Original killed stopped horizon reward converges to true upper-first payoff
ProvedAvramDividend.Classical.avram_ws01b_true_horizon_split_reward_tendsto_upper_terminalavramdividendbvdominated_convergencekilledoptional_stopping
On every true original spectrally negative Lévy sample path, the source finite-horizon completed-up-exit plus surviving qscale reward converges at growing natural horizons to the genuine upper-first discounted source terminal payoff; handles finite upper/lower exits and the never-exit atom.
Preamble
import Mathlib import Definitions.Def_AvramDividend_Classical_ScaleFunction import Theorems.Thm_AvramDividend_Classical_avram_ws01b_true_never_exit_discounted_scale_decay import Theorems.Thm_AvramDividend_Classical_chat04_two_sided_strict_exit_no_finite_tie open AvramDividend.Classical MeasureTheory Filter Topology Set open scoped NNReal ENNReal
Formal statement
theorem AvramDividend.Classical.avram_ws01b_true_horizon_split_reward_tendsto_upper_terminal
{Ω : Type*} [mΩ : MeasurableSpace Ω] {P : Measure Ω}
{𝓕 : Filtration ℝ≥0 mΩ}
(X : SpectrallyNegativeLevy P 𝓕)
(q : ℝ) (hq : 0 < q)
(W : ℝ → ℝ) (hW : IsScaleFunction X q W)
(a x : ℝ) (ha : 0 < a) (hx0 : 0 ≤ x) (hxa : x ≤ a)
(ω : Ω) :
Tendsto (fun n : ℕ =>
let U : ℝ≥0∞ := twoSidedExitUpTime X (a-x) ω
let D : ℝ≥0∞ := twoSidedExitDownTime X (-x) ω
(if U < D ∧ U.toNNReal ≤ (n : ℝ≥0) then
Real.exp (-(q * (U.toNNReal : ℝ))) * W (x + X.X U.toNNReal ω)
else 0) +
(if ((n : ℝ≥0) : ℝ≥0∞) < U ∧ ((n : ℝ≥0) : ℝ≥0∞) < D then
Real.exp (-(q * (n : ℝ))) * W (x + X.X (n : ℝ≥0) ω)
else 0)) atTop
(𝓝 (if twoSidedExitUpTime X (a-x) ω <
twoSidedExitDownTime X (-x) ω then
Real.exp (-(q * ((twoSidedExitUpTime X (a-x) ω).toNNReal : ℝ))) *
W (x + X.X (twoSidedExitUpTime X (a-x) ω).toNNReal ω)
else 0)) := by sorrySource
Canonical original no-finite-tie and never-exit discounted scale decay, finite first-exit clock exhaustion at natural horizons.