Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 4.4, p. 14 — α is bounded and β square integrable, so u* ∈ L² and X* ∈ L²(Ω; C(0, T; ℝ))

Disproved
TimeInconsLQ.Deterministic.feedback_bounded

by mikedeng1 · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

feedback-controllinear-quadraticp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1stochastic-control

Consider the scalar-state problem with deterministic coefficients of §4 under the standing assumptions, with G≥h>0G\ge h>0G≥h>0 and one of the cases (i), (ii), (iii) of Theorem 4.4. Let (M,N)(M,N)(M,N) be a positive solution pair of (4.9) on [0,T][0,T][0,T] and Φ\PhiΦ a solution of (4.8), and let α,β\alpha,\betaα,β be the feedback coefficients (4.4).

Claim. The following hold.

  1. α\alphaα is essentially bounded on [0,T][0,T][0,T], and β\betaβ is square integrable on [0,T][0,T][0,T].
  2. On every probability space carrying a standard ddd-dimensional Brownian motion and for every initial state x0x_0x0​, the closed-loop equation
dXs∗=[AsXs∗+Bs′(αsXs∗+βs)+bs]ds+[CsXs∗+Ds(αsXs∗+βs)+σs]′dWs,X0∗=x0,dX^*_s=[A_sX^*_s+B_s'(\alpha_sX^*_s+\beta_s)+b_s]ds+[C_sX^*_s+D_s(\alpha_sX^*_s+\beta_s)+\sigma_s]'dW_s,\qquad X^*_0=x_0,dXs∗​=[As​Xs∗​+Bs′​(αs​Xs∗​+βs​)+bs​]ds+[Cs​Xs∗​+Ds​(αs​Xs∗​+βs​)+σs​]′dWs​,X0∗​=x0​,

has a solution X∗X^*X∗ with continuous paths on [0,T][0,T][0,T] and Esup⁡s≤T∣Xs∗∣2<∞\mathbb E\sup_{s\le T}|X^*_s|^2<\inftyEsups≤T​∣Xs∗​∣2<∞ (that is, X∗∈LF2(Ω;C(0,T;R))X^*\in L^2_{\mathcal F}(\Omega;C(0,T;\mathbb R))X∗∈LF2​(Ω;C(0,T;R))), whose feedback control u∗=αX∗+βu^*=\alpha X^*+\betau∗=αX∗+β lies in LF2(0,T;Rl)L^2_{\mathcal F}(0,T;\mathbb R^l)LF2​(0,T;Rl).

This is the integrability step of the proof of Theorem 4.4: the linear feedback is admissible and its state is well defined.

Formalization Note. The paper states that α\alphaα and β\betaβ are "uniformly bounded". Under the paper's assumptions (σ∈L2\sigma\in L^2σ∈L2 only, B,C,DB,C,DB,C,D essentially bounded) α\alphaα is essentially bounded but β\betaβ is in general only square integrable, since it contains MsDs′σsM_sD_s'\sigma_sMs​Ds′​σs​; we state the true version, which suffices for the conclusion. "X∗∈L2(Ω;C)X^*\in L^2(\Omega;C)X∗∈L2(Ω;C)" is stated as existence of a solution with these properties, because the substrate's notion of SDE solution already contains sup⁡sE∣Xs∣2<∞\sup_s\mathbb E|X_s|^2<\inftysups​E∣Xs​∣2<∞ and u∗∈L2u^*\in L^2u∗∈L2 is not implied by it.

Preamble
import Mathlib
import Definitions.Def_Peng1990_SMP_Stochastic
import Definitions.Def_TimeInconsLQ_Deterministic_Model
import Definitions.Def_TimeInconsLQ_Deterministic_Riccati
import Definitions.Def_TimeInconsLQ_Deterministic_Ansatz

open MeasureTheory ProbabilityTheory Filter Topology Matrix
open scoped ENNReal NNReal
Formal statement
namespace TimeInconsLQ.Deterministic

/-- Proof of Theorem 4.4 (p. 14): in each of the three cases, for a positive solution pair
`(M, N)` of (4.9) and a solution `Φ` of (4.8), the gain `α` of (4.4) is essentially bounded on
`[0, T]` and `β` is square integrable on `[0, T]` (bounded when `σ` is); hence, on every
probability space with a Brownian motion, the closed-loop equation has a solution `X*` with
continuous paths and `E sup_{s ≤ T}|X*_s|² < ∞`, whose feedback control is in `L²_𝓕(0, T; ℝˡ)`. -/
theorem feedback_bounded {Ω : Type*} [MeasurableSpace Ω] {l d : ℕ} (c : Coeffs l d)
    (T : ℝ≥0) (hT : 0 < T) (hc : CoeffsStanding c T) (hGh : c.h ≤ c.G) (hh : 0 < c.h)
    (hcase : Case_i c T ∨ Case_ii c T ∨ Case_iii c T)
    (M N Φ : ℝ≥0 → ℝ) (hMN : IsPosSol49 c T M N) (hΦ : IsSol48 c T M N Φ) :
    (∃ K : ℝ, ∀ᵐ s ∂(volume.restrict (Set.Icc (0 : ℝ) T)),
      ‖alpha c T M N s.toNNReal‖ ≤ K) ∧
    ∫⁻ s in Set.Icc (0 : ℝ) T, ‖beta c M Φ s.toNNReal‖ₑ ^ 2 < ⊤ ∧
    ∀ (P : Measure Ω), IsProbabilityMeasure P →
    ∀ (W : ℝ≥0 → Ω → Fin d → ℝ) (hW : Peng1990.SMP.IsStdBrownian P W) (x₀ : Fin 1 → ℝ),
      ∃ X, IsClosedLoop (toData c P W hW T hT x₀)
          (fun s _ => alpha c T M N s) (fun s _ => beta c M Φ s) X ∧
        TimeInconsLQ.Sufficient.Admissible (toData c P W hW T hT x₀) (fbControl c T M N Φ X) ∧
        (∀ᵐ ω ∂P, ContinuousOn (fun s => X s ω 0) (Set.Iic T)) ∧
        ∫⁻ ω, (⨆ s ∈ Set.Iic T, ‖X s ω 0‖ₑ ^ 2) ∂P < ⊤ := by sorry

end TimeInconsLQ.Deterministic
Source
Hu, Jin, Zhou, Time-Inconsistent Stochastic Linear–Quadratic Control, arXiv:1111.0818v1, p. 14, proof of Theorem 4.4
Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 7, 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