Analysis in Hilbert-Kunz theory
Cheng Meng
TL;DR
The paper develops a unifying framework for characteristic $p$ invariants by introducing the multivariate $h$-function and establishing integration formulas that express $h_{R,I,f}$ as RS integrals against kernel functions $D_oldsymbol{}}$, enabling computations across composite polynomials. Building on Han–Monsky representation rings and Teixeira’s $p$-fractal structure, the authors derive both fixed- and limit-characteristic formulas, prove convergence of derivatives, and provide tools to compute limit invariants like $e_{HK}$ and $s$ for families of hypersurfaces. As applications, they prove the Watanabe–Yoshida inequality for all odd primes, characterize strict inequality cases, and obtain asymptotics for Fermat hypersurfaces, alongside explicit $h$-functions for several singularities. These results yield new computational methods and conceptual links between Hilbert–Kunz theory, $F$-signature, and singularity invariants, with potential implications for understanding regularity and singularities in positive characteristic.
Abstract
This paper focuses on a numerical invariant for local rings of characteristic $p$ called $h$-function, that recovers several important invariants, including the Hilbert-Kunz multiplicity, $F$-signature, $F$-threshold, and $F$-signature of pairs. In this paper, we prove some integration formulas for the $h$-function of hypersurfaces defined by polynomials of the form $φ(f_1,\ldots,f_s)$, where $φ$ is a polynomial and $f_i$ are polynomials in independent sets of variables. We demonstrate some applications of these integration formulas, including the following three applications. First, we establish the asymptotic behavior of the Hilbert-Kunz multiplicity for Fermat hypersurfaces of degree 3, extending the degree 2 case previously resolved by Gessel and Monsky. Second, we prove an inequality conjectured by Watanabe and Yoshida holds for all odd primes, generalizing a result of Trivedi. We give a characterization of the cases where the inequality is strict. Third, we generalize an inequality initially established by Caminata, Shideler, Tucker, and Zerman.
