Problem of determining the speed of sound (Q749407)

Let us examine the problem of determining the coefficient \(q(x,y)>0\) in the half-plane \({\mathbb{R}}^ 2_+=\{(x,y)\in {\mathbb{R}}^ 2| y>0\}\) which enters the wave equation \[ (1)\quad u_{tt}=q^ 2(u_{xx}+u_{yy}),\quad (x,y)\in {\mathbb{R}}^ 2_+,\quad t\in {\mathbb{R}}^ 1. \] through the following conditions: the function u(x,y,t) is a generalized solution of the mixed problem for Eq. (1) with boundary conditions \[ (2)\quad u|_{t<0}\equiv 0,\quad u_ y|_{y=0}=g(x)\delta (t)+h(x,t)\theta (t), \] and for \(y=0\) takes on the following values: \[ (3)\quad u|_{y=0}=f(x,t)\theta (t). \] Here \(\delta\) (t) is the Dirac delta function; \(\theta\) (t) is the Heaviside function; g, h, and f are given functions with g(x)\(\neq 0\) for \(| x| \leq s_ 0\), \(s_ 0>0\). We call the pair of functions u,q a solution of the problem if they satisfy conditions (1)-(3). There are many papers both by Soviet and non-Soviet authors devoted to the analysis of problem(1)-(3) and related problems in the case where q does not depend on the variable x. For the case where \(q=q(x,y)\), the literature for the most part contains only the kinematical aspect of the inverse problem, which somewhat differs from (1)-(3) by the presence of an additional parameter, as well as in the form of the boundary conditions. However, the applicability of the kinematic approach is limited to those velocities which correspond to the propagation of sound outside a waveguide. The presence of a waveguide in the medium makes the solution of the inverse kinematic problem nonunique. Thus it is of considerable interest to use the full dynamic information of the same form as in (3). We examine below questions connected with the solvability of the inverse problem (1)-(3). We show that this problem, formulated in somewhat different terms, is locally uniquely solvable on a class of functions analytic in x. We write out an estimate of stability and formulate a theorem concerning the global uniqueness of the solution.