Theorem 2.2, p.97 — and are convex
ProvedKallMayerSLP.ProbConstr.theorem_2_2chance-constraintsp2o-gran-per-chapterp2o-plan-bookp2o-strengthenedp2o-v1probability-constraintsstochastic-programming
Let be a random vector in with an arbitrary distribution , and let , be measurable random data. Then the feasible sets
of the joint probability constraint are convex.
Without assumptions on the distribution these are the only levels at which convexity of is guaranteed; at the constraint asks for a "fat solution" that satisfies the random system almost surely.
Differences from the source:
- generalization: T(ξ) = T + Σ_j T^j ξ_j and h(ξ) = h + Σ_j h^j ξ_j affine in ξ (Notations, p.ix) → T(ξ), h(ξ) any functions of ξ whose entries are Borel measurable; reason: affine maps are continuous, hence measurable, so the book's data are an instance
- encoding: the distribution ℙ_ξ of ξ on (ℝ^r, 𝓑^r) → a probability measure μ on Fin r → ℝ with its Borel σ-algebra; probabilities are (μ A).toReal; reason: the real value of a probability is exact, so every instance is the same
Preamble
import Mathlib import Definitions.Def_KallMayerSLP_ProbConstr_feasibleSet open MeasureTheory
Formal statement
namespace KallMayerSLP.ProbConstr
/-- **Theorem 2.2** (Kall). `B(α)` is convex for `α = 0` and `α = 1`.
Kall & Mayer 2005, Chapter 2, Theorem 2.2, p.97 (PDF 107).
**Formalization Note.** `μ` is any probability measure on `ℝ^r` (the book: "without any assumptions concerning
the probability distribution of ξ"). The data `T(ξ)`, `h(ξ)` are any Borel measurable functions
of `ξ` (entrywise); the book's affine data (Notations, p.ix) are continuous, hence measurable, so
the statement contains the book's. -/
theorem theorem_2_2 {r m n : ℕ} (μ : Measure (Fin r → ℝ)) [IsProbabilityMeasure μ]
(T : (Fin r → ℝ) → Matrix (Fin m) (Fin n) ℝ) (h : (Fin r → ℝ) → Fin m → ℝ)
(hT : ∀ i j, Measurable fun ξ => T ξ i j) (hh : ∀ i, Measurable fun ξ => h ξ i) :
Convex ℝ (feasibleSet μ T h 0) ∧ Convex ℝ (feasibleSet μ T h 1) := by sorry
end KallMayerSLP.ProbConstr
Source
Kall & Mayer, Stochastic Linear Programming: Models, Theory, and Computation (Kluwer/Springer 2005), Chapter 2, Chapter 2, §2.1, Theorem 2.2, p.97 (PDF 107)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.