Proposition 4.1 — integrated tails of and
ProvedKallMayerSLP.ICC.proposition_4_1convexityintegrated-chance-constraintsp2o-gran-per-chapterp2o-plan-bookp2o-v1polyhedrastochastic-programming
Let be a real random variable on a probability space whose expected value exists, with distribution function , and let . Write and . Then both and exist, and
These formulas explain the word "integrated": an expected shortfall is an integral of tail probabilities.
Formalization Note " exists" is integrability of . The statement also asserts that the two improper integrals exist, as Lebesgue integrals over and .
Differences from the source:
- encoding: 𝔼[ϑ] exists, i.e. ∫|t| dF_ϑ(t) is finite (p.147) → Integrable ϑ μ, with the existence of all four integrals as explicit clauses; reason: the Bochner integral would otherwise be 0; same instances
Preamble
import Mathlib import Definitions.Def_KallMayerSLP_ICC_distFn open MeasureTheory
Formal statement
namespace KallMayerSLP.ICC
/-- **Proposition 4.1 (integrated tails).**
Kall & Mayer, *Stochastic Linear Programming* (2005), Chapter 2, Proposition 4.1 and (4.1), p.148
(PDF 158). If `𝔼[ϑ]` exists, then for all `z ∈ ℝ` both `𝔼[(ϑ − z)⁺]` and `𝔼[(ϑ − z)⁻]` exist, and
`𝔼[(ϑ − z)⁺] = ∫_z^∞ (1 − F_ϑ(t)) dt`, `𝔼[(ϑ − z)⁻] = ∫_{−∞}^z F_ϑ(t) dt`.
**Formalization Note.** "`𝔼[ϑ]` exists" is the finiteness of `∫|t| dF_ϑ(t)` (p.147), i.e.
`Integrable ϑ μ`. Existence of both expectations and of both improper integrals is stated
explicitly, as a Bochner integral would otherwise be `0`. `u⁺ = max{0, u}`, `u⁻ = max{0, −u}`. -/
theorem proposition_4_1 {Ω : Type*} [MeasurableSpace Ω] (μ : Measure Ω) [IsProbabilityMeasure μ]
(ϑ : Ω → ℝ) (hϑ : Integrable ϑ μ) (z : ℝ) :
Integrable (fun ω => max 0 (ϑ ω - z)) μ ∧
Integrable (fun ω => max 0 (-(ϑ ω - z))) μ ∧
IntegrableOn (fun t => 1 - distFn μ ϑ t) (Set.Ioi z) ∧
IntegrableOn (distFn μ ϑ) (Set.Iic z) ∧
∫ ω, max 0 (ϑ ω - z) ∂μ = ∫ t in Set.Ioi z, (1 - distFn μ ϑ t) ∧
∫ ω, max 0 (-(ϑ ω - z)) ∂μ = ∫ t in Set.Iic z, distFn μ ϑ t := by sorry
end KallMayerSLP.ICC
Source
Kall & Mayer, Stochastic Linear Programming (2005), Chapter 2, Proposition 4.1 and (4.1), p.148 (PDF 158)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.