Floating point numbers.

This provides methods for computing over and parsing 64-bit floating pointer numbers.