Lemma 1 — covariance inequality for bounded variables under strong mixing (Doukhan)
ProvedTPG.lemma_1_covariance_inequalityLet be a probability space and let be sub--algebras. Their strong mixing coefficient is
Let be a real random variable measurable with respect to and one measurable with respect to , both bounded, with .
Lemma.
This inequality converts a bound on mixing coefficients into a bound on covariances of bounded functionals of separated parts of a process; the source uses it in the consistency proof of the variance estimator.
Formalization Note The formal statement takes real numbers , , with and almost surely and for all , , and concludes . Since such are nonnegative and the bound is monotone in each, this is equivalent to the displayed inequality with the essential suprema and the supremum.
import Mathlib open MeasureTheory ProbabilityTheory open scoped BigOperators ENNReal
namespace TPG
theorem lemma_1_covariance_inequality {Ω : Type} [mΩ : MeasurableSpace Ω]
(μ : Measure Ω) [IsProbabilityMeasure μ]
(mU mV : MeasurableSpace Ω) (hU : mU ≤ mΩ) (hV : mV ≤ mΩ)
(X Y : Ω → ℝ) (hX : Measurable[mU] X) (hY : Measurable[mV] Y)
(a b α : ℝ) (hXa : ∀ᵐ ω ∂μ, |X ω| ≤ a) (hYb : ∀ᵐ ω ∂μ, |Y ω| ≤ b)
(hα : ∀ A B : Set Ω, MeasurableSet[mU] A → MeasurableSet[mV] B →
|μ.real (A ∩ B) - μ.real A * μ.real B| ≤ α) :
|cov[X, Y; μ]| ≤ 4 * α * a * b := by sorry
end TPGRead-back
What the Lean code literally says, in plain math · claude-opus-5-5
Objects quantified over. The statement is asserted for all of the following.
- An arbitrary set carrying a -algebra , and a measure on that is a probability measure, i.e. .
- Two further -algebras and on the same set , each contained in : and . Nothing else is assumed about how and relate to one another (they may coincide, and no independence is assumed).
- Two functions such that is measurable with respect to and is measurable with respect to . In both cases the target carries its Borel -algebra, and measurability is in the everywhere sense (not merely up to a null set).
- Three real numbers , , . No sign condition on , or is written as a hypothesis.
Hypotheses.
-
for -almost every .
-
for -almost every .
-
For every pair of subsets with and ,
Here denotes the measure of read as a real number (the value in converted to , with sent to ; since is a probability measure every value lies in and this convention is never triggered).
Conclusion.
where the covariance is defined by
All three integrals are Lebesgue (Bochner) integrals with respect to on , with the convention that the integral of a non-integrable function is . The inequality is non-strict, and the right-hand side is the product of the four real numbers , , , .
Edge cases and literal details.
- The number is any real number that bounds the quantity in hypothesis 3 for all pairs ; it is not required to be the supremum of those quantities.
- Hypothesis 3 ranges over all pairs with , , including , , , . Taking shows that hypothesis 3 can only hold when .
- Because is a probability measure (so is non-empty and a -almost-everywhere statement holds at some point), hypotheses 1 and 2 can only hold when and . These sign facts are consequences of the hypotheses, not separate assumptions.
- The bounds and are almost-everywhere bounds on and ; the values or are allowed (then or almost everywhere and the right-hand side is ), and is allowed (then the right-hand side is ).
- No integrability or moment hypothesis is stated separately; and are measurable and almost everywhere bounded on a probability space.
Confirmed by the mission captain (proposal self-audit).