Uniform ergodicity gives exponentially fast -mixing
ProvedMarkovChainCLT.uniformlyErgodic_phiMixing_expLet be a Harris ergodic Markov chain with invariant distribution . If is uniformly ergodic, i.e. for a constant not depending on the starting point and some , then the stationary chain is uniformly (-) mixing at an exponential rate: there are and with
that is, .
A Harris ergodic chain is uniformly ergodic if and only if it is uniformly mixing (Ibragimov & Linnik 1971, pp. 367-368), and as a consequence of the strong Markov property a -mixing Harris ergodic chain automatically enjoys exponentially fast -mixing (Bradley 1986, Theorem 4.2).
This is the quantitative half of the mission's Theorem 2(iv), stated for a general measurable state space. It is the form in which uniform ergodicity is consumed by Corollary 5: exponential decay of makes Billingsley's summability condition (eq. (13)) immediate, which is exactly the source's remark that "if is uniformly ergodic the coefficients decrease exponentially and (13) is obvious".
import Definitions.Def_MarkovErgodicity import Definitions.Def_MarkovChainPathMeasure import Definitions.Def_MixingCoefficients open MeasureTheory ProbabilityTheory Filter open scoped ENNReal NNReal Topology ProbabilityTheory
theorem MarkovChainCLT.uniformlyErgodic_phiMixing_exp {X : Type*} [MeasurableSpace X]
(P : Kernel X X) [IsMarkovKernel P] (π : Measure X) [IsProbabilityMeasure π]
(hP : HarrisErgodic P π) (huni : UniformlyErgodic P π) :
∃ c θ : ℝ, 0 ≤ c ∧ 0 < θ ∧ ∀ n : ℕ, 1 ≤ n →
phiMixingCoef (chainMeasure P π) (fun i ω => ω i) n ≤ c * Real.exp (-θ * n) := by sorry