For univariate polynomials, I can compute the squarefree variant with the same roots as $p / \text{gcd}(p, dp/dx)$, which is simple enough.
Then I learned about radical of ideals generated by multivariate polynomials, and it seems that it is non-trivial to get the generator set of the radical ideal, at least for the general case.
But it seems to me that if I could compute the squarefree version of each multivariate polynomial in the generating set, I would get the same variety, and since there is no $f$ such as $f^{m>1} = g$ if $g$ is squarefree, then this ideal must be radical.
Is my reasoning correct? If I can compute the squarefree variants of my polynomials with an algorithm such as this, then they generate the radical ideal?