Homogenization and corrector result for a coupled parabolic hyperbolic system (Q2287279)

The authors prove the existence of a unique solution to a coupled parabolic-hyperbolic system which intends to simulate a dune erosion phenomenon under the action of a shallow water motion. They also prove an homogenization result when a small parameter associated to the tide period goes to 0. Following their previous work [Discrete Contin. Dyn. Syst., Ser. S 9, No. 5, 1521--1551 (2016; Zbl 1356.35169)], they write the problem in a non-dimensional form as \(\frac{\partial v^{\epsilon }}{\partial t}+\frac{1}{\epsilon ^{4}}\sum_{j=1}^{2}A^{j}\cdot \frac{\partial v^{\epsilon }}{\partial x_{j}}=\frac{1}{\epsilon }h(v^{\epsilon })-\frac{1}{\epsilon ^{4}}P(v^{\epsilon },z^{\epsilon })\), \(\frac{\partial z^{\epsilon }}{\partial t}- \frac{1}{\epsilon ^{2}}\nabla \cdot (\mathcal{A}^{\epsilon }\nabla z^{\epsilon })=\frac{1}{\epsilon ^{2}}\nabla \cdot \mathcal{C}^{\epsilon }\), where \(v^{\epsilon }=\left( \begin{array}{c} m^{\epsilon } \\ q_{1}^{\epsilon } \\ q_{2}^{\epsilon } \end{array} \right) \), \(A^{1},A^{2}\) are \(3\times 3\) matrices whose coefficients depend on \(v^{\epsilon }\), \[h(v^{\epsilon })=-\frac{f}{\epsilon }v^{\epsilon }+ \frac{1}{\epsilon }\left( \begin{array}{c} 0 \\ -\frac{kq_{1}}{m^{\epsilon }+b^{1}} \\ -\frac{kq_{2}}{m^{\epsilon }+b^{1}} \end{array}\right),\, P(v^{\epsilon },z^{\epsilon })=\left( \begin{array}{c} 0 \\ d^{1}(m+b^{1})\frac{\partial z^{\epsilon }}{\partial x} \\ d^{1}(m+b^{1})\frac{\partial z^{\epsilon }}{\partial y} \end{array} \right), \] \(\mathcal{A}^{\epsilon }=\frac{a(1-b\epsilon m^{\epsilon })\left\vert q^{\epsilon }\right\vert ^{3}}{\left\vert m^{\epsilon }+b^{1}\right\vert ^{3}}\) and \(\mathcal{C}^{\epsilon }=\frac{c(1-b\epsilon m^{\epsilon })\left\vert q^{\epsilon }\right\vert ^{2}q^{\epsilon }}{ (m^{\epsilon }+b^{1})^{3}}\), for different constants. The authors observe that this system is singular when \(\epsilon =0\) or degenerate for values of \( \epsilon \) which cancel coefficients of the equations. They add the initial conditions \(v^{\epsilon }(0,x)=v^{0}\), \(z^{\epsilon }(0,x)=z^{0}\) for initial data which satisfy regularity properties and uniform bounds. They first prove the existence of a unique solution \((\overline{v}^{\epsilon }, \overline{z}^{\epsilon })\in C([0,T];H^{s}(\mathbb{T}^{2};\mathbb{R} ^{3}))\cap C^{1}([0,T];H^{s-1}(\mathbb{T}^{2};\mathbb{R}^{3}))\times L^{\infty }([0,T];H^{p}(\mathbb{T}^{2}))\) which satisfies uniform bounds with respect to \(\epsilon \). For the proof, the authors adapt the tools they used in their previous paper [loc. cit.], proving new estimates on \(z^{\epsilon }\). Moving to the homogenization result, the authors observe that \(v^{\epsilon }=\left( \begin{array}{c} v \\ z^{\epsilon } \end{array} \right) \) is the solution of the equation \(\frac{\partial w^{\epsilon }}{ \partial t}+\frac{1}{\epsilon ^{4}}\sum_{j=1}^{2}L^{j}\cdot \frac{\partial w^{\epsilon }}{\partial x_{j}}-\frac{1}{\epsilon ^{2}}\sum_{j=1}^{2}M^{j} \cdot \frac{\partial w^{\epsilon }}{\partial x_{j}}-\frac{1}{\epsilon ^{2}} \sum_{j=1}^{2}N^{j}\cdot \frac{\partial ^{2}w^{\epsilon }}{\partial x_{j}^{2} }=\frac{1}{\epsilon ^{2}}E(w^{\epsilon })\) for some matrix operators \( L^{1},L^{2},M^{1},M^{2},N^{1},N^{2}\) and \(E\). Their second main result proves that \((\overline{v}^{\epsilon },\overline{z}^{\epsilon })\) three-scale converges to a function \((\overline{v}^{0},\overline{z}^{0})\in L^{\infty }([0,T);L_{\#}^{\infty }([0,1];L_{\#}^{\infty }([0,1];H^{l}( \mathbb{T}^{2};\mathbb{R}^{4}))))\), with \(l=\min (p,s)\), and which satisfies an equation written as \(\frac{\partial \overline{w}_{0}}{\partial \tau } +\sum_{j=1}^{2}L_{0}^{j}\cdot \frac{\partial \overline{w}_{0}}{\partial x_{j} }=0\) for some matrix operators \(L_{0}^{j}\). Under further hypotheses on the orders of the operators \(L^{j},M^{j},N^{j}\) and \(E\) with respect to \( \epsilon \), the authors build three correctors to this homogenization problem. They first recall the notions of two- and three-scale and the associated classical compactness result and the proof of the three-scale convergence is a direct consequence of the properties of this three-scale convergence. The verification of the corrector results is also quite immediate.