The Fuglede spectral conjecture holds for convex planar domains (Q1429362)

A bounded open domain \(\Omega\) in the Euclidean space is called \textit{spectral} if the Hilbert space \(L^2(\Omega)\) admits a basis whose elements are exponentials, i.e., functions of the form \(e_\lambda(x) := e^{2\pi i \lambda\cdot x}\), for some collection \(\Lambda\) of \(d\)-dimensional frequency vectors, called a \textit{spectrum}. A domain is said to \textit{tile} the Euclidean space \textit{by translations} with a set \(T\) of vectors if its translates by \(T\) are overlapping at most on Lebesgue measure \(0\) and almost all (Lebesgue) of the Euclidean space is covered by these translates. Fuglede's conjecture states that the class of sets which tile the Euclidean space is the same as the class of spectral domains. This conjecture is now known to be false in this generality. It is however true in special cases. For instance, sets which tile a space by a \textit{lattice} are known to be the same as sets which possess a lattice spectrum (this is due to Fuglede himself). It is also true that if a convex set tiles, then it also tiles by a lattice, hence it is spectral by the result of the previous sentence. It remains an open question whether any spectral convex domain is also a tile. In this paper the authors prove that this is the case in dimension \(d=2\), hence the only convex spectral domains in the plane are parallelograms and some hexagons. As is usual in this field, the proof uses Fourier Analysis.