Recurrence and transience for the frog model on trees (Q1681592)

The frog model is a system of interacting random walks in discrete time on a rooted graph. At time \(t=0\) the graph contains one particle (= frog) at the root and a configuration of sleeping particles on some vertices of the graph, with one sleeping particle per vertex. The particle at the root performs a nearest neighbor random walk, waking sleeping particles as it visits them. The awaked particles performs their own independent random walks waking the visited sleeping particles. The frog model depicts the spread of rumors or infections, where informed or infected agents are referred to as awakened frogs. Another context of the frog model is a lattice gas model with active particles that spread throughout the graph igniting the contacted inert particles. The main theorem of the paper states that the frog model on the tree \(T_d\) is a.s. recurrent for \(d=2\) and is a.s. transient for \(d\geq 5\). This result answers a question posed in [\textit{O. S. M. Alves} et al., Ann. Appl. Probab. 12, No. 2, 533--546 (2002; Zbl 1013.60081)].