Thomas wrote:
> A Taylor Series is definitely the way to go if you are 
> rolling your own.

Not if you want a good sine function! You need to consider 
errors and convergence.

http://en.wikipedia.org/wiki/Generating_trigonometric_tables

BTW, it's only necessary to compute sin(x) for x in [0, 
pi/2] because other values of x can be shifted into this 
interval. And if you're using a series method, check the 
conditions for convergence. Cosines can be calculated from 
sines as they're basically the same function but shifted.

All computed sine functions are approximations. So it's a 
question of how much error you can tolerate, and how it's 
distributed. For example, you may be concerned that sin(0) 
== 0 exactly and sin(pi/2) == 1 exactly. A Taylor series 
will guarantee the first but not the second.

Have a look at the CORDIC algorithm:

http://en.wikipedia.org/wiki/CORDIC

C code here:

http://www.dcs.gla.ac.uk/~jhw/cordic/

David

Reply via email to