This is a rather basic question that I have never really thought about until now, when I was forced to think about a sheafy version. This condition is needed for example to define the relative Proj construction, and I don't understand what exactly about it is making something 'projective.' Here's an example result who's proof I understand algebraically, but not geometrically. I would like to explain geometrically what this proof is about.

Consider the well-known result that if $f: Z \to X$ is a projective, birational map between quasiprojective varieties, there is a coherent sheaf of ideals $\mathscr{I}$ on $X$ so that $Z \simeq \tilde{X}$, the blow up with respect to $I$. This is proven in Hartshorne and the proof sketch is as follows.

Since $f$ is projective we have a map $i: Z \hookrightarrow \mathbb{P}^n_X$ and an invertible sheaf $\mathscr{L} = i^\ast \mathcal{O}_X(1)$ on $Z$. We define $\mathscr{S} = \bigoplus_{d \geq 0} \mathscr{L}^d$. Since $\mathscr{L}$ is coherent, each power is and the sum is quasicoherent. The condition to make $Z$ the Proj of something is that $\mathscr{S}$ needs to be generated in degree $1$. This isn't the case, so we work affinely and find a related algebra corresponding to a degree $D$ Veronese embedding so that after composing $i$ above with the Veronese embedding, the condition holds.

Some more technicalities follow in this argument concerning whether or not we have a coherent sheaf of ideals, and arranging this to be the case, but the primary issue has already arisen.

What is this Veronese embedding doing, geometrically? I understand how it takes degree $D$ things to hyperplanes - this isn't quite what I'm looking for and is a little bit on-the-nose. Stated another way, what is composing with this embedding fixing about this object in projective space that is detected about being generated in degree $1$, that isn't there in general? Can we see visually/geometrically what fails in the Proj construction when this condition isn't met?

I apologize if this question is still not clear, and I worry that more rambling is only making it worse, so I'll stop here and try to clarify if anyone doesn't understand what I'm asking.


When you have a graded ring $A$ finitely generated in degree 1 as a $A_0$-algebra, then what you'll have is a surjection of graded rings $A_0[X_1,\ldots,X_n] \twoheadrightarrow A$. This is what gives you the closed embedding $\mathrm{Proj} A \hookrightarrow \mathbb{P}^n_{A_0}$.

You can then also realize $A$ as a quotient of $A_0[X_1,\ldots,X_n]$ (so really cut out by equations!). Under this embedding, $A_1$ remembers the restriction of linear forms to $A$.

So I think this also answers your second question about the Veronese embedding. Composing with the Veronese embedding ensures now that your degree 1 component is equal to the linear forms on the ambient projective space restricted to your variety.

  • $\begingroup$ Great answer. Possible minor typo, unless I've misunderstood something: I think you have $\mathbb{P}^{n}_{A}$ written where you should have $\mathbb{P}^{n}_{A_{0}}$. $\endgroup$ – Alex Wertheim Feb 23 at 2:35
  • $\begingroup$ @AlexWertheim You're absolutely right! Thanks! $\endgroup$ – loch Feb 23 at 2:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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