Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 2.22 — D_{R⋆}(−z_{1:t} ‖ −z_{1:t−1}) ≤ η Σᵢ wₜ[i]zₜ[i]² when ηzₜ[i] ≥ −1

Proved
OnlineLearningOCO.NormalizedEG.bregman_step_bound

by mikedeng1 · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

exponentiated-gradientonline-convex-optimizationonline-learningp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1regret

Let d≥1d \ge 1d≥1, η>0\eta > 0η>0, let z1,z2,⋯∈Rdz_1, z_2, \dots \in \mathbb R^dz1​,z2​,⋯∈Rd be loss vectors, let wtw_twt​ be the normalized-EG weights wt[i]=e−ηz1:t−1[i]/∑je−ηz1:t−1[j]w_t[i] = e^{-\eta z_{1:t-1}[i]} / \sum_j e^{-\eta z_{1:t-1}[j]}wt​[i]=e−ηz1:t−1​[i]/∑j​e−ηz1:t−1​[j], and let R⋆(θ)=1ηlog⁡(∑ieηθ[i])R^\star(\theta) = \frac1\eta\log\big(\sum_i e^{\eta\theta[i]}\big)R⋆(θ)=η1​log(∑i​eηθ[i]). If at round ttt we have ηzt[i]≥−1\eta z_t[i] \ge -1ηzt​[i]≥−1 for every iii, then

DR⋆(−z1:t ∥ −z1:t−1)≤η∑iwt[i] zt[i]2.D_{R^\star}(-z_{1:t}\,\|\,-z_{1:t-1}) \le \eta \sum_{i} w_t[i]\, z_t[i]^2 .DR⋆​(−z1:t​∥−z1:t−1​)≤ηi∑​wt​[i]zt​[i]2.

This is the per-round bound that, summed over ttt and combined with Lemma 2.20, gives the local-norm regret bound of Theorem 2.22. The right-hand side is η∥zt∥t2\eta\|z_t\|_t^2η∥zt​∥t2​ for the local norm ∥z∥t=∑iwt[i]z[i]2\|z\|_t = \sqrt{\sum_i w_t[i] z[i]^2}∥z∥t​=∑i​wt​[i]z[i]2​ (p. 153).

Formalization Note Rounds are numbered from 000: wmWeights η z t is the paper's wt+1w_{t+1}wt+1​, and cumSum z t is the paper's z1:tz_{1:t}z1:t​. The Bregman divergence is that of R⋆R^\starR⋆ with the softmax as its gradient (bregmanLSE).

Preamble
import Mathlib
import Definitions.Def_OnlineLearningOCO_NormalizedEG_Duality
import Definitions.Def_UnderstandingML_Online
Formal statement
namespace OnlineLearningOCO.NormalizedEG

open UnderstandingML

/-- §2.8, proof of Theorem 2.22, pp. 153–154. If `η > 0` and `η z_t[i] ≥ −1` for every `i`, the
Bregman divergence of the log-sum-exp conjugate `R⋆(θ) = (1/η) log ∑_i e^{η θ[i]}` along the
normalized-EG run is bounded by the local norm of `z_t`:
`D_{R⋆}(−z_{1:t} ‖ −z_{1:t−1}) ≤ η ∑_i w_t[i] z_t[i]²`.
Rounds are 0-based: `wmWeights η z t` is the paper's `w_{t+1}`. -/
theorem bregman_step_bound {d : ℕ} (hd : 0 < d) (η : ℝ) (hη : 0 < η) (z : ℕ → Fin d → ℝ) (t : ℕ)
    (hz : ∀ i, -1 ≤ η * z t i) :
    bregmanLSE η (-(cumSum z (t + 1))) (-(cumSum z t)) ≤
      η * ∑ i, wmWeights η z t i * z t i ^ 2 := by sorry

end OnlineLearningOCO.NormalizedEG
Source
Shalev-Shwartz, Online Learning and Online Convex Optimization, Found. Trends Mach. Learn. 4(2) (2011) 107–194, pp. 153–154, §2.8, proof of Theorem 2.22 (claim on p. 153, last display on p. 154)
Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 7, 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