Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 3.5.1 — Subadditivity of matrix cgfs

Proved
TroppMatrixConcentration.trace_cgf_subadditivity

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

matrix-concentrationrandom-matrices

Let X1,…,XNX_1,\ldots,X_NX1​,…,XN​ be independent measurable Hermitian random d×dd\times dd×d matrices, with d≥1d\ge1d≥1. For any real θ\thetaθ for which each matrix exponential eθXke^{\theta X_k}eθXk​ is integrable,

Etr⁡exp⁡(∑kθXk)≤tr⁡exp⁡(∑klog⁡EeθXk).\mathbb E\operatorname{tr}\exp\left(\sum_k\theta X_k\right)\le\operatorname{tr}\exp\left(\sum_k\log\mathbb E e^{\theta X_k}\right).Etrexp(k∑​θXk​)≤trexp(k∑​logEeθXk​).

The finite-mgf hypothesis makes explicit the source's standing regularity convention. The formula retains the matrix logarithms inside the sum.

Preamble
import Definitions.Def_TroppMatrixConcentration_probability
import Definitions.Def_TroppMatrixConcentration_dilation
import Mathlib.Analysis.Convex.Function

open MeasureTheory ProbabilityTheory
open scoped Matrix.Norms.L2Operator

Formal statement
namespace TroppMatrixConcentration

theorem trace_cgf_subadditivity {Ω : Type*} [MeasurableSpace Ω]
    (μ : Measure Ω) [IsProbabilityMeasure μ] {d N : ℕ} [NeZero d]
    (X : Fin N → Ω → Matrix (Fin d) (Fin d) ℂ) (θ : ℝ)
    (hMeas : ∀ k, Measurable (X k))
    (hHerm : ∀ k, ∀ᵐ ω ∂μ, (X k ω).IsHermitian)
    (hIndep : iIndepFun X μ)
    (hExp : ∀ k, Integrable (fun ω => matrixExp (θ • X k ω)) μ) :
    (∫ ω, traceExp (∑ k, θ • X k ω) ∂μ) ≤
      traceExp (cumulantSum μ X θ) := 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; Lemma 3.5.1, equation (3.5.1), printed pp. 35–36.
Read-back

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

For every measurable space (Ω,F)(\Omega,\mathcal F)(Ω,F) with a probability measure μ\muμ, every positive integer ddd, every nonnegative integer NNN, every real number θ\thetaθ, and every family of measurable maps Xk:Ω→Md(C)X_k:\Omega\to M_d(\mathbb C)Xk​:Ω→Md​(C) indexed by k∈{0,…,N−1}k\in\{0,\ldots,N-1\}k∈{0,…,N−1}, assume that each XkX_kXk​ is Hermitian μ\muμ-almost everywhere, that the family is jointly independent under μ\muμ, and that each map ω↦exp⁡(θXk(ω))\omega\mapsto \exp(\theta X_k(\omega))ω↦exp(θXk​(ω)) is Bochner integrable with respect to μ\muμ (equivalently here, its operator norm has finite integral). Define exp⁡(A)=∑n=0∞An/n!\exp(A)=\sum_{n=0}^{\infty}A^n/n!exp(A)=∑n=0∞​An/n!, T(A)=Re⁡tr⁡(exp⁡(A))T(A)=\operatorname{Re}\operatorname{tr}(\exp(A))T(A)=Retr(exp(A)), L(A)L(A)L(A) to be the real continuous functional calculus of the total real logarithm at AAA, and C=∑k=0N−1L(∫Ωexp⁡(θXk(ω)) dμ(ω))C=\sum_{k=0}^{N-1}L\bigl(\int_\Omega\exp(\theta X_k(\omega))\,d\mu(\omega)\bigr)C=∑k=0N−1​L(∫Ω​exp(θXk​(ω))dμ(ω)). The logarithm used in this calculus is ℓ(t)=log⁡∣t∣\ell(t)=\log|t|ℓ(t)=log∣t∣ for t≠0t\ne0t=0 and ℓ(0)=0\ell(0)=0ℓ(0)=0; for Hermitian AAA, L(A)L(A)L(A) applies ℓ\ellℓ to its real eigenvalues, and the total functional-calculus definition returns the zero matrix if AAA is not self-adjoint or if ℓ\ellℓ is not continuous on its real spectrum. Then ∫ΩT(∑k=0N−1θXk(ω)) dμ(ω)≤T(C)\int_\Omega T\bigl(\sum_{k=0}^{N-1}\theta X_k(\omega)\bigr)\,d\mu(\omega)\le T(C)∫Ω​T(∑k=0N−1​θXk​(ω))dμ(ω)≤T(C). All matrix integrals are Bochner integrals and the left-hand integral is a real Bochner integral, with the convention that such an integral is zero when its integrand is not integrable; integrability of the left-hand integrand is not a separate hypothesis. There is no sign restriction on θ\thetaθ, no assumption that the matrices commute, and no hypothesis that the XkX_kXk​ themselves are integrable. The dimension d=0d=0d=0 is excluded, whereas N=0N=0N=0 is included: the family hypotheses are then vacuous, both sums are zero, and the asserted inequality is d≤dd\le dd≤d; when θ=0\theta=0θ=0, it likewise reduces to d≤dd\le dd≤d.

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