At Tue, 6 May 2008 19:29:05 -0700 (PDT),
Joseph Pang wrote:
> Yes, my comments still stand. Using an approximate sampling for the
> tail of the base layer (or strip) should not be muddled with the
> pre-calculations of the Ziggurat layers. In fact, I would have far
> less objection if the pre-calculations were done with the exact erfc
> function, while keeping the current approximate algorithm for the
> tail generation intact. Actually, this is what I was suggesting in
> my original email.
I have forwarded your report to the original author of the routine.
--
Brian Gough
GNU Scientific Library -
http://www.gnu.org/software/gsl/
