UNPKG

double-double

Version:

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

10 lines (9 loc) 340 B
/** * Returns the result of summing an array of double-double-precision floating * point numbers naively (i.e. not using pairwise addition to reduce error a bit). * * * an error bound is given by: (n-1)(1+ϵ), * where ϵ <= 3u^2 + 13u^3, u = 0.5 * Number.EPSILON */ declare function ddSum(qs: number[][]): number[]; export { ddSum };