Proof of Theorem 4.5, p. 1493 — Π_q(𝒜) ⊆ 𝒰 iff there exist s₁, …, s_m, t₁, …, t_m with e′(s_k + t_k) ≥ v_k and s_{k,i} + t_{k,j} ≤ (u′_k a_j)q_i
ProvedRiskUncSets.InnerApprox.permutohull_subset_polytope_ifflinear-optimizationlp-dualityp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1permutohullrobust-optimization
Let , let be any vector, and let be the polyhedron (14). Then if and only if there exist such that
Containment of the permutohull in each half-space is a minimization over doubly stochastic matrices, and the displayed system is its dual (an assignment problem and its dual). This turns the containment constraint of Theorem 4.5 into the linear constraints of (15).
Formalization Note The vector is arbitrary in , with no sign or ordering condition: the linear program (15) applies this equivalence at , which may have negative entries. Indices are -based in Lean. For both sides hold trivially.
Preamble
import Mathlib import Definitions.Def_RiskUncSets_InnerApprox_Setting noncomputable section
Formal statement
namespace RiskUncSets.InnerApprox
/-- Proof of Theorem 4.5: `Π_q(𝒜) ⊆ 𝒰` iff there are `s_k, t_k ∈ ℝᴺ` with
`e′(s_k + t_k) ≥ v_k` and `s_{k,i} + t_{k,j} ≤ (u_k′ a_j) q_i`; for every `q ∈ ℝᴺ`. -/
theorem permutohull_subset_polytope_iff {N n m : ℕ} (q : Fin N → ℝ)
(a : Fin N → Fin n → ℝ) (u : Fin m → Fin n → ℝ) (v : Fin m → ℝ) :
permutohull q a ⊆ polytope u v ↔
∃ s t : Fin m → Fin N → ℝ, (∀ k, v k ≤ ∑ i, (s k i + t k i)) ∧
∀ k i j, s k i + t k j ≤ (u k ⬝ᵥ a j) * q i := by sorry
end RiskUncSets.InnerApprox
Source
Bertsimas & Brown, Constructing uncertainty sets for robust linear optimization, Oper. Res. 57(6) (2009), p. 1493, proof of Theorem 4.5
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.