Periodic Hypersurfaces and Lee-Yang Polynomials

Abstract: We prove a Fourier criterion for a $\mathbb{Z}^n$-periodic $C^{1+\epsilon}$-hypersurface $\Sigma\subset\mathbb{R}^n$ to be algebraic in the sense that $\Sigma=\Sigma(p):={x\in\mathbb{R}^n : p(e^{2\pi i x_1},\dots,e^{2\pi i x_n})=0}$ for some polynomial $p$. More precisely, each direction $\ell\in\mathbb{R}^n$ defines a periodic measure $m_{\ell}$ on $\Sigma$ by $dm_{\ell}(x)=|\langle\ell,\hat n(x)\rangle|\,d\sigma(x)$, with $\hat n(x)$ the unit normal to $\Sigma$ and $\sigma$ its $n-1$ dimensional surface measure. Our Fourier criterion for $\Sigma$ is that there exists $\ell\in\mathbb{R}^n$ with Q-linearly independent entries and a closed cone $C\subset\mathbb{R}^n$ for which $\langle\ell,v\rangle>0$ for all nonzero $v\in C$, such that for $k\in\mathbb{Z}^n$ we have $\widehat{m}{\ell}(k)=\int{\Sigma/\mathbb{Z}^n}e^{2\pi i\langle k,x\rangle}\,dm_{\ell}(x)=0$ for every $k\notin C\cup(-C)$. Using Meyers terminology this says $(m_{\ell},C)$ is a "lighthouse". Under this hypothesis one shows $\Sigma=\Sigma(p)$, and after a suitable monomial change $p$ can be taken as a Lee-Yang polynomial. The proof relies on a recent characterization of all one-dimensional Fourier quasicrystals.