chickadee » math » fp2ulp

fp2ulp x2 x1procedure
fp2ulp-error x2 x1 rprocedure
x2
flonum
x1
flonum
r
flonum

Like fpulp and fpulp-error, but for double-double flonums.

The unit in last place of a double-double is that of the higher-order of the pair, shifted 52 bits right.

Examples:

> (fp2ulp 1.0 0.0)
4.930380657631324e-32
> (let-values ([(x2 x1)  (fl2 1/7)])
    (fp2ulp-error x2 x1 1/7))
0.07142857142857142