Abstract:Time in-homogeneous cyclic Markov chain Monte Carlo (MCMC) samplers, including deterministic scan Gibbs samplers and Metropolis within Gibbs samplers, are extensively used for sampling from multi-dimensional distributions. We establish a multivariate strong invariance principle (SIP) for Markov chains associated with these samplers. The rate of this SIP essentially aligns with the tightest rate available for time homogeneous Markov chains. The SIP implies the strong law of large numbers (SLLN) and the central limit theorem (CLT), and plays an essential role in uncertainty assessments. Using the SIP, we give conditions under which the multivariate batch means estimator for estimating the covariance matrix in the multivariate CLT is strongly consistent. Additionally, we provide conditions for a multivariate fixed volume sequential termination rule, which is associated with the concept of effective sample size (ESS), to be asymptotically valid. Our uncertainty assessment tools are demonstrated through various numerical experiments.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to establish the multivariate strong invariance principle (SIP) for time - inhomogeneous cyclic Markov chains (such as deterministic scan Gibbs samplers and Metropolis within Gibbs samplers). Specifically, the author aims to: 1. **Establish the multivariate strong invariance principle**: For time - inhomogeneous cyclic Markov chains, the author establishes the multivariate strong invariance principle (SIP), and its convergence rate is comparable to the best rate of time - homogeneous Markov chains. SIP is not only stronger than the strong law of large numbers (SLLN) and the central limit theorem (CLT), but also plays a key role in uncertainty assessment. 2. **Prove the consistency of the batch - mean estimator**: Using SIP, the author proves the strong consistency of the batch - mean estimator for estimating the asymptotic covariance matrix in the multivariate CLT. 3. **Provide an effective termination rule**: The author also provides a multivariate fixed - volume sequential termination rule based on the concept of effective sample size (ESS) and proves that this rule is asymptotically effective under certain conditions. ### Main contributions - **Theoretical contributions**: Through a new regenerative construction method, the paper successfully extends the existing SIP results of time - homogeneous Markov chains to time - inhomogeneous cyclic Markov chains. - **Application contributions**: The author shows through numerical experiments that time - inhomogeneous samplers are more efficient than their natural time - homogeneous counterparts in some cases. ### Mathematical formulas 1. **Definition of SIP**: \[ \left\| \sum_{t = 1}^n (f(X_t(\omega))-\theta)-\Gamma B(n)(\omega) \right\| \leq M(\omega)\phi(n) \] where \(B(n)=\sum_{t = 1}^n C(t)\), \(C(t)\) is an independent and identically distributed standard normal random vector, \(\Gamma\) is a \(d\times d\) constant matrix satisfying \(\Gamma\Gamma^{\top}=\Sigma\), \(\phi(n)\) is a non - negative increasing function, and \(\frac{\phi(n)}{\sqrt{n}}\to0\) as \(n\to\infty\). 2. **Autocorrelation matrix**: \[ \psi(j, l)=\begin{cases} E_{\pi}\{(f(X_j)-\theta)(f(X_{j + l})-\theta)^{\top}\}, & l\geq0\\ E_{\pi}\{(f(X_{j - l})-\theta)(f(X_j)-\theta)^{\top}\}, & l < 0 \end{cases} \] 3. **Asymptotic covariance matrix**: \[ \Sigma=\sum_{l = -\infty}^{+\infty}\sum_{j = 0}^{k - 1}\frac{1}{k}\psi(j, l) \] ### Conclusion By establishing the multivariate strong invariance principle of time - inhomogeneous cyclic Markov chains, the paper provides a powerful tool for uncertainty assessment and proves the consistency of the batch - mean estimator and the effectiveness of the fixed - volume sequential termination rule. These results are of great significance in practical applications, especially in Bayesian statistics and sampling of complex distributions.

Multivariate strong invariance principle and uncertainty assessment for time in-homogeneous cyclic MCMC samplers

Importance Sampling of Interval Markov Chains

Multicanonical sequential Monte Carlo sampler for uncertainty quantification

Dimension-free Relaxation Times of Informed MCMC Samplers on Discrete Spaces

A Vine-Copula Based Adaptive Mcmc Sampler for Efficient Inference of Dynamical Systems

THE BEST RATES IN STRONG INVARIANCE-PRINCIPLE

Convergence Bounds for Sequential Monte Carlo on Multimodal Distributions using Soft Decomposition

ON THE ERROR RATE OF IMPORTANCE SAMPLING WITH RANDOMIZED QUASI-MONTE CARLO*

Achieving High Convergence Rates by Quasi-Monte Carlo and Importance Sampling for Unbounded Integrands

On importance sampling and independent Metropolis-Hastings with an unbounded weight function

Overlapping Batch Confidence Intervals on Statistical Functionals Constructed from Time Series: Application to Quantiles, Optimization, and Estimation

On the convergence conditions of Laplace importance sampling with randomized quasi-Monte Carlo

An investigation into Markov chain Monte Carlo algorithms for Subset simulation: Emphasizing uncertainty analysis

A large deviation principle for the empirical measures of Metropolis-Hastings chains

Convergence of sequential Markov Chain Monte Carlo methods: I. Nonlinear flow of probability measures

Asymptotic bias of inexact Markov Chain Monte Carlo methods in high dimension

A large deviation principle for the empirical measures of Metropolis–Hastings chains

Bayesian Annealed Sequential Importance Sampling: An Unbiased Version of Transitional Markov Chain Monte Carlo

Bounding Wasserstein distance with couplings

Piecewise deterministic sampling with splitting schemes