4
$\begingroup$

Let $X \subset \mathbb{C}^n$ be a smooth complex affine variety of dimension $m$. Let $$I_X \subset \mathbb{C}[z_1, \ldots, z_n]$$ be its ideal. Note that it is not always possible to find a set of generators $$I_X = \langle g_1, \ldots, g_r \rangle$$ such that $r = n - m$ is the codimension. In general, we only have $r \ge n - m$. But can we do this locally? That is:

Question. Can we cover $X$ by distinguished open sets $X_f$ such that by viewing $X_f$ as an affine variety in $\mathbb{C}^{n + 1}$, its ideal is generated by $n + 1 - m$ elements?

If not, is there a result of that flavour that I'm missing?

$\endgroup$
4
  • 2
    $\begingroup$ No, that is not even possible locally. There is a connectedness theorem, usually attributed to Hartshorne, that says that such varieties (set-theoretic local complete intersections, or even set-theoretic Cohen-Macaulay varieties) are "locally connected away from codimension $>1$", i.e., remove a codimension 2 closed subset does not (locally) disconnect the variety. So a surface in $4$-space that has a unique singular point, and where that point has two local branches, gives a counterexample. $\endgroup$ Commented Aug 25, 2022 at 15:35
  • $\begingroup$ What is true is that you can cover $X$ by open affines $X_f$ each of which is a union of some of the irreducible components in such a set-theoretic local complete intersection. In other words, after forming the union of $X$ with other irreducible subvarieties (whose dimension equals the dimension of $X$), the union is a set-theoretic local complete intersection. This fact is one of the cornerstones of dimension theory. $\endgroup$ Commented Aug 25, 2022 at 15:38
  • 1
    $\begingroup$ @JasonStarr Thanks! Just to make sure: did you see that I'm assuming $X$ to be smooth? $\endgroup$ Commented Aug 25, 2022 at 17:24
  • $\begingroup$ Sorry, I missed that. Every smooth subvariety is a local complete intersection. $\endgroup$ Commented Aug 25, 2022 at 19:01

1 Answer 1

7
$\begingroup$

I am just writing an answer to correct my mistake!

This is true for smooth subvarieties, roughly by the Jacobian criterion. For each closed point, the rank of the Jacobian matrix at that point equals the codimension $m$. So if you choose $m$ defining relations whose partial derivative vectors give $m$ linearly independent rows in the Jacobian matrix (or columns, depending how you write things), then those $m$ relations locally define $X_f$.

My comment above was answering a different question: this is not true if you allow $X$ to be singular. There are "topological" consequences of being a set-theoretic local complete intersection, e.g., the connectedness theorem of Hartshorne. This prevents certain singular varieties from being set-theoretic complete intersections (notice, this connectedness theorem does not prevent curves in $3$-space from being set-theoretic complete intersections, and it is open whether all curves in $3$-space are set-theoretic complete intersections).

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .