Abstract:This paper considers the speed of convergence (mixing) of a finite Markov kernel $P$ with respect to the Kullback-Leibler divergence (entropy). Given a Markov kernel one defines either a discrete-time Markov chain (with the $n$-step transition kernel given by the matrix power $P^n$) or a continuous-time Markov process (with the time-$t$ transition kernel given by $e^{t(P-\mathrm{Id})}$). The contraction of entropy for $n=1$ or $t=0+$ are characterized by the famous functional inequalities, the strong data processing inequality (SDPI) and the modified log-Sobolev inequality (MLSI), respectively. When $P=KK^*$ is written as the product of a kernel and its adjoint, one could also consider the ``half-step'' contraction, which is the SDPI for $K$, while the ``full-step'' contraction refers to the SDPI for $P$. The work [DMLM03] claimed that these contraction coefficients (half-step, full-step, and continuous-time) are generally within a constant factor of each other. We disprove this and related conjectures by working out a number of different counterexamples. In particular, we construct (a) a continuous-time Markov process that contracts arbitrarily faster than its discrete-time counterpart; and (b) a kernel $P$ such that $P^{m+1}$ contracts arbitrarily better than $P^m$. Hence, our main conclusion is that the four standard inequalities comparing five common notions of entropy and variance contraction are generally not improvable. In the process of analyzing the counterexamples, we survey and sharpen the tools for bounding the contraction coefficients and characterize properties of extremizers of the respective functional inequalities. As our examples range from Bernoulli-Laplace model, random walks on graphs, to birth-death chains, the paper is also intended as a tutorial on computing MLSI, SDPI and other constants for these types of commonly occurring Markov chains.

Contractivity of a Markov operator on the space of normalised positive distributions

Stability Analysis Of Neutral Stochastic Delay Differential Equations By A Generalisation Of Banach'S Contraction Principle

Quantitative contraction rates for Markov chains on general state spaces

Entropy Contractions in Markov Chains: Half-Step, Full-Step and Continuous-Time

Probabilistic Contraction Analysis of Iterated Random Operators

Contraction of Markovian Operators in Orlicz Spaces and Error Bounds for Markov Chain Monte Carlo

Mostly contracting random maps

Contraction analysis of nonlinear random dynamical systems

Spectral theoretic characterisation of Markov chain convergence

On contraction of functional differential equations with Markovian switching

Some Limit Properties of Markov Chains Induced by Stochastic Recursive Algorithms

On the Markov commutator

On regularity of functions of Markov chains

Contracting differential equations in weighted Banach spaces

Convergence to stable laws for a class of multidimensional stochastic recursions

Compressibility and Stochastic Stability of Monotone Markov Chain

Annealed central limit theorem for a class of contractive random dynamical systems

Contractivity of Distributed Optimization and Nash Seeking Dynamics

k-Contraction: Theory and Applications

Operator-Theoretical Treatment of Ergodic Theorem and Its Application to Dynamic Models in Economics