On Mon, Jul 14, 2025 at 10:14 AM fabioromano1 <d...@openjdk.org> wrote: > > On Sat, 12 Jul 2025 09:18:27 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: > > > > Removed useless instruction > > The integers are closed under positive powers, and not under positive nthRoots, this is true. However, the problem remains, as if the root degree `n` is negative, then for radicands that are not 1, -1 or 0 we should return 0, and zero raised to a negative exponent is undefined in real numbers, and so the remainder is. > Good point. But by that argument nthRoot(int n) should still be well-defined. And the comment for nthRootAndRemainder still looks wrong to me
>From the spec, we can take the answer to be "(x.signum() * floor(abs(nthRoot(x, n)))), where nthRoot(x, n) denotes the real nth root of x treated as a real". This means that e.g. the (-3)rd root of 2, is floor(abs(nthRoot(2, -3))), nthRoot(2, -3) is presumably 1 / nthRoot(2, 3), which is 0.79... So if we take this as the definition, the answer is clearly 0, which is perfectly well-defined. Again, I'm not arguing for a functional change. I just don't like the spec comments.
