Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition 5.15 — the minimax model satisfies I, I.1, I.2 or D, D.2, with scalar α

Proved
BertsekasShreve.Monotone.prop5_15_minimax_assumptions

by mikedeng1 · Oct 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

dynamic-programmingminimax-controlp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1

Let WWW be a set, W(x,u)⊆WW(x,u)\subseteq WW(x,u)⊆W, g:S×C×W→[−∞,∞]g : S\times C\times W\to[-\infty,\infty]g:S×C×W→[−∞,∞], f:S×C×W→Sf : S\times C\times W\to Sf:S×C×W→S and α>0\alpha>0α>0. Consider an abstract monotone dynamic programming model on SSS, CCC with constraint sets U(x)U(x)U(x) such that W(x,u)W(x,u)W(x,u) is nonempty for every x∈Sx\in Sx∈S, u∈U(x)u\in U(x)u∈U(x), whose mapping is the minimax mapping

H(x,u,J)=sup⁡w∈W(x,u){g(x,u,w)+αJ[f(x,u,w)]}H(x,u,J)=\sup_{w\in W(x,u)}\bigl\{g(x,u,w)+\alpha J[f(x,u,w)]\bigr\}H(x,u,J)=w∈W(x,u)sup​{g(x,u,w)+αJ[f(x,u,w)]}

(sums with ∞−∞=∞\infty-\infty=\infty∞−∞=∞) and whose terminal function is J0(x)=0J_0(x)=0J0​(x)=0 for all x∈Sx\in Sx∈S.

  1. If
0≤g(x,u,w)∀x∈S, u∈U(x), w∈W,0\le g(x,u,w)\qquad\forall x\in S,\ u\in U(x),\ w\in W,0≤g(x,u,w)∀x∈S, u∈U(x), w∈W,

then Assumptions I, I.1 and I.2 are satisfied, with the scalar in I.2 equal to α\alphaα. 2. If

g(x,u,w)≤0∀x∈S, u∈U(x), w∈W,g(x,u,w)\le 0\qquad\forall x\in S,\ u\in U(x),\ w\in W,g(x,u,w)≤0∀x∈S, u∈U(x), w∈W,

then Assumptions D and D.2 are satisfied, with the scalar in D.2 equal to α\alphaα.

Assumption D.1 is absent from part 2: without further hypotheses the supremum does not commute with decreasing limits, which is why only part of the chapter's theory applies to minimax control under D.

Formalization Note Part 1 is stated for every MonotoneDP.Increase.Model and part 2 for every MonotoneDP.Decrease.Model whose fields are H = BertsekasShreve.FiniteHorizon.minimaxH Wset g f α (the shared definition of Section 2.3.5's mapping, sum taken with badd, i.e. ∞−∞=∞\infty-\infty=\infty∞−∞=∞) and Jbar ≡ 0; such models exist because the mapping is monotone in JJJ. The section's standing assumptions α>0\alpha>0α>0 and W(x,u)≠∅W(x,u)\ne\varnothingW(x,u)=∅ are explicit hypotheses.

Preamble
import Mathlib
import Definitions.Def_MonotoneDP_Increase_Model
import Definitions.Def_MonotoneDP_Increase_Assumptions
import Definitions.Def_MonotoneDP_Decrease_Model
import Definitions.Def_MonotoneDP_Decrease_Assumptions
import Definitions.Def_BertsekasShreve_FiniteHorizon_SpecificModels
Formal statement
namespace BertsekasShreve.Monotone

/-- Bertsekas & Shreve (1996), p. 90, Proposition 5.15: for the minimax mapping
`H(x, u, J) = sup_{w ∈ W(x, u)} {g(x, u, w) + α J[f(x, u, w)]}` of Section 2.3.5 (with `α > 0` and
`W(x, u)` nonempty for `x ∈ S`, `u ∈ U(x)`, the standing assumptions of that section) and
`J₀ ≡ 0`:
(a) if `0 ≤ g(x, u, w)` for all `x ∈ S`, `u ∈ U(x)`, `w ∈ W`, then I, I.1 and I.2 hold, the scalar
in I.2 being `α`;
(b) if `g(x, u, w) ≤ 0` for all `x ∈ S`, `u ∈ U(x)`, `w ∈ W`, then D and D.2 hold, the scalar in
D.2 being `α`. -/
theorem prop5_15_minimax_assumptions {S C W : Type*}
    (Wset : S → C → Set W) (g : S → C → W → EReal) (f : S → C → W → S) (α : ℝ) (hα : 0 < α) :
    (∀ m : MonotoneDP.Increase.Model S C, m.H = BertsekasShreve.FiniteHorizon.minimaxH Wset g f α → m.Jbar = (fun _ => 0) →
      (∀ x : S, ∀ u ∈ m.U x, (Wset x u).Nonempty) →
      (∀ x : S, ∀ u ∈ m.U x, ∀ w : W, 0 ≤ g x u w) →
        m.AssumptionI ∧ m.AssumptionI1 ∧ m.AssumptionI2 α) ∧
    (∀ m : MonotoneDP.Decrease.Model S C, m.H = BertsekasShreve.FiniteHorizon.minimaxH Wset g f α → m.Jbar = (fun _ => 0) →
      (∀ x : S, ∀ u ∈ m.U x, (Wset x u).Nonempty) →
      (∀ x : S, ∀ u ∈ m.U x, ∀ w : W, g x u w ≤ 0) →
        m.AssumptionD ∧ m.AssumptionD2 α) := by sorry

end BertsekasShreve.Monotone
Source
Bertsekas & Shreve, Stochastic Optimal Control: The Discrete-Time Case, Athena Scientific 1996, p. 90, Proposition 5.15; mapping from p. 38, Section 2.3.5, Eq. (29) of Chapter 2
Human review
  • Endorsed by Shuze Chen · Oct 5, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 5, 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