Proof of Theorem 3.1, p. 11 — |Λ′(t)| ≤ κ for |X| ≤ κ
ProvedSAARate.SharpLD.cgf_deriv_abs_lelarge-deviationsmoment-generating-functionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1probability
Let be a real random variable on a probability space with almost surely. Let be its logarithmic moment generating function (cumulant generating function). Then is differentiable on and, for every ,
In the paper for a fixed direction , and is Assumption (B). The bound is the first step towards the strong convexity of the rate function .
Formalization Note is Mathlib's real-valued cgf, which is the true here because every exponential moment of a bounded variable is finite. The statement is made for an arbitrary bounded measurable rather than for , which makes it more general. The differentiability conjunct makes explicit that deriv is the true derivative.
Preamble
import Mathlib open MeasureTheory ProbabilityTheory Filter Topology
Formal statement
namespace SAARate.SharpLD
theorem cgf_deriv_abs_le {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
[IsProbabilityMeasure P] (X : Ω → ℝ) (hX : Measurable X) (κ : ℝ)
(hXκ : ∀ᵐ ω ∂P, |X ω| ≤ κ) :
Differentiable ℝ (cgf X P) ∧
∀ t : ℝ, deriv (cgf X P) t = (∫ ω, X ω * Real.exp (t * X ω) ∂P) / mgf X P t ∧
|deriv (cgf X P) t| ≤ κ := by sorry
end SAARate.SharpLD
Source
Shapiro & Homem-de-Mello, On Rate of Convergence of Optimal Solutions of Monte Carlo Approximations of Stochastic Programs, preprint (SPEPS copy, edoc.hu-berlin.de), p. 11, proof of Theorem 3.1, the display |Λ′(t)| ≤ E[|X|e^{tX}]/E[e^{tX}] ≤ κ
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.