Inverse problems for the zeros of the Wigner function

Abstract

In this work we consider the inverse problem of determining the properties of a Wigner function from the set of its zeros (the nodal set). The previous state of the art of the problem is Hudson's theorem, which shows that an empty nodal set is associated only with generalized Gaussians. We extend this analysis to non-Gaussian functions. Our first main result states that, if the nodal set of the Wigner distribution of a function $f$ is bounded, then $f$ is equal to a generalized Gaussian times a polynomial. An immediate consequence of this result is that any open set is a uniqueness set for Wigner functions with bounded nodal set. Our second main result shows that the only Wigner function vanishing on a circle of radius $\sqrt{\hbar/2}$ and centered at the origin is the Wigner distribution of the first Hermite function. We prove similar results for the second and third Hermite functions. We also derive for Wigner functions a counterpart of the sign uncertainty principle of J. Bourgain, L. Clozel and J.-P. Kahane, which says that if the negative part of a Wigner function is contained in a ball, then the radius of the ball has a lower bound. Finally, we obtain various constraints on Wigner distributions whose bounded nodal sets contain circles, ellipses or line segments. As a by-product of our work we prove several non-trivial results about the zeros of Laguerre polynomials.