Abstract
Annals of Applied Probability 2007, Vol. 17, No. 1, 284-304 We compare convergence rates of Metropolis--Hastings chains to multi-modal
target distributions when the proposal distributions can be of ``local'' and
``small world'' type. In particular, we show that by adding occasional
long-range jumps to a given local proposal distribution, one can turn a chain
that is ``slowly mixing'' (in the complexity of the problem) into a chain that
is ``rapidly mixing.'' To do this, we obtain spectral gap estimates via a new
state decomposition theorem and apply an isoperimetric inequality for
log-concave probability measures. We discuss potential applicability of our
result to Metropolis-coupled Markov chain Monte Carlo schemes.
