Stability for the inverse random source scattering problems in elastic waves

Inverse problems for PDEs (35R30) PDEs with randomness, stochastic partial differential equations (35R60) Stochastic partial differential equations (aspects of stochastic analysis) (60H15) Inverse problems (including inverse scattering) in optics and electromagnetic theory (78A46)

Review text

The authors study the elastic wave equation in a homogeneous, isotropic, and normalized medium, modeled by the Navier equation with an external source \(\mathbf{f}\) supported in a bounded domain \(D\subset\mathbb{R}^{3}\):\N\[\N\Delta^{*}\mathbf{u} + \omega^{2}\mathbf{u} = \mathbf{f} \quad \text{in }\mathbb{R}^{3},\N\]\Nwhere \(\Delta^{*} = \mu\Delta + (\lambda + \mu)\nabla\nabla\cdot\) and \(\mathbf{u}:\mathbb{R}^{3}\rightarrow\mathbb{R}^{3}\) denotes the vector-valued displacement field satisfying the Kupradze radiation condition at infinity. Here, \(\omega>0\) denotes the angular frequency and \((\lambda,\mu)\) are Lamé constants, with \(\mu>0\) and \(2\lambda+3\mu>0\).\N\NIn Section~2, the authors study the inverse problem of recovering the ``unknown strength \(\mathbb{M}\) from the \textit{multi-frequency} near-field measurements\N\[\N\{\mathbf{u}|_{\partial B_{R}} , \left( \mu\partial_{\nu}\mathbf{u} + (\lambda+\mu)(\nabla\cdot \mathbf{u})\nu \right)|_{\partial B_{R}} : \omega \ge \omega_{0} \}\N\]\Nfor some \(\omega_{0}>0\), where \(\nu\) denotes the unit normal vector on \(\partial B_{R}\). In Theorem~2.1, the authors show that the result holds under the following assumption, providing an explicit formula; a corresponding stability estimate is given in Theorem~2.7:\N\NAssumption~(A). The random source \(\mathbf{f}\) is a real-valued microlocally isotropic Gaussian random function of order \(-m\) with zero expectation, where \(m>2\). Let \(M>0\) be a constant. The covariance operator has the matrix symbol \(\mathbb{C}(x,\xi)=[c_{ij}]_{1\le i,j\le 3}\) with the principal symbol \(\mathbb{M}(x)|\xi|^{-m}\) satisfying\N\begin{itemize}\N\item [(i)] \(m_{ij}(x) \in C_{0}^{\infty}(D)\), \(|m_{ij}(x)|\le M\) for \(x\in D\);\N\item [(ii)] \(|c_{ij}(x,\xi)| \le M(1+|\xi|)^{-m}\) for \(\xi\in\mathbb{R}^{3}\) and \(x\in D\);\N\item [(iii)] \(|\mathbb{A}(x,\xi)| \le M|\xi|^{-(m+1)}\) for \(|\xi|\ge 1\) and \(x\in D\), where \(\mathbb{A}(x,\xi)=\mathbb{C}(x,\xi) - \mathbb{M}(x)|\xi|^{-m}\).\N\end{itemize}\N\NI would like to emphasize that the concept of an ``incident field does not apply in this manuscript. The notation \(\mathbb{u}^{\mathrm{inc}}\) used by the authors is misleading, as it merely denotes a field satisfying the Navier equation \(\Delta^{*}\mathbb{u}^{\mathrm{inc}} + \omega^{2}\mathbb{u}^{\mathrm{inc}}=0\) in \(\mathbb{R}^{3}\), and cannot satisfy the Kupradze radiation condition unless it is trivial. I would also like to point out that the measurement \(\left( \mu\partial_{\nu}\mathbf{u} + (\lambda+\mu)(\nabla\cdot \mathbf{u})\nu \right)|_{\partial B_{R}}\) is referred to as the traction.\N\NIn Section~3, the authors assume that the random source \(\mathbf{f}\) is a random function driven by an additive white noise which takes the form\N\[\N\mathbf{f} = \sigma \dot{\mathbf{W}}_{x},\N\]\Nwhere \(\sigma(x) = \mathrm{diag}\,(\sqrt{\sigma_{1}(x)},\sqrt{\sigma_{2}(x)},\sqrt{\sigma_{3}(x)})\) is a deterministic diagonal matrix function with \(\sigma_{j}\ge 0\) supported on \(D\), and \(\mathbf{W}(x)=(W_{1}(x),W_{2}(x),W_{3}(x))^{\intercal}\) is a three-dimensional three-parameter Brownian sheet where \(W_{1}\), \(W_{2}\) and \(W_{3}\) are three independent one-dimensional three-parameter Brownian sheets. They study the inverse problem of determining the standard deviation \(\sigma\) of \(\mathbf{f}\) from \textit{mono-frequency} near-field measurements\N\[\N\{\mathbf{u}|_{\partial B_{R}} , \left( \mu\partial_{\nu}\mathbf{u} + (\lambda+\mu)(\nabla\cdot \mathbf{u})\nu \right)|_{\partial B_{R}} \}\N\]\Nfor some given fixed \(\omega>0\). A stability estimate is given in Theorem~3.2.

Cites work

Cited in

(2)

This page was built for publication: Stability for the inverse random source scattering problems in elastic waves

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6962711)