Monday, December 8, 2008

Arbitrary Precision

I think it'd be cool to implement arbitrary precision trigonometric functions using lazily evaluated Taylor series. (Is this how they're done?) Pretty bad performance, but at least you have a reliable algorithm.

I've also always wondered how TI calculators know that sin(pi) = 0. No floating-point calculation I've seen in a general-purpose programming language can figure it out.

No comments: