Section 2, Eq. (15) — under ,
ProvedRobbinsSeqDesign.ForcedSampling.expected_average_tendsto_maxLet and be two populations whose expectations and exist (are finite), let and be disjoint sequences of density , and sample according to Robbins' rule . Then
where is the expectation under the sampling process generated by and .
This is the expectation form of display (14); Robbins notes that (14) "can be shown to imply" it. It turns the almost-sure statement into the statement about the expected average reward that the loss (9) is built from.
Formalization Note. Finite expectations are the hypothesis that both reward laws are integrable, and nothing more. The expectation is the integral against the trajectory law banditTrajMeasure ν (rule σ); under the integrability hypothesis is integrable for every , so the integral is not a junk value.
import Mathlib import Definitions.Def_RobbinsSeqDesign_ForcedSampling_Rule import Definitions.Def_RobbinsSeqDesign_ForcedSampling_Loss
namespace RobbinsSeqDesign.ForcedSampling
open MeasureTheory ProbabilityTheory Filter Topology BanditAlgorithm
/-- Robbins (1952), Section 2, Eq. (15), p. 532: for any two populations A, B whose
expectations `α`, `β` exist (are finite) and any pair of sequences (13), under the rule `R̄`,
`E(S_n / n) → max(α, β)`. -/
theorem expected_average_tendsto_max (ν : StochasticBandit 2)
(hint : ∀ i, Integrable id (ν.P i)) (σ : ForcedSchedule) :
Tendsto (fun n : ℕ => ∫ ω, partialSum n ω / (n : ℝ) ∂(banditTrajMeasure ν (rule σ)))
atTop (𝓝 (max (banditArmMean ν 0) (banditArmMean ν 1))) := by sorry
end RobbinsSeqDesign.ForcedSampling
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.