Exponential attractors for the strongly damped wave equations (Q2655103)

The authors study the asymptotic behaviour of solutions of the following strongly damped wave equation: \[ \begin{aligned} u_{tt}-\Delta u_t-\Delta u+f(u)=g(x), &\quad x\in\Omega,\;g\in H^{-1}(\Omega),\\ (u(0),u_t(0))=(u_0,v_0), &\quad u=0\text{ on }\partial\Omega.\end{aligned}\tag{1} \] Here, \(\Omega\) is a smooth bounded domain \(\subset\mathbb R^3\), while \(f\in C^1(R)\) with \(f(0)=0\) satisfies \[ |f'(s)|\leq C(1+|s|^\varphi),\quad s\in\mathbb R\text{ and some }p,\quad 0\leq p\leq 4\tag{2} \] and in addition \[ \liminf_{|s|\to\infty} s^{-1}f(s)>-\lambda_1\tag{3} \] with \(\lambda_1\) the first eigenvalue of \(-\Delta\) on \(H^1_0(\Omega)\). The basic spaces underlying the analysis are \[ {\mathcal H}=H^1_0(\Omega)\times L^2(\Omega),\quad {\mathcal V}=H^1_0(\Omega)\times H^1_0(\Omega);\tag{4} \] one sets \(\|\;\|=\|\;\|_{L^2(\Omega)}\). The authors note that under the assumptions (2) and (3) equation (1) generates a \(C_0\)-solution semigroup \(S(t)\), \(t\geq 0\) on the phase space \(\mathcal H\). The authors first recall known results about the asymptotics of solutions to (1)--(3). They then proceed to state and prove their main results. Theorem 3.5 eg. asserts that the semigroup \(S(t)\), \(t\geq 0\) admits a global attractor \({\mathcal A}_{\mathcal H}\subset H^1_0(\Omega)\times L^2(\Omega)\) which remains bounded in \(H^1_0(\Omega)\times H^1_0(\Omega)\). Theorem 3.5 is extended by Theorem 3.7 which asserts that \(S(t)\), \(t\geq 0\) admits an attractor \(Q\subset {\mathcal V}\) which is compact and invariant in \(\mathcal V\) and attracts every in \(\mathcal H\) bounded set in the \(\mathcal V\)-norm. While the above results merely assume \(g\in H^{-1}(\Omega)\), more can be said if \(g\in L^2(\Omega)\). Thus let \(g\in L^2(\Omega)\). Theorem 4.1 then asserts that there is a set \({\mathcal E}\subset {\mathcal V}\), compact in \(\mathcal V\) and bounded in \(H^2(\Omega)\times H^1_0(\Omega)\) such that: (a) \(S(t){\mathcal E}\subset {\mathcal E}\) for \(t\geq 0\), (b) \(\mathcal E\) has a finite fractal dimension in \(\mathcal V\), (c) there exists an increasing \(\varphi:R_+\to R_+\) and \(\alpha>0\) such that \[ \text{dist}_{\mathcal V}(S(t)B,{\mathcal E})\leq \varphi(\lambda)t^{-\frac12}e^{\alpha t},\quad t>0, \] for any subset \(B\subset {\mathcal H}\) with \(\sup_{z\in B}\|z\|_{\mathcal H}<\lambda\). The proofs are based on a series of lemmas of independent interest.