proposal to change acosh



The Common Lisp standard is 
 
     acosh (x)  = 2 log( sqrt((x+1) /2 ) + sqrt((x-1) / 2))

(See  
http://www.franz.com/support/documentation/6.2/ansicl/dictentr/sinhcosh.htm
.)  But in Maxima 

     (%i14) acosh(x), logarc;

     (%o14) log(sqrt(x^2-1)+x)

I often get these things wrong, but I believe:

(1)    The function x -> log(sqrt(x^2 -1) + x) is  discontinuous along the 
imaginary axis,

(2)     Except for the origin, the function x ->  = 2 log( sqrt((x+1) /2 ) 
+ sqrt((x-1) / 2)) is 
          continuous on the imaginary axis.

(3)      Both functions are antiderivaties of x -> 1/sqrt(x^2 -1), but 
            x ->  = 2 log( sqrt((x+1) /2 ) + sqrt((x-1) / 2)) is valid on 
a larger set.

(4)  Abramowitz and Stegun Equation 4.6.21 clearly makes the restriction x 
> 1 for
       acosh(z) = log(z + sqrt(z^2 - 1)),   z = x + i y.  See

      
http://www.convertit.com/Go/ConvertIt/Reference/AMS55.ASP?Res=150&Page=87&Submit=Go
 
(5)  Since we last talked about such things, GCL has fixed some bugs in 
its acosh function; see
(there may be other important messages, but I can't find them right now)

http://www.math.utexas.edu/pipermail/maxima/2002/002955.html
http://www.math.utexas.edu/pipermail/maxima/2002/002949.html

Maybe 5.9.3 would be a good time to:

(a)  make the Maxima acosh function the same as the CL acosh function.

(b) move numerical evaluation for exponetial-like functions to CL.  Then 
Maxima would
not have to use rectform to evaluate things like sin(1.0 + 5.0 * %i). 

(c) check for other branch cut inconsistencies.


Barton