Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 3.6, p. 722 — Ent_π(μH_t) ≤ Ent_π(μ)e^{−2αt}, and ≤ Ent_π(μ)e^{−4αt} when (K, π) is reversible

Proved
LogSobolevMC.Entropy.theorem_3_6

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

entropy-decaylog-sobolev-inequalitymarkov-chainsp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1relative-entropy

Let KKK be a Markov kernel on a finite set X\mathcal XX with invariant probability π\piπ, π(x)>0\pi(x)>0π(x)>0 for all xxx, and let α\alphaα be its log-Sobolev constant,

α=inf⁡{E(f,f)L(f): L(f)≠0},\alpha=\inf\Big\{\frac{\mathcal E(f,f)}{\mathcal L(f)}:\ \mathcal L(f)\neq0\Big\},α=inf{L(f)E(f,f)​: L(f)=0},

where E(f,f)=⟨(I−K)f,f⟩π\mathcal E(f,f)=\langle (I-K)f,f\rangle_\piE(f,f)=⟨(I−K)f,f⟩π​ and L(f)=∑x∣f(x)∣2log⁡(∣f(x)∣2/∥f∥22)π(x)\mathcal L(f)=\sum_x|f(x)|^2\log\big(|f(x)|^2/\|f\|_2^2\big)\pi(x)L(f)=∑x​∣f(x)∣2log(∣f(x)∣2/∥f∥22​)π(x). Let Ht=e−t(I−K)H_t=e^{-t(I-K)}Ht​=e−t(I−K) and, for a probability measure μ\muμ on X\mathcal XX, let μHt(y)=∑xHt(x,y)μ(x)\mu H_t(y)=\sum_x H_t(x,y)\mu(x)μHt​(y)=∑x​Ht​(x,y)μ(x). Write Entπ(ν)=∑xν(x)log⁡ν(x)π(x)\mathrm{Ent}_\pi(\nu)=\sum_x\nu(x)\log\frac{\nu(x)}{\pi(x)}Entπ​(ν)=∑x​ν(x)logπ(x)ν(x)​ for the relative entropy. Then for every probability measure μ\muμ:

Entπ(μHt)≤Entπ(μ) e−2αtfor all t>0;\mathrm{Ent}_\pi(\mu H_t)\le \mathrm{Ent}_\pi(\mu)\,e^{-2\alpha t}\qquad\text{for all }t>0;Entπ​(μHt​)≤Entπ​(μ)e−2αtfor all t>0;
  1. if (K,π)(K,\pi)(K,π) is reversible (π(x)K(x,y)=π(y)K(y,x)\pi(x)K(x,y)=\pi(y)K(y,x)π(x)K(x,y)=π(y)K(y,x) for all x,yx,yx,y), then
Entπ(μHt)≤Entπ(μ) e−4αtfor all t>0.\mathrm{Ent}_\pi(\mu H_t)\le \mathrm{Ent}_\pi(\mu)\,e^{-4\alpha t}\qquad\text{for all }t>0.Entπ​(μHt​)≤Entπ​(μ)e−4αtfor all t>0.

The log-Sobolev constant is thus an exponential rate of convergence to stationarity in relative entropy, for every initial distribution. Combined with Pinsker's inequality it gives total variation bounds such as (3.3).

Formalization Note The chain is not assumed irreducible: the statement holds for every stochastic KKK with a positive invariant probability (for a reducible chain α=0\alpha=0α=0 and the claim is that entropy does not increase). α\alphaα is a real infimum, which is 000 on a one-point space; no positivity of α\alphaα is assumed.

Preamble
import Mathlib
import Definitions.Def_mm_basic
import Definitions.Def_mm_continuous
import Definitions.Def_LogSobolevMC_Entropy_Setting
Formal statement
namespace LogSobolevMC.Entropy

theorem theorem_3_6 {V : Type*} [Fintype V] [DecidableEq V]
    (K : Matrix V V ℝ) (hK : MarkovMixing.IsStochastic K)
    (π : V → ℝ) (hπ : MarkovMixing.IsStationary K π) (hπpos : ∀ x, 0 < π x)
    (μ : V → ℝ) (hμ : MarkovMixing.IsDist μ) :
    (∀ t : ℝ, 0 < t →
        relEnt π (measHeat K t μ) ≤ relEnt π μ * Real.exp (-2 * LogSobolevMC.ChiSquare.logSobolev K π * t)) ∧
      (MarkovMixing.DetailedBalance K π → ∀ t : ℝ, 0 < t →
        relEnt π (measHeat K t μ) ≤
          relEnt π μ * Real.exp (-4 * LogSobolevMC.ChiSquare.logSobolev K π * t)) := by sorry

end LogSobolevMC.Entropy
Source
Diaconis and Saloff-Coste, Logarithmic Sobolev inequalities for finite Markov chains, Ann. Appl. Probab. 6 (1996), p. 722, Theorem 3.6
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 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