Geometric ergodicity + detailed balance exponential -mixing (Jones Thm 2(iii))
ProvedMarkovChainCLT.rho_mixing_exp_of_geometric_reversibleLet be a Markov chain with transition kernel , Harris ergodic with invariant probability . Suppose the chain is geometrically ergodic and reversible with respect to (detailed balance, the source's eq. (8)). Then the stationary chain is asymptotically uncorrelated with exponentially fast -mixing: there exist and such that
Reversibility holds by construction for Metropolis–Hastings samplers, so this result (Roberts–Rosenthal 1997) is the gateway to second-moment CLTs for the most common MCMC algorithms.
Formalization Note "Harris ergodic" is encoded by its total-variation characterization: is invariant for and for every starting point (equivalent to the classical aperiodic, -irreducible, positive Harris recurrent definition; the "every " quantifier is exactly the Harris property). Convergence in distribution is weak convergence of laws, and is read as the point mass at , which absorbs the source's "" caveat.
import Definitions.Def_MarkovErgodicity import Definitions.Def_MarkovChainPathMeasure import Definitions.Def_MixingCoefficients open MeasureTheory ProbabilityTheory Filter open scoped ENNReal NNReal Topology ProbabilityTheory /-- **Theorem 2, part 3** (Roberts–Rosenthal 1997): a geometrically ergodic chain satisfying detailed balance is asymptotically uncorrelated, with exponentially fast ρ-mixing. -/
theorem MarkovChainCLT.rho_mixing_exp_of_geometric_reversible {X : Type*} [MeasurableSpace X]
(P : Kernel X X) [IsMarkovKernel P] (π : Measure X) [IsProbabilityMeasure π]
(hP : HarrisErgodic P π) (hgeo : GeometricallyErgodic P π)
(hrev : Kernel.IsReversible P π) :
∃ c θ : ℝ, 0 ≤ c ∧ 0 < θ ∧ ∀ n : ℕ, 1 ≤ n →
rhoMixingCoef (chainMeasure P π) (fun i ω => ω i) n ≤ c * Real.exp (-θ * n) := by sorry