Tree-like approximation is a computational method commonly used for dynamic properties on quenched finite network realizations, including empirical networks. Percolation cluster sizes, epidemic thresholds, and Ising/Potts partition functions and magnetization are among such properties. That method is exact only when the network is a tree: removal of one inner node leaves the network fragmented, with each fragment again separable in the same way. A generalization of this recursive separation is called tree-decomposition of width k, allowing a set of up to k nodes as a separator in each step. In this talk, I show (i) how to find useful tree-decompositions and (ii) how to employ these to efficiently compute exact properties of the Ising model and other stochastic processes with detailed balance.
Meeting ID: 838 2931 8876
Tobias Galla Contact form