ddAddDd

Returns the result of adding two double-double-precision floating point numbers.

  • relative error bound: 3u^2 + 13u^3, i.e. fl(a+b) = (a+b)(1+ϵ), where ϵ <= 3u^2 + 13u^3, u = 0.5 * Number.EPSILON

  • the error bound is not sharp - the worst case that could be found by the authors were 2.25u^2

ALGORITHM 6 of https://hal.archives-ouvertes.fr/hal-01351529v3/document

Parameters

x

a double-double precision floating point number

y

another double-double precision floating point number