Proof of Theorem 4.21, p. 32 — ν^π_s ≤ ν, µ|]−∞,s] ⪯C ν^π_s, hence ν^{π_lc}_s = S^ν(µ|]−∞,s]) ⪯C ν^π_s
ProvedMartOT.Product.curtain_target_leastconvex-orderleft-curtain-couplingmartingale-optimal-transportp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be finite Borel measures on with finite first moments, let be the left-curtain coupling of and , and let be any martingale transport plan. For put
the image of the mass under . Then for every :
- ;
- ;
- .
Since , item 3 says that the left-curtain coupling sends each left part of to the convex-order-least possible target among all martingale plans. This comparison is what makes optimal for every cost of product form with decreasing.
Formalization Note enters through its defining property IsLeftCurtain μ ν πlc; nothing is assumed about its existence.
Preamble
import Mathlib import Definitions.Def_MartOT_Var_Setting
Formal statement
namespace MartOT.Product
open MeasureTheory
/-- Proof of **Theorem 4.21** (p. 32): for every martingale plan `π ∈ Π_M(μ, ν)` and every `s`,
`ν^π_s ≤ ν` and `μ|]−∞,s] ⪯C ν^π_s`, hence `ν^{π_lc}_s = S^ν(μ|]−∞,s]) ⪯C ν^π_s`. -/
theorem curtain_target_least (μ ν : Measure ℝ) (hμν : MartOT.Var.ConvexLE μ ν)
(πlc π : Measure (ℝ × ℝ)) (hlc : MartOT.Var.IsLeftCurtain μ ν πlc) (hπ : MartOT.Var.IsMartingalePlan μ ν π)
(s : ℝ) :
MartOT.Var.targetUpTo π s ≤ ν ∧ MartOT.Var.ConvexLE (μ.restrict (Set.Iic s)) (MartOT.Var.targetUpTo π s) ∧
MartOT.Var.ConvexLE (MartOT.Var.targetUpTo πlc s) (MartOT.Var.targetUpTo π s) := by sorry
end MartOT.Product
Source
arXiv:1208.1509v2, §4.4, proof of Theorem 4.21, p. 32
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.