§2, p. 3 — the state equation (2.1) has a unique solution (uniqueness part)
OpenTimeInconsLQ.Sufficient.state_uniqueUniqueness of the state process of the time-inconsistent stochastic LQ problem.
Work under the standing assumptions of §2: is a standard -dimensional Brownian motion on with its natural filtration , is a bounded deterministic matrix function on , the coefficients are essentially bounded progressively measurable processes, and . Fix an admissible control . If and are both solutions of the state equation
then for every ,
This is the uniqueness half of the paper's remark (p. 3) that for every control the state equation (2.1) has a unique solution. It allows statements proved along one state process of (such as the expansion of Proposition 3.1) to be transferred to any other state process of the same control.
Formalization Note A solution of (2.1) is the model's IsState: a progressively measurable process with satisfying the integral form of (2.1) almost surely at every time , with the stochastic integrals being Itô integrals in the sense of Peng1990.SMP.IsItoIntegral.
import Mathlib import Definitions.Def_Peng1990_SMP_Stochastic import Definitions.Def_TimeInconsLQ_Sufficient_Model
namespace TimeInconsLQ.Sufficient
open MeasureTheory ProbabilityTheory Filter Topology
open scoped ENNReal NNReal Matrix
/-- §2, p. 3 (uniqueness part of "for any control `u ∈ L²_𝓕(0, T; ℝˡ)` there exists a unique
solution `X`" of the state equation (2.1)): under the standing assumptions, any two state
processes `X`, `X'` of the same admissible control `u` agree almost surely at every time
`s ∈ [0, T]`. -/
theorem state_unique {Ω : Type*} [MeasurableSpace Ω] {n l d : ℕ} (M : Data Ω n l d)
(hM : Standing M) (u : ℝ≥0 → Ω → Fin l → ℝ) (hu : Admissible M u)
(X X' : ℝ≥0 → Ω → Fin n → ℝ) (hX : IsState M u X) (hX' : IsState M u X') :
∀ s : ℝ≥0, s ≤ M.T → X s =ᵐ[M.P] X' s := by sorry
end TimeInconsLQ.Sufficient