Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 6.1(iv) — stochastic fictitious play in a supermodular game with a unique rest point converges almost surely

Proved
StochFictPlay.Supermodular.sfp_tendsto_unique_rest_point_ae

by mikedeng1 · Sep 28, 2026 · Mathlib 0df444a (Lean v4.33.1)

game-theorylearning-in-gamesp2o-batch-p200bp2o-gran-per-chapterp2o-plan-paperp2o-v1stochastic-approximationstochastic-fictitious-playsupermodular-games

Let GGG be a strictly supermodular game with p≥2p \ge 2p≥2 players, each with at least one strategy, and let each player's shock density fαf^\alphafα satisfy the conditions of Theorem 2.1. Suppose the perturbed best response dynamic (P) has a unique rest point in Σ\SigmaΣ:

RP(P)={x∗}.RP(P) = \{x^*\}.RP(P)={x∗}.

Then standard stochastic fictitious play converges to x∗x^*x∗ almost surely: on every probability space, for every family of shocks εtα\varepsilon^\alpha_tεtα​ with densities fαf^\alphafα that is independent over time and across players, and for every initial pure profile, the empirical frequencies ZtZ_tZt​ satisfy

P(lim⁡t→∞Zt=x∗)=1.P\Big(\lim_{t\to\infty} Z_t = x^*\Big) = 1.P(t→∞lim​Zt​=x∗)=1.

This is a global convergence result for learning in games with strategic complementarities; it concerns the random process itself, not only its mean dynamic.

Formalization Note The process is the one of the SFP definition, defined pathwise from the shocks; densities may differ across players. The first and third sentences of the paper's Theorem 6.1(iv) (invariant manifolds, and the low-dimensional case) are not part of this statement.

Preamble
import Mathlib
import Definitions.Def_StochFictPlay_Supermodular_ChoiceModel
import Definitions.Def_StochFictPlay_Supermodular_Game
import Definitions.Def_StochFictPlay_Supermodular_Dynamics
import Definitions.Def_StochFictPlay_Supermodular_SFP

open MeasureTheory ProbabilityTheory Filter Topology
open scoped ENNReal
Formal statement
namespace StochFictPlay.Supermodular

/-- Theorem 6.1(iv), unique rest point clause (Hofbauer–Sandholm 2002, manuscript p. 24). Let `G`
be a strictly supermodular game with `p ≥ 2` players, and let each player's shock density `f α`
meet the conditions of Theorem 2.1. If the perturbed best response dynamic `(P)` has exactly one
rest point `x*` in `Σ`, then standard stochastic fictitious play converges to `x*` almost
surely: on every probability space, for every family of shocks `ε_t^α` with densities `f α`,
independent over time and across players, and every initial pure profile, the beliefs
`Z_t = (1/t) ∑_{u ≤ t} ζ_u` satisfy `P(lim_{t→∞} Z_t = x*) = 1`. -/
theorem sfp_tendsto_unique_rest_point_ae {p : ℕ} (n : Fin p → ℕ) (hp : 2 ≤ p)
    (hn : ∀ α, 1 ≤ n α)
    (u : (α : Fin p) → Profile n → ℝ) (hu : IsStrictlySupermodular u)
    (f : (α : Fin p) → (Fin (n α) → ℝ) → ℝ≥0∞) (hf : ∀ α, IsRegularDensity (f α))
    (xs : Mixed n) (hRP : restPoints (pField f u) (mixedProfiles n) = {xs})
    {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω) [IsProbabilityMeasure P]
    (ε : ℕ → (α : Fin p) → Ω → (Fin (n α) → ℝ)) (hε : IsShockFamily P f ε)
    (s₁ : Profile n) :
    ∀ᵐ ω ∂P, Tendsto (fun t : ℕ => sfpBelief u s₁ ε t ω) atTop (𝓝 xs) := by sorry

end StochFictPlay.Supermodular
Source
Hofbauer and Sandholm, On the Global Convergence of Stochastic Fictitious Play, Econometrica 70 (2002); authors' manuscript of February 21, 2002, p. 24, Theorem 6.1(iv), second sentence (proof p. 35)
Human review
  • Endorsed by Shuze Chen · Oct 1, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 1, 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