1
$\begingroup$

Let $Z =C \cup F$ be a disjoint scheme union of closed subschemes of Pn.

Let $p_C$ be the Hilbert polynomial; assume

(1) $F$ finite, reduced,

(2) all irreducible components of $C$ are positive dimensional

Q. does it follow that $C$ is the sole subscheme of $Z$ with same Hilbert polynomial $p_C$?

OK if $C$ is integral: indeed, if $C'$ is a subscheme of $Z$ with $p_{C'} = p_C$, we get

$C'_{red} = C \cup F'$, disjoint union, with $F'$ finite, hence

$p_{C'_red }=p_C+deg F'\leq p_{C'}=p_C$,

thus $F'$ is empty and $C'=C$.

I wish it were true say for $C$ a local complete intersection.

$\endgroup$

1 Answer 1

2
$\begingroup$

Equivalently, does $C$ have a subscheme $C'$ such that $p_C - p_{C'}$ is finite?

$C$ satisfies Serre's condition S1 iff $C$ is the closure of the union of the generic points of its geometric components ("no embedded components"). If $C'$ misses one of those generic points, $p_C - p_{C'}$ will grow like the Hilbert polynomial of the missing component, so by your positive-dimensionality condition the difference wouldn't be finite. One place to read about S1 is the book [Eisenbud], which includes the exercise "Show $X$ reduced $\iff$ it is R0 (generically reduced) + S1."

Local complete intersection $\implies$ Cohen-Macaulay $\implies$ S1, so you're good.

For a non-example, let $C = \{[x,y,z] : x^2 = xy = 0\}$, $C' = \{[x,y,z] : x = 0\}$.

$\endgroup$
3
  • $\begingroup$ $C'$ is assumed to be a subscheme of $Z$, not necessarily of $C$ proper. Of course if this were so, the equality of hilbpols would imply $C=C'$ at once. The difficulty is how to control possible embedded points of $C'$. $\endgroup$ Aug 29, 2013 at 15:39
  • $\begingroup$ Sorry, you didn't use $C'$ in your question, and I didn't notice you were using it in your "OK if $C$ is integral" section. My $C'$ is not your $C'$. Since you assume $Z$ and $F$ disjoint, your question is equivalent to "Does $C$ have a subscheme only finitely much smaller?" to which one then adds points from $F$. $\endgroup$ Sep 1, 2013 at 17:19
  • $\begingroup$ I think that mod a typo you've clarified it all. $\endgroup$ Sep 2, 2013 at 20:34

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.