On the behavior of solutions to certain parabolic SPDE's driven by Wiener processes. (Q1879534)

The structure and stability properties of the global attractor associated with semilinear stochastic PDEs of the form \[ du(t,x)= (\text{div} (k\nabla u(t,x))+ g(u(t,x))\,dt +\sum _{j=1}^r h_j(u(t,x))\,dW_j(t),\quad (t,x)\in R_+\times D, \] with the initial condition \(u(0,x)=\varphi (x)\) and the Neumann boundary conditions \((\partial u(t,x)/\partial n(k))\) \(=0,\;(t,x)\in R_+ \times \partial D,\) are studied, where \(\partial /\partial n(k)\) denotes the conormal derivative, \(D\subset R^d\) is a bounded domain, \(k\) stands for a symmetric positive definite and sufficiently regular matrix, \(\{W_i\} \) is an \(r\)-dimensional Wiener process and for given \(0<u_1<u_2\) it is assumed that \(g, h_j \in \mathcal C^2([u_1, u_2])\) and \(g(u_1)=g(u_2)=h_j(u_1)=h_j(u_2)=0\). The initial datum \(\varphi \) is an \(L^2(D)\)-valued random field that evaluates between \(u_1\) and \(u_2\), so the solution \(u\) remains there as well. The long-time behavior of the solutions is investigated in detail. In particular, when the noise term of the equation is in a sense subordinated to the nonlinear part of the drift, the corresponding Lyapunov exponents are determined explicitly.