Positive-spatial-level first-passage increments have stationary law under Brownian strong Markov
ProvedKyprianou.Ch01.inverseGaussianPositiveBarrierStationaryIncrementV1brownian-passageinverse-gaussiankyprianou-ch01rootwardstationary-incrementsstrong-markov
For a unit-volatility positive-drift Brownian Lévy process and strictly positive barrier s and spatial step h, the level-indexed first-upper-passage clock increment T(s+h)-T(s) has the same distribution as T(h)-T(0). This is the nontrivial positive-positive stationarity step. A sound proof uses the Brownian strong Markov property at the finite passage time T(s), spatial homogeneity X_{T(s)}=s, and deterministic first-passage time-shift additivity. The other stationary-increment boundary cases are handled separately by the generic positive-pair reduction.
Preamble
import Mathlib import Definitions.Def_Kyprianou_Ch01_McKeanBrownianStopping import Definitions.Def_Kyprianou_Ch01_ExplicitLevyExamples import Definitions.Def_LeanLevy_Processes_LevyProcess open MeasureTheory ProbabilityTheory Set Filter open scoped NNReal ENNReal set_option autoImplicit false
Formal statement
theorem Kyprianou.Ch01.inverseGaussianPositiveBarrierStationaryIncrementV1
{Ω : Type*} [MeasurableSpace Ω]
(P : Measure Ω) [IsProbabilityMeasure P]
(X : ℝ≥0 → Ω → ℝ)
(b : ℝ) (hb : 0 < b)
(hB : Kyprianou.Ch01.IsLinearBrownian P X 1 b)
(hMeas : ∀ t : ℝ≥0, Measurable (X t))
(hContinuous : ∀ ω, Continuous (fun t : ℝ≥0 => X t ω))
(s h : ℝ≥0) (hs : 0 < s) (hh : 0 < h) :
let T : ℝ≥0 → Ω → ℝ :=
fun u ω =>
if u = 0 then (0 : ℝ)
else (match Kyprianou.Ch01.firstUpperPassage X (u : ℝ) ω with
| none => (0 : ℝ)
| some t => (t : ℝ))
IdentDistrib (ProbabilityTheory.increment T s (s + h))
(ProbabilityTheory.increment T 0 h) P P := by sorrySource
Kyprianou (2006), Chapter 1 §1.2.5, strong Markov property at first Brownian passage.