Remarks on the determination of the Lorentzian metric by the lengths of geodesics or null-geodesics
Gregory Eskin
TL;DR
Problem: determine the Lorentzian metric on $\mathbb{R}\times\mathbb{R}^n$ from lengths of space-time geodesics. Approach: formulate the problem in a Hamiltonian framework with energy $H$ and derive that the geodesic length data $R(g,T,\hat y,\hat\eta)$ is tied to $H$ via $R=\sqrt{2H(y,\eta_0,\eta)}\,T$, then develop perturbation estimates for null-geodesics to compare nearby metrics. Main results: (i) a pointwise rigidity theorem showing equality of geodesic lengths fixes the metric at the base point (and boundary derivatives) when metrics are close, and (ii) a corresponding rigidity result for null-geodesics showing equality of Euclidean lengths implies equality of the underlying Hamiltonians; Significance: the work advances inverse boundary problems in Lorentzian geometry by proving local uniqueness from geodesic-length data.
Abstract
We consider a Lorentzian metric in $\mathbb{R}\times\mathbb{R}^n$. We show that if we know the lengths of the space-time geodesics starting at $(0,y,η)$ when $t=0$, then we can recover the metric at $y$. We prove the rigidity of Lorentzian metrics. We also prove a variant of the rigidity property for the case of null-geodesics: if two metrics are close and if corresponding null-geodesics have equal Euclidian lengths then the metrics are equal.
