accurate 0.3.1

(more or less) accurate floating point algorithms
Documentation