Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition C.3 (first part), p. 35 — F_h(p, y) ≥ ½K ‖Q(y) − p‖² for all y

Proved
ReinfRegGames.StrictStable.proposition_C_3_bound

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

choice-mapconvex-analysisfenchel-couplingp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1penalty-function

Let hhh be a KKK-strongly convex penalty function on the simplex Δ=Δ(B)\Delta=\Delta(B)Δ=Δ(B) (Definition 2.1), with conjugate h∗(y)=max⁡x∈Δ{⟨y∣x⟩−h(x)}h^*(y)=\max_{x\in\Delta}\{\langle y|x\rangle-h(x)\}h∗(y)=maxx∈Δ​{⟨y∣x⟩−h(x)} and Fenchel coupling Fh(p,y)=h(p)+h∗(y)−⟨y∣p⟩F_h(p,y)=h(p)+h^*(y)-\langle y|p\rangleFh​(p,y)=h(p)+h∗(y)−⟨y∣p⟩. Let p∈Δp\in\Deltap∈Δ. Then for every score vector yyy and x=Q(y)x=Q(y)x=Q(y),

Fh(p,y) ≥ 12K ∥x−p∥22.F_h(p,y)\ \ge\ \tfrac12K\,\|x-p\|_2^2 .Fh​(p,y) ≥ 21​K∥x−p∥22​.

The Fenchel coupling therefore controls the distance of the choice Q(y)Q(y)Q(y) from ppp: if Fh(x∗,y)F_h(x^*,y)Fh​(x∗,y) is small, Q(y)Q(y)Q(y) is close to x∗x^*x∗. This is how the proof of Theorem 5.2 shows that the sublevel set Uε∗U^*_\varepsilonUε∗​ of Fh(x∗,⋅)F_h(x^*,\cdot)Fh​(x∗,⋅) is mapped by QQQ into a small neighbourhood of x∗x^*x∗.

Formalization Note Only the first claim of Proposition C.3 is stated here; the clause "Fh(p,y)→0F_h(p,y)\to0Fh​(p,y)→0 iff Q(y)→pQ(y)\to pQ(y)→p" is not part of this item, and (C.11) is a separate item. The norm in the strong convexity inequality (2.5) is the Euclidean one; the page leaves it unspecified, and since KKK is existential in Definition 2.1 the class of penalty functions does not depend on this choice, but the constant KKK does.

Preamble
import Mathlib
import Definitions.Def_agt_games
import Definitions.Def_ReinfRegGames_Extinction_Model
Formal statement
namespace ReinfRegGames.StrictStable

theorem proposition_C_3_bound {B : Type*} [Fintype B] (h : (B → ℝ) → ℝ) (K : ℝ) (hpen : ReinfRegGames.Extinction.IsPenalty h K)
    (p : B → ℝ) (hp : p ∈ stdSimplex ℝ B) (y x : B → ℝ) (hx : ReinfRegGames.Extinction.IsChoice h y x) :
    1 / 2 * K * ReinfRegGames.Extinction.sqDist x p ≤ ReinfRegGames.Extinction.fenchelCoupling h p y := by sorry

end ReinfRegGames.StrictStable
Source
Mertikopoulos & Sandholm, Learning in games via reinforcement and regularization, arXiv:1407.6267v2, p. 35, Proposition C.3, first claim (proof (C.12)–(C.14))
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