23
3

Unbiased time-average estimators for Markov chains

Abstract

We consider a time-average estimator fkf_{k} of a functional of a Markov chain. Under a coupling assumption, we show that the expectation of fkf_{k} has a limit μ\mu as the number of time-steps goes to infinity. We describe a modification of fkf_{k} that yields an unbiased estimator f^k\hat f_{k} of μ\mu. It is shown that f^k\hat f_{k} is square-integrable and has finite expected running time. Under certain conditions, f^k\hat f_{k} can be built without any precomputations, and is asymptotically at least as efficient as fkf_{k}, up to a multiplicative constant arbitrarily close to 11. Our approach provides an unbiased estimator for the bias of fkf_{k}. We study applications to volatility forecasting, queues, and the simulation of high-dimensional Gaussian vectors. Our numerical experiments are consistent with our theoretical findings.

View on arXiv
Comments on this paper