Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 2.4.1 — coupling implies convergence in variation

Proved
PalmQueueing.Loynes.coupling_convergence

by mikedeng1 · Oct 1, 2026 · Mathlib 0df444a (Lean v4.33.1)

couplingprobabilityqueueingstationary-process

Theorem 2.4.1. If {Xn}\{X_n\}{Xn​} couples with a θ\thetaθ-compatible sequence {Z∘θn}\{Z\circ\theta^n\}{Z∘θn}, then the sequence {Xn+k}n≥0\{X_{n+k}\}_{n \ge 0}{Xn+k​}n≥0​ converges in variation to {Z∘θn}\{Z\circ\theta^n\}{Z∘θn} as kkk tends to ∞\infty∞:

lim⁡k→∞∣P~X,k−P~Z,k∣=0,(2.4.1)\lim_{k\to\infty} \big| \widetilde{P}_{X,k} - \widetilde{P}_{Z,k} \big| = 0 , \tag{2.4.1}k→∞lim​​PX,k​−PZ,k​​=0,(2.4.1)

where P~X,k\widetilde{P}_{X,k}PX,k​ is the law of the whole shifted trajectory X~k=(Xk,Xk+1,… )\widetilde{X}_k = (X_k, X_{k+1}, \dots)Xk​=(Xk​,Xk+1​,…) on the infinite product space.

The convergence is in variation on the law of the entire trajectory, not of a single coordinate and not merely in distribution — which is what makes it the right notion of "reaches the stationary regime". The book derives it from the coupling inequality

∣P~X,k−P~Z,k∣≤P(N>k),(2.4.2)|\widetilde{P}_{X,k} - \widetilde{P}_{Z,k}| \le P(N > k) , \tag{2.4.2}∣PX,k​−PZ,k​∣≤P(N>k),(2.4.2)

obtained by splitting each event on {X~k=Z~k}\{\widetilde{X}_k = \widetilde{Z}_k\}{Xk​=Zk​}; finiteness of the coupling time NNN then gives (2.4.1), with no further hypothesis. In particular, if {Zn}\{Z_n\}{Zn​} is θ\thetaθ-stationary, ∣P~X,k−P~Z∣→0|\widetilde{P}_{X,k} - \widetilde{P}_Z| \to 0∣PX,k​−PZ​∣→0 (2.4.3).

The statement quantifies the measurable set inside the index, which is exactly the supremum defining the distance in variation being small. EEE is any measurable space: p.99 takes (E,E)(E, \mathcal{E})(E,E) Polish and introduces the metric d∞d_\inftyd∞​ on E∞E^\inftyE∞ only to identify that space's Borel field with the product σ\sigmaσ-field, which is used directly here.

Formalization Note. The coupling time NNN is an N\mathbb{N}N-valued random variable (measurable), and Xn=ZnX_n = Z_nXn​=Zn​ for all n≥Nn \ge Nn≥N holds PPP-almost surely.

Preamble
import Mathlib
import Definitions.Def_PalmQueueing_Loynes_Coupling
Formal statement
/-!
# Theorem 2.4.1: coupling implies convergence in variation (§2.4.1, p.99)
-/

namespace PalmQueueing.Loynes

open MeasureTheory Filter Topology

variable {Ω : Type*} [MeasurableSpace Ω]

/-- **Theorem 2.4.1** (§2.4.1, p.99). If `{X_n}` couples with a `θ`-compatible sequence
`{Z ∘ θⁿ}`, then the sequence `{X_{n+k}}_{n ≥ 0}` converges in variation to `{Z ∘ θⁿ}` as `k`
tends to `∞`:

`(2.4.1)  lim_{k→∞} | P̃_{X,k} − P̃_{Z,k} | = 0`,

where `P̃_{X,k}` is the law on `(E^∞, E^∞)` of the whole shifted trajectory
`X̃_k = (X_k, X_{k+1}, …)`.

The convergence is **in variation on the law of the entire trajectory**, not of one coordinate and
not in distribution: `|P̃_{X,k} − P̃_{Z,k}| = sup_{C} |P̃_{X,k}(C) − P̃_{Z,k}(C)|` over all
measurable `C ⊆ E^∞`. That is why it is stated with the set quantified inside the index `K`, which
is exactly the supremum being small. The book derives it from the **coupling inequality**
`(2.4.2) |P̃_{X,k} − P̃_{Z,k}| ≤ P(N > k)`, which is finite-`N` plus nothing else.

`E` is any measurable space here. p.99 takes `(E, ℰ)` Polish with its Borel field and introduces
the metric `d_∞` on `E^∞` only to identify that space's Borel field with the product σ-field; the
statement uses the product σ-field directly, so the metric is not needed. -/
theorem coupling_convergence {E : Type*} [MeasurableSpace E]
    (P : Measure Ω) [IsProbabilityMeasure P]
    (θ : Ω → Ω) (X Z : ℕ → Ω → E)
    (hXmeas : ∀ n, Measurable (X n)) (hZmeas : ∀ n, Measurable (Z n))
    (hZcomp : IsShiftCompatible θ Z)
    (hcouple : Couple P X Z) :
    ∀ ε : ℝ, 0 < ε → ∃ K : ℕ, ∀ k : ℕ, K ≤ k →
      ∀ C : Set (ℕ → E), MeasurableSet C →
        |(P {ω | (fun n : ℕ => X (n + k) ω) ∈ C}).toReal
          - (P {ω | (fun n : ℕ => Z (n + k) ω) ∈ C}).toReal| ≤ ε := by sorry

end PalmQueueing.Loynes
Source
Baccelli & Bremaud, Elements of Queueing Theory: Palm Martingale Calculus and Stochastic Recurrences, 2nd ed., Springer 2003, p. 99, Theorem 2.4.1
Human review
  • Endorsed by Shuze Chen · Oct 5, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 5, 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