UNPKG

double-double

Version:

Pure double-double precision functions *with strict error bounds*.

15 lines (14 loc) 574 B
/** * Returns the result of dividing two double-double-precision floating point * numbers, i.e. returns x/y. * * * relative error bound: 15u^2 + 56u^3, i.e. fl(a/b) = (a/b)(1+ϵ), * where ϵ <= 15u^2 + 56u^3, u = 0.5 * Number.EPSILON * * the largest error found was 8.465u^2 * * * ALGORITHM 17 of https://hal.archives-ouvertes.fr/hal-01351529v3/document * @param x a double-double precision floating point number * @param y another double-double precision floating point number */ declare function ddDivDd(x: number[], y: number[]): number[]; export { ddDivDd };