On Thu, 20 Jul 2006, John J. Trammell wrote:

> On Thu, Jul 20, 2006 at 04:48:14PM -0500, Chuck Cole wrote:
>> Isn't a simple ratio approximation similar to the one 355/113 about
>> that good ?  That particular one is good to about 1 part per ten
>> million (about 7 places) and is much faster in hardware !!
>
> I generally prefer 4 * atan2(1,1).  Less to remember....  :-)

I usually specify "pi" or "Pi" depending on the program I'm using.  You 
could make it one character shorter than that, if you wanted to.

Here are some formulas for pi including those of Ramanujan (numbers 57, 
74-77, 96-112):

http://mathworld.wolfram.com/PiFormulas.html

Rational approximations:

http://www.isi.edu/~johnh/ABOUT/FEATURES/RATIONAL_PI/index.html

So 355/113 seems to have some value.  You can get better with this:

104348/33215

But you have to memorize more digits of the formula than it gives back in 
digits of pi!  If you want to get pi right, just memorize it.

Mike