Kac's theorem: independence via characteristic functions
ProvedFamousTheorems.kac_independence_char_fun_6bKac's theorem: independence via characteristic functions. Let be random variables on a probability space, where takes values in a separable real Hilbert space . Then are independent if and only if, for all ,
where is the characteristic function.
This is a convenient test for independence: the joint characteristic function factors exactly when the variables are independent. It is used to show that uncorrelated jointly Gaussian variables are independent and in Fourier-analytic proofs of limit theorems.
Formalization note. Mathlib's ProbabilityTheory.iIndepFun_iff_charFun_pi. The random vector is viewed in WithLp 2 (∀ i, E i), the product with the inner product . charFun μ t is , and P.map Y is the law of . Each is assumed almost everywhere measurable.
import Mathlib
namespace FamousTheorems
open MeasureTheory
theorem kac_independence_char_fun_6b {Ω ι : Type*} {mΩ : MeasurableSpace Ω} {P : Measure Ω} [Fintype ι] [IsProbabilityMeasure P]
{E : ι → Type*} {mE : ∀ i, MeasurableSpace (E i)} [∀ i, NormedAddCommGroup (E i)]
[∀ i, InnerProductSpace ℝ (E i)] [∀ i, CompleteSpace (E i)] [∀ i, BorelSpace (E i)]
[∀ i, SecondCountableTopology (E i)] {X : ∀ i, Ω → E i} (hX : ∀ i, AEMeasurable (X i) P) :
ProbabilityTheory.iIndepFun X P ↔
∀ t : WithLp 2 (∀ i, E i),
charFun (P.map fun ω => WithLp.toLp 2 fun i => X i ω) t = ∏ i, charFun (P.map (X i)) (t.ofLp i) := by sorry
end FamousTheorems