A Poisson count with exact cell-conditional marked exponential moments has the compound-Poisson Laplace transform
ProvedAvramDividend.Classical.avram_a30_poisson_count_conditional_mark_Laplace_20261116conditional-expectationlaplace-transformlevy-processespoisson
An arbitrary integrable real-valued random variable f on its original source Ω has the compound-Poisson Laplace expectation if it admits a measurable natural-valued source count K, and its integrals on every count cell K=k are the Poisson(r) masses times m^k. This theorem uses a disjoint measurable count partition and the Proved Poisson probability-generating function, and does not assume or construct an independently randomised mark stream. The integrability of k↦m^k under Poisson(r) is explicit.
Preamble
import Mathlib import Theorems.Thm_AvramDividend_Classical_chat06_poisson_count_pgf_general open MeasureTheory ProbabilityTheory Set open scoped NNReal ENNReal noncomputable section
Formal statement
theorem AvramDividend.Classical.avram_a30_poisson_count_conditional_mark_Laplace_20261116
{Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
(K : Ω → ℕ) (hK : Measurable K)
(r : ℝ≥0) (m : ℝ)
(f : Ω → ℝ) (hf : Integrable f P)
(hm : Integrable (fun k : ℕ => m ^ k) (poissonMeasure r))
(hCell : ∀ k : ℕ,
(∫ ω in {ω : Ω | K ω = k}, f ω ∂P) =
(poissonMeasure r).real {k} * m ^ k) :
(∫ ω, f ω ∂P) = Real.exp ((r : ℝ) * (m - 1)) := by sorrySource
Pinned Mathlib MeasureTheory.integral_iUnion, MeasureTheory.integral_countable; Proved AvramDividend.Classical.chat06_poisson_count_pgf_general.