library for multiple-precision floating-point computations with exact rounding
