω) (f n) (OmegaHat n ω) (thetaHat n ω) (j n) - thetaStar n (j n))
          / Real.sqrt (popVar P (L n) (f n) (OmegaStar n) (thetaStar n) (j n))
        + Real.sqrt (n : ℝ) *
          (thetaStar11 (f n) (OmegaStar n) (thetaStar n) *ᵥ grad (L n ω) (thetaStar n)) (j n)
          / Real.sqrt (popVar P (L n) (f n) (OmegaStar n) (thetaStar n) (j n))) atTop 0 ∧
    TendstoInMeasure P (fun (n : ℕ) (ω : Ω) =>
        Real.sqrt (n : ℝ) *
          ((estThetaD (L n ω) (f n) (OmegaHat n ω) (thetaHat n ω) (i n) - thetaStar n (i n))
            - (estThetaD (L n ω) (f n) (OmegaHat n ω) (thetaHat n ω) (j n) - thetaStar n (j n)))
          / Real.sqrt (popVarPair P (L n) (f n) (OmegaStar n) (thetaStar n) (i n) (j n))
        + Real.sqrt (n : ℝ) *
          ((thetaStar11 (f n) (OmegaStar n) (thetaStar n) *ᵥ grad (L n ω) (thetaStar n)) (i n)
            - (thetaStar11 (f n) (OmegaStar n) (thetaStar n) *ᵥ grad (L n ω) (thetaStar n)) (j n))
          / Real.sqrt (popVarPair P (L n) (f n) (OmegaStar n) (thetaStar n) (i n) (j n)))
      atTop 0 := by sorry
end LagrangeRank.Debias
Source
Liu, Fang, Lu, Lagrangian Inference for Ranking Problems, arXiv:2110.00151v1, pp. 40–41, proof of Theorem 3.1 (Appendix A.1), closing displays
Human review
  • Endorsed by Shuze Chen · Oct 23, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 23, 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