Inviscid limit of compressible viscoelastic equations with the no-slip boundary condition (Q2687331)

The authors consider the two-dimensional compressible viscoelastic equations written in the half plane \(\mathbb{R}_{+}^{2}\) as: \[ \begin{cases} \partial_{t}\rho^{\varepsilon}+\nabla \cdot (\rho^{\varepsilon}\mathbf{u}^{\varepsilon})=0,\\ \rho^{\varepsilon}\partial_{t}\mathbf{u}^{\varepsilon}+\rho^{\varepsilon}\mathbf{u}^{\varepsilon}\cdot \nabla \mathbf{u}^{\varepsilon}-\varepsilon \mu \Delta \mathbf{u}^{\varepsilon}-\varepsilon (\mu +\lambda )\nabla \operatorname{div}\mathbf{u}^{\varepsilon}+\nabla p(\rho^{\varepsilon})=\operatorname{div}(\rho^{\varepsilon}\mathbf{F}^{\varepsilon}\mathbf{F}^{\varepsilon \intercal}),\\ \partial_{t}\mathbf{F}^{\varepsilon}+\mathbf{u}^{\varepsilon}\cdot \nabla \mathbf{F}^{\varepsilon}=\nabla \mathbf{u}^{\varepsilon}\cdot \mathbf{F}^{\varepsilon}, \end{cases} \] where \(\rho^{\varepsilon}\) denotes the density, \(\mathbf{u}^{\varepsilon}=(u^{\varepsilon},v^{\varepsilon})\) the velocity, \(\mathbf{F}^{\varepsilon}=(F_{1}^{\varepsilon},F_{2}^{\varepsilon})^{\intercal}\) the deformation tensor matrix with \(F_{1}^{\varepsilon}=(1+f_{1}^{\varepsilon}, f_{2}^{\varepsilon})\), \(F_{2}^{\varepsilon}=(f_{3}^{\varepsilon},1+f_{4}^{\varepsilon})\), \(\varepsilon \mu\) and \(\varepsilon \lambda\) the viscosity coefficients which satisfy \(\mu >0\) and \((\mu +\lambda )>0\), \(\varepsilon \in (0,1)\) a small parameter. The pressure \(p(\rho )\) is a function of the density \(\rho\) given by: \(p(\rho )=\rho^{\gamma}\), \(\gamma \geq 1\) (the isentropic case), where \(\gamma\) is the adiabatic constant. The initial data \(\rho^{\varepsilon}(0,x,y)=\rho_{0}(x,y)\), \(\mathbf{u}^{\varepsilon}(0,x,y)= \mathbf{u}_{0}(x,y)\), \(\mathbf{F}^{\varepsilon}(0,x,y)=\mathbf{F}_{0}(x,y)\) are imposed, together with the no-slip boundary condition \(\mathbf{u}^{\varepsilon}(t,x,0)=0\) on the velocity. The authors prove that the limit problem when \(\varepsilon \rightarrow 0\) is governed by inviscid elastodynamic equations \[ \begin{cases} \partial_{t}\rho^{0}+\nabla \cdot (\rho^{0}\mathbf{u}^{0})=0,\\ \rho^{0}\partial_{t}\mathbf{u}^{0}+\rho^{0}\mathbf{u}^{0}\cdot \nabla \mathbf{u}^{0}+\nabla p(\rho^{0})=\operatorname{div}(\rho^{0}\mathbf{F}^{0}\mathbf{F}^{0\intercal}),\\ \partial_{t}\mathbf{F}^{0}+\mathbf{u}^{0}\cdot \nabla \mathbf{F}^{0}=\nabla \mathbf{u}^{0}\cdot \mathbf{F}^{0}. \end{cases} \] If \(m>8\) and if the initial data \((\rho_{0},\mathbf{u}_{0},\mathbf{F}_{0})\) are small enough in the sense: \[ \left\Vert (\rho_{0}-1,\mathbf{u}_{0},\mathbf{F}_{0}- \mathbb{I}_{2\times 2})\right\Vert_{m}^{2}+\left\Vert \partial_{y}(\rho_{0},\mathbf{u}_{0},\mathbf{F}_{0})\right\Vert_{m-1}^{2}+\left\Vert \partial_{y}(\nabla \rho_{0},\nabla \mathbf{u}_{0},\nabla \mathbf{F}_{0})\right\Vert_{m-2}^{2}\leq \sigma_{0}, \] for some sufficiently small positive constant \(\sigma_{0}\), and \(\rho_{0}det(\mathbf{F}_{0})=1\), \(\operatorname{div}(\rho_{0}\mathbf{F}_{0}^{\intercal})=0\), there exists a time \(T>0\) independent of \(\varepsilon\) and a unique solution \(U^{\varepsilon}=(\rho^{\varepsilon},\mathbf{u}^{\varepsilon},\mathbf{F}^{\varepsilon})\in \Lambda^{m}(T)\) to the original problem, such that the following estimate holds for \(t\in \lbrack 0,T]\), \(N_{m}(t)+\left\Vert (\rho^{\varepsilon}-1, \mathbf{u}^{\varepsilon},\mathbf{F}^{\varepsilon}-\mathbb{I}_{2\times 2})(t)\right\Vert_{1,\infty}+\left\Vert \nabla (\rho^{\varepsilon}, \mathbf{u}^{\varepsilon},\mathbf{F}^{\varepsilon})(t)\right\Vert_{1,\infty}\leq C\sigma_{0}\), where \(C>0\) is some constant independent of \(\varepsilon\), and there exists a function \(U^{0}=(\rho^{0},\mathbf{u}^{0}, \mathbf{F}^{0})\in \Lambda^{m}(T)\) satisfying: \[ \lim_{\varepsilon \rightarrow 0}\sup_{t\in \lbrack 0,T]}\left\Vert (U^{\varepsilon}-U^{0},\partial_{y}(U^{\varepsilon}-U^{0}))(.,t)\right\Vert_{L^{\infty}(\mathbb{R}_{+}^{2})}=0, \] and \(U^{0}=(\rho^{0},\mathbf{u}^{0}, \mathbf{F}^{0})\) is a unique solution to the ideal compressible elastodynamic equations with the same initial data \((\rho_{0},\mathbf{u}_{0},\mathbf{F}_{0})\) and the no-slip boundary condition. Here \(\Lambda^{m}(T)=\{(\rho -1,\mathbf{u},\mathbf{F}-\mathbb{I}_{2\times 2})\in \mathcal{ H}_{co}^{m}\), \(\partial_{y}(\rho ,\mathbf{u},\mathbf{F})\in \mathcal{H}_{co}^{m-1}\), \(\partial_{y}^{2}(\rho ,\mathbf{u},\mathbf{F})\in \mathcal{H}_{co}^{m-2}\}\), where \(\mathcal{H}_{co}^{m}([0,t]\times \mathbb{R}_{+}^{2})=\{f:Z^{\alpha}f\in L^{\infty}([0,t],L^{2}(\mathbb{R}_{+}^{2})),\ \left\vert \alpha \right\vert \leq m\}\), the conormal derivative operators \(Z^{\alpha}\) being defined from the operators \(Z_{0}=\partial_{t}\), \(Z_{1}=\partial_{x}\), \(Z_{2}=\varphi (y)\partial_{y}\), through \(Z^{\alpha}=Z_{0}^{\alpha_{0}}Z_{1}^{\alpha_{1}}Z_{2}^{\alpha_{2}}\), with \(\alpha =(\alpha_{0},\alpha_{1},\alpha_{2})\) and \(\left\vert \alpha \right\vert =\alpha_{0}+\alpha_{1}+\alpha_{2}\), the weight function \(\varphi (y)\) satisfying \(\varphi (0)=0\), \(\varphi^{\prime}(0)>0\), \(\left\Vert \partial_{y}^{i}\varphi \right\Vert_{L^{\infty}}\leq C\) (\(i=0,\ldots ,m\) for some integer \(m>0\)), and \(\varphi (y)\) having uniform lower and upper positive bounds away from the physical boundary. The norm \(\left\Vert f(t)\right\Vert_{m}^{2}\) is defined as \(\sum_{\left\vert \alpha \right\vert \leq m}\left\Vert Z^{\alpha}f(t,.)\right\Vert_{L^{2}(\mathbb{R}_{+}^{2})}^{2}\) and \(N_{m}(t)\) is an energy associated to the original problem. For the proof, the authors establish estimates of \(\partial_{y}v^{\varepsilon}\), \(\partial_{y}u^{\varepsilon}\), \(\partial_{y}f_{2}^{\varepsilon}\), \(\partial_{y}p^{\varepsilon}\), \(\partial_{y}f_{3}^{\varepsilon}\), \(\partial_{y}f_{4}^{\varepsilon}\), and \(\partial_{y}f_{1}^{\varepsilon}\), then of the second order normal derivatives.