Proof of Theorem 6, p. 909 — if (17) fails, some nondecreasing sequence converges in probability but not almost surely
ProvedStochIneqPO.Monotone.exists_counterexample_of_not_regular17Let be a complete separable metric space with a closed partial ordering and its Borel -algebra, and suppose that does not satisfy condition (17). Then there exist a probability space and measurable maps with for every , such that
- converges in probability to a constant: there is with for every ;
- does not converge almost surely: there is no with for -almost every .
This is the "only if" half of the second part of Theorem 6: without (17), almost sure convergence is not equivalent to the other three modes for all nondecreasing sequences.
Formalization Note The paper says "take any probability space" carrying events with and ; such events do not exist on every probability space (e.g. a one-point space), so the statement asserts the existence of a suitable probability space, which is what the argument establishes. The probability space is taken in Lean's lowest universe Type.
import Mathlib import Definitions.Def_StochIneqPO_Monotone_Regular17 import Definitions.Def_StochIneqPO_Monotone_ConvergenceModes
namespace StochIneqPO.Monotone
open MeasureTheory Filter
theorem exists_counterexample_of_not_regular17 {E : Type*} [MetricSpace E] [CompleteSpace E]
[SecondCountableTopology E] [MeasurableSpace E] [BorelSpace E]
[PartialOrder E] [OrderClosedTopology E]
(h17 : ¬ Regular17 E) :
∃ (Ω : Type) (_ : MeasurableSpace Ω) (μ : Measure Ω), IsProbabilityMeasure μ ∧
∃ X : ℕ → Ω → E, (∀ i, Measurable (X i)) ∧ (∀ ω, Monotone fun i => X i ω) ∧
(∃ z : E, TendstoInMeasure μ X atTop (fun _ => z)) ∧ ¬ ConvergesAS μ X := by sorry
end StochIneqPO.Monotone
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.