First-hand information for everyone
What the paper shows, in plain terms, is a new bound on how large certain automorphic functions can get. The authors prove a uniform “power‑saving” improvement over the standard or “trivial” bound for the maximum size (the sup‑norm) of minimal weight vectors that live in any cuspidal representation of GL(n,Z)ackslash GL(n,R). The improvement is measured in terms of the archimedean data of the representation, meaning its spectral parameters and the size of its minimal K‑type (the smallest type of symmetry under the compact subgroup). A key technical ingredient is a new set of decay estimates for generalized spherical functions, the kernel functions that appear in the non‑spherical harmonic analysis on GL(n).
The sup‑norm problem asks how big a normalized joint eigenfunction of the natural invariant operators can get, usually in terms of its spectral data. For non‑compact spaces like GL(n,Z)ackslash GL(n,R) one typically fixes a compact region to avoid extremely large peaks near the cusps. Previous high‑rank results mainly treated spherical forms, meaning functions that are invariant on the right under the compact subgroup. This paper drops that restriction. It treats arbitrary K‑types, that is, vector‑valued automorphic forms that transform nontrivially under the compact subgroup, and it gives bounds that are uniform across these more general types.
Technically the authors first develop analytic tools for the non‑spherical spectral transform. They prove a Paley–Wiener type theorem adapted to K‑types, build a spectral localizer, and perform a careful localization analysis of generalized spherical functions. Using these pieces they obtain uniform decay bounds for those generalized spherical functions. Combining these analytic inputs with spectral summation methods yields the main theorems. The general statement (Theorem 1 in the paper) gives a uniform power‑saving over the trivial bound for minimal K‑type cusp forms on any compact set. In a concrete growing‑parameter family (Theorem 2) they obtain a stronger power‑saving in the spectral aspect; symmetric square lifts of holomorphic cusp forms are given as a natural example that fits this family.
A methodological novelty of the paper is how it handles the arithmetic counting that usually drives sup‑norm bounds. Instead of attacking matrix counting by arithmetic (diophantine) methods, the authors reverse the usual pretrace formula and analyze the counting problem by analytic means. This requires applying the pretrace formula more than once, with absolute values and Cauchy–Schwarz steps in between. The authors compare this multi‑directional use of spectral summation to the Deshouillers–Iwaniec treatment of the spectral large sieve.
Why this matters: improved sup‑norm bounds feed into several problems in analytic number theory and arithmetic geometry. Examples include bounds for L‑functions, estimates for the number of nodal domains, equidistribution of zeros, and other arithmetic counting questions. Important caveats are stated in the paper itself. The power saving obtained in full generality is described as very small. The statements are proved for minimal K‑type vectors and under a unitary central character (which the authors may assume by a harmless twist). For clarity of exposition they also assume the representation is not a discrete series, though they say this can be removed. Finally, because of the known cusp behavior in non‑compact spaces, the bounds are given on fixed compact sets rather than on the whole space.