Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 8.1.4 — Joint convexity of matrix relative entropy

Proved
TroppMatrixConcentration.ch8_entropy_joint_convex

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

convexitymatrix-analysismatrix-concentration

For every positive matrix dimension, the set of pairs (A,H)(A,H)(A,H) of positive-definite complex matrices is convex, and the function (A,H)↦D(A;H)(A,H)\mapsto D(A;H)(A,H)↦D(A;H) is convex on this set. Thus for any two such pairs and any 0≤t≤10\le t\le10≤t≤1,

D(tA1+(1−t)A2; tH1+(1−t)H2)≤tD(A1;H1)+(1−t)D(A2;H2).D(tA_1+(1-t)A_2;\,tH_1+(1-t)H_2)\le tD(A_1;H_1)+(1-t)D(A_2;H_2).D(tA1​+(1−t)A2​;tH1​+(1−t)H2​)≤tD(A1​;H1​)+(1−t)D(A2​;H2​).

The domain includes arbitrary noncommuting, unnormalized positive-definite matrices. This is the joint convexity statement supporting Lieb’s theorem.

Preamble
import Definitions.Def_TroppMatrixConcentration_ch8_entropy

open scoped Matrix.Norms.L2Operator ComplexOrder

Formal statement
namespace TroppMatrixConcentration

theorem ch8_entropy_joint_convex {d : ℕ} [NeZero d] :
    ConvexOn ℝ {P : Matrix (Fin d) (Fin d) ℂ × Matrix (Fin d) (Fin d) ℂ |
      P.1.PosDef ∧ P.2.PosDef} (fun P => ch8_relativeEntropy P.1 P.2) := 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; Theorem 8.1.4 and Section 8.8, printed pp. 120, 137–138.
Read-back

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

For every natural number d≠0d\ne 0d=0, let SSS be the set of ordered pairs (A,H)(A,H)(A,H) of complex d×dd\times dd×d positive-definite matrices (each matrix is Hermitian and its quadratic form is strictly positive on every nonzero vector). Define L(X)L(X)L(X) to be the continuous functional calculus of the real logarithm applied to XXX; for a positive-definite matrix, this applies the real logarithm to each positive eigenvalue. Define the real-valued function D(A,H)=Re⁡tr⁡(A(L(A)−L(H))−(A−H))D(A,H)=\operatorname{Re}\operatorname{tr}\bigl(A(L(A)-L(H))-(A-H)\bigr)D(A,H)=Retr(A(L(A)−L(H))−(A−H)). Then SSS is convex over R\mathbb RR, and for every (A1,H1),(A2,H2)∈S(A_1,H_1),(A_2,H_2)\in S(A1​,H1​),(A2​,H2​)∈S and every pair of real numbers a,b≥0a,b\ge 0a,b≥0 satisfying a+b=1a+b=1a+b=1, one has D(aA1+bA2,aH1+bH2)≤aD(A1,H1)+bD(A2,H2)D(aA_1+bA_2,aH_1+bH_2)\le aD(A_1,H_1)+bD(A_2,H_2)D(aA1​+bA2​,aH1​+bH2​)≤aD(A1​,H1​)+bD(A2​,H2​). Convexity of SSS asserts that both matrices in this combined pair are again positive definite. The weights may equal zero, including the two endpoint choices; dimension zero is excluded. No normalization of the traces or commutativity of the matrices is assumed.

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