ddMultDouble
ddMultDouble1(y: number, x: number[]): number[]
Defined in double-mixed-double-double/dd-mult-double.ts:21
Returns the product of a double-double-precision floating point number and a double.
slower than ALGORITHM 8 (one call to fastTwoSum more) but about 2x more accurate
relative error bound: 1.5u^2 + 4u^3, i.e. fl(a+b) = (a+b)(1+ϵ), where ϵ <= 1.5u^2 + 4u^3, u = 0.5 * Number.EPSILON
the bound is very sharp
probably prefer
ddMultDouble2due to extra speedALGORITHM 7 of https://hal.archives-ouvertes.fr/hal-01351529v3/document
Parameters:
| Name | Type | Description |
|---|---|---|
y | number | a double |
x | number[] | a double-double precision floating point number |
Returns: number[]
ddMultDouble2
▸ ddMultDouble2(y: number, x: number[]): number[]
Defined in double-mixed-double-double/dd-mult-double.ts:59
Returns the product of a double-double-precision floating point number and a double.
faster than ALGORITHM 7 (one call to fastTwoSum less) but about 2x less accurate
relative error bound: 3u^2, i.e. fl(ab) = (ab)(1+ϵ), where ϵ <= 3u^2, u = 0.5 * Number.EPSILON
the bound is sharp
probably prefer this algorithm due to extra speed
ALGORITHM 8 of https://hal.archives-ouvertes.fr/hal-01351529v3/document
Parameters:
| Name | Type | Description |
|---|---|---|
y | number | a double |
x | number[] | a double-double precision floating point number |
Returns: number[]