Spanning forests on random planar lattices (Q833378): Difference between revisions

The generating function of spanning forests in a graph \(G\) can be represented as a Grassmann integral of the exponential of a local fermionic weight involving a Gaussian form. The parameter \(t\) which appears in the generating function of the forests on a lattice to count the number of trees in a forest is related to the coupling constant in the \(O(n)\)-invariant \(\sigma\)-model. The same generating function can be obtained by a suitable limit \(q \rightarrow 0\) of the \(q\)-state Potts model defined on the same graph \(G\). Recently, a combinatorial perturbative equivalence of the generating function with the \(O(n)\) model in the limit \(n \rightarrow -1\) have been found. In the paper, a standard random-matrix approach to the \(O(n)\) model on the ensemble of degree-\(k\) random planar lattices is described. The problem of counting configuration of self-avoiding loops represents a combinatorial invariant of the \(O(n)\) loop-gas model especially in the case \(k = 3\) and for this case the similarities of the studied partition function with Kostov solution have been found. It is also described how to construct a perturbative expansion for the partition function at every order in \(t\) using the combinatorial results. This general prescription is a slight modification of the technique of Cauchy integrals. In the thermodynamic limit, the leading contribution to the finite sum of finite-dimensional Cauchy integrals comes from terms for which resummation can be obtained exactly.