Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Equation (3.5), p. 8 — linear lower bound H(x, u) ≥ (r/n)|u| − r‖Ψ‖∞

Proved
SemilinearHJ.Verification.eq_3_5

by mikedeng1 · Oct 24, 2026 · Mathlib 0df444a (Lean v4.33.1)

hamilton-jacobioptimal-controlp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1state-constraintsviscosity-solutions

Let X⊂RnX\subset\mathbb R^nX⊂Rn be a polytope with tangent space TXTXTX, let Ψ:X→R+n\Psi:X\to\mathbb R^n_+Ψ:X→R+n​ be Lipschitz continuous and r>0r>0r>0. Let H(x,u)=max⁡v∈TX∩Br(⟨u,v⟩−⟨Ψ(x),[v]+⟩)H(x,u)=\max_{v\in TX\cap B_r}\big(\langle u,v\rangle-\langle\Psi(x),[v]_+\rangle\big)H(x,u)=maxv∈TX∩Br​​(⟨u,v⟩−⟨Ψ(x),[v]+​⟩) be the maximized Hamiltonian (3.4), ∣⋅∣|\cdot|∣⋅∣ the ℓ1\ell^1ℓ1 norm and ∥Ψ∥∞=max⁡x∈X∣Ψ(x)∣\|\Psi\|_\infty=\max_{x\in X}|\Psi(x)|∥Ψ∥∞​=maxx∈X​∣Ψ(x)∣. Then for every x∈Xx\in Xx∈X and every u∈TXu\in TXu∈TX,

H(x,u) ≥ rn ∣u∣−r∥Ψ∥∞.H(x,u)\ \ge\ \frac rn\,|u|-r\|\Psi\|_\infty .H(x,u) ≥ nr​∣u∣−r∥Ψ∥∞​.

This coercivity bound is what makes subsolutions of the Hamilton–Jacobi system Lipschitz (Equation (4.7)) and drives the Lipschitz estimate in the proof of Theorem 3.2.

Formalization Note HHH is defined for every u∈Rnu\in\mathbb R^nu∈Rn but depends only on the TXTXTX-component of uuu; the paper's u∈Rmu\in\mathbb R^mu∈Rm is a vector of TXTXTX, which is what the statement quantifies over.

Differences from the source:

  • reading: u∈Rmu\in\mathbb R^mu∈Rm with m=dim⁡Xm=\dim Xm=dimX → u∈TXu\in TXu∈TX (vectorSpan ℝ X); reason: the paper identifies Rm\mathbb R^mRm with TXTXTX (p. 4); for u⊥TXu\perp TXu⊥TX the bound is false when m<nm<nm<n, since then H(x,u)=0H(x,u)=0H(x,u)=0.
  • encoding: Ψ:X→R+n\Psi:X\to\mathbb R^n_+Ψ:X→R+n​ → Ψ:Rn→Rn\Psi:\mathbb R^n\to\mathbb R^nΨ:Rn→Rn, nonnegative and Lipschitz on XXX; reason: values off XXX are never used.
  • encoding: max⁡\maxmax in (3.4) and in ∥Ψ∥∞\|\Psi\|_\infty∥Ψ∥∞​ → real sSup of the image of a compact nonempty set under a continuous function; reason: the supremum is attained, so it is the maximum.
  • reading: (3.4) maximizes H(x,u,v)=⟨u,v⟩−L(x,v)\mathcal H(x,u,v)=\langle u,v\rangle-L(x,v)H(x,u,v)=⟨u,v⟩−L(x,v) over v∈TX∩Brv\in TX\cap B_rv∈TX∩Br​, which (3.3) equates with ⟨u,v⟩−⟨Ψ(x),[v]+⟩\langle u,v\rangle-\langle\Psi(x),[v]_+\rangle⟨u,v⟩−⟨Ψ(x),[v]+​⟩ → HHH is the maximum of ⟨u,v⟩−⟨Ψ(x),[v]+⟩\langle u,v\rangle-\langle\Psi(x),[v]_+\rangle⟨u,v⟩−⟨Ψ(x),[v]+​⟩ over all v∈TX∩Brv\in TX\cap B_rv∈TX∩Br​ (not only v∈TX(x)v\in TX(x)v∈TX(x)), and the bound is stated for every x∈Xx\in Xx∈X; reason: the two forms agree on X∘X^\circX∘; at a boundary point they differ, and with L=+∞L=+\inftyL=+∞ off TX(x)TX(x)TX(x) the bound fails at a vertex (X=[0,1]X=[0,1]X=[0,1], x=0x=0x=0, Ψ≡1\Psi\equiv1Ψ≡1, r=1r=1r=1, u=−10u=-10u=−10: H=0<9H=0<9H=0<9), so the formula reading is the one under which (3.5) holds on XXX as printed.
Preamble
import Mathlib
import Definitions.Def_SemilinearHJ_Verification_Setting
Formal statement
namespace SemilinearHJ.Verification

open scoped ENNReal MeasureTheory

theorem eq_3_5 {n : ℕ} (X : Set (E n)) (hX : IsPolytope X)
    (Ψ : E n → E n) (hΨL : ∃ K : NNReal, LipschitzOnWith K Ψ X) (hΨ0 : ∀ x ∈ X, ∀ i, 0 ≤ Ψ x i)
    (r : ℝ) (hr : 0 < r) :
    ∀ x ∈ X, ∀ u ∈ vectorSpan ℝ X, r / n * l1 u - r * psiNorm X Ψ ≤ ham X Ψ r x u := by sorry

end SemilinearHJ.Verification
Source
Sandholm, Tran & Arigapudi, Hamilton-Jacobi Equations with Semilinear Costs and State Constraints, with Applications to Large Deviations in Games, author's manuscript (July 4, 2018; Math. Oper. Res. doi:10.1287/moor.2020.1114), p. 8, Equation (3.5)
Human review
  • Endorsed by Shuze Chen · Oct 24, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 24, 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