Spatial asymptotics for the parabolic Anderson models with generalized time-space Gaussian noise (Q282522)

This fairly long paper provides new and precise results on the spatial asymptotic behavior for the parabolic Anderson equation \[ \frac{\partial u}{\partial t}(t,x) = \frac{1}{2} \Delta u(t,x) + V(t,x)u(t,x) \] on \(\mathbb{R}_+ \times \mathbb{R}^d \) with initial condition \(u_0(x)\). Here, \(V(t,x)\) is a homogeneous generalized Gaussian noise. Its covariance is \(\gamma_0(s-t)\gamma(x-y)\), with \(\gamma_0\) and \(\gamma\) taking various forms such as \(\delta_0(.)\) or \(|.|^{-\alpha}\) leading to white or fractional noise. The point is finding the precise value of \[ \lim_{R\rightarrow \infty} (\log R)^{-2/\beta}\log \max_{|x|\leq R} u(t,x), \] where \(\beta =4-\alpha\) when \(\gamma(x)=|x|^{-\alpha}\) and \(\beta=3\) when \(d=1\) and \(\gamma(x)=\delta_0(x)\). The methods of proof are different for the case \(\gamma_0(t)=|t|^{-\alpha_0}\), where a Feynman-Kac representation is available, and for the case \(\gamma_0(t)=\delta_0(t)\), where the Dalang-Walsh theory of martingale measures is used. The paper ends with a comparison of quenched spatial asymptotics and quenched time asymptotics in the case of a time-independent potential.