Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 11.5, p. 183 — exp-concave losses on the shrunken competitor set

Proved
ModernOnlineLearning.Aggregating.restricted_prior_step

by mikedeng1 · Oct 9, 2026 · Mathlib 0df444a (Lean v4.33.1)

exp-concavityonline-learningp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1weighted-average

Let d,T≥1d,T\geq1d,T≥1, α>0\alpha>0α>0, V⊆RdV\subseteq\mathbb R^dV⊆Rd be convex, u∈Vu\in Vu∈V, and each ℓt\ell_tℓt​ be α\alphaα-exp-concave on VVV. Put r=(T/d)/(T/d+1)r=(T/d)/(T/d+1)r=(T/d)/(T/d+1) and V′={ru+(1−r)w:w∈V}V'=\{ru+(1-r)w:w\in V\}V′={ru+(1−r)w:w∈V}. For every x∈V′x\in V'x∈V′ there is w∈Vw\in Vw∈V with x=ru+(1−r)wx=ru+(1-r)wx=ru+(1−r)w, and for 1≤t≤T1\leq t\leq T1≤t≤T,

e−αℓt(x)≥re−αℓt(u)+(1−r)e−αℓt(w)≥re−αℓt(u).e^{-\alpha\ell_t(x)}\geq r e^{-\alpha\ell_t(u)}+(1-r)e^{-\alpha\ell_t(w)}\geq r e^{-\alpha\ell_t(u)}.e−αℓt​(x)≥re−αℓt​(u)+(1−r)e−αℓt​(w)≥re−αℓt​(u).

Consequently,

e−α∑t=1Tℓt(x)≥rTe−α∑t=1Tℓt(u)≥e−de−α∑t=1Tℓt(u).e^{-\alpha\sum_{t=1}^T\ell_t(x)}\geq r^T e^{-\alpha\sum_{t=1}^T\ell_t(u)}\geq e^{-d}e^{-\alpha\sum_{t=1}^T\ell_t(u)}.e−α∑t=1T​ℓt​(x)≥rTe−α∑t=1T​ℓt​(u)≥e−de−α∑t=1T​ℓt​(u).

This gives the loss comparison on the portion of the prior retained near a fixed competitor.

Formalization Note The hypotheses d,T≥1d,T\geq1d,T≥1 make the scaling and the printed division by ddd well defined.

Preamble
import Mathlib
import Definitions.Def_ModernOnlineLearning_Aggregating_ExpConcave
import Definitions.Def_ModernOnlineLearning_Aggregating_ShrunkSet
set_option autoImplicit false
Formal statement
namespace ModernOnlineLearning.Aggregating

/-- The two exponential inequalities in the proof of Theorem 11.5, p. 183. -/
theorem restricted_prior_step {d : ℕ} (hd : 0 < d) (T : ℕ) (hT : 0 < T)
    (V : Set (EuclideanSpace ℝ (Fin d))) (hV : Convex ℝ V)
    (u : EuclideanSpace ℝ (Fin d)) (hu : u ∈ V)
    (α : ℝ) (hα : 0 < α)
    (ℓ : ℕ → EuclideanSpace ℝ (Fin d) → ℝ)
    (hℓ : ∀ t ∈ Finset.Icc 1 T, ExpConcaveOn V α (ℓ t)) :
    ∀ x ∈ shrunkSet V u T, ∃ w ∈ V,
      x = (((T : ℝ) / (d : ℝ)) / ((T : ℝ) / (d : ℝ) + 1)) • u +
          (1 / ((T : ℝ) / (d : ℝ) + 1)) • w ∧
      (∀ t ∈ Finset.Icc 1 T,
        Real.exp (-α * ℓ t x) ≥
          (((T : ℝ) / (d : ℝ)) / ((T : ℝ) / (d : ℝ) + 1)) *
            Real.exp (-α * ℓ t u) +
          (1 / ((T : ℝ) / (d : ℝ) + 1)) * Real.exp (-α * ℓ t w) ∧
        (((T : ℝ) / (d : ℝ)) / ((T : ℝ) / (d : ℝ) + 1)) *
            Real.exp (-α * ℓ t u) +
          (1 / ((T : ℝ) / (d : ℝ) + 1)) * Real.exp (-α * ℓ t w) ≥
          (((T : ℝ) / (d : ℝ)) / ((T : ℝ) / (d : ℝ) + 1)) *
            Real.exp (-α * ℓ t u)) ∧
      Real.exp (-α * ∑ t ∈ Finset.Icc 1 T, ℓ t x) ≥
        ((((T : ℝ) / (d : ℝ)) / ((T : ℝ) / (d : ℝ) + 1)) ^ T) *
        Real.exp (-α * ∑ t ∈ Finset.Icc 1 T, ℓ t u) ∧
      Real.exp (-α * ∑ t ∈ Finset.Icc 1 T, ℓ t x) ≥
        Real.exp (-α * ∑ t ∈ Finset.Icc 1 T, ℓ t u) / Real.exp (d : ℝ) := by sorry

end ModernOnlineLearning.Aggregating
Source
Orabona, arXiv:1912.13213v10, proof of Theorem 11.5, p. 183, first two display chains
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