For simplicity let's suppose $n=2$ and $\Bbb{P}^{n}_k = \operatorname{Proj}k[X,Y,Z]$ and $\Bbb{P}^{n\vee}_k = \operatorname{Proj}k[A,B,C]$. With the segre embeding

$$ \operatorname{Proj}k[X,Y,Z] \times \operatorname{Proj}k[A,B,C]\to\operatorname{Proj}k[AX,AY,\cdots, CZ] $$

and define "the incident variety" $\Sigma$ (or, "the incident correspondence", only find this terminology in Vakil's FOAG,No. 13.4.1) be the closed subscheme in $\operatorname{Proj}k[AX,AY,\cdots, CZ]$ vanishing $AX+BY+CZ$, i.e.,

$$ \Sigma := V(AX+BY+CZ) = \operatorname{Proj}\frac{k[AX,AY,\cdots, CZ]}{AX+BY+CZ} $$

Prove that $\Sigma \to \operatorname{Proj}k[A,B,C]$ is the projectivization of some locally free sheaf of rank 2 on $\operatorname{Proj}k[A,B,C]$, i.e., there is some locally free sheaf $\mathscr{E} $of rank 2 on $\operatorname{Proj}k[A,B,C]$, such that $\Sigma \to \operatorname{Proj}k[A,B,C]$ is the relative proj of $\operatorname{Sym}\mathscr{E}$:

$$ \Sigma = \mathcal{Proj}\operatorname{Sym}\mathscr{E} \to \operatorname{Proj}k[A,B,C] $$

I saw this question from David Eisenbud, Joe Harris, The geometry of schemes, Exercise III-25, Page104

The background for it is that I am continuing trying to gain some understanding the scheme theoretic dual projective space $\Bbb{P}_k^{n\vee}$, following a question I posted Two definitions of scheme theoretic dual projective space.

The following is what I have tried:

Looking at it in the affine open subscheme $\operatorname{Spec} k[U,V]$ where $U=B/A, V=C/A$, I got

$$ \operatorname{Proj}\frac{k[AX,AY,\cdots, CZ]}{AX+BY+CZ} \to \operatorname{Spec} k[U,V] $$ Then "since $A \neq 0$" (this makes no sense scheme theoretically, I just want to show what I have tried), I got (from no reason) $$ \operatorname{Proj}\frac{k[U,V][X,Y,Z]}{X+UY+VZ} \to \operatorname{Spec} k[U,V] $$

Then I failed to describe the left hand side of the above map as the symmetric sheaf of some locally free sheaf over $\operatorname{Spec} k[U,V]$, making less gluing them into a locally free sheaf on $\operatorname{Proj}k[A,B,C]$.

Is the above argument a right direction? If so, how to make it rigorous and conitue the argument? As in the comment of the answer here, the sheaf $\mathscr{E}$ seems to be some "equivalent class (maybe quotient out $\mathscr{O^*}$?) of the sheaf $\mathscr{O}(1)$", in such a way it corresponds to the inutition "A point of the dual projective space is an equivalence class of linear forms on the original vector space, which defines a hyperplane in the usual way." that Zhen Lin said. But I failed to make it concide with the above argument.

Thanks for reading, any help would be appreciated.

  • 1
    $\begingroup$ Your $n$s and your variables don't match up - $\operatorname{Proj} k[x,y,z]\cong\Bbb P^2$, not $\Bbb P^3$. Also, one tip for organization: it would be nice to put the problem before your work - it seems that in this post you start working on things first and then say what the problem is, and that's a little harder to follow. $\endgroup$
    – KReiser
    May 28, 2023 at 21:12
  • $\begingroup$ @KReiser Thank you. I fixed some typos and reorganized the text. $\endgroup$
    – onRiv
    May 29, 2023 at 0:14

1 Answer 1


Still not totally accurate, but anyway I update the rest that I have tried as an answer, since it seems giving the desired sheaf.

By replacing $X$ to $-UY-VZ$, the scheme $\operatorname{Proj}\frac{k[U,V][X,Y,Z]}{X+UY+VZ}$ in fact is just $\operatorname{Proj}k[U,V][Y,Z]$. Hence on the affine open subscheme $\operatorname{Spec} k[U, V]$, the incident variety $\Sigma$ is the projectivization of the rank 2 free sheaf with generators $Y$ and $Z$.

Changing $U, V$ back to $B/A$ and $C/A$, the transition map from $\operatorname{Spec} k[B/A, C/A]$ to $\operatorname{Spec}k[A/B, C/B]$, by $Y = -A/B \cdot X -C/B \cdot Z$, is

\begin{aligned} \left(\begin{array}{l} Y \\ Z \end{array}\right) & =\left(\begin{array}{c} -\frac{A}{B} X-\frac{C}{B} Z \\ Z \end{array}\right) \\ & =\left(\begin{array}{cc} -\frac{A}{B} & -\frac{C}{B} \\ 0 & 1 \end{array}\right)\left(\begin{array}{l} X \\ Z \end{array}\right) \end{aligned}

and the other two transition maps have similar forms.

Hence "symbolically", it has the same form with the classic case treating dual projective space as linear forms.


You must log in to answer this question.

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