Long-time tails in the parabolic Anderson model with bounded potential (Q1872202)

The parabolic Anderson model is the Euclidean-time (or diffusion) version of the Schrödinger equation with a random potential: \[ \partial_tu(t,z)=k\triangle^du(t,z)+\xi(z) u(t,z),\;(t,z)\in (0,\infty)\times Z^d, \quad u(0,z)=1, \;z\in Z^d,\tag{1} \] where \(\partial_t\) is the time derivative, \(u:[0,\infty)\times Z^d\to [0,\infty)\) is a function, \(k>0\) is a diffusion constant, \(\triangle^d\) is the discrete Laplacian \([\triangle^df](z)=\sum_{y\sim z}(f(y)-f(z))\) (here \(y\sim z\) denotes that \(y\) and \(z\) are nearest neighbors), and \(\xi=(\xi(z))_{z\in Z^d}\) is a random i.i.d. potential. The main assumption is that \(\text{ess sup} \xi(0)=0.\) The subject of the paper concerning (1) is the large time behavior of the \(p\)th moment \(\langle u(t,0)^p\rangle\) for all \(p>0\) and the almost-sure asymptotics of \(u(t,0)\), depending on the thickness of the distribution \(\text{Prob}(\xi(0)\in \cdot)\) close to its essential supremum. Both the asymptotics of the moments of \(u(t,0)\) and the almost-sure asymptotics of \(u(t,0)\) as \(t\to\infty\) are identified in terms of variational problems. Lifshitz tails for the random Schrödinger operator \(-k\triangle-\xi\) at the bottom of its spectrum are established. For the class of \(\xi\) distributions, the Lifshitz exponent ranges from \(d/2\) to \(\infty\); the power law is typically accompanied by lower-order corrections.