Theorem 7 — c₁ ≤ a′ ≤ c₂ on [t₀, ∞) gives Π_{0,c₂}(x) ≤ P{(X − t)/a(t) ≤ x | X > t} ≤ Π_{0,c₁}(x)
ProvedBalkemaDeHaan.FiniteT.theorem_7Let be a distribution function which has a positive, differentiable density for , let , and let be real numbers such that
If is a random variable with distribution , then for and all real ,
Here is the family of limit laws (generalized Pareto for , exponential for , bounded support for ). The theorem gives explicit two-sided bounds on the normed residual life distribution at every finite age , not only in the limit . Note the orientation: the upper bound on the derivative gives the lower bound on the distribution function.
Formalization Note The derivative of is a named function with HasDerivWithinAt on . is . For , is completed by beyond (see the definition). No sign condition on is added.
import Mathlib import Definitions.Def_BalkemaDeHaan_FiniteT_Pi0 import Definitions.Def_BalkemaDeHaan_FiniteT_ResidualLife open MeasureTheory ProbabilityTheory
namespace BalkemaDeHaan.FiniteT
/-- Theorem 7, pp. 801–802: let `F = cdf μ` have a positive, differentiable density `f` for
`t ≥ t₀`, and let `c₁ ≤ (d/dt)((1 - F(t))/F′(t)) ≤ c₂` for `t ≥ t₀`. Then for `t ≥ t₀`
and all `x`, `Π_{0,c₂}(x) ≤ P{(X - t)/a(t) ≤ x | X > t} ≤ Π_{0,c₁}(x)`, where
`a(t) = (1 - F(t))/F′(t)` and `P{(X - t)/a(t) ≤ x | X > t} = F_t(x a(t))`. -/
theorem theorem_7 (μ : Measure ℝ) [IsProbabilityMeasure μ] (t₀ : ℝ) (f f' a' : ℝ → ℝ)
(hdens : HasPosDiffDensity μ t₀ f f')
(ha' : ∀ t ≥ t₀, HasDerivWithinAt (normA μ f) (a' t) (Set.Ici t₀) t)
(c₁ c₂ : ℝ) (hc : ∀ t ≥ t₀, c₁ ≤ a' t ∧ a' t ≤ c₂) :
∀ t ≥ t₀, ∀ x : ℝ,
pi0 c₂ x ≤ BalkemaDeHaan.ParetoBounds.residualLife μ t (x * normA μ f t) ∧
BalkemaDeHaan.ParetoBounds.residualLife μ t (x * normA μ f t) ≤ pi0 c₁ x := by sorry
end BalkemaDeHaan.FiniteT
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.