The law of total expectation (tower property)
ProvedFamousTheorems.law_of_total_expectationThe law of total expectation. Let be a measure on , a sub-σ-algebra with σ-finite on , and a function with values in a real Banach space. Then the conditional expectation of given has the same integral as :
This is the simplest case of the tower property of conditional expectation. It is the formal basis of conditioning arguments in probability, of the martingale property, and of the law of total variance.
Formalization note. Mathlib's MeasureTheory.integral_condExp. MeasureTheory.condExp m μ f (notation μ[f|m]) is the conditional expectation and μ.trim hm is the restriction of μ to m. No integrability hypothesis is needed: for non-integrable f, both the Bochner integral and Mathlib's conditional expectation are defined as .
Preamble
import Mathlib
Formal statement
namespace FamousTheorems
theorem law_of_total_expectation {α E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E] [CompleteSpace E] {m m₀ : MeasurableSpace α}
{μ : @MeasureTheory.Measure α m₀} (f : α → E) (hm : m ≤ m₀) [MeasureTheory.SigmaFinite (μ.trim hm)] :
∫ x, MeasureTheory.condExp m μ f x ∂μ = ∫ x, f x ∂μ := by sorry
end FamousTheoremsSource
Listed in Mathlib's curated theorem manifests (docs/1000.yaml); formalized in Mathlib as `MeasureTheory.integral_condExp`. Proof here reduces to that Mathlib result.