The purpose of this proposal is to attack one of the most fundamental inverse problems, called the Calderón problem, from multiple angles. The Calderón problem models recovery of information from indirect observations. Indirect observations are characteristic, for example, in medical imaging, exploration geophysics, and non-destructive testing. The Lorentzian Calderón problem gives a mathematical model that captures many essential features of such physical problems when traveling waves are used. Its sister, the elliptic, or Riemannian, version of the Calderón problem models probing with fixed frequency waves.
The overall objectives are
1. Solve the Lorentzian Calderón problem without curvature bounds
2. Solve inverse problems for nonlinear elliptic equations in non-product geometries
3. Construct counterexamples to the Lorentzian Calderón problem
More specific goals, as formulated in B2 part of the proposal, are
A. Solve the Lorentzian Calderón problem in geometries without null cut points
B. Solve inverse problems for nonlinear elliptic equations in near-Euclidean geometries
C. Find a counterexample to the Lorentzian Calderón problem with full data.
From the geometrical point of view, it could be said that the theory of inverse problems is mature in the case of nonlinear hyperbolic equations, while it is still in its youth in the case of non-linear elliptic and linear hyperbolic equations, the Lorentzian Calderón problem being a prototype of the latter case. The disparity between these cases can be explained heuristically by saying that the richer is the set of solutions to a partial differential equation, the easier is the associated inverse problem. A guiding principle in the project is to proceed from richer solution sets to poorer ones. We hope that advances in relation to the objectives 1 and 2 will eventually help us to make progress in the context of the Riemannian Calderón problem, whose the solution set is poorer than those for the linear hyperbolic and non-linear elliptic problems. The Riemannian problem has remained open for more than 30 years, while the Lorentzian version generalizes the even more classical inverse problem studied by Gelfand and Levitan.
