Full shifted-path Esscher-weighted reward factor at arbitrary finite Levy stopping times
ProvedAvramDividend.Classical.avram_ws01e_b_finiteStopping_full_futurePath_esscher_reward_factoravram-classicalesscherfinite-stoppingfull-future-pathregenerationstrong-markovwiener-hopf-prerequisite
At any genuinely finite Levy stopping time tau and for any measurable stopped-sigma history weight H, the entire shifted future Levy path admits the exact deterministic-time Esscher weighted strong Markov reward factor at every future lag u: E_P[H exp(theta*(X_(tau+u)-X_tau)-u psi(theta)) F(shifted path)]=E_P[H] E_P[exp(theta X_u-u psi(theta))F(original path)]. It includes every measurable nonnegative functional F of the entire future path, not merely fixed-time increments or finite cylinders. The result is a full-future-path tilt component; stopped Esscher prefix weight and random Exp(p) horizon still need to be incorporated to reach Qp regeneration.
Preamble
import Mathlib import Theorems.Thm_AvramDividend_Classical_levy_finiteStopping_historyWeight_futureReward_factor open AvramDividend.Classical MeasureTheory ProbabilityTheory Set open scoped NNReal ENNReal
Formal statement
theorem AvramDividend.Classical.avram_ws01e_b_finiteStopping_full_futurePath_esscher_reward_factor
{Ω : Type*} [mΩ : MeasurableSpace Ω] {P : Measure Ω}
{𝓕 : Filtration ℝ≥0 mΩ}
(X : SpectrallyNegativeLevy P 𝓕)
(τ : Ω → ℝ≥0)
(hτ : ∀ s : ℝ≥0,
MeasurableSet[𝓕 s] {ω : Ω | τ ω ≤ s})
(pastWeight : Ω → ℝ≥0∞)
(hp : Measurable pastWeight)
(hpPast : Measurable[
(isStoppingTime_withTop_of_le_events τ hτ).measurableSpace] pastWeight)
(θ : ℝ) (hθ : 0 ≤ θ) (u : ℝ≥0)
(reward : (ℝ≥0 → ℝ) → ℝ≥0∞)
(hr : Measurable reward) :
(∫⁻ ω, pastWeight ω *
(ENNReal.ofReal (Real.exp
(θ * (X.X (τ ω + u) ω - X.X (τ ω) ω) -
(u : ℝ) * X.ψ θ)) *
reward (fun t : ℝ≥0 =>
X.X (τ ω + t) ω - X.X (τ ω) ω)) ∂P) =
(∫⁻ ω, pastWeight ω ∂P) *
(∫⁻ ω, ENNReal.ofReal (Real.exp
(θ * X.X u ω - (u : ℝ) * X.ψ θ)) *
reward (fun t : ℝ≥0 => X.X t ω) ∂P) := by sorrySource
Direct mathematically meaningful Esscher specialisation of the accepted original-law full stopped-history-weighted strong Markov factor b995a471-5065-4673-ba97-d0f13ebfc768, together with measurable_pi_apply and ENNReal.measurable_ofReal for the exponential reward.