Uniqueness theorems for multidimensional inverse problems with unbounded coefficients (Q1122712)

The author proves five uniqueness results for inverse problems related to elliptic equations. All of them rely on the following Proposition. For any \(k\geq 0\) and any function \(q\in L^ 2({\mathbb{R}}^ 3)\) with Supp \(q\subset D\), D being a bounded domain, then the equation \[ (1)\quad [\Delta^ 2+k^ 2-q(x)]u=0\quad in\quad {\mathbb{R}}^ 3 \] admits a solution of the form \(u(x,z)=\exp (iz\cdot x)[1+R(x,z)]\) \((z\in {\mathbb{C}}^ 3\), \(z\cdot z=k^ 2)\) such that \(\| R(\cdot,z)\|_{L^{\infty}(D)}\leq c| z|^{-1/2} \ell n| z|\) as \(z\to \infty\), \(z\cdot z=k^ 2\), Im \(z\neq 0.\) The first uniqueness result concerns the recovering of the potential \(q\in L^ 2({\mathbb{R}}^ 3)\) (with Supp \(q\subset D)\) in (1) under the assumption that the scattering amplitude \[ (2)\quad A(\theta ',\theta,k)=-(4\pi)^{-1}\int_{{\mathbb{R}}^ 3}q(x)\exp (-ik\theta '\cdot x)u(\theta,k,x)dx \] is known. In (2) \(\theta,\theta '\in S^ 2\) (the unit sphere in \({\mathbb{R}}^ 3)\) and \(u(\theta,k,x)=\exp (ik\theta \cdot x)+v(\theta,k,x).\) A second uniqueness result for the potential q in (1) is proved when the following additional information is prescribed: \[ (3)\quad u=f\quad on\quad \partial D,\quad \partial u/\partial N=g\quad on\quad \partial D, \] for any f running through \(C^ 1(\partial D).\) A similar result for equation (1) with left-hand side substituted by - \(\delta\) (x-y) and \(D\subset {\mathbb{R}}^ 3_-=\{x\in {\mathbb{R}}^ 3:\quad x_ 3<0\}\) is obtained when we are given the additional information \(u(x,y)=\phi (x,y),\) \(x,y\in {\mathbb{R}}^ 3_-.\) A further uniqueness result is related to the identification of the unknown coefficient \(\sigma \in H^ 2(\Omega)\) in the overdetermined boundary value problem \[ (4)\quad \nabla \cdot (\sigma (x)\nabla u)=0\quad in\quad D,\quad u=f\quad on\quad \partial D,\quad \partial u/\partial N=g\quad on\quad \partial D. \] Here \(D\subset {\mathbb{R}}^ 3\) is a bounded domain with a smooth boundary \(\partial D\), N is the outward normal versor to \(\partial D\), \(\sigma (x)\geq C>0\) \(\forall x\in {\bar \Omega}\) and f runs through \(C^ 1(\partial \Omega)\). The last uniqueness result is concerned with the determination of the pair \((a_ 1,a_ 2)\) in the overdetermined problem (5) \(\Delta_ xu+k^ 2_ ja_ 1(x)u+\nabla_ x\cdot (a_ 2(x)\nabla_ xu)=-\delta (x-y)\) in \({\mathbb{R}}^ 3\), \(j=1,2,\) (6) \(u(x,y,k_ j)=\phi_ j(x,y)\), \(x,y\in \partial {\mathbb{R}}^ 3_-\), \(j=1,2,\) where \(0<k_ 1<k_ 2\), \(0<1+a_ 2(x)\) \(\forall x\in {\mathbb{R}}^ 3\), Supp \(a_ j\subset D\) \((j=1,2)\), D being a bounded domain.