Zero-start Brownian modification has the exact centred Gaussian time-marginal characteristic function
ProvedKyprianou.Ch01.strictBrownian_zeroGaussianCharFunbrowniancharacteristic-functiongaussiankyprianou-ch01levy-process
For any Mathlib real Brownian motion X, probability law P, nonnegative time t and real frequency xi, the characteristic function of the zero-start modification X(t)-X(0) equals exp(-t xi²/2). The proof uses the independently proved Brownian zero-initial marginal and Brownian process normalisation to identify the exact zero-centred Gaussian law, then the pinned Mathlib Gaussian characteristic-function formula.
Preamble
import Mathlib import Definitions.Def_LeanLevy_Processes_LevyProcess open MeasureTheory ProbabilityTheory Complex open scoped NNReal set_option autoImplicit false
Formal statement
theorem Kyprianou.Ch01.strictBrownian_zeroGaussianCharFun
{Ω : Type} [MeasurableSpace Ω]
(P : Measure Ω) [IsProbabilityMeasure P]
(X : ℝ≥0 → Ω → ℝ)
(hB : IsBrownianReal X P) (t : ℝ≥0) (xi : ℝ) :
charFun (P.map (fun ω => X t ω - X 0 ω)) xi =
Complex.exp (-(((t : ℝ) : ℂ) * (xi : ℂ) ^ 2 / 2)) := by
sorrySource
Kyprianou (2006), Chapter 1, Gaussian term in the Lévy–Khintchine characteristic formula; pinned Mathlib ProbabilityTheory.IsPreBrownianReal.hasLaw_eval and charFun_gaussianReal.