Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Product body for the dimension-dependent lower bound

Definition
AltRegret_Supplement_productBody

by iris · Nov 17, 2026 · Mathlib 0df444a (Lean v4.33.1)

algorithmic-game-theoryonline-convex-optimizationonline-learningp2o-gran-per-chapterp2o-plan-paperp2o-v1regret

For d≥2d\ge2d≥2, set r=1/dr=1/\sqrt dr=1/d​. The body used in Theorem 10 is

X=rB2(1)×⋯×rB2(1)⏟⌊d/2⌋ copies×[−r,r]d−2⌊d/2⌋⊆Rd.X=\underbrace{rB_2(1)\times\cdots\times rB_2(1)}_{\lfloor d/2\rfloor\text{ copies}} \times[-r,r]^{d-2\lfloor d/2\rfloor}\subseteq\mathbb R^d.X=⌊d/2⌋ copiesrB2​(1)×⋯×rB2​(1)​​×[−r,r]d−2⌊d/2⌋⊆Rd.

It is compact, convex, full-dimensional, and contained in Bd(1)B_d(1)Bd​(1). The associated exact lower-bound parameters are m=min⁡{⌊d/2⌋,⌊T/16⌋}m=\min\{\lfloor d/2\rfloor,\lfloor T/16\rfloor\}m=min{⌊d/2⌋,⌊T/16⌋} and K=⌊log⁡16⌊T/m⌋⌋K=\lfloor\log_{16}\lfloor T/m\rfloor\rfloorK=⌊log16​⌊T/m⌋⌋.

This is the specific body on which Theorem 10 establishes the dimension dependence.

Definition code
import Mathlib
import Definitions.Def_AltRegret_Supplement_ContinuousAAHedgeRun

namespace AltRegret.Supplement

/-- Coordinate `i`, extended by zero outside `Fin d`. -/
def coord {d : ℕ} (x : Point d) (i : ℕ) : ℝ :=
  if h : i < d then x ⟨i, h⟩ else 0

/-- The radius `r = 1 / √d` in §4.2. -/
noncomputable def blockRadius (d : ℕ) : ℝ := 1 / Real.sqrt d

/-- The product of `⌊d/2⌋` radius-`1/√d` planar balls and, for odd `d`, one interval. -/
noncomputable def productBody (d : ℕ) : Set (Point d) :=
  {x | (∀ j : Fin (d / 2),
      coord x (2 * j.1) ^ 2 + coord x (2 * j.1 + 1) ^ 2 ≤ blockRadius d ^ 2) ∧
    (d % 2 = 1 → |coord x (d - 1)| ≤ blockRadius d)}

/-- `m = min{⌊d/2⌋, ⌊T/16⌋}` from Theorem 10. -/
def lowerBlockCount (d T : ℕ) : ℕ := min (d / 2) (T / 16)

/-- `K = ⌊log₁₆ ⌊T/m⌋⌋` from Theorem 10. -/
noncomputable def lowerLogCount (d T : ℕ) : ℕ :=
  ⌊Real.logb 16 (T / lowerBlockCount d T : ℕ)⌋₊

end AltRegret.Supplement
Source
Tao, Zheng, Optimal Alternating Regret for Online Learning and Games, arXiv:2608.24731, §4.2, p.15

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