A note on Schwartz functions and modular forms (Q2187420)

The notorious sphere packing problem asks for the greatest density achieved by a packing of equal-radius spheres in Euclidean space of a given dimension. \textit{H. Cohn} and \textit{N. Elkies} [Ann. Math. (2) 157, No. 2, 689--714 (2003; Zbl 1041.52011)] found linear programming bounds which yield an upper bound to the sphere packing problem given any admissible function satisfying certain inequalities. \textit{M. S. Viazovska} [Ann. Math. (2) 185, No. 3, 991--1015 (2017; Zbl 1373.52025)] constructed such an admissible function for which the Cohn-Elkies bound equals the packing density of the \(E_8\) root lattice, solving the sphere packing problem in dimension 8. By similar methods, [\textit{H. Cohn} et al., Ann. Math. (2) 185, No. 3, 1017--1033 (2017; Zbl 1370.52037)] solved the problem in dimension 24. In this paper the authors investigate the extent to which Viazovska's construction works in other dimensions that are multiples of 8. In particular they develop a general recipe to construct radial Schwartz functions with specified zeros using modular and quasi-modular forms, which recovers the Schwartz functions used in the solution of the sphere packing problem in dimensions 8 and 24. In other dimensions the upper bounds obtained through this argument are weaker than previously known upper bounds. Related results were obtained independently by \textit{A. S. Feigenbaum} et al. [Math. Proc. Camb. Philos. Soc., 2021].