Riemann-Hilbert Characterisation of Rational Functions with a General Distribution of Poles on the Extended Real Line Orthogonal with Respect to Varying Exponential Weights: Multi-Point Pad\'e Approximants and Asymptotics

TL;DR

This paper develops a Riemann-Hilbert framework for analyzing orthogonal rational functions with arbitrary poles on the extended real line, deriving asymptotics and approximation results for multi-point Padé approximants using complex analysis and variational methods.

Contribution

It introduces a novel Riemann-Hilbert characterization of orthogonal rational functions with arbitrary poles and applies steepest descent techniques to derive their asymptotics and approximation properties.

Findings

Derived uniform asymptotics of ORFs and their coefficients.

Established existence and regularity of equilibrium measures.

Obtained asymptotic error estimates for multi-point Padé approximants.

Abstract

Given $K$ arbitrary poles, which are neither necessarily distinct nor bounded, on the extended real line, a corresponding ordered base of rational functions orthogonal with respect to varying exponential weights is constructed: this gives rise to a $K$-fold family of orthogonal rational functions (ORFs). The ORF problem is characterised as a family of $K$ matrix Riemann-Hilbert problems (RHPs) on the extended real line, and a corresponding family of $K$ energy minimisation (variational) problems containing external fields with singular points is formulated, and the existence, uniqueness, and regularity properties of the associated family of equilibrium measures is established. The family of $K$ equilibrium measures is used to derive a family of $K$ model matrix RHPs on the extended real line that are amenable to asymptotic analysis via the Deift-Zhou non-linear steepest-descent method: this is used to derive uniform asymptotics, in a certain double-scaling limit, of the ORFs and their leading coefficients, as well as related, important objects, in the entire complex plane. A family of $K$ multi-point Pad\'e approximants (MPAs) for the Markov-Stieltjes transform is also presented, and uniform asymptotics, in a certain double-scaling limit, are obtained for the corresponding MPAs and their associated errors in approximation (MPA error terms) in the entire complex plane.