Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 37, p. 24 — for A with orthonormal columns, ‖A(Ỹ − Y*)‖_F ≤ 2√ε‖B − AY*‖_F

Proved
InputSparsity.Regress.lemma_37

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

least-squaresp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1randomized-linear-algebrasketchingsubspace-embedding

Let r≥1r\ge1r≥1 and ε>0\varepsilon>0ε>0. Let A∈Rn×kA\in\mathbb R^{n\times k}A∈Rn×k have orthonormal columns (A⊤A=IkA^\top A=I_kA⊤A=Ik​) with k≤rk\le rk≤r, let B∈Rn×d′B\in\mathbb R^{n\times d'}B∈Rn×d′, and let S∈Rt×nS\in\mathbb R^{t\times n}S∈Rt×n. Suppose that

  1. SSS is a subspace embedding for AAA with parameter η∈[0,1/2]\eta\in[0,1/2]η∈[0,1/2]: ∣∥SAx∥22−∥Ax∥22∣≤η∥Ax∥22\bigl|\|SAx\|_2^2-\|Ax\|_2^2\bigr|\le\eta\|Ax\|_2^2​∥SAx∥22​−∥Ax∥22​​≤η∥Ax∥22​ for all xxx;
  2. Y∗Y^*Y∗ minimizes ∥AY−B∥F2\|AY-B\|_F^2∥AY−B∥F2​;
  3. SSS satisfies the approximate matrix multiplication event of Lemma 32 for the pair (A, B−AY∗)(A,\,B-AY^*)(A,B−AY∗) with error parameter ε/r\sqrt{\varepsilon/r}ε/r​: ∥A⊤S⊤S(B−AY∗)−A⊤(B−AY∗)∥F2≤εr∥A∥F2∥B−AY∗∥F2\|A^\top S^\top S(B-AY^*)-A^\top(B-AY^*)\|_F^2\le\frac{\varepsilon}{r}\|A\|_F^2\|B-AY^*\|_F^2∥A⊤S⊤S(B−AY∗)−A⊤(B−AY∗)∥F2​≤rε​∥A∥F2​∥B−AY∗∥F2​;
  4. Y~\tilde YY~ minimizes ∥S(AY−B)∥F2\|S(AY-B)\|_F^2∥S(AY−B)∥F2​.

Then

∥A(Y~−Y∗)∥F≤2ε ∥B−AY∗∥F.\|A(\tilde Y-Y^*)\|_F\le2\sqrt\varepsilon\,\|B-AY^*\|_F .∥A(Y~−Y∗)∥F​≤2ε​∥B−AY∗∥F​.

This is the core estimate of the sketch-and-solve analysis: the sketched solution is close to the true one in the geometry of AAA.

Formalization Note "For S,A,B,Y∗S,A,B,Y^*S,A,B,Y∗ and Y~\tilde YY~ as in Theorem 36" is spelled out. With orthonormal columns rank⁡A=k\operatorname{rank}A=krankA=k, so "rank at most rrr" is k≤rk\le rk≤r. The embedding parameter η\etaη plays the role of the paper's ϵ02\epsilon_0^2ϵ02​, and ϵ0≤1/2\epsilon_0\le1/\sqrt2ϵ0​≤1/2​ becomes η≤1/2\eta\le1/2η≤1/2: the proof (Appendix A, p. 37) bounds ∥A⊤S⊤SA−I∥2\|A^\top S^\top SA-I\|_2∥A⊤S⊤SA−I∥2​ by ϵ02\epsilon_0^2ϵ02​ and uses ϵ02≤1/2\epsilon_0^2\le1/2ϵ02​≤1/2; under the norm convention ∥SAx∥2=(1±ϵ0)∥Ax∥2\|SAx\|_2=(1\pm\epsilon_0)\|Ax\|_2∥SAx∥2​=(1±ϵ0​)∥Ax∥2​ of §1.2 with ϵ0=1/2\epsilon_0=1/\sqrt2ϵ0​=1/2​ the constant 222 would fail. The Lemma 32 event is the inequality it is used as in the proof (p. 37), with ≤\le≤ for the printed <<<.

Preamble
import Mathlib
import Definitions.Def_InputSparsity_Regress_Basic
Formal statement
namespace InputSparsity.Regress
open Matrix

theorem lemma_37 {n k d' t : ℕ} (r : ℕ) (A : Matrix (Fin n) (Fin k) ℝ)
    (B : Matrix (Fin n) (Fin d') ℝ) (S : Matrix (Fin t) (Fin n) ℝ) (ε η : ℝ)
    (Ystar Ytil : Matrix (Fin k) (Fin d') ℝ)
    (hr : 1 ≤ r) (hA : InputSparsity.Embed.HasOrthonormalCols A) (hk : k ≤ r)
    (hε : 0 < ε) (hη₀ : 0 ≤ η) (hη : η ≤ 1 / 2)
    (hemb : IsSubspaceEmbedding S A η)
    (hYstar : IsLSMinimizer A B Ystar)
    (hamm : AMMEvent S A (B - A * Ystar) (Real.sqrt (ε / r)))
    (hYtil : IsSketchedMinimizer S A B Ytil) :
    Real.sqrt (InputSparsity.Embed.frobSq (A * (Ytil - Ystar))) ≤
      2 * Real.sqrt ε * Real.sqrt (InputSparsity.Embed.frobSq (B - A * Ystar)) := by sorry

end InputSparsity.Regress
Source
Clarkson and Woodruff, Low Rank Approximation and Regression in Input Sparsity Time, arXiv:1207.6365v4, p. 24, Lemma 37 (proof in Appendix A, p. 37)
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