Nonequilibrium steady-state dynamics of Markov processes on graphs
We propose an analytic approach for the steady-state dynamics of Markov processes on locally tree-like graphs. It is based on the definition of a probability distribution for infinite edge trajectories in terms of infinite matrix products. For homogeneous ensembles on regular graphs, the distribution is parametrized by a single $d\times d\times r^2$ tensor, where $r$ is the number of states per variable and $d$ is the matrix-product bond dimension. While the approach becomes exact in the large-$d$ limit, it usually produces extremely accurate results even for small $d$. The $d^2r^2$ parameters are found by solving a fixed point equation, for which we provide an efficient belief-propagation procedure. We apply it to a variety of models, including Ising-Glauber dynamics with symmetric and asymmetric couplings and the SIS model. Even for small $d$, the results are compatible with Monte Carlo estimates and accurately reproduce known exact solutions. The method gives access to accurate temporal correlations which, in some regimes, may be virtually impossible to estimate by sampling.
