Documentation

Exchangeability.Probability.CondExpBasic

Basic Helper Lemmas for Conditional Expectation #

This file provides basic helper lemmas for working with conditional expectations, σ-finiteness, and indicator functions.

These are foundational utilities extracted from the main CondExp.lean file to improve compilation speed.

Main components #

σ-Finiteness #

Indicators #

Helper lemmas for σ-finiteness and indicators #

Note: Some lemmas in this section explicitly include {m m₀ : MeasurableSpace Ω} as parameters to work with multiple measurable space structures (e.g., for trimmed measures). This makes the section variable [MeasurableSpace Ω] unused for those lemmas, requiring set_option linter.unusedSectionVars false.

If μ is finite, then any trim of μ is σ-finite.

theorem Exchangeability.Probability.indicator_iUnion_tsum_of_pairwise_disjoint {Ω : Type u_1} [MeasurableSpace Ω] (f : Set Ω) (hdisj : Pairwise (Function.onFun Disjoint f)) :
(fun (ω : Ω) => (⋃ (i : ), f i).indicator (fun (x : Ω) => 1) ω) = fun (ω : Ω) => ∑' (i : ), (f i).indicator (fun (x : Ω) => 1) ω

For pairwise disjoint sets, the indicator of the union equals the pointwise tsum of indicators (for ℝ-valued constants).

theorem Exchangeability.Probability.indicator_tsum_le_one_of_pairwise_disjoint {Ω : Type u_1} [MeasurableSpace Ω] (f : Set Ω) (hdisj : Pairwise (Function.onFun Disjoint f)) (x : Ω) :
∑' (i : ), (f i).indicator (fun (x : Ω) => 1) x 1

For pairwise disjoint sets, the tsum of indicators is bounded by 1 at each point. This follows from the fact that at most one indicator is 1 at any point.

theorem Exchangeability.Probability.measure_tsum_eq_measure_iUnion {α : Type u_3} [MeasurableSpace α] (μ : MeasureTheory.Measure α) (f : Set α) (hf_meas : ∀ (i : ), MeasurableSet (f i)) (hdisj : Pairwise (Function.onFun Disjoint f)) :
∑' (i : ), μ (f i) = μ (⋃ (i : ), f i)

For pairwise disjoint measurable sets, the tsum of measures equals the measure of the union.

For pairwise disjoint measurable sets under a probability measure, the tsum of measures is at most 1.