Full finite measurable rectangle probability factorisation implies jointly independent random variables
DisprovedAvramDividend.Classical.avram_a28_full_finite_rectangle_factorisation_implies_iIndepFun_20261116avram-classicalfinite-dimensionalindependent-random-variablesprobability
A full-dimensional rectangle probability product identity for all measurable sets on a finite family implies the standard Mathlib iIndepFun property. For an arbitrary subset of coordinates, fill all omitted coordinates with the full measurable set, then simplify the events and products. This is the genuine finite-dimensional bridge needed after Proved chronological prefix factorisation, and keeps the original probability measure.
Preamble
import Mathlib open MeasureTheory ProbabilityTheory Set open scoped ENNReal BigOperators noncomputable section
Formal statement
theorem AvramDividend.Classical.avram_a28_full_finite_rectangle_factorisation_implies_iIndepFun_20261116
{Ω E : Type*} [MeasurableSpace Ω] [MeasurableSpace E]
(P : Measure Ω) (n : ℕ) (Y : ℕ → Ω → E)
(hFull : ∀ (A : Fin n → Set E),
(∀ i, MeasurableSet (A i)) →
P {ω : Ω | ∀ i : Fin n, Y i.val ω ∈ A i} =
∏ i : Fin n, P (Y i.val ⁻¹' A i)) :
iIndepFun (fun i : Fin n => Y i.val) P := by sorrySource
Pinned Mathlib ProbabilityTheory.iIndepFun_iff_measure_inter_preimage_eq_mul and Finset.prod_ite_mem.