On a positive-probability original Gaussian triplet event even the zero-martingale-noise fixed-band stopped cash budget is impossible
ProvedAvramDividend.Classical.avram_gaussianB_original_triplet_positive_event_refutes_zero_noise_stopped_cash_budgetavramdividendcounterexamplegaussianoriginal-tripletstopped-cashzero-martingale
For a genuine original c=2, sigma=1, nu=0 Gaussian Lévy process and admissible equal-cap barrier, there exists a measurable source event of strictly positive Gaussian probability on which the uncharged discounted stopped quadratic terminal-plus-dividend cash upper bound with defect 1/1936 is false even for N=0. Applies the Proved original B180 source refutation to the actual Proved Gaussian long tube with the zero noise benchmark, without inventing stochastic martingale L1 convergence.
Preamble
import Mathlib.Tactic.NormNum import Mathlib.Probability.BrownianMotion.Basic import Definitions.Def_AvramDividend_Classical_SpectrallyNegativeLevy import Definitions.Def_AvramDividend_Classical_DividendStrategy import Theorems.Thm_AvramDividend_Classical_avram_gaussianB_original_triplet_positive_measurable_long_mesh_event import Theorems.Thm_AvramDividend_Classical_avram_gbB180_original_gaussian_positive_event_ae_fixedband_cash_budget_refuted open MeasureTheory ProbabilityTheory AvramDividend.Classical Set open scoped NNReal ENNReal noncomputable section
Formal statement
theorem AvramDividend.Classical.avram_gaussianB_original_triplet_positive_event_refutes_zero_noise_stopped_cash_budget
[IsProbabilityMeasure gaussianLimit] :
∃ (𝓕 : Filtration ℝ≥0
(inferInstance : MeasurableSpace (ℝ≥0 → ℝ)))
(Y : SpectrallyNegativeLevy gaussianLimit 𝓕),
Y.c = 2 ∧ Y.σ = 1 ∧ Y.ν = 0 ∧
Y.Standing ∧ ¬ Y.BoundedVariation ∧
IsAdmissibleLe Y (1 / 8 : ℝ) (ENNReal.ofReal (1 / 8 : ℝ))
(barrierStrategy Y (1 / 8 : ℝ) (1 / 8 : ℝ)) ∧
∃ E : Set (ℝ≥0 → ℝ), MeasurableSet E ∧
0 < gaussianLimit E ∧
¬ (∀ᵐ ω ∂(gaussianLimit.restrict E), Real.exp (-(1 / 100000 : ℝ)) *
(2 * (riskProcess Y (1 / 8 : ℝ) (barrierStrategy Y (1 / 8 : ℝ) (1 / 8 : ℝ)) (1 / 1000 : ℝ≥0) ω) - 4 * (riskProcess Y (1 / 8 : ℝ) (barrierStrategy Y (1 / 8 : ℝ) (1 / 8 : ℝ)) (1 / 1000 : ℝ≥0) ω) ^ 2) +
(∫⁻ t in paymentTimes
(ruinTime Y (1 / 8 : ℝ) (barrierStrategy Y (1 / 8 : ℝ) (1 / 8 : ℝ)) ω) ∩ Iio ((1 / 1000 : ℝ≥0) : ℝ),
ENNReal.ofReal (Real.exp (-((1 / 100 : ℝ) * t)))
∂(dividendMeasure (barrierStrategy Y (1 / 8 : ℝ) (1 / 8 : ℝ)) ω)).toReal - (3 / 16 : ℝ) ≤ (1 / 1936 : ℝ)) := by sorry