Boundary local integrability of rational functions in two variables
Greg Knese
TL;DR
This work provides a complete local theory for the boundary integrability of rational functions in two variables, focusing on quotients $Q/P$ with $P$ nonvanishing in the bi-upper half-plane but vanishing at the origin. By replacing $P$ with a robust local model $[P]$ and exploiting a two-variable branch-matching framework via the auxiliary polynomial $A+tB$, the authors derive explicit necessary-and-sufficient criteria for $Q/P$ to belong to $L^{\mathfrak{p}}_{loc}$, parameterized by Puiseux-type data $L_j$ and $q_j$. They give precise dimension counts for the relevant local ideals and furnish a constructive basis for the quotient $\mathbb{C}\{x,y\}/(P,\bar P)$, enabling concrete test criteria and computable invariants. Applications include a result that every bounded rational function on the bidisk has derivatives on the torus in $L^1$, and a new, self-contained proof of Kollár’s description of the integrally closed ideal associated to $P$. The approach, rooted in a local model for stable polynomials and refined one- and two-variable analyses, offers a flexible toolkit for addressing other local questions about stable polynomials.
Abstract
Motivated by studying boundary singularities of rational functions in two variables that are analytic on a domain, we investigate local integrability on $\mathbb{R}^2$ near $(0,0)$ of rational functions with denominator non-vanishing in the bi-upper half-plane but with an isolated zero (with respect to $\mathbb{R}^2$) at the origin. Building on work of Bickel-Pascoe-Sola, we give a necessary and sufficient test for membership in a local $L^{p}(\mathbb{R}^2)$ space and we give a complete description of all numerators $Q$ such that $Q/P$ is locally in a given $L^{p}$ space. As applications, we prove that every bounded rational function on the bidisk has partial derivatives belonging to $L^1$ on the two-torus. In addition, we give a new proof of a conjecture, started in Bickel-Knese-Pascoe-Sola and completed by Kollár, characterizing the ideal of $Q$ such that $Q/P$ is locally bounded. A larger takeaway from this work is that a local model for stable polynomials we employ is a flexible tool and may be of use for other local questions about stable polynomials.
