Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 2.31 — one projected subgradient step

Proved
ModernOnlineLearning.OGD.lemma_2_31

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

online-learningonline-subgradient-descentp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1regret

Let VVV be nonempty, closed, and convex; let xt,u∈Vx_t,u\in Vxt​,u∈V and ηt>0\eta_t>0ηt​>0. If gtg_tgt​ is a subgradient of ℓt\ell_tℓt​ at xtx_txt​ and xt+1=ΠV(xt−ηtgt)x_{t+1}=\Pi_V(x_t-\eta_tg_t)xt+1​=ΠV​(xt​−ηt​gt​), then

ηt(ℓt(xt)−ℓt(u))≤ηt⟨gt,xt−u⟩≤12∥xt−u∥2−12∥xt+1−u∥2+ηt22∥gt∥2.\eta_t(\ell_t(x_t)-\ell_t(u))\le\eta_t\langle g_t,x_t-u\rangle\le\frac12\|x_t-u\|^2-\frac12\|x_{t+1}-u\|^2+\frac{\eta_t^2}{2}\|g_t\|^2.ηt​(ℓt​(xt​)−ℓt​(u))≤ηt​⟨gt​,xt​−u⟩≤21​∥xt​−u∥2−21​∥xt+1​−u∥2+2ηt2​​∥gt​∥2.

This extends the per-round OGD estimate to subgradient choices.

Formalization Note Losses are real valued on VVV, and the subgradient inequality is required only against points of VVV. A full-space subgradient as in Definition 2.20 implies this relative condition.

Preamble
import Mathlib
import Definitions.Def_ModernOnlineLearning_OGD_Defs
Formal statement
namespace ModernOnlineLearning.OGD

/-- Orabona, Lemma 2.31, p. 19: the projected subgradient step obeys the
same one-step regret inequalities. -/
theorem lemma_2_31 {E : Type*} [NormedAddCommGroup E] [InnerProductSpace ℝ E]
    [CompleteSpace E] (V : Set E) (hV : V.Nonempty) (hc : IsClosed V)
    (hv : Convex ℝ V) (loss : E → ℝ) (x xn g u : E) (η : ℝ)
    (hx : x ∈ V) (hu : u ∈ V) (hη : 0 < η)
    (hg : IsSubgradientOn V loss x g)
    (hxn : OnlineConvexOpt.FirstOrder.IsMetricProjection V (x - η • g) xn) :
    η * (loss x - loss u) ≤ η * inner ℝ g (x - u) ∧
      η * inner ℝ g (x - u) ≤
        (1 / 2 : ℝ) * ‖x - u‖ ^ 2 - (1 / 2 : ℝ) * ‖xn - u‖ ^ 2 +
          (η ^ 2 / 2) * ‖g‖ ^ 2 := by sorry

end ModernOnlineLearning.OGD
Source
Orabona, arXiv:1912.13213v10, Lemma 2.31, p. 19
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