Numerical evaluation of the Gauss hypergeometric function by power summations
HTML articles powered by AMS MathViewer
- by Jurgen A. Doornik PDF
- Math. Comp. 84 (2015), 1813-1833 Request permission
Abstract:
Numerical evaluation of the Gauss hypergeometric function ${}_2F_1(a,b;c;z)$, with complex parameters $a,b,c$ and complex argument $z$ is notoriously difficult. Carrying out the summation that defines the function may fail, even for moderate values of $z$. Formulae are available to transform the effective argument in the series, potentially leading to a numerically successful summation. Unfortunately, these transformations have a singularity when $b-a$ or $c-a-b$ is an integer, and suffer numerical instability near that. This singularity has to be removed analytically after collecting powers in $z$.
The contributions in this paper are fourfold. First, analytical expressions are provided that remove the singularity from Bühring’s $1/(z-z_0)$ transformation. This is more difficult, because the singularity occurs twice, and it is necessary to collect powers of $z_0$, as well as $z$. The resulting expression has a three-term recursion, like the original. Next, improved expressions are derived for the cases that have been addressed before. We study a transformation that converges outside $|z-0.32| > 0.32$ for ${\mathcal {R}}z>0$, which is tighter than the $|z-0.5| > 0.5$ which is normally considered. Finally, we derive an improved algorithm for the numerical evaluation of ${}_2F_1$.
References
- M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions, Dover Publications Inc., New York, 1970.
- Wolfgang Bühring, An analytic continuation of the hypergeometric series, SIAM J. Math. Anal. 18 (1987), no. 3, 884–889. MR 883575, DOI 10.1137/0518066
- J. A. Doornik, Conversion of high-period random numbers to floating point, ACM Transactions on Modeling and Computer Simulation 17 (2007).
- —, Object-oriented Matrix Programming using Ox, 7th ed., Timberlake Consultants Press, London, 2013.
- Robert C. Forrey, Computing the hypergeometric function, J. Comput. Phys. 137 (1997), no. 1, 79–100. MR 1481885, DOI 10.1006/jcph.1997.5794
- Amparo Gil, Javier Segura, and Nico M. Temme, The ABC of hyper recursions, J. Comput. Appl. Math. 190 (2006), no. 1-2, 270–286. MR 2209508, DOI 10.1016/j.cam.2005.01.041
- Amparo Gil, Javier Segura, and Nico M. Temme, Numerically satisfactory solutions of hypergeometric recursions, Math. Comp. 76 (2007), no. 259, 1449–1468. MR 2299782, DOI 10.1090/S0025-5718-07-01918-7
- José L. López and Nico M. Temme, New series expansions of the Gauss hypergeometric function, Adv. Comput. Math. 39 (2013), no. 2, 349–365. MR 3082518, DOI 10.1007/s10444-012-9283-y
- N. Michel and M. V. Stoitsov, Fast computation of the Gauss hypergeometric function with all its parameters complex with application to the Pöschl-Teller-Ginocchio potential wave functions, Comput. Phys. Comm. 178 (2008), no. 7, 535–551. MR 2585243, DOI 10.1016/j.cpc.2007.11.007
- A. B. Olde Daalhuis, Hypergeometric function, NIST Handbook of Mathematical Functions, Cambridge University Press, New York, 2010.
- S. L. Skorokhodov, A regularization method for computing the hypergeometric function $F(a,b;c;z)$ in a neighborhood of the singular points $z=1$ and $z=\infty$, Zh. Vychisl. Mat. Mat. Fiz. 41 (2001), no. 12, 1808–1832 (Russian, with Russian summary); English transl., Comput. Math. Math. Phys. 41 (2001), no. 12, 1718–1741. MR 1882970
- Nico M. Temme, Numerical aspects of special functions, Acta Numer. 16 (2007), 379–478. MR 2417932, DOI 10.1017/S0962492906330012
Additional Information
- Jurgen A. Doornik
- Affiliation: Institute for New Economic Thinking at the Oxford Martin School, University of Oxford
- Email: jurgen.doornik@nuffield.ox.ac.uk
- Received by editor(s): May 1, 2013
- Received by editor(s) in revised form: September 25, 2013, and October 25, 2013
- Published electronically: December 3, 2014
- Additional Notes: This research was supported in part by grants from the Open Society Foundations and the Oxford Martin School.
- © Copyright 2014 American Mathematical Society
- Journal: Math. Comp. 84 (2015), 1813-1833
- MSC (2010): Primary 33C05, 65D20
- DOI: https://doi.org/10.1090/S0025-5718-2014-02905-0
- MathSciNet review: 3335893