Lemma 2.1: with uniform marginals, weak convergence equals uniform convergence of joint cdfs
ProvedPermLimits.Cauchy.weak_conv_iff_uniform_cdfp2o-batch-p200ap2o-gran-per-chapterp2o-plan-paperp2o-v1permutation-limitsprobability
Let and be probability measures on , with random points , and joint distribution functions , . Assume for every . Then
For general measures on the square only the implication from right to left holds; the uniform marginals are what make the joint cdfs uniformly equicontinuous. The lemma translates weak convergence into a metric statement, which is how rectangular distance enters the theory.
Formalization Note Weak convergence is tested against all continuous functions on the compact square. Uniform convergence of to is TendstoUniformly on . No marginal assumption is made on , as in the source.
Preamble
import Mathlib import Definitions.Def_PermLimits_Shared_LimitMeasure import Definitions.Def_PermLimits_Shared_LimitConvergence open PermLimits.Shared
Formal statement
namespace PermLimits.Cauchy
open MeasureTheory Filter unitInterval
/-- **Lemma 2.1** (Hoppen et al., *Limits of permutation sequences*, arXiv:1103.5844v2, p. 7).
Let `(μ_n)` and `μ` be probability measures on `[0,1]²`, with joint distribution functions
`F_n` and `F`, and assume that for every `n` both marginals of `μ_n` are uniform on `[0, 1]`.
Then `μ_n ⇒ μ` (convergence in distribution `(X_n, Y_n) →ᵈ (X, Y)`) if and only if
`‖F_n − F‖_∞ = sup_{x,y ∈ [0,1]} |F_n(x, y) − F(x, y)| → 0`.
**Formalization Note.** Weak convergence is the paper's (9) with continuous test functions on the
compact square (`WeakConvMeasures`). `‖F_n − F‖_∞ → 0` is `TendstoUniformly` of the joint
distribution functions on `[0,1]²`. No marginal assumption is made on `μ`, as in the paper. -/
theorem weak_conv_iff_uniform_cdf (μs : ℕ → Measure (I × I)) (μ : Measure (I × I))
(hμs : ∀ n, IsProbabilityMeasure (μs n)) [IsProbabilityMeasure μ]
(hX : ∀ n, (μs n).map Prod.fst = volume) (hY : ∀ n, (μs n).map Prod.snd = volume) :
WeakConvMeasures μs μ ↔
TendstoUniformly (fun n (p : I × I) => jointCDF (μs n) p.1 p.2)
(fun p : I × I => jointCDF μ p.1 p.2) atTop := by sorry
end PermLimits.Cauchy
Source
Hoppen, Kohayakawa, Moreira, Ráth, Sampaio, Limits of permutation sequences, arXiv:1103.5844v2, https://arxiv.org/abs/1103.5844v2, p. 7, Lemma 2.1
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.