This is the mail archive of the
gsl-discuss@sourceware.org
mailing list for the GSL project.
difficult region for gsl_sf_hyperg_1F1
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: GSL Discuss Mailing List <gsl-discuss at sourceware dot org>
- Date: Tue, 15 Dec 2009 17:50:33 +0000
- Subject: difficult region for gsl_sf_hyperg_1F1
We have gsl_sf_hyperg1F1(-37.8, 2.01, 103.58) => -3.16661e+14 +/-
3.31+19 from the series. Correct answer is -6.21927211009e17(*).
Does anyone know a better algorithm for this region?
http://savannah.gnu.org/bugs/?28267
(*) In pari:
poch(a,x) = { gamma(a+x)/gamma(a) }
t(a,b,x,k) = { (poch(a,k)/poch(b,k)) * (x^k)/(k!) }
suminf(k=0,t(-37.8,2.01,103.58,k))