Let $Z$ be a local complete intersection subscheme of dimension $m$ in $\mathbb{P}^{2m+1}$. Let $P$ be the Hilbert polynomial of $Z$. Denote by $\operatorname{Hilb}_P$ the Hilbert scheme of local complete intersection subschemes with Hilbert polynomial $P$. Is it true that for a generic subscheme $Z'$ in $\operatorname{Hilb}_P$, $I(Z')$ can be generated by polynomials that define smooth hypersurfaces in $\mathbb{P}^{2m+1}$ i.e., does there exists polynomials $P_1, ..., P_n$ such that $I(Z')=(P_1,...,P_n)$ and the zero locus of $P_i$ is smooth for all $i$?

A slightly weaker condition would be to ask if a generic hypersurface in $\operatorname{pr}_2 \operatorname{Hilb}_{P,Q}$ is smooth where $\operatorname{Hilb}_{P,Q}$ is the flag Hilbert scheme of pairs $(Z \subset X)$ where $Z$ is a local complete intersection subscheme of dimension $m$ with Hilbert polynomial $P$ contained is a hypersurface $X$ in $\mathbb{P}^{2m+1}$ with Hilbert polynomial $Q$ and $\operatorname{pr}_2$ denotes the natural projection map.

  • 1
    $\begingroup$ A necessary condition is that the tangent space at any point $P\in Z$ has dimension $\le 2m$. $\endgroup$
    – rita
    Commented Nov 15, 2012 at 15:10
  • $\begingroup$ @Rita: That is what I was thinking as well, but I cannot think of a local complete intersection with a singular point of embedding dimension $2m+1$ for which I can prove the same is true after a generic deformation (I have been trying to produce "rigid" nonreduced structures on a smooth scheme of dimension $m$, but the computations are beyond me). Since you are an expert on this, are there configurations of subschemes (e.g., configurations of degree 6 Del Pezzo surfaces in $\mathbb{P}^5$) that are LCI of embedding dimension $2m+1$ and also "rigid"? $\endgroup$ Commented Nov 15, 2012 at 15:21
  • $\begingroup$ @Jason: I had not seen your comment earlier. Unfortunately, I have no idea. $\endgroup$
    – rita
    Commented Dec 14, 2012 at 12:16

1 Answer 1


I believe that the answer to the first question is no for the reason that a local complete intersection is pretty far from a complete intersection. Take $Z$ to be $3$ points in $\mathbb P^3$ which aren't on a line. Then the Hilbert scheme of local complete intersections is a dense open subset of the Hilbert scheme of $3$ points in $\mathbb P^3$. However, the complete intersections will have to be the intersection of a cubic hypersurface with two hyperplanes, for degree reasons. Thus, the complete intersections will all lie on a single line in $\mathbb P^3$, which is not true for 3 generic points.

If you assumed that $Z$ is a complete intersection, not just a local complete intersection, then what you want is probably true.

  • 6
    $\begingroup$ @Dustin: The OP does not ask whether $Z$ is a complete intersection. He asks whether $Z$ is an intersection of some (presumably huge) number $n$ of smooth hypersurfaces. The integer $n$ need not equal $m+1$, the codimension. The answer to the question is yes for $Z$ any finite collection of reduced points in $\mathbb{P}^3$. $\endgroup$ Commented Nov 15, 2012 at 15:15
  • $\begingroup$ @Starr: Is there some results when $Z$ is not zero dimensional? $\endgroup$ Commented Nov 15, 2012 at 16:41
  • 1
    $\begingroup$ @Jason Starr: probably everybody thinks the question is over an algebraically closed fields, but this is not stated explicitely in the OP. Over an imperfect field, a reduced point can not be contained in a smooth hypersurface if its residue field has too high inseparability degree. $\endgroup$
    – Qing Liu
    Commented Nov 15, 2012 at 22:04
  • $\begingroup$ @Qing Liu: The OP explicitly states that he is taking a generic element of the corresponding Hilbert scheme. Thus the zero dimensional scheme will be smooth over the ground field. $\endgroup$ Commented Nov 16, 2012 at 3:28
  • 1
    $\begingroup$ yes we all assumed he meant complete intersection! $\endgroup$
    – roy smith
    Commented Feb 22, 2013 at 4:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.