A bilinear strategy for Calderón's problem (Q1982542)

The paper deals with the Calderón problem, which is the problem to determine the conductivity from the boundary data. More precisely, let \(\Omega \subset \mathbb{R}^d\) and consider the problem \[ \mathrm{div}(\gamma \nabla u)=0\, ,\qquad \mathrm{in}\ \Omega\, , \] with \[ u_{|\partial \Omega}=f\, , \] where \(\gamma\) is the conductivity and \(f\) is the potential at the boundary. To the problem above, one associates the Dirichlet-to-Neumann map \(\Lambda_\gamma\) which takes the trace \(u_{|\partial \Omega}\) of a function \(u\) such that \(\mathrm{div}(\gamma \nabla u)=0\) to \(\gamma \partial_\nu u\), where \(\partial_\nu u\) is the normal derivative. The Calderón problem can be then reformulated as the issue of determining \(\gamma\) from the knowledge of \(\Lambda_\gamma\). The author proves that, for \(d= 5,6\), the conductivity \(\gamma\) is uniquely determined by \(\Lambda_\gamma\) when \(\gamma\) is in \(W^{1+(d-5)/(2p)+,p}(\Omega)\), for \(d \leq p <\infty\). The proof is based on an extension of Tao's bilinear theorem.