No label defined (Q4244040)

In the paper the following abstract framework for inverse problems is considered. Given the space \({\mathcal P}\) of parameters, \({\mathcal D}\) of data, and \({\mathcal U}\) of solutions, the equation \[ \Lambda(p,u)= d \] is considered, where \(\Lambda:{\mathcal P}\times{\mathcal U}\to{\mathcal D}\) is a given operator. The minimization problems the author deals with are \[ \inf\{F_k(Cu)+ G_k(p):p\in{\mathcal P},\;\Lambda(p,u)= d_k\},\leqno{(\text{IP})_k} \] where \(k\) runs in the set of integers, \(C\) is an observation operator from \({\mathcal U}\) into a space of observations \({\mathcal Z}\), and \(F_k:{\mathcal Z}\to \overline{\mathbb{R}}\), \(G_k:{\mathcal P}\to \overline{\mathbb{R}}\) are given mappings. Under suitable assumptions on the convergences \(d_k\to d_\infty\), \(F_k\to F_\infty\), \(G_k\to G_\infty\) (the last one involving the \(\Gamma\)-convergence) it is shown that the sets of solutions of \((\text{IP})_k\) converge to the set of solutions of \((\text{IP})_\infty\) as well as their minimal values. The abstract framework is then applied to first-order systems with accretive operators (Section 4) and to second-order systems (Section 5). In particular, Section 6 contains an example of an inverse problem governed by the one-dimensional wave equation.