Uniqueness properties for discrete equations and Carleman estimates (Q524266)

In the continuum setting, the uncertainty principle is well-known and the classical proof is based on complex analysis arguments from solutions of the free Schrödinger equations. In more recent papers, the uncertainty principle was proved by using real variable methods, and the proof was extended to the solutions of the linear Schrödinger equation with a bounded potential term. The main technique in the proof is the log-convexity property for solutions with Gaussian decay and Carleman estimates. In the discrete setting, the first author has developed the complex analysis arguments that suggest that the discrete version of the Gaussian is the product of modified Bessel functions. Also log-convexity properties were proved for solutions of the discrete Schrödinger equation with discrete Gaussian decay. In this paper, the authors try a different approach in order to improve the previous results. The use of the log-convexity properties and Carleman estimates, allowed them to deduce the uniqueness result for the linear discrete Schrödinger equation with a bounded discrete potential. In the one-dimensional setting, the result is sharp and improves the previous work of Jaming-Lyubarskii-Malinnikova-Perfekt. In higher dimensions. the result is not sharp. Using Carleman estimates, the authors give a lower bound for solutions to the linear discrete Schrödinger equation in both dynamic and stationary settings. This allowed them to prove uniqueness results, under some assumptions on the decay of the solutions.