Stability of an inverse problem for the discrete wave equation and convergence results (Q2344558): Difference between revisions

Let \(\Omega\) be a smooth bounded domain of \(\mathbb R^d\). Consider the inverse problem of recovering the potential \(q=q(x)\) in the wave equation \(\partial_{tt}y-\Delta y+qy=f\) in \((0,T) \times \Omega\); \(y=f_{\partial}\) on \((0,T) \times \partial\Omega\); \(y(0,\cdot)=y^0, \partial_t y(0,\cdot)=y^1\) in \(\Omega\), from additional knowledge of the flux \({\mathcal M}[q]=\partial_{\nu}y[q]\) on \((0,T) \times \Gamma_0\). Here, \(y=y(t,x)=y[q](t,x)\), \(\Gamma_0 \subset \partial \Omega\) is a non-empty open subset and \(\nu\) denotes the unit outward normal vector on \(\partial \Omega\). The authors are motivated by Lipschitz and logarithmic type stability estimates \[ C^{-1} \| q^a-q^b\|_{L^2(\Omega)} \leq \| {\mathcal M}[q^a]- {\mathcal M}[q^b] \|_{H^1(0,T;L^2(\Gamma_0))} \leq C \| q^a-q^b\|_{L^2(\Omega)} \] and \[ \| q^a-q^b\|_{L^2(\Omega)} \leq C[\log(2+C \| {\mathcal M}[q^a]- {\mathcal M}[q^b] \|_{H^1(0,T;L^2(\Gamma_0))}^{-1} )]^{-1/(1+\alpha)}, \] \(\alpha>0\), recently [\textit{L. Baudouin} and \textit{S. Ervedoza}, SIAM J. Control Optim. 51, No. 1, 556--598 (2013; Zbl 1264.35279)] obtained for the original problem [\textit{L. F. Ho}, C. R. Acad. Sci., Paris, Sér. I 302, 443--446 (1986; Zbl 0598.35060)] under corresponding assumptions on \(\Gamma_0\) . The goal of the paper is to derive counterparts of these estimates for finite-difference space approximations of the wave equation discretized on a uniform mesh. The discrete stability results, when compared with their continuous counterparts, include new terms depending on the discretization parameter \(h\). From these stability results, the authors design a numerical method to compute convergent approximations of the continuous potential.