The Gauss circle problem for Penrose tilings
Alan Haynes, Christopher Lutsko
TL;DR
This work resolves a Gauss-circle-type counting problem for the vertex set of unit-length rhombic Penrose tilings by representing Penrose vertices as cut-and-project sets in a 4-dimensional lattice and applying Fourier-analytic Poisson summation with carefully smoothed indicators. The main result shows that, for $R\ge 2$, the number of Penrose-tiling vertices in the disk $B_R$ satisfies $\#(V\cap B_R)=\pi C_P R^2 + O(R^{2/3}(\log R)^{2/3})$, where $C_P=\frac{\phi}{5}\sqrt{10+2\sqrt{5}}\approx1.231$ and $\phi=(1+\sqrt{5})/2$, with the density constant $\mathrm{dens}(\mathcal{L})=\frac{4}{25\sqrt{5}}$. The analysis combines the cut-and-project framework (Minkowski embedding of an algebraic lattice, pentagonal windows) with a delicate, anisotropic smoothing and a careful treatment of Fourier coefficients, especially in directions perpendicular to tiling edges. By bounding the dual-lattice contribution in three regimes and optimizing smoothing parameters, the authors achieve a uniform error bound valid for all non-singular window choices, enabling limits that yield the Penrose tiling result and connecting to the classical Gauss circle problem. The explicit constants arise from window areas and lattice density, reinforcing the link between quasicrystal geometry and lattice-point discrepancy theory, and suggesting extensions to other aperiodic tilings such as Ammann-Beenker tilings.
Abstract
Let $B_R$ denote the closed Euclidean ball of radius $R$ in the plane. In this paper we prove that, if $V$ is the set of vertices of any unit length rhombic Penrose tiling then, for $R\ge 2$, \[\#(V\cap B_R)=πC_P R^2 + O(R^{2/3}(\log R)^{2/3}),\] where $C_P\approx 1.231$ is a constant.
