Multidimensional inverse problems: Uniqueness theorems (Q749782): Difference between revisions

A method is given to prove uniqueness theorems for a number of multidimensional inverse problems. Applications of the method give the following results: (1) The surface data u(x,y,k) known for all \(x,y\in P=\{x:\) \(x_ 3=0\}\) and a fixed \(k>0\) determine \(v(x)\in L^ 2(D)\) uniquely. Here \([\nabla^ 2+k^ 2+k^ 2v(x)]u=-\delta (x-y)\) in \({\mathbb{R}}^ 3\), \(D\subset {\mathbb{R}}^ 3_ -=\{x:\) \(x_ 3<0\}\), D is a bounded domain throughout, \(v(x)=0\) outside D, \(v=\bar v.\) (2) The surface data u(x,y,k) known for all x,y\(\in P\) and at \(k=k_ j\), \(j=1,2\), \(k_ 1\neq k_ 2\), \(k_ j>0\), determine \(a_ j(x)\), \(j=1,2\) uniquely. Here \(a_ 1\in L^ 2(D)\), \(a_ 2\in W^{2,2}(D),\) \(\nabla^ 2u+k^ 2u+k^ 2a_ 1(x)u+\nabla \cdot [a_ 2(x)\nabla u]=- \delta (x-y)\) in \({\mathbb{R}}^ 3\), D is the same as in 1), \(1+a_ 2>0\), \(a_ 1=\bar a_ 1\), \(W^{p,\ell}(D)\) are the Sobolev spaces. (3) The data \(\{\) f,\(\sigma\) \(h\}\), where f runs through all of \(C^ 1(\Gamma)\), determine \(\sigma (x)\in W^{2,2}(D)\) uniquely. Here \(\Gamma =\partial D\) is smooth, \(\nabla \cdot (\sigma (x)\nabla u)=0\) in D, \(u=f\), \(u_ N=h\) on \(\Gamma\), N is the outer unit normal to \(\Gamma\). (4) The scattering amplitude \(A(\theta ',\theta,k)\) known for all \(\theta ',\theta \in S^ 2\) and a fixed \(k>0\) determines the potential \(q\in L^ 2(D)\) uniquely. Here \(D\subset {\mathbb{R}}^ 3\), \(q(x)=\bar q(x)\), the bar stands for complex conjugate, and \(q=0\) outside D. (5) A necessary and sufficient condition for an operator \(Lu=\sum_{| \alpha | \leq \ell}a_{\alpha}\partial^{\alpha}u(x)\), \(x\in {\mathbb{R}}^ n\), \(a_{\alpha}=\bar a_{\alpha}=const\), to have the property C is given. The property C is defined as follows. Let \(D\subset {\mathbb{R}}^ n\) be an arbitrary bounded domain and \(f\in L^ p(D)\), \(P\geq 1\). Suppose that (*) \(\int_{D}f(x)\psi_ j(x)\psi_{j'}(x)dx=0\forall \psi_ j,\psi_{j'}\in N_ D(L),\) where \(N_ D(L)=\{u:\) \(Lu=0\) in \(D\}\). If (*) implies that \(f=0\) then we say that L has property C (completeness of the set of products of solutions to the equation \(Lu=0)\). We give sufficient conditions for \(L=\nabla^ 2+k^ 2-q(x)\) to have property C and use it to prove uniqueness theorems in 1)-4). Property C can be defined for a pair \((L,L_ 1)\) in which case \(\psi_{j'}\) in (*) should be substituted by \(\phi_{j'}\in N_ D(L_ 1)\).