Asymptotics of the density of parabolic Anderson random fields (Q2078012)

This work studies the shape of the density \(\rho(t,x;y)\) of the solution to the stochastic partial differential equation \[\frac{\partial}{\partial t}u(t,x)=\frac{1}{2}\Delta u(t,x)+lu(t,x)\diamond \dot{W}(t,x), \,\,t\geq 0, \,\,x\in\mathbb{R}^d, \] where \(l\) is a positive constant, \(\diamond\) denotes the Wick product, and \(\dot W\) is a Gaussian noise with covariance structure of the form \[E(\dot{W}(t,x)\dot{W}(s,y))=\gamma_0(\vert t-s\vert)\gamma(\vert x-y\vert). \] It is assumed that for some \(\alpha_0\in[0,1)\), \(\gamma_0\) is a function equivalent up to positive constants to the function \(f(t)=t^{-\alpha_0}\), and for some \(\alpha\in(0,2)\), \(\gamma\) is a function that \(\sup_{c,x:c>0,\gamma(x)\neq 0}\frac{\gamma(cx)}{c^{-\alpha}\gamma(x)}<\infty\) and \(\inf_{x:\vert x\vert\leq\varepsilon}\gamma(x)\geq \varepsilon^{-\alpha}\), \(\varepsilon>0\). Using Malliavin calculus tools two main results are proved: Theorem 6.2 stating that the density of the solution satisfies the following right tail estimation: for any positive, big enough \(y\), the density \(\rho(t,x;y)\) is equivalent, up to positive constants to the function \[f(t,y)=t^{-\frac{4-2\alpha_0}{4}}\exp\left(-t^{-\frac{4-2\alpha_0}{4}}(\log y)^{\frac{4-\alpha}{2}}\right). \] Theorem 6.4 stating that the density satisfies that for any positive, small enough \(y\), the density \(\rho(t,x;y)\) is upperly bounded by a positive constant times the function \[f(t,y)=t^{-\frac{4-2\alpha_0}{4}}\exp\left(-(\log y)^{2}\right). \]