Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

One knows that the support $S$ of a coherent sheaf on a noetherian scheme is closed. E.g. on an affine scheme $X=Spec(A)$ and $F$ corresponding to a finitely generated $A$-module $M$, then the closed subset which corresponds to $S$ is just $V(Ann(M))$.

One often says that $S$ is endowed with the structure of a closed subscheme by taking the sheaf of ideals $Ann(F)$ defined as the kernel of $\mathcal O_X \rightarrow Hom_{\mathcal O_X}(F,F)$.

Now my question: this is not the (unique) reduced subscheme structure, isn't it? Can one anyhow describe the reduced structure?

Thanks

share|improve this question
1  
Could you explain a bit about how you came to this question? As is, it's not clear what constitutes a description. It sounds like you wouldn't be satisfied with "it's the subscheme structure corresponding to the radical of $Ann(F)$" but without knowing what sort of answer you would be satisfied with, it's hard to know what else to say. –  Anton Geraschenko Aug 25 '11 at 6:39
    
My problem was to know whether the structure induced by $Ann(F)$ is already reduced. If you read the first answer of mathoverflow.net/questions/34621/… it looks as if. That was confusing me a bit. –  Descartes Aug 25 '11 at 7:37

2 Answers 2

up vote 3 down vote accepted

No it isn't the reduced induced closed subscheme structure in general. For example, let $A={\bf Z}$, $M={\bf Z}/4{\bf Z}$. Then ${\rm Ann}(M)=4{\bf Z}$ and the prime ideal defining $S$ (with its reduced structure) is $2{\bf Z}={\rm rad}({\rm Ann}(M))$. So if $S$ is endowed with the reduced structure, it is isomorphic to ${\rm Spec}({\bf Z}/2{\bf Z})$ and if it is endowed with the structure given by the annihilator then it is isomorphic to ${\rm Spec}({\bf Z}/4{\bf Z})$.

share|improve this answer
    
Perfect, thank you, Damian! –  Descartes Aug 25 '11 at 9:39

The answer to your second question is pleasantly general and simple.

Given a completely general scheme $X$ (no noetherian, separation, ...hypothesis) and an arbitrary closed subspace $F\subset |X|$ of its underlying topological space, there is a unique closed reduced subscheme $Y\subset X$ whose underlying set is $|Y|=F$. Here is the proof:
i) If $X=Spec A$ is affine, $Y$ is given by the reduced ideal $I=\bigcap_{x\in F} j_x \;$
[as usual, for $x\in SpecA, j_x \subset A$ denotes the ideal corresponding to the point $x$],
ii) If $X$ is not affine, the reduced scheme $Y=V_{sch}(\mathcal I)$ is obtained by the unique ideal sheaf $\mathcal I\subset \mathcal O_X$ restricting on each open affine $U=Spec A$ to the ideal sheaf $\tilde I$ associated to the $I$ above.

Reference EGA 1, Chap.1 , §5, Proposition (5.2.1)

Addendum: the scheme structure on the support of a sheaf.
For reference purpose, let me describe the schematic structure on the support of a sheaf in a fairly general setting.
The situation is that we have a completely arbitrary scheme $X$ (no noetherian assumption) and a quasi-coherent sheaf $\mathcal F$ of $\mathcal O_X$-Modules of finite type on $X$. ($\mathcal F$ needn't be coherent and so this applies to those strange schemes where $\mathcal O_X$ is not coherent!)
Then there exists a smallest closed subscheme $i:Y\hookrightarrow X$ with underlying set $|Y|=supp(\mathcal F)$ and a sheaf of finite type $\mathcal F'$ of $\mathcal O_Y$-Modules with support $|Y|$ such that $i_* \mathcal F'=\mathcal F$.
Of course if $X=SpecA$ then $\mathcal F=\tilde M$ for some finitely generated $A$-module $M$, then we have $Y=V_{sch}(annM)$ and $\mathcal F'=\widetilde {M^\prime}$ , where $M^\prime$ is $M$ seen as an $A/annM$-module.
Although no coherence is requested of $\mathcal F$, some finiteness condition is necessary, else $supp M$ wouldn't even be closed: just look at the $\mathbb Z$-module $\mathbb Q$ whose support is the non-closed generic point of $Spec(\mathbb Z)$

share|improve this answer
    
I guess you don' mean that $F'$ is of finite type (but rather quasi-coherent, not more). –  Martin Brandenburg Sep 16 '11 at 15:20

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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