Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 9 — for every α > 1, O(⌈α/(α−1)⌉ n^{⌈α/(α−1)⌉+2}) candidates include an α-approximate solution of (7)

Proved
ConstrNestedLogit.Scheme.theorem_9

by mikedeng1 · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

approximationassortment-optimizationknapsacknested-logitp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Consider a nest iii of the nested logit model with preference weights vij>0v_{ij}>0vij​>0, revenues rijr_{ij}rij​ (of any sign), no within-nest no-purchase weight, and space constraints Ci={S:∑j∈Swij≤ci}\mathcal C_i=\{S:\sum_{j\in S}w_{ij}\le c_i\}Ci​={S:∑j∈S​wij​≤ci​} with 0<wij≤ci0<w_{ij}\le c_i0<wij​≤ci​ and ci≥0c_i\ge0ci​≥0. Write Vi(S)=∑j∈SvijV_i(S)=\sum_{j\in S}v_{ij}Vi​(S)=∑j∈S​vij​ and Ri(S)=∑j∈Srijvij/Vi(S)R_i(S)=\sum_{j\in S}r_{ij}v_{ij}/V_i(S)Ri​(S)=∑j∈S​rij​vij​/Vi​(S) (with Ri(∅)=0R_i(\emptyset)=0Ri​(∅)=0). For every α>1\alpha>1α>1, with q=⌈α/(α−1)⌉q=\lceil\alpha/(\alpha-1)\rceilq=⌈α/(α−1)⌉, there is a collection {Ait:t∈Ti}⊆Ci\{A_i^t:t\in\mathcal T_i\}\subseteq\mathcal C_i{Ait​:t∈Ti​}⊆Ci​ with

∣Ti∣≤5 (q+1) nq+2+1|\mathcal T_i|\le 5\,(q+1)\,n^{q+2}+1∣Ti​∣≤5(q+1)nq+2+1

such that for every u≥0u\ge0u≥0 some member S^\hat SS^ of the collection is an α\alphaα-approximate solution of problem (7):

Vi(S)(Ri(S)−u)≤α Vi(S^)(Ri(S^)−u)for every S∈Ci.V_i(S)\big(R_i(S)-u\big)\le \alpha\,V_i(\hat S)\big(R_i(\hat S)-u\big)\qquad\text{for every } S\in\mathcal C_i .Vi​(S)(Ri​(S)−u)≤αVi​(S^)(Ri​(S^)−u)for every S∈Ci​.

Combined with Theorem 4 and Theorem 2 of the paper, this yields an α\alphaα-approximation for the space-constrained assortment problem through a linear program with 1+m1+m1+m variables and O(m⌈α/(α−1)⌉n⌈α/(α−1)⌉+2)O(m\lceil\alpha/(\alpha-1)\rceil n^{\lceil\alpha/(\alpha-1)\rceil+2})O(m⌈α/(α−1)⌉n⌈α/(α−1)⌉+2) constraints.

Formalization Note The paper's O(⌈α/(α−1)⌉n⌈α/(α−1)⌉+2)O(\lceil\alpha/(\alpha-1)\rceil n^{\lceil\alpha/(\alpha-1)\rceil+2})O(⌈α/(α−1)⌉n⌈α/(α−1)⌉+2) is pinned to the explicit bound 5(q+1)nq+2+15(q+1)n^{q+2}+15(q+1)nq+2+1, uniform in qqq like the paper's: for n≥1n\ge1n≥1, at most (q+1)nq(q+1)n^q(q+1)nq sets JJJ of size at most qqq, times at most 2n(n+1)+1≤5n22n(n+1)+1\le5n^22n(n+1)+1≤5n2 pieces of [0,∞)[0,\infty)[0,∞) (the open intervals between the at most n(n+1)n(n+1)n(n+1) intersection points of the lines hijh_{ij}hij​ and fijf_{ij}fij​, and the points themselves); the +1+1+1 covers n=0n=0n=0. The collection is fixed before uuu. vij>0v_{ij}>0vij​>0, wij>0w_{ij}>0wij​>0 and ci≥0c_i\ge0ci​≥0 are added; wij≤ciw_{ij}\le c_iwij​≤ci​ is the paper's assumption. Products are Fin n, nests a finite type, and the within-nest no-purchase weight of the published model is set to zero.

Preamble
import Mathlib
import Definitions.Def_NestedLogitVariants_LP_Model
import Definitions.Def_ConstrNestedLogit_Scheme_Model
Formal statement
namespace ConstrNestedLogit.Scheme

open NestedLogitVariants.LP in
/-- Theorem 9, p. 40: under space constraints, for every `α > 1`, with `q = ⌈α/(α − 1)⌉`, there is
a collection `A` of space-feasible assortments of nest `i` with at most
`5 (q + 1) n^(q + 2) + 1` members (the paper's `|T_i| = O(⌈α/(α − 1)⌉ n^{⌈α/(α−1)⌉+2})`)
that contains an `α`-approximate solution of problem (7) for every `u ≥ 0`. -/
theorem theorem_9 {ι : Type*} {n : ℕ}
    (I : NestedLogitVariants.LP.Instance ι n) (hvnp : ∀ i, I.vnp i = 0)
    (hv : ∀ i j, 0 < I.v i j) (w : ι → Fin n → ℝ) (c : ι → ℝ)
    (hwc : ∀ i j, w i j ≤ c i) (hw : ∀ i j, 0 < w i j) (hc : ∀ i, 0 ≤ c i) (i : ι)
    (α : ℝ) (hα : 1 < α) :
    ∃ A : Finset (Finset (Fin n)), (∀ S ∈ A, ConstrNestedLogit.Space.spaceFeasible w c i S) ∧
      A.card ≤ 5 * (⌈α / (α - 1)⌉₊ + 1) * n ^ (⌈α / (α - 1)⌉₊ + 2) + 1 ∧
      ∀ u : ℝ, 0 ≤ u → ∃ S ∈ A, ∀ S', ConstrNestedLogit.Space.spaceFeasible w c i S' →
        V I i S' * (R I i S' - u) ≤ α * (V I i S * (R I i S - u)) := by sorry

end ConstrNestedLogit.Scheme
Source
Gallego & Topaloglu, Constrained Assortment Optimization for the Nested Logit Model, Management Science (2014), DOI 10.1287/mnsc.2014.1931; authors' manuscript of Sept. 11, 2013, p. 40, Theorem 9
Human review
  • Endorsed by Shuze Chen · Oct 8, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 8, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me