Re: Controlling ('rand') random number upper limit

[Chris Rehm <javadisciple@xxxxxxxxxxxxx>]

     His message did show that he was using 32768 as the divisor. So, 
he'll get a value between 0/32768 and 32767/32768.
   This gives me a chance to use the only thing I've ever liked about 
Windows, the calculator it comes with. So he'll get a value ranging from 
0 to .999969482421875 (barring any typos on my part).
   That should give as flat a distribution across his own range as the 
generator has across its own range, provided he does not wish to 
generate numbers larger than 32767. If he does, he will need a different 
random number generator or to append multiple results from this 
generator bitwise and then divide by the new range.

Jim Langston wrote:

> Chris, Paul,
> 
> That being said, Paul's solution can still be screwed, depending
> on how the system is adjusting for math.  One out of 32767 times
> it will return a value 1 greater than his range.
> 
> For a range 1 to 10 (normal return):
> 32766 / 32767 = 0.999969481 * 10 = 9.99969481.  Truncate returns

[snip] s/b 32766/32768

> Regards,
> 
> Jim Langston
> 
-- 

   Chris Rehm
<mailto:javadisciple@earthlink.net>