Solvability and asymptotic analysis of a generalization of the Caginalp phase field system (Q1948018)

The authors consider the initial and boundary value problem \[ \begin{gathered} w_{tt}-\alpha\Delta w_t-\beta\Delta w+u_t=f\quad \text{in } \Omega\times(0,T)\tag{1} \\ u_t-\Delta u+\gamma(u)+g(u)\ni w_t\quad \text{in } \Omega\times(0,T) \end{gathered} \] \[ \partial_nw=\partial_nu=0\quad {\text{on}}\;\;\Gamma\times(0,T) \tag{3} \] \[ w(\cdot,0)=w_0,\;\;w_t(\cdot,0)=v_0,\;\;u(\cdot,0)=u_0\quad {\text{in}}\;\;\Omega, \tag{4} \] where \(\Omega\subset{\mathbb R}^3\) is a bounded domain with smooth boundary \(\Gamma\), \(T>0\) represents some finite time, and \(\partial_n\) denotes the outward normal derivative on \(\Gamma\). Moreover, \(\alpha\) and \(\beta\) are two positive parameters, \(\gamma:{\mathbb R}\to2^{\mathbb R}\) is a maximal monotone graph, \(g:{\mathbb R}\to{\mathbb R}\) is a Lipschitz continuous function, \(f\) in Equation (1) is a given source term, and \(w_0\), \(u_0\), \(u_0\) stand for initial data. The inclusion which takes place in (2) is due to the present of the, possibly multivalued, graph \(\gamma\). Equations (1)--(2) yield a system of phase-field type, such systems have been introduced in order to include phase-dissipation effects in the dynamics of moving interfaces arising in thermally induced phase transitions. In this case, they move from the following expression for the total free energy \[ \Psi(\theta,u)=\intop_\Omega\left(-{1\over2}\theta^2-\theta u+ \phi(u)+G(u)+{1\over2}|\nabla u|^2\right) \tag{5} \] where the variables \(\theta\) and \(u\) denote the (relative) temperature and order parameter, respectively. Let us notice from the beginning that \(w\) here represents the thermal displacement variable, related to \(\theta\), by \[ w(\cdot,t)=w_0+(1*\theta)(\cdot,t)=w_0*\intop_0^t\theta(\cdot,s)\,ds \tag{6} \] In (5), \(\Phi:\cdot[0,+\infty]\to{\mathbb R}\), \(t\in[0,T]\), is a convex and lower semicontinuous function such that \(\phi(0)=0=\min\Phi\), and its subdifferential \(\partial\phi\) coincides with \(\gamma\), while \(G\) stands for a smooth, in general concave, function such that \(G=g\). A typical example form of \(\phi\) and \(G\) is the double obstacle case \[ \phi(u)=I_{[-1,+1]}(u)=\begin{cases} 0&\text{if }|u|\leq1,\\+\infty&\text{if }|u|>1.\end{cases}\tag{7} \] So that the two wells of the sum \(\phi(u)+G(u)\) are located in \(-1\) and \(+1\) and one of the two is preferred as minimum of the potential in (5) according to where the temperature \(\theta\) is negative or positive noticed the presence of the terms \(-\theta u\) besides \(\phi(u)+G(u)\) in the expression of \(\psi\). In the formula (7) the subdifferential of the indicator function of the interval \([-1,-1]\) reads \[ \xi\in\partial I_{[-1,+1]}(u){\text{ if and only if }} \xi\;\begin{cases} \leq0&\text{if } u=-1\\=0&\text{if } |u|<1\\ \geq0&\text{if } u=+1\end{cases} \] In the paper the authors interested in the study of existence, uniqueness, regularity of the solution to the initial-boundary value problem (1)--(4) where \(\gamma\) is an arbitrary maximal monotone graph, possibly multivalued, singular and with bounded domain. They considered in (5) the case of a multiwell potential, because the function \(u\to-w_tu+\phi(u)+G(u)\) may exhibit different wells according especially to the shape of \(G\) in the region where \(\phi\not=+\infty\). Additionally they considered the singular case for a convex potential \(\phi\) with the same interval \([-1,+1]\) as domain, but with \[ \phi(u)=k_1((1+u)\ln(1+u)+(1-u)\ln(1-u)) \] for \[ g(u)=G'(u)=-2k_0u \] where the meaningful constants \(k_0,k_1\) have to satisfy \(0<k_1<k_0\). The main goal of this paper, was to proved the existence and uniqueness of weak solutions to (1)--(4) and the continuous dependence on the data. Next, they showed some regularity estimates and they obtained the existence of strong solutions. They also studied the asymptotic behaviour of the problems as \(\beta\to0\), obtained convergence of solutions to the problem with \(\beta=0\) which corresponds with the case of Green and Naghdi (formula cf. (11)) in the paper. Additionally, they also proved two error estimates of the difference solution in suitable norms, showing a linear rate of convergence in both estimates. They investigated the analogous limit \(\alpha\to0\) to obtain the second type proposed by Green and Naghdi (cf. formula (1.12) in the paper). In the proof they used the method of Sobolev spaces, method of energy estimates and the Faedo-Galerkin methods.