Inverse boundary problems for polyharmonic operators with unbounded potentials (Q294068)

The authors impose the following data \(\Omega(\subset \mathbb R^n)\) a bounded open set endowed with a smooth boundary \(\partial \Omega\), \(L_{m,V}=(-\Delta)^m+V\), for \(n>2m\), such that \(V\) is a complex-valued potential belonging to \(L_{n/(2m)}(\Omega)\), \((-\Delta)^m\) is the classic polyharmonic operator, and \(f=(f_j)_{0\leq j\leq m-1}\in \prod_{j=0}^{m-1}H^{m-j-1/2}(\partial\Omega)\), where \(H^{\bullet}(\partial\Omega)\) represent Sobolev spaces on \(\partial\Omega\). Next, they recall that the following boundary value problem \[ (BVP):\;L_{m,V}u=0\text{ in }\Omega\text{ and }\Big(\nabla^ju.\nu\Big)_{0\leq j\leq m-1}=f\text{ on }\partial\Omega, \] has a unique weak solution \(u\in H^m(\Omega)\), the Sobolev space of order \(m\) on \(\Omega\), and \(\nabla^j\), \(\nu\) stand for the jth covariant derivative operator and the normal vector, respectively. Then, they associated to \((BVP)\) the Dirichlet-to-Neumann map (D-N), i.e., the following sesquilinear form \[ <\Lambda_Vf,\overline{\gamma v}>=\sum_{|\alpha|=m}\frac{m!}{\alpha!}\int_\Omega D^\alpha u(x)\overline{D^{\alpha}v(x)}dx+\int_\Omega u(x)V(x)\overline{v(x)}dx, \] such that \(v\in H^m(\Omega)\). The main result of the article is to recover in a unique way \(V\) from D-N, which takes the name ``the inverse boundary value problem for \(L_{m,V}\)''. Precisely, the authors state that \(V_1=V_2\) in \(\Omega\) whenever \(\Lambda_{V_1}=\Lambda_{V_2}\) such that zero does not belong to the spectrum associated to the operators \(L_{m,V_1}, L_{m,V_2}\) (Theorem 1.1). The proof is essentially based on Proposition 4.1 (and its associated Remark 4.2) which states that if \(V\) belongs to an explicit proper subspace of \(L_{2/(2m)}(\mathbb R^n)\), then a solution of \(L_{m,V}u=0\) in \(\mathbb R^n\) is given as \(u(x)=\exp(x.\zeta)(1+r(x))\) with \(\zeta\in \mathbb C^n\) such that \(\zeta.\zeta=0\) which it comes back to seek \(r\) such that \(P_{\zeta,V}r=-V\) in \(\mathbb R^n\) where \(P_{\zeta,V}\) represents the operator \((-\Delta-2\zeta.\nabla)^m+V\). In the proof of Proposition 4.1, the authors look for \(r\) in terms of \(V\) and of \(G_\zeta^{(m)}\), the convolution of a tempered fundamental solution, of \(P_{\zeta,0}\), by \(f\), a function belonging to \(L_{2,\sigma}(\mathbb R^n)=L_2(\mathbb R^n,(1+|x|^2)^\sigma dx)\) where \(\sigma\) is a real number, and \(P_{\zeta,0}G_\zeta^{(m)}f=f\) in \(\mathbb R^n\) (Proposition 3.4). In the matter of the behaviour of \(r\) when \(|\zeta|\) is close to infinity, the authors state that the \(L_{2n/(n-2m)}(\mathbb R^n)\)-norm of \(r\) is \(O(1)\) (Proposition 4.1), the proof is based on the assessment of \(G_\zeta^{(m)}\) in adapted \(L_p\) spaces (Proposition 3.12), the proof of this latter proposition is based on using a uniform Carleman-type inequality (Proposition 2.2). Also, the authors state that for \(K\), a compact set of \(\mathbb R^n\), the \(L_2(K)\)-norm of \(r\) tends to zero whenever \(|\zeta|\) is close to infinity (Proposition 4.1), the proof also relies on an estimation of \(G_\zeta^{(m)}\) (Proposition 3.11).