Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition 3.2.1 — Tail bounds for eigenvalues

Proved
TroppMatrixConcentration.ch3_laplace_tails

by tc · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

matrix-concentrationrandom-matrices

For a measurable, almost-surely Hermitian random matrix YYY on a probability space, whenever eθYe^{\theta Y}eθY is Bochner integrable,

P{λmax⁡(Y)≥t}≤e−θtEtr⁡eθY(θ>0),\mathbb P\{\lambda_{\max}(Y)\ge t\}\le e^{-\theta t}\mathbb E\operatorname{tr}e^{\theta Y}\quad(\theta>0),P{λmax​(Y)≥t}≤e−θtEtreθY(θ>0), P{λmin⁡(Y)≤t}≤e−θtEtr⁡eθY(θ<0).\mathbb P\{\lambda_{\min}(Y)\le t\}\le e^{-\theta t}\mathbb E\operatorname{tr}e^{\theta Y}\quad(\theta<0).P{λmin​(Y)≤t}≤e−θtEtreθY(θ<0).

Both bounds hold for every real threshold ttt. These pointwise parameter bounds provide the source's infimum bounds over parameters with finite matrix mgf.

Preamble
import Definitions.Def_TroppMatrixConcentration_probability

open MeasureTheory ProbabilityTheory
open scoped Matrix.Norms.L2Operator

Formal statement
namespace TroppMatrixConcentration

theorem ch3_laplace_tails {Ω : Type*} [MeasurableSpace Ω]
    (μ : Measure Ω) [IsProbabilityMeasure μ] {d : ℕ} [NeZero d]
    (Y : Ω → Matrix (Fin d) (Fin d) ℂ) (θ : ℝ)
    (hMeas : Measurable Y) (hHerm : ∀ᵐ ω ∂μ, (Y ω).IsHermitian)
    (hExp : Integrable (fun ω => matrixExp (θ • Y ω)) μ) :
    (0 < θ → ∀ t : ℝ,
      (μ {ω | t ≤ lambdaMax (Y ω)}).toReal ≤
        Real.exp (-θ * t) * (∫ ω, traceExp (θ • Y ω) ∂μ)) ∧
    (θ < 0 → ∀ t : ℝ,
      (μ {ω | lambdaMin (Y ω) ≤ t}).toReal ≤
        Real.exp (-θ * t) * (∫ ω, traceExp (θ • Y ω) ∂μ)) := by sorry

end TroppMatrixConcentration
Source
Joel A. Tropp, An Introduction to Matrix Concentration Inequalities, arXiv:1501.01571v1 (7 January 2015); https://arxiv.org/abs/1501.01571v1; Proposition 3.2.1, printed pp. 32–33.
Read-back

What the Lean code literally says, in plain math · gpt-6

For every measurable space Ω\OmegaΩ, every probability measure μ\muμ on it, every nonzero natural number ddd, every measurable map Y:Ω→Cd×dY:\Omega\to\mathbb C^{d\times d}Y:Ω→Cd×d (with the Borel measurable structure on matrices), and every real number θ\thetaθ, suppose that Y(ω)Y(\omega)Y(ω) equals its conjugate transpose for μ\muμ-almost every ω\omegaω and that the matrix-valued function ω↦exp⁡(θY(ω))\omega\mapsto\exp(\theta Y(\omega))ω↦exp(θY(ω)) is Bochner integrable with respect to μ\muμ. Here exp⁡(A)\exp(A)exp(A) is the normed-algebra matrix exponential, σR(A)={r∈R:rI−A is not invertible}\sigma_{\mathbb R}(A)=\{r\in\mathbb R:rI-A\text{ is not invertible}\}σR​(A)={r∈R:rI−A is not invertible} is the spectrum over the real scalars, λmax⁡(A)=sup⁡σR(A)\lambda_{\max}(A)=\sup\sigma_{\mathbb R}(A)λmax​(A)=supσR​(A) and λmin⁡(A)=inf⁡σR(A)\lambda_{\min}(A)=\inf\sigma_{\mathbb R}(A)λmin​(A)=infσR​(A) are the real-valued supremum and infimum used in the definitions, and trExp⁡(A)=Re⁡tr⁡(exp⁡(A))\operatorname{trExp}(A)=\operatorname{Re}\operatorname{tr}(\exp(A))trExp(A)=Retr(exp(A)). Then both of the following implications hold:

θ>0⟹∀t∈R,μ{ω:t≤λmax⁡(Y(ω))}≤e−θt∫ΩRe⁡tr⁡(exp⁡(θY(ω))) dμ(ω),\theta>0\quad\Longrightarrow\quad \forall t\in\mathbb R,\qquad \mu\{\omega:t\le\lambda_{\max}(Y(\omega))\} \le e^{-\theta t}\int_\Omega\operatorname{Re}\operatorname{tr}\bigl(\exp(\theta Y(\omega))\bigr)\,d\mu(\omega),θ>0⟹∀t∈R,μ{ω:t≤λmax​(Y(ω))}≤e−θt∫Ω​Retr(exp(θY(ω)))dμ(ω), θ<0⟹∀t∈R,μ{ω:λmin⁡(Y(ω))≤t}≤e−θt∫ΩRe⁡tr⁡(exp⁡(θY(ω))) dμ(ω).\theta<0\quad\Longrightarrow\quad \forall t\in\mathbb R,\qquad \mu\{\omega:\lambda_{\min}(Y(\omega))\le t\} \le e^{-\theta t}\int_\Omega\operatorname{Re}\operatorname{tr}\bigl(\exp(\theta Y(\omega))\bigr)\,d\mu(\omega).θ<0⟹∀t∈R,μ{ω:λmin​(Y(ω))≤t}≤e−θt∫Ω​Retr(exp(θY(ω)))dμ(ω).

The measures in these inequalities are converted from extended nonnegative real values to real numbers; they are finite because μ\muμ is a probability measure. The threshold ttt may be any real number, including zero and negative values, and both tail events include equality. Dimension zero is excluded. Hermitian symmetry is required only almost everywhere; the functions in the events are defined on all matrices by the real-spectrum supremum and infimum above, including any exceptional non-Hermitian values, with the total real supremum and infimum conventions (in particular, value zero for an empty real spectrum). When θ=0\theta=0θ=0, neither antecedent holds, so the asserted conjunction imposes no tail inequality.

Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by tc · Oct 7, 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