On Wed, 23 Jul 2025 16:11:20 GMT, fabioromano1 <d...@openjdk.org> wrote:
>> This PR implements nth root computation for BigIntegers using Newton method.
>
> fabioromano1 has updated the pull request incrementally with one additional
> commit since the last revision:
>
> More accurate lower bound for doubles values tests
Please merge branch `master` into yours every now and then.
src/java.base/share/classes/java/math/MutableBigInteger.java line 1956:
> 1954: rToN1 = BigInteger.unsignedLongPow(rLong, n - 1);
> 1955: r1 = ((n - 1) * rLong + Long.divideUnsigned(x,
> rToN1)) / n;
> 1956: } while (r1 < rLong); // Terminate when non-decreasing.
Here (and below as well) it would be useful to have the same variable names as
in the paper, where possible.
No need to rename `x` to m and `n` to k, though, they can remain.
src/java.base/share/classes/java/math/MutableBigInteger.java line 2000:
> 1998: int shiftLack = n - shiftExcess;
> 1999: shift += shiftLack; // shift is long, no overflow
> 2000: rad /= Double.parseDouble("0x1p" + shiftLack);
Suggestion:
rad /= Math.scalb(1.0, shiftLack);
src/java.base/share/classes/java/math/MutableBigInteger.java line 2026:
> 2024: int correctBits = ((Double.PRECISION - 1) - radExp
> - 1) / n + 1;
> 2025: rootShift -= correctBits;
> 2026: approx *= Double.parseDouble("0x1p" + correctBits);
Suggestion:
approx *= Math.scalb(1.0, correctBits);
-------------
PR Comment: https://git.openjdk.org/jdk/pull/24898#issuecomment-3109471808
PR Review Comment: https://git.openjdk.org/jdk/pull/24898#discussion_r2226198312
PR Review Comment: https://git.openjdk.org/jdk/pull/24898#discussion_r2226198452
PR Review Comment: https://git.openjdk.org/jdk/pull/24898#discussion_r2226198568
