Computational complexity of Markov chain Monte Carlo methods for finite Markov random fields

SUMMARY This paper studies the computational complexity of Markov chain Monte Carlo algorithms with finite-valued Markov random fields on a finite regular lattice as target distributions. We state conditions under which the complexity for approximate convergence is polynomial in n, the number of variables. Approximate convergence takes time O(n log n) as n -+ oo if the target field satisfies certain spatial mixing conditions. Otherwise, if the field has a potential with finite interaction range independent of n, the complexity is exponential in n", with y < 1, which is still more favourable than enumerating all the states. When the interaction range grows with n, the algorithms can converge exponentially in n. Analogous results are provided for an expectation approximated by an average along the chain.

[1]  W. K. Hastings,et al.  Monte Carlo Sampling Methods Using Markov Chains and Their Applications , 1970 .

[2]  L. Thomas,et al.  Bound on the mass gap for finite volume stochastic ising models at low temperature , 1989 .

[3]  Michael Aizenman,et al.  Rounding effects of quenched randomness on first-order phase transitions , 1990 .

[4]  P. Diaconis,et al.  Geometric Bounds for Eigenvalues of Markov Chains , 1991 .

[5]  Charles J. Geyer,et al.  Practical Markov Chain Monte Carlo , 1992 .

[6]  D. Stroock,et al.  The logarithmic sobolev inequality for discrete spin systems on a lattice , 1992 .

[7]  Peter W. Glynn,et al.  Stationarity detection in the initial transient problem , 1992, TOMC.

[8]  C. Hwang,et al.  Optimal Spectral Structure of Reversible Stochastic Matrices, Monte Carlo Methods and the Simulation of Markov Random Fields , 1992 .

[9]  G. Parisi,et al.  Simulated tempering: a new Monte Carlo scheme , 1992, hep-lat/9205018.

[10]  Mark Jerrum,et al.  Polynomial-Time Approximation Algorithms for the Ising Model , 1990, SIAM J. Comput..

[11]  P. Matthews.  A slowly mixing Markov chain with implications for Gibbs sampling , 1993 .

[12]  C. Hwang,et al.  Convergence rates of the Gibbs sampler, the Metropolis algorithm and other single-site updating dynamics , 1993 .

[13]  Richard L. Tweedie,et al.  Markov Chains and Stochastic Stability , 1993, Communications and Control Engineering Series.

[14]  Peter Green,et al.  Spatial statistics and Bayesian computation (with discussion) , 1993 .

[15]  F. Martinelli.  On the two-dimensional dynamical Ising model in the phase coexistence region , 1994 .

[16]  J. Rosenthal.  Minorization Conditions and Convergence Rates for Markov Chain Monte Carlo , 1995 .

[17]  F. Martinelli,et al.  Approach to equilibrium of Glauber dynamics in the one phase region , 1994 .

[18]  Nicholas G. Polson,et al.  Sampling from log-concave distributions , 1994 .

[19]  F. Martinelli,et al.  For 2-D lattice spin systems weak mixing implies strong mixing , 1994 .

[20]  A. Frigessi,et al.  A dynamical phase transition in a caricature of a spin glass , 1994 .

[21]  Kerrie Mengersen,et al.  [Bayesian Computation and Stochastic Systems]: Rejoinder , 1995 .

[22]  J. Besag,et al.  Bayesian Computation and Stochastic Systems , 1995 .

[23]  Bin Yu,et al.  Regeneration in Markov chain samplers , 1995 .

[24]  C. Robert.  Convergence Control Methods for Markov Chain Monte Carlo Algorithms , 1995 .

[25]  R. Tweedie,et al.  Rates of convergence of the Hastings and Metropolis algorithms , 1996 .

[26]  R. Tweedie,et al.  Geometric convergence and central limit theorems for multidimensional Hastings and Metropolis algorithms , 1996 .

[27]  J. Propp,et al.  Exact sampling with coupled Markov chains and applications to statistical mechanics , 1996 .