fastTwoSumError
fastTwoSumError(
a,b,sum):number
Defined in: two-sum.ts:43
The exact rounding error of sum, where sum is the binary64 result of a + b and |a| >= |b| is already known — three additions instead of
twoSumError’s six.
Dekker’s fast two-sum. The magnitude precondition is not checked: violating it returns a wrong answer rather than an error, so only use it where the ordering is established by construction.
Parameters
Section titled “Parameters”number
The larger-magnitude term.
number
The smaller-magnitude term.
number
The binary64 result of a + b.
Returns
Section titled “Returns”number
The part of a + b that sum dropped.