Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
โ† Formalpedia

(C.5), p. 48 โ€” Reg_ฯ€(T, v) โ‰ค ๐”ผ_ฯ€ ฮฃ_{โ„“=1}^L (1 + V(S_โ„“))(R(S*, v) โˆ’ R(S_โ„“, v))

Proved
MNLBandit.WellSeparated.display_C_5

by mikedeng1 ยท Oct 6, 2026 ยท Mathlib 0df444a (Lean v4.33.1)

assortment-optimizationmnl-banditp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1regret

Consider the MNL-Bandit problem under Assumption 4.1 with revenues riโˆˆ[0,1]r_i\in[0,1]riโ€‹โˆˆ[0,1], and run Algorithm 1 with any tie-breaking rule. Let LLL be the number of epochs started within the first TTT customers, Sโ„“S_\ellSโ„“โ€‹ the assortment of epoch โ„“\ellโ„“, and V(S)=โˆ‘jโˆˆSvjV(S)=\sum_{j\in S}v_jV(S)=โˆ‘jโˆˆSโ€‹vjโ€‹. Then

Regฯ€(T,v)โ‰คEฯ€[โˆ‘โ„“=1L(1+V(Sโ„“))(R(Sโˆ—,v)โˆ’R(Sโ„“,v))].\mathrm{Reg}_\pi(T,\mathbf v)\le\mathbb E_\pi\Big[\sum_{\ell=1}^{L}\big(1+V(S_\ell)\big)\big(R(S^*,\mathbf v)-R(S_\ell,\mathbf v)\big)\Big].Regฯ€โ€‹(T,v)โ‰คEฯ€โ€‹[โ„“=1โˆ‘Lโ€‹(1+V(Sโ„“โ€‹))(R(Sโˆ—,v)โˆ’R(Sโ„“โ€‹,v))].

Given its assortment, the length of an epoch is geometric with mean 1+V(Sโ„“)1+V(S_\ell)1+V(Sโ„“โ€‹). The display rewrites the per-customer regret as a per-epoch regret, the form in which the epoch-wise bounds of Lemmas 4.3, 6.2 and Corollary C.1 apply.

Formalization Note The paper writes "===". With a finite horizon the last epoch is cut at time TTT, so it contributes at most its expected full length. Only "โ‰ค\leโ‰ค" holds, and this is all the proof of Theorem 3 uses. The expectation is over histories of length TTT, a finite sum.

Preamble
import Mathlib
import Definitions.Def_ChoiceCDLP_MNL_mnlObjective
import Definitions.Def_MNLBandit_WellSeparated_Setting
Formal statement
namespace MNLBandit.WellSeparated

open ChoiceCDLP.MNL

/-- Display (C.5), p. 48 (the epoch rewriting of the regret, as an inequality), for Algorithm 1
under Assumption 4.1 and `r_i โˆˆ [0, 1]`:
`Reg_ฯ€(T, v) โ‰ค ๐”ผ_ฯ€ โˆ‘_{โ„“=1}^{L} (1 + V(S_โ„“))(R(S*, v) โˆ’ R(S_โ„“, v))`, `V(S) = โˆ‘_{jโˆˆS} v_j`, where `L`
is the number of epochs started within the first `T` customers. -/
theorem display_C_5 (N : โ„•) (v r : Fin N โ†’ โ„) (๐’ฎ : Finset (Finset (Fin N))) (h๐’ฎ : ๐’ฎ.Nonempty)
    (sel : (Finset (Fin N) โ†’ โ„) โ†’ Finset (Fin N))
    (hTU : MNLBandit.UCB.IsTU ๐’ฎ) (hdown : MNLBandit.UCB.DownClosed ๐’ฎ) (hv : โˆ€ i, 0 โ‰ค v i โˆง v i โ‰ค 1)
    (hr : โˆ€ i, r i โˆˆ Set.Icc (0 : โ„) 1) (hsel : MNLBandit.UCB.IsArgmaxSel ๐’ฎ sel) (T : โ„•) :
    MNLBandit.UCB.regret v r ๐’ฎ h๐’ฎ (alg1 r sel) T โ‰ค
      โˆ‘ h : Fin T โ†’ Option (Fin N), histProb v (alg1 r sel) T h *
        โˆ‘ โ„“ โˆˆ Finset.Icc 1 (epochsStarted r sel h),
          (1 + โˆ‘ j โˆˆ epochSet r sel h โ„“, v j) *
            (MNLBandit.UCB.optRevenue v r ๐’ฎ h๐’ฎ - revenue v r (epochSet r sel h โ„“)) := by sorry

end MNLBandit.WellSeparated
Source
Agrawal, Avadhanula, Goyal, Zeevi, MNL-Bandit: A Dynamic Learning Approach to Assortment Selection, arXiv:1706.03880v2, p. 48, (C.5) (the same rewriting as (A.14), p. 37)
Human review
  • Endorsed by Shuze Chen ยท Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 ยท Oct 6, 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
ยฉ 2026 Prove2Me