Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 1, p. 6 — under HEC, Q_i(u, λ) = G_i(u, λ) − Σ_{k∈W_i^−} (λ_i/L_k^+)·G_k(u, λ)

Proved
HetExponomial.ChoiceProb.theorem_1

by mikedeng1 · Oct 22, 2026 · Mathlib 0df444a (Lean v4.33.1)

choice-probabilityclosed-formdiscrete-choicep2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1random-utility

Theorem 1 (closed-form HEC choice probabilities). Consider n≥2n\ge 2n≥2 choices N={1,…,n}N=\{1,\dots,n\}N={1,…,n}. Choice iii has random utility ui−ziu_i-z_iui​−zi​, where ui∈Ru_i\in\mathbb Rui​∈R is its ideal utility. The errors zj∼Exp(λj)z_j\sim\mathrm{Exp}(\lambda_j)zj​∼Exp(λj​), λj>0\lambda_j>0λj​>0, are independent across choices. Let

Qi(u,λ)=P{ui−zi≥uj−zj  ∀j∈N, j≠i}Q_i(\mathbf u,\boldsymbol\lambda)=P\{u_i-z_i\ge u_j-z_j\ \ \forall j\in N,\ j\ne i\}Qi​(u,λ)=P{ui​−zi​≥uj​−zj​  ∀j∈N, j=i}

be the probability that choice iii is utility-maximizing. Let Bi={j:uj≥ui}B_i=\{j: u_j\ge u_i\}Bi​={j:uj​≥ui​}, Bi+={j:uj>ui}B_i^+=\{j: u_j>u_i\}Bi+​={j:uj​>ui​}, Wi−={j:uj<ui}W_i^-=\{j: u_j<u_i\}Wi−​={j:uj​<ui​}, Li=∑j∈BiλjL_i=\sum_{j\in B_i}\lambda_jLi​=∑j∈Bi​​λj​, Li+=∑j∈Bi+λjL_i^+=\sum_{j\in B_i^+}\lambda_jLi+​=∑j∈Bi+​​λj​, and

Gi(u,λ)=λiLiexp⁡[−∑j∈Biλj(uj−ui)].G_i(\mathbf u,\boldsymbol\lambda)=\frac{\lambda_i}{L_i}\exp\Big[-\sum_{j\in B_i}\lambda_j(u_j-u_i)\Big].Gi​(u,λ)=Li​λi​​exp[−j∈Bi​∑​λj​(uj​−ui​)].

Then for every i∈Ni\in Ni∈N,

Qi(u,λ)=Gi(u,λ)−∑k∈Wi−λiLk+⋅Gk(u,λ).Q_i(\mathbf u,\boldsymbol\lambda)=G_i(\mathbf u,\boldsymbol\lambda)-\sum_{k\in W_i^-}\frac{\lambda_i}{L_k^+}\cdot G_k(\mathbf u,\boldsymbol\lambda).Qi​(u,λ)=Gi​(u,λ)−k∈Wi−​∑​Lk+​λi​​⋅Gk​(u,λ).

The theorem is the closed form of the HEC choice probabilities. It underlies every later result of the paper: continuity and derivatives (Theorem 2), consumer surplus (Theorem 3), and the pricing analysis. Ties among the ideal utilities are allowed. When all uju_juj​ are equal it reduces to Qi=λi/(λ1+⋯+λn)Q_i=\lambda_i/(\lambda_1+\dots+\lambda_n)Qi​=λi​/(λ1​+⋯+λn​). As an illustration (Table 1 of the paper), u=(1,2,3)u=(1,2,3)u=(1,2,3) with λ=(10,1,0.1)\lambda=(10,1,0.1)λ=(10,1,0.1) gives Q≈(27.13%,57.59%,15.28%)Q\approx(27.13\%,57.59\%,15.28\%)Q≈(27.13%,57.59%,15.28%).

Formalization Note QiQ_iQi​ is the probability of the event above under the product of the laws Exp(λj)\mathrm{Exp}(\lambda_j)Exp(λj​). It is neither the integral (1) nor the right-hand side. The paper defines Lk+L_k^+Lk+​ only when uk<umax⁡u_k<u_{\max}uk​<umax​; the Lean function is total, with value 000 at a maximal choice. In the theorem it is evaluated only at k∈Wi−k\in W_i^-k∈Wi−​, where i∈Bk+i\in B_k^+i∈Bk+​ gives Lk+≥λi>0L_k^+\ge\lambda_i>0Lk+​≥λi​>0, so no extra hypothesis is needed. The hypothesis n≥2n\ge 2n≥2 is the paper's standing assumption (p. 4).

Preamble
import Mathlib
import Definitions.Def_HetExponomial_ChoiceProb_Model
Formal statement
namespace HetExponomial.ChoiceProb

theorem theorem_1 {n : ℕ} (hn : 2 ≤ n) (u lam : Fin n → ℝ) (hlam : ∀ j, 0 < lam j) (i : Fin n) :
    choiceProb u lam i =
      G u lam i - ∑ k ∈ strictlyWorse u i, lam i / Lplus u lam k * G u lam k := by sorry

end HetExponomial.ChoiceProb
Source
Alptekinoğlu & Semple, Heteroscedastic Exponomial Choice, SSRN 3232788 (version of August 8, 2020; forthcoming in Operations Research), p. 6, Theorem 1, equation (3)
Human review
  • Endorsed by Shuze Chen · Oct 22, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 22, 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