Abstract
We find two series expansions for Legendre's second incomplete elliptic integral $E(\lambda, k)$ in terms of recursively computed elementary functions. Both expansions converge at every point of the unit square in the $(\lambda, k)$ plane. Partial sums of the proposed expansions form a sequence of approximations to $E(\lambda, k)$ which are asymptotic when $\lambda$ and/or $k$ tend to unity, including when both approach the logarithmic singularity $\lambda=k=1$ from any direction. Explicit two-sided error bounds are given at each approximation order. These bounds yield a sequence of increasingly precise asymptotically correct two-sided inequalities for $E(\lambda, k)$. For the reader's convenience we further present explicit expressions for low-order approximations and numerical examples to illustrate their accuracy. Our derivations are based on series rearrangements, hypergeometric summation algorithms and extensive use of the properties of the generalized hypergeometric functions including some recent inequalities.
Introduction
Elliptic integrals are fundamental tools in classical physics and engineering, appearing in contexts ranging from pendulum motion to electromagnetic field calculations. However, the incomplete elliptic integral of the second kind, $E(\lambda, k)$, presents significant computational challenges as the amplitude $\lambda$ and modulus $k$ both approach unity. This region, known as a logarithmic singularity, often causes existing asymptotic expansions to suffer from convergence issues or directional sensitivity.
This paper addresses these limitations by developing two new series expansions for $E(\lambda, k)$ that converge across the entire unit square. These expansions are specifically designed to remain highly accurate near the $(1,1)$ singularity, providing explicit two-sided error bounds that outperform existing approximations in the literature.
Methodology
The derivation of these expansions utilizes a combination of classical analysis and modern symbolic computation. The authors represent the integral using Appell’s hypergeometric function $F_1$ and the Gauss hypergeometric function ${}_2F_1$. By rearranging known expansions from Byrd-Friedman and Carlson, they derived two distinct types of convergent series.
To manage the complexity of the terms, the researchers employed the HolonomicFunctions.m package in Mathematica and specialized commands in Maple to derive linear recurrence relations. This allowed the expansions to be expressed in terms of recursively computed elementary functions. Furthermore, advancements in inequality theory for generalized hypergeometric functions (specifically ${}_3F_2$ and ${}_4F_3$) were used to establish rigorous two-sided error bounds.
Key Insights and Results
The primary contribution of this work is the achievement of global convergence within the unit square, coupled with a "sandwich" effect provided by explicit upper and lower bounds for the series remainders. This allows for a two-sided inequality where the potential error is precisely known.
Numerical experiments demonstrated that these new methods are significantly more accurate than standard benchmarks like the Carlson-Gustafson and López approximations. For instance, second-order refined approximations—created by taking a weighted average of the bounds—achieved relative errors as low as $10^{-11}$ near the singularity (e.g., $\lambda=0.99, k=0.999$). Crucially, these approximations remain robust regardless of the path taken toward the $(1,1)$ singularity.
Conclusion and Future Work
While the proposed expansions offer unprecedented precision and computational efficiency through recursive implementation, some areas for further research remain. The authors identified an "optimal" weight $\delta \approx 67/187$ for refined approximations through numerical testing, but finding the exact value that minimizes error across the entire domain remains an open problem. Additionally, while the expansions are theoretically valid for complex variables, future work could extend the numerical verification beyond the real unit square to the wider complex plane.
Summary updated