Lemma 1.4 (1.11), p. 749 — log Ee^{hX} ≤ hμ + e^tβh²/2 + α_t(e^{hb} − 1 − hb)/b^t for h > t/b
ProvedNagaevLD.FukNagaev.lemma_1_4_bcumulant-generating-functionlarge-deviationsmoment-generating-functionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be a random variable with for some , finite second moment, , , and , where and . Then for ,
This is the second half of Lemma 1.4; the extra term controls the contribution of large positive values of , which is where the polynomial part of the Fuk–Nagaev bound comes from.
Formalization Note is , not . is assumed, as for (1.10). is a real power.
Preamble
import Mathlib import Definitions.Def_NagaevLD_FukNagaev_Setting
Formal statement
namespace NagaevLD.FukNagaev
open MeasureTheory ProbabilityTheory
theorem lemma_1_4_b {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω) [IsProbabilityMeasure P]
(X : Ω → ℝ) (hX : Measurable X) (b : ℝ) (hb : 0 < b) (hXb : P {ω | b < X ω} = 0)
(t : ℝ) (ht : 2 ≤ t)
(hL2 : MemLp X 2 P) (h : ℝ) (hht : t / b < h) :
Real.log (∫ ω, Real.exp (h * X ω) ∂P) ≤
h * (∫ ω, X ω ∂P) + Real.exp t * (∫ ω, X ω ^ 2 ∂P) * h ^ 2 / 2 +
(∫ ω in {ω | 0 ≤ X ω}, X ω ^ t ∂P) * (Real.exp (h * b) - 1 - h * b) / b ^ t := by sorry
end NagaevLD.FukNagaev
Source
Nagaev, Large deviations of sums of independent random variables, Ann. Probab. 7 (1979), p. 749, Lemma 1.4, (1.11)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.