Inverse problem for one-dimensional wave equation with matrix potential (Q2422502)

Under consideration is the inverse problem of recovering the matrix potential \(a(x,t)\) in the system of wave equations \[ \Box u- a(x,t)u=0\ (x>0,t>0),\ \ u(x,0)=0,\ u_{t}(x,0)=0,\ u(0,t)=f(t). \] using additional measurements of the response \(u_{x}(0,t)=g(t)\) for given \(f(t)\). Here \(u(x,t)\) is a vector of length \(n\). Let \(\Omega=\{(x,t):\ t>x>0\}\). Given a collection of the data \(f_{j}\), \(j=1,2,\ldots, n\) and corresponding responses \(g_{j}\). It is proved that if \[ \begin{gathered} \Box u_{ij}- a_{i}(x,t)u_{ij}=0\ ((x,t)\in \Omega), \ u_{ij}(x,0)=0,\ u_{ijt}(x,0)=0,\\ u_{1j}(0,t)=u_{2j}(0,t)=f_{j}(t), \ a_{1t}(x,t)=a_{2t}(x,t),\ \\ u_{1jx}(0,t)=u_{2jx}(0,t)=g_{j}(t), u_{1}(x.x)=u_{2}(x,x), det f(0)\neq 0, \end{gathered} \] where \(j=1,2,\ldots,n\) and \(f\) is the matrix with the columns \(f_{j}\), then \(a_{1}=a_{2}\) and \(u_{1j}=u_{2j}\) for all \(j=1,2,\ldots,n\). The proof relies on Carleman estimates and solutions \(u(x,t)\) as well as the matrices \(a(x,t)\) and the functions \(f_{j},g_{j}\) are assumed to be sufficiently smooth.