-1

Let $X = \mathbb{P}^{3}$ and $\mathcal{E}$ and $\mathcal{G}$ locally free sheaves on $X$ such that $\text{rank}(\mathcal{E}) = e$ and $\text{rank}(\mathcal{G}) = g$.

Definition: The degeneracy scheme $\text{Sing}(\psi)$ of the $\psi : \mathcal{E} \longrightarrow \mathcal{G}$ is the zero scheme of the associated global section $\omega_{\psi} \in H^{0}(X, \bigwedge^{g}(\mathcal{E}^{\vee}) \otimes \text{det}(\mathcal{G}))$.

If $\psi : \mathcal{E} \longrightarrow \mathcal{G}$ is a generically surjective morphism, then:

1) Is it possible to have $\text{Sing}(\psi) = \emptyset$?

2) Is it possible to have $\text{Sing}(\psi) = \lbrace p_{1}, \dots, p_{k} \rbrace$?

Thanks in advance.

Allan Ramos
  • 249
  • 1
  • 18
  • Have you tried anything on this problem? There's a clear and obvious choice of $\psi$, $\mathcal{E}$, and $\mathcal{G}$ which gives you an affirmative answer to question 1. (What global sections do you know that have no zeroes?) – KReiser Nov 17 '19 at 19:58
  • @KReiser. For item 1 we can choose $\mathcal{E} = \mathcal{G} = \mathcal{O}_{X}$. – Allan Ramos Nov 17 '19 at 21:36
  • Great! Next time please try to think this through for yourself (and put your progress/results in the question if you can't solve it all the way) before posting your question. That way people will be more likely to answer your question without you needing to put a bounty on it. – KReiser Nov 17 '19 at 21:42
  • Dear @KReiser. I am trying to show item 2, but so far, to no avail. – Allan Ramos Nov 17 '19 at 21:45

1 Answers1

2

Question 1 has been resolved in the comments: take $\psi$ to be the identity morphism of $\mathcal{O}_{\Bbb P^n}$.

For 2, we use an alternate definition of the degeneracy locus which will be easier to work with. We define the degeneracy locus as the subscheme cut out locally by the maximal minors of the matrix which locally describe the morphism. In fact, we can pick a morphism which is globally determined by a matrix. As the ideal of a finite set of $M$ points in $\Bbb P^n$ can be written down using $n$ polynomials $f_1,\cdots,f_n$ of degree $M$ (see here, for instance), we let $\mathcal{E}=\mathcal{O}^n$ and $\mathcal{G}=\mathcal{O}(M)$ and describe the morphism globally by the matrix $\begin{pmatrix} f_1 & \cdots & f_n\end{pmatrix}$. This is clearly generically surjective (if any of the $f_i$ are invertible at a point, it's surjective there, and this is a generic condition). The common zero locus of the maximal minors (aka the $f_i$) is just $V(f_1,\cdots,f_n)=\{p_1,\cdots,p_M\}$.

This latter idea works in general: if $I$ is an ideal generated by several homogeneous polynomials $f_i$ of the same degree $d$ so that $V(I)$ is a proper subvariety, then we can express $V(I)$ as a degeneracy locus of the morphism $\mathcal{O}^n\to\mathcal{O}(d)$ where the map on each coordinate is just given by multiplication by $f_i$.

KReiser
  • 65,137
  • KReiser. Ok, I will calmly read your answer. Thank you. – Allan Ramos Nov 17 '19 at 21:57
  • KReiser. Thank you very much for your answer. I posted a question here in the forum (https://math.stackexchange.com/q/3434683) where I do as you say above by posting what I tried to do. Could you analyze my question as well as what I tried to do in that question? – Allan Ramos Nov 17 '19 at 23:02
  • Again. Thank you very much . – Allan Ramos Nov 17 '19 at 23:28