(28), A.3 Case 1, pp. 45–46 — if ẑ_i(ϵ) has a fractional component, 2ŷ_i ≥ (v_i0+ϵ)^γ_i [K_i(ϵ)/(v_i0+ϵ) − 2x̂]
ProvedNestedLogitVariants.PartialCompetitive.ineq_28Consider a nested logit instance with for every nest and at least one product per nest, and let be an optimal solution of the linear program (4) over the candidate collections . Fix a nest and a capacity , and suppose that the greedy solution of the continuous knapsack problem (11) has a fractional component , . Then
This is Case 1 of the proof of Theorem 10: the constraints of (4) for and for the singleton together dominate the continuous knapsack value, hence the knapsack value.
Formalization Note The page states (28) at a maximizer of the right side of (10) at ; the argument applies to every capacity , and that is the form stated here, which is what feasibility in (10) requires. The page writes the right side as ; for the two forms are equal. The hypothesis is the paper's nonempty .
import Mathlib import Definitions.Def_NestedLogitVariants_PartialCompetitive_Model import Definitions.Def_NestedLogitVariants_PartialCompetitive_Knapsack
namespace NestedLogitVariants.PartialCompetitive
/-- Inequality (28), A.3 Case 1, pp. 45–46: if `ẑ_i(ε)` has a fractional component `k`, then
`2 ŷ_i ≥ (v_{i0} + ε)^{γ_i} [K_i(ε)/(v_{i0} + ε) − 2 x̂]`. Stated for every `ε ≥ 0`. -/
theorem ineq_28 {ι : Type*} [Fintype ι] [DecidableEq ι] {n : ℕ} (I : Instance ι n)
(hI : I.Standing)
(hγ : ∀ i, I.γ i ≤ 1) (hn : 0 < n)
(xh : ℝ) (yh : ι → ℝ) (hopt : LP4Optimal I (candidates I) xh yh)
(i : ι) (ε : ℝ) (hε : 0 ≤ ε) (k : Fin n) (hk : zhat I i ε k ∈ Set.Ioo (0 : ℝ) 1) :
(I.vnp i + ε) ^ I.γ i * (Kval I i ε / (I.vnp i + ε) - 2 * xh) ≤ 2 * yh i := by sorry
end NestedLogitVariants.PartialCompetitive
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.