Re: acosh issues



Camm Maguire wrote:
> Greetings, and thanks for these pointers!  There are probably other
> functions that need attention as well.  In any case, we should be
> encouraged that the original version basically had the scaling issue
> down.  
> 
> Here is what I've just committed:
> 
> (defun acosh (x)
>   (* 2 (log (+ (sqrt (/ (1+ x) 2)) (sqrt (/ (1- x) 2))))))
> 

Although this doesn't help in the case where x is real, but you might be 
interested in looking at the routines in CMUCL (see src/irrat.lisp) for 
these functions for complex args.  They're translations of the routines 
in Kahan's article that Barton references.

Ray