An Inverse Problem with Partial Neumann Data and $L^{n/2}$ Potentials
Leonard Busch, Leo Tzou
TL;DR
The paper proves uniqueness for the inverse Schrödinger problem with partial Neumann data for potentials in $L^{n/2}(\Omega)$. It develops an explicit Green's function and CGO solutions by constructing a global parametrix from small- and large-frequency components, together with boundary-correcting Poisson kernels, all in a flat, semiclassical framework. This approach yields a robust integral identity from CGO solutions, leading to $q_+=q_-$ under partial data and extending Calderón-type results to partial Neumann data. The methodology relies on boundary flattening, semiclassical pseudodifferential calculus, and careful Green's function construction, offering a direct route to uniqueness without Carleman estimates. The results have potential implications for practical inverse boundary problems where measurements are restricted to parts of the boundary and the conductivity is rough.
Abstract
We consider a partial data inverse problem with unbounded potentials. Rather than rely on functional analytic arguments or Carleman estimates, we construct an explicit Green's function with which we construct complex geometric optics (CGO) solutions and show unique determinability of potentials in $L^{n/2}$ for the Schrödinger equation with partial Neumann data.
