At Tue, 12 Apr 2011 18:33:04 -0400, James Ward wrote: > > I did some timing tests, and it looks like Moler & Morrision is about > twice as slow as gsl_hypot(), up to the same accuracy (1e-15). > > http://steve.hollasch.net/cgindex/math/pythag-root.txt > > suggests this is due to M&M being cubically convergent, while > gsl_hypot() is quadratically convergent. So ... never mind! >
Interesting read! Thanks for that. _______________________________________________ Bug-gsl mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-gsl
