memLp compensated
ProvedMathFin.memLp_compensatedcontext-reviewlevystochcalcsource-import
The source theorem MathFin.memLp_compensated is extracted from MathFin.Foundations.PoissonRandomMeasure. The original Lean proposition is retained verbatim, and its proof is tracked separately for remote certification.
Preamble
import Definitions.Def_MathFin_Foundations_PoissonRandomMeasure_F000
/-
Copyright (c) 2026 Raphael Coelho. All rights reserved.
Released under Apache 2.0 license as described in the file LICENSE.
Authors: Raphael Coelho
-/
/-!
# Poisson random measure and the compensated increment
A *Poisson random measure* `N` on the time-mark space `[0, ∞) × E` with σ-finite
mark-intensity `ν` is the driving object of Lévy-driven stochastic calculus: `N(·, B)`
is `Poisson`-distributed with mean `ν̂(B)` (`ν̂ := Leb[0,∞) ⊗ ν`), and counts on disjoint
regions are independent. The *compensated* measure `Ñ(B) := N(B) − ν̂(B)` is the mean-zero,
`L²` object one integrates against.
This module provides the object as a hypothesis-bundling structure (its fields are the
Applebaum Def. 2.3.1 properties) plus the two Poisson moment facts the isometry needs —
`𝔼[N(B)] = ν̂(B)` and `𝔼[N(B)²] = ν̂(B)² + ν̂(B)` — hence `𝔼[Ñ(B)] = 0`, `𝔼[Ñ(B)²] = ν̂(B)`.
Mathlib carries the Poisson law (`poissonMeasure`, `integral_poissonMeasure`)
but *not* its mean or variance; we derive them here from the total-mass identity by a single
index shift (`(n+1)·pmf(n+1) = r·pmf(n)`), with no exp-series differentiation.
## Provenance
The `PoissonRandomMeasure` structure's field shape is consulted from
`cgarryZA/LevyStochCalc` (Apache-2.0), itself a faithful rendering of Applebaum Def. 2.3.1;
the isometry developed on top (see `PoissonCompensatedIsometryAdapted` / `…IntegralL2`) is our
own — LevyStochCalc states that isometry as its cited axiom #6, we prove it. References:
Applebaum, *Lévy Processes and Stochastic Calculus*, CUP 2009, Def. 2.3.1 / Thm 4.2.3.
-/
@[expose] public section
namespace MathFin
open MeasureTheory ProbabilityTheory
open scoped NNReal ENNReal
/-! ## §1 Moments of the scalar Poisson law (Mathlib gap-fill) -/
/-! ## §2 The Poisson random measure and its compensated increment -/
variable {Ω : Type*} [MeasurableSpace Ω] {E : Type*} [MeasurableSpace E]
/-! ### Compensated-increment moments (transport of the scalar moments through `poisson_law`) -/
variable {P : Measure Ω} [IsProbabilityMeasure P] {ν : Measure E} [SigmaFinite ν]
Formal statement
theorem _root_.MathFin.memLp_compensated (N : PoissonRandomMeasure P ν) {D : Set (ℝ × E)}
(hD : MeasurableSet D) (hfin : referenceIntensity ν D ≠ ⊤) :
MemLp (N.compensated D) 2 P := by sorry
Source