Metastability in the Greenberg-Hastings model (Q1313066)

The Greenberg-Hastings model (GHM) is a multitype cellular automaton living on a lattice such as \(\mathbb{Z}^ 2\), and which evolves as follows. At each time \(t\), a vertex \(x\) is in a state \(\gamma_ t(x)\) taking values in \(\{0,1,2,\dots,\kappa-1\}\); \(\kappa\) is the number of available `colours'. Colours update according to the colours of vertices nearby. That is, if \(\gamma_ t(x)\neq 0\), then \(\gamma_{t+1}(x)=\gamma_ t(x)\bmod\kappa\). If \(\gamma_ t(x)=0\), then \(\gamma_{t+1}(x)=0\) unless at least \(\theta\) vertices in the set \(x+N\) have colour 1 at time \(t\), in which case \(\gamma_{t+1}(x)=1\). Here, \(N\) is some finite set of vertices, and \(\theta\) is an integer. The above deterministic evolution rule is combined with a random initial distribution, to provide a dynamical system. The question is to understand the behaviour of the system for given \(\kappa,N,\theta\), and initial distribution \(\mu\). This paper is directed at the GHM on a finite box \(B_ L\) having side- length \(L\), and towards understanding how large \(L\) must be for given \(\kappa\), in order that the process survives forever (i.e., does not become extinct by absorption in the state of all 0). It is shown that \(L\) must be of order \(e^{\beta\kappa}\) for suitable \(\beta\), in order that survival have large probability. More specifically, the main theorems may be summarised as follows. First, if \(\theta>{1\over 2}| N|\), then the GHM dies out. For the next two theorems, we start the process in uniform product measure. If \(0<\alpha<1/2\) and \(\alpha| N|\leq\theta\leq| N|\), \(\alpha\kappa\geq 5\), then there exists \(c=c_ \alpha>0\) such that the process dies out with large probability when \(L<e^{c\theta\kappa}\). On the other hand, if \(N\) is a box of volume \((2\rho+1)^ 2\) and \(\theta\leq{1\over 4}\rho^ 2\), then there exists a constant \(C>0\) such that, if \(L>e^{C\rho^ 2\kappa}\), then the probability of survival approaches 1 as either \(\rho\to\infty\) or \(\kappa\to\infty\). The next result (Theorem 3) applies to the `basic' GHM (with \(N=\{x:\| x\|_ \infty=1\}\) and \(\theta=1)\). For this process it is shown that the process on \(B_ L\) dies out if \(L<e^{c\kappa}\), and survives if \(L>e^{C\kappa}\), where \(c\) and \(C\) are positive constants. This is combined with a type of metastability theorem for a GHM on the infinite lattice \(\mathbb{Z}^ 2\). There is a variety of further discussion, including of different initial measures and of a model with random dynamics.