Theorem 4.2 — stability of submartingale property Bellman's principle
ProvedMultiperiodRisk.Bellman.stable_tfaeLet be a probability space with a filtration whose initial σ-algebra is -trivial, fix a horizon , and let be a closed convex set of test probabilities on , absolutely continuous with respect to , with . For a value process and a stopping time let
and let be the generalized Snell envelope of Theorem 4.1. The following are equivalent:
- is stable (Definition 3.1);
- for every and every value process , is a -submartingale;
- for every value process , a.s. for ;
- (Bellman's principle) for every value process and all stopping times ,
The theorem identifies stability under pasting as exactly the property of the set of test probabilities that makes the risk-adjusted value time consistent: the value can be computed by backward induction, and replacing the future of a position by its risk-adjusted value does not change today's value.
Formalization Note Two conventions the page leaves implicit are made explicit. (a) is -trivial: the proof evaluates as the number , and without it the implications (1)–(3) stability fail (for and a two-atom , statements 2–3 hold for every while stability can fail). (b) , implied by the paper's convention . . All equalities of risk-adjusted values hold -a.s. because essential infima are defined up to null sets.
Retired 2026-10-27: a Lean-vs-paper fidelity audit found this statement weaker than, or narrower than, the cited result. A corrected statement will be resubmitted.
import Mathlib import Definitions.Def_MultiperiodRisk_Bellman_StopOps
namespace MultiperiodRisk.Bellman
open MeasureTheory
/-- Theorem 4.2: for a closed convex set of test probabilities with `𝒫ᵉ ≠ ∅` and an initial
σ-algebra `ℱ_0` that is trivial under `P₀`, stability is equivalent to each of
(i) `Ψ(X)` is a `ℚ`-submartingale for every `ℚ ∈ 𝒫` and every value process `X`;
(ii) `Ψ = Ψ̄` on value processes;
(iii) Bellman's principle `Ψ_σ(X) = Ψ_σ(X^{τ-} + Ψ_τ(^τX) 1_{[τ,N]})` for stopping times `σ ≤ τ`. -/
theorem stable_tfae {Ω : Type*} {m : MeasurableSpace Ω} (P₀ : Measure Ω)
[IsProbabilityMeasure P₀] (ℱ : Filtration ℕ m) (N : ℕ)
(hℱ₀ : ∀ s, MeasurableSet[ℱ 0] s → P₀ s = 0 ∨ P₀ s = 1)
(D : TestSet P₀ ℱ N) (hPe : (Pe D).Nonempty) :
List.TFAE
[ IsStable D,
∀ f ∈ D.set, ∀ X : ℕ → Ω → ℝ, IsValueProcess P₀ ℱ N X →
Submartingale (fun n => PsiN D X (min n N)) ℱ (Q P₀ f),
∀ X : ℕ → Ω → ℝ, IsValueProcess P₀ ℱ N X →
∀ n ≤ N, PsiN D X n =ᵐ[P₀] PsiBar D X n,
∀ X : ℕ → Ω → ℝ, IsValueProcess P₀ ℱ N X →
∀ (σ τ : Ω → WithTop ℕ) (hσ : IsBddStoppingTime ℱ N σ)
(hτ : IsBddStoppingTime ℱ N τ), (∀ ω, σ ω ≤ τ ω) →
Psi D X σ hσ.1 =ᵐ[P₀] Psi D (bellmanProcess D X τ hτ.1) σ hσ.1 ] := by sorry
end MultiperiodRisk.Bellman
Read-back
What the Lean code literally says, in plain math · claude-opus-5-5
Let be a measurable space, a probability measure on it, and a filtration of sub-σ-algebras of indexed by the natural numbers. Let be a horizon. The statement assumes:
- (Triviality of under .) Every -measurable set has or .
- (Test set.) is an object of the bundle-defined type "test set" built from , and . Its definition is not shown here. The statement uses only its underlying collection , whose elements are written .
- (Non-emptiness.) The set is nonempty. This is a bundle-defined set attached to , and its definition is not shown.
The docstring describes the test set as "closed convex". No closedness or convexity hypothesis appears in the statement itself. Whether these properties hold depends entirely on the unseen definition of a test set.
The conclusion is that the following four propositions are pairwise logically equivalent. Each one implies every other one.
-
Stability. satisfies the bundle-defined predicate "stable" (definition not shown).
-
Submartingale property. Take any and any , , that satisfies the bundle-defined predicate "value process" relative to , , . For every such pair, the real-valued process
is a submartingale with respect to under the measure . Here:
- is the bundle-defined map , which gives a random variable for each time index.
- is a bundle-defined measure on .
- The process is frozen at its time- value for all .
- "Submartingale" means three things. (a) For each , the time- value is strongly -measurable. (b) For each , it is -integrable. (c) For all , the time- value is at most the -conditional expectation of the time- value given , -almost everywhere.
- Agreement of the two operators. For every value process and every ,
where is the bundle-defined map .
-
Bellman principle. Take any value process and any two random times that satisfy:
- both satisfy the bundle-defined predicate "bounded stopping time" relative to and ;
- for every , not merely almost every .
For all such , , ,
Here:
- is the bundle-defined map . It also takes as input the first component of the proof that is a bounded stopping time.
- is the bundle-defined process . It likewise takes the first component of the bounded-stopping-time proof for .
- The docstring describes as . That description is not visible in the statement.
Measures used in each condition.
- The almost-everywhere equalities in (3) and (4) are with respect to .
- The submartingale inequalities in (2) are with respect to each .
- The triviality hypothesis on is imposed only under . Nothing is assumed about how behaves on .
Degenerate cases.
- cannot be empty, because is a probability measure.
- If has one element, or is the trivial σ-algebra, the triviality hypothesis holds automatically.
- No value processes. If no satisfies the value-process predicate, then (2), (3) and (4) are all vacuously true. The theorem then asserts outright that is stable.
- Empty . If is empty, (2) is vacuously true, and the theorem asserts that (1), (3) and (4) all hold. Whether this case is compatible with depends on the unseen definitions.
- Empty . If can never be nonempty for any test set, the theorem is vacuous.
- Horizon .
- In (2), the process is the constant-in-time process . The condition reduces to three requirements on : it is -measurable, it is -integrable, and -a.e. it is at most its own -conditional expectation given each .
- In (3), only is tested.
- In (4), only stopping times allowed by the bounded-stopping-time predicate with bound are tested.
- Times after . The process is defined for all times . Whether values after matter depends on the unseen definitions.
- Infinite random times. The random times take values in . Whether the value is excluded depends on the unseen "bounded stopping time" predicate.
- Default conditional expectation. The -conditional expectation given is the formal default whenever restricted to is not σ-finite. In that case, condition (c) of (2) requires the time- value to be -a.e.
- Nature of . No hypothesis states that is a probability measure, or even finite. That depends on the unseen definitions of and of the test set.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.