Comparison inequalities and fastest-mixing Markov chains
arXiv:1109.6075 · doi:10.1214/12-AAP886
Abstract
We introduce a new partial order on the class of stochastically monotone Markov kernels having a given stationary distribution on a given finite partially ordered state space . When in this partial order we say that and satisfy a comparison inequality. We establish that if and are reversible and for , then . In particular, in the time-homogeneous case we have for every if and are reversible and , and using this we show that (for suitable common initial distributions) the Markov chain with kernel mixes faster than the chain with kernel , in the strong sense that at every time the discrepancy - measured by total variation distance or separation or -distance - between the law of and is smaller than that between the law of and . Using comparison inequalities together with specialized arguments to remove the stochastic monotonicity restriction, we answer a question of Persi Diaconis by showing that, among all symmetric birth-and-death kernels on the path , the one (we call it the uniform chain) that produces fastest convergence from initial state 0 to the uniform distribution has transition probability 1/2 in each direction along each edge of the path, with holding probability 1/2 at each endpoint.
Published in at http://dx.doi.org/10.1214/12-AAP886 the Annals of Applied Probability (http://www.imstat.org/aap/) by the Institute of Mathematical Statistics (http://www.imstat.org)
References in corpus (5)
Cited by in corpus (5)
- Exclusive electroproduction off protons in the resonance region at photon virtualities 0.4~GeV ~GeV
- Fastest Mixing Reversible Markov Chain: Clique Lifted Graphs and Subgraphs
- Stochastic Orderings of Multivariate Elliptical Distributions
- Geometric Bounds on the Fastest Mixing Markov Chain
- Comparison of hit-and-run, slice sampling and random walk Metropolis