A variational approach to branching random walk in random environment (Q2365747)

The model studied in the present paper is a simplified version of the model in [\textit{A. Greven} and \textit{F. den Hollander}, Probab. Theory Relat. Fields 91, No. 2, 195-249 (1992; Zbl 0744.60079)]. With each \(x\in Z\) a random probability measure \(F_ x\) on \(N\cup\{0\}\) is associated (called the offspring distribution at site \(x)\). The sequence \(F=(F_ x)_{x\in Z}\) is assumed to be i.i.d. \((F\) plays the role of a random environment). Given \(F\), define a discrete-time Markov process \((\eta_ n)\) on \((N\cup\{0\})^ Z\) with \(\eta_ n=(\eta_ n(x))_{x\in Z}\) (here, \(\eta_ n(x)\) is the number of particles at site \(x\) at time \(n)\) which evolves as follows. At time \(n=0\), one particle is placed at every site, i.e., \(\eta_ 0(x)\equiv 1\). Given state \(\eta_ n\) at time \(n\), each particle is independently replaced by a new generation. The size of a new generation descending from a particle at \(x\) has distribution \(F_ x\). Immediately after creation each new particle independently decides to either stay at the site where it is or to jump to the right neighbouring site (the jump probability being \(h\) not depending on \(x)\). The resulting sequence of particle numbers make up state \(\eta_{n+1}\) at time \(n+1\). The authors obtain a variational representation of the (exponential) growth rate (depending on \(h)\) of the particle density at time \(n\) as \(n\to\infty\). It turns out that the analysis of the variational formula can be carried out in complete detail which enables the authors to give a precise description of the phase diagram.