Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

App. A, pp. 14–15 — max((nd/16) tanh(ηd/8), min(d log 2/(12η), nd/12)) ≥ min(0.04 nd, 0.01 d^{3/2}√n)

Proved
OnlineCombOpt.Exp2LB.appA_eta_optimization

by mikedeng1 · Oct 8, 2026 · Mathlib 0df444a (Lean v4.33.1)

exponential-weightsinequalitylower-boundp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

For all integers n,d≥0n,d\ge 0n,d≥0 and every learning rate η>0\eta>0η>0,

max⁡(nd16tanh⁡(ηd8), min⁡(dlog⁡212η, nd12))  ≥  min⁡(0.04 nd, 0.01 d3/2n).\max\Bigl(\frac{nd}{16}\tanh\Bigl(\frac{\eta d}{8}\Bigr),\ \min\Bigl(\frac{d\log 2}{12\eta},\ \frac{nd}{12}\Bigr)\Bigr)\;\ge\;\min\bigl(0.04\,nd,\ 0.01\,d^{3/2}\sqrt n\bigr).max(16nd​tanh(8ηd​), min(12ηdlog2​, 12nd​))≥min(0.04nd, 0.01d3/2n​).

This is the optimization over the learning rate in the proof of Theorem 1: whatever η\etaη is, one of the two adversaries of App. A forces a regret of at least min⁡(0.04 nd, 0.01 d3/2n)\min(0.04\,nd,\,0.01\,d^{3/2}\sqrt n)min(0.04nd,0.01d3/2n​), which equals 0.01 d3/2n0.01\,d^{3/2}\sqrt n0.01d3/2n​ when n≥dn\ge dn≥d.

Formalization Note The paper takes the minimum over η∈[0,+∞)\eta\in[0,+\infty)η∈[0,+∞), reading dlog⁡2/(12η)d\log 2/(12\eta)dlog2/(12η) as +∞+\infty+∞ at η=0\eta=0η=0; here the chain is stated pointwise for each η>0\eta>0η>0, because Lean reads dlog⁡2/(12⋅0)d\log2/(12\cdot 0)dlog2/(12⋅0) as 000. The constants 0.040.040.04 and 0.010.010.01 are the reals 4/1004/1004/100 and 1/1001/1001/100, and d3/2=ddd^{3/2}=d\sqrt dd3/2=dd​.

Preamble
import Mathlib
import Definitions.Def_OnlineCombOpt_Exp2LB_Setting

open Finset
Formal statement
namespace OnlineCombOpt.Exp2LB

/-- App. A, pp. 14–15 (Audibert, Bubeck, Lugosi, arXiv:1204.4710v2), the optimization over the
learning rate, pointwise in `η > 0`:
`max((nd/16) tanh(ηd/8), min(d log 2/(12η), nd/12)) ≥ min(0.04 nd, 0.01 d^{3/2} √n)`. -/
theorem appA_eta_optimization (n d : ℕ) (η : ℝ) (hη : 0 < η) :
    min ((4 : ℝ) / 100 * n * d) ((1 : ℝ) / 100 * d * Real.sqrt d * Real.sqrt n) ≤
      max ((n : ℝ) * d / 16 * Real.tanh (η * d / 8))
        (min ((d : ℝ) * Real.log 2 / (12 * η)) ((n : ℝ) * d / 12)) := by sorry

end OnlineCombOpt.Exp2LB
Source
Audibert, Bubeck, Lugosi, Regret in Online Combinatorial Optimization, arXiv:1204.4710v2, App. A, pp. 14–15, the displays "sup R_n ≥ max(…)" (p. 14) and "A = min_{η∈[0,+∞)} max(…) ≥ … ≥ min(0.04 nd, 0.01 d^{3/2}√n)" (p. 15)
Human review
  • Endorsed by Shuze Chen · Oct 9, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 9, 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