Stochastic domination: the contact process, Ising models and FKG measures

We prove for the contact process on $Z^d$, and many other graphs, that the upper invariant measure dominates a homogeneous product measure with large density if the infection rate $\lambda$ is sufficiently large. As a consequence, this measure percolates if the corresponding product measure percolates. We raise the question of whether domination holds in the symmetric case for all infinite graphs of bounded degree. We study some asymmetric examples which we feel shed some light on this question. We next obtain necessary and sufficient conditions for domination of a product measure for ``downward'' FKG measures. As a consequence of this general result, we show that the plus and minus states for the Ising model on $Z^d$ dominate the same set of product measures. We show that this latter fact fails completely on the homogenous 3-ary tree. We also provide a different distinction between $Z^d$ and the homogenous 3-ary tree concerning stochastic domination and Ising models; while it is known that the plus states for different temperatures on $Z^d$ are never stochastically ordered, on the homogenous 3-ary tree, almost the complete opposite is the case. Next, we show that on $Z^d$, the set of product measures which the plus state for the Ising model dominates is strictly increasing in the temperature. Finally, we obtain a necessary and sufficient condition for a finite number of variables, which are both FKG and exchangeable, to dominate a given product measure.

[1]  Roberto H. Schonmann,et al.  Second order large deviation estimates for ferromagnetic systems in the phase coexistence region , 1987 .

[2]  Thomas M. Liggett,et al.  The survival of contact processes , 1978 .

[3]  R. Schonmann,et al.  Domination by product measures , 1997 .

[4]  Jeff Kahn,et al.  Some conditional correlation inequalities for percolation and related processes , 2006, Random Struct. Algorithms.

[5]  T. Liggett,et al.  A strong correlation inequality for contact processes and oriented percolation , 1997 .

[6]  T. Liggett,et al.  Stochastic Interacting Systems: Contact, Voter and Exclusion Processes , 1999 .

[7]  Hans-Otto Georgii,et al.  Gibbs Measures and Phase Transitions , 1988 .

[8]  Thomas M. Liggett,et al.  SURVIVAL AND COEXISTENCE IN INTERACTING PARTICLE SYSTEMS , 1994 .

[10]  Olle Häggström,et al.  Infinite clusters in dependent automorphism invariant percolation on trees , 1997 .

[11]  Thomas M. Liggett,et al.  Survival of Discrete Time Growth Models, with Applications to Oriented Percolation , 1995 .

[12]  Geoffrey Grimmett,et al.  Probability and Phase Transition , 1994 .

[13]  Colin J. Thompson,et al.  Mathematical Statistical Mechanics , 1972 .

[14]  Steif,et al.  DYNAMICAL STABILITY OF PERCOLATION FOR SOME INTERACTING PARTICLE SYSTEMS AND ε -MOVABILITY , 2004 .

[16]  John C. Wierman,et al.  Substitution Method Critical Probability Bounds for the Square Lattice Site Percolation Model , 1995, Combinatorics, Probability and Computing.

[17]  Russell Lyons,et al.  Stationary determinantal processes: Phase multiplicity, Bernoullicity, entropy, and domination , 2002 .

[18]  F. Papangelou GIBBS MEASURES AND PHASE TRANSITIONS (de Gruyter Studies in Mathematics 9) , 1990 .

[19]  Jean Bricmont,et al.  Percolation in strongly correlated systems: The massless Gaussian field , 1987 .

[20]  Percolation, Path Large Deviations and Weakly Gibbs States , 2000 .

[21]  Russell Lyons,et al.  Group-invariant Percolation on Graphs , 1999 .