Maximal attractor for the coupled Cahn-Hilliard equations (Q1348728): Difference between revisions

The authors investigate the asymptotic behaviour of solutions of the Cahn-Hilliard equations, which after two transformations assume the form \[ \partial_t \varphi=M_1 \partial^2_x\bigl(-K\partial_x^2 \varphi+ \varphi^3 -(a \varphi +b)u\bigr), \] \[ C\partial_t u+u^2(a\varphi+b) \varphi_t=M_2 u^2 \partial_x^2 u,\;x\in[0,1], \] \[ (\partial_x \varphi)(\zeta)= (\partial^3_x \varphi)(\zeta)= (\partial_xu) (\zeta)=0,\;\zeta=0,1, \] \[ \varphi(0,x)= \varphi_0 (x),\;u(0,x)= u_0(x)>0. \] Here \(a,M_1,M_2,K>0,b\) are constants, while \(\varphi, u\) have a suitable physical interpretation. The aim is to construct attractors. This program encounters a number of difficulties as explained by the authors in the introduction. In order to overcome them, they introduce the set \[ H=\bigl\{ (\varphi,u)\in H^2[0,1]\times H^1[0,1];\;u(x)>0,\;x\in[0,1],\;\partial_x \varphi (\zeta) =0,\;\zeta=0,1 \bigr\}. \] While \(H\) is a metric space under the metric induced by \(H^2[0,1] \times H^1[0,1]\), it is not complete. Therefore subsets \(U_{\beta_1 \beta_2}\) of \(H\) have to be introduced where \(\beta_1>0\), \(\beta_2 \in\mathbb{R}\) and \[ \begin{multlined} U_{\beta_1\beta_2}=\\ \left\{ (\varphi,u)\in H; u\geq \delta >0, \int^1_0 \bigl(F_1 (\varphi)+ \ln(u)\bigr) dx\leq\beta_1, \int^1_0 \bigl(\tfrac 12 a\varphi^2+ b\varphi-u^{-1} \bigr)dx \geq\beta_2\right\} \end{multlined} \] with \(F_1(\varphi)= \frac 12(K_1\partial_x \varphi)^2+4^{-1} \varphi^4\), and \(\delta \geq 0\) a fixed constant. Further, define \[ B_{\beta_1 \beta_2}= \bigl\{(\varphi, u) \in U_{\beta_1\beta_2},\|\varphi|_{H^2} \leq C,\;\|u|_{H^1}\leq C' \bigr\} \] with certain specific constants \(C,C'\). One of the main theorems, Theorem 2.1, splits into three parts. Part (i) states that for initial conditions \((\varphi_0,u_0)\in H\) the associated solution \((\varphi(t),u(t))\) exists globally and remains in \(H\); thus \(H\) is an invariant set for the solution semigroup \(S(t)\), \(t\geq 0\). Part (i) has been proven by the authors in an earlier paper, given in the references. Part (ii) states that for every bounded set \(B\subseteq U_{\beta_1 \beta_2}\) there exists \(t_0=t_0(B)\) such that \(\cup S(t)B\), \(t\geq t_0\) is relatively compact in \(H\). Part (iii) states that orbits starting from a bounded set \(B \subseteq U_{\beta_1\beta_2}\) reenter \(U_{\beta_1 \beta_2}\) after some time \(t(B)\), that the sets \(B_{\beta_1\beta_2}\) are absorbing in \(U_{\beta_1 \beta_2} \) and that the \(\omega\)-limit set of \(B_{\beta_1 \beta_2}\) is a maximal compact attractor which attracts bounded sets of \(U_{\beta_1 \beta_2}\). There are further theorems of a similar type.