A growth model in a random environment (Q1872296)

The article deals with the following model called oriented digital boiling. Given \(t = 0, 1 , \dots , \) there is attached a (occupied) set \({\mathcal A}_t \subset {\mathbb Z}^2\), which is defined as follows: \({\mathcal A}_t = \{ (x,y) \in {\mathbb Z}^2 \;|\;x \in {\mathbb Z}, \;y \leq h_t (x)\}\). The height function \(h_t\) evolves according to the rule \[ h_{t+1}(x) = \max\{ h_t (x-1), h_t + \varepsilon_{x,t} \}. \] Here \(\varepsilon_{x,t}\) are independent Bernoulli random variables, with \(P (\varepsilon_{x,t}=1) = p_x\). The initial state of this model is specified by the condition \(h_0 (x) = 0\) for \(x =0\), and \(h_0 (x) = - \infty\) for \(x \neq 0\). In their previous paper [J. Stat. Phys. 102, 1085-1132 (2001; Zbl 0989.82030)] the authors analyzed the homogeneous case \(p_x \equiv p\). In this paper, they consider the case where \(p_x\) are independent identically distributed random variables with common distribution \(P(p_x \leq s) = F(s)\). For this case, the asymptotic shape is identified explicitly. It is proved that, in the pure regime, the fluctuations about this shape, normalized by \(\sqrt{t}\), are asymptotically normal.