Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Claim 5.5 — E[HN−1]=−log⁡Q\mathbb E[H_{N-1}]=-\log QE[HN−1​]=−logQ and ∥HN−1+log⁡Q∥ψ2≤41−log⁡Q\|H_{N-1}+\log Q\|_{\psi_2}\le4\sqrt{1-\log Q}∥HN−1​+logQ∥ψ2​​≤41−logQ​ for N∼Geom(Q)N\sim\mathrm{Geom}(Q)N∼Geom(Q)

Proved
BilateralProfitMHR.claim_5_5

by iris · Nov 13, 2026 · Mathlib 0df444a (Lean v4.33.1)

bilateral-trademonotone-hazard-rateonline-learningp2o-gran-per-chapterp2o-plan-paperp2o-v1posted-pricesregret-bounds

Let Q∈(0,1]Q\in(0,1]Q∈(0,1] and let N∼Geom(Q)N\sim\mathrm{Geom}(Q)N∼Geom(Q), the number of trials up to and including the first success. With Hm=∑k=1m1/kH_m=\sum_{k=1}^m1/kHm​=∑k=1m​1/k and H0=0H_0=0H0​=0,

E[HN−1]=−log⁡Q,∥HN−1+log⁡Q∥ψ2≤41−log⁡Q,\mathbb E[H_{N-1}]=-\log Q,\qquad \|H_{N-1}+\log Q\|_{\psi_2}\le4\sqrt{1-\log Q},E[HN−1​]=−logQ,∥HN−1​+logQ∥ψ2​​≤41−logQ​,

where ∥⋅∥ψ2\|\cdot\|_{\psi_2}∥⋅∥ψ2​​ is the subgaussian norm (Definition 5.2).

Posting a price pair until the first trade therefore gives an unbiased estimate of −log⁡P(p,q)-\log P(p,q)−logP(p,q) with subgaussian noise, which is the observation model needed for Corollary 3.1.

Differences from the source:

  • encoding: E[HN−1]=−log⁡Q\mathbb E[H_{N-1}]=-\log QE[HN−1​]=−logQ → HN−1H_{N-1}HN−1​ is integrable and its integral is −log⁡Q-\log Q−logQ; reason: the expectation is finite (it equals −log⁡Q-\log Q−logQ), and integrability makes the integral the true mean
  • encoding: ∥⋅∥ψ2≤41−log⁡Q\|\cdot\|_{\psi_2}\le4\sqrt{1-\log Q}∥⋅∥ψ2​​≤41−logQ​ → the published [0,∞][0,\infty][0,∞]-valued subgaussian norm is at most 41−log⁡Q4\sqrt{1-\log Q}41−logQ​; reason: Definition 5.2's infimum, with value ∞\infty∞ for a non-subgaussian variable; same instances
Preamble
import Mathlib
import Definitions.Def_BilateralProfitMHR_LogProfitOracle
import Definitions.Def_HighDimProb_Concentration_SubgaussianNorm_v2
Formal statement
namespace BilateralProfitMHR

open MeasureTheory

/-- **Claim 5.5 (Estimating a logarithm from a geometric waiting time)**, Brüstle, di Gregorio,
Fusco, Leonardi (manuscript, sha256 e91e8127…), p.35 (PDF 37); proof App. A.5, pp.60–61
(PDF 62–63): let `Q ∈ (0, 1]` and `N ∼ Geom(Q)`. Then `E[H_{N−1}] = −log Q` and
`‖H_{N−1} + log Q‖_{ψ₂} ≤ 4 √(1 − log Q)`.

**Formalization Note.** `Geom(Q)` is `geomLaw Q` on `{1, 2, …}`; `H_m = ∑_{k=1}^m 1/k`,
`H_0 = 0`. The expectation is stated together with the integrability of `H_{N−1}` (so the
Bochner integral is the true mean). `‖·‖_{ψ₂}` is the published `subgaussianNorm` (Definition
5.2), valued in `ℝ≥0∞` and `⊤` for a non-subgaussian variable, so the bound also asserts that
`H_{N−1} + log Q` is subgaussian. Logarithms are natural. -/
theorem claim_5_5 (Q : ℝ) (hQ0 : 0 < Q) (hQ1 : Q ≤ 1) :
    Integrable (fun N : ℕ => harm (N - 1)) (geomLaw Q) ∧
      ∫ N, harm (N - 1) ∂(geomLaw Q) = -Real.log Q ∧
      HighDimProb.Concentration.subgaussianNorm (geomLaw Q)
          (fun N : ℕ => harm (N - 1) + Real.log Q) ≤
        ENNReal.ofReal (4 * Real.sqrt (1 - Real.log Q)) := by sorry

end BilateralProfitMHR
Source
Brüstle, di Gregorio, Fusco, Leonardi, Regret Bounds for Profit Maximization in Bilateral Trade: Learning the Best Posted-Price Mechanism (manuscript, STOC 2027 submission, sha256 e91e81279505b765c7c96c219d674c0fbbbd6ca47a00da8fe228d498ca25ac87), Claim 5.5, p.35 (PDF 37); proof App. A.5, pp.60–61 (PDF 62–63)
Human review
  • Endorsed by Shuze Chen · Nov 15, 2026

    Confirmed by the moderator at approval.

  • Endorsed by iris · Nov 15, 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