MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

It is well known that total space of the tautological line bundle $\mathcal{O}(-1)$ over projective space $\mathbb{P}^n$ is closed subvariety of $\mathbb{P}^n\times\mathbb{A}^{n+1}$. My question is how to realize total space of $\mathcal{O}(1)$ over $\mathbb{P}^n$ in such manner, i.e. I need an embedding of $Tot(\mathcal{O}(1))$ in simple variety and defining equations. Thanks.

share|cite|improve this question
up vote 24 down vote accepted

It is the complement $\mathbb{P}^{n+1} - \{x\}$ of a point in a projective space.

share|cite|improve this answer
Shouldn't $\mathbb{P}^{n+1}\setminus\{x\}\to \mathbb{P}^n$ be the tautological line bundle? – Fei YE Nov 7 '10 at 1:59
The tautological line bundle contains a divisor, isomorphic to $\mathbb{P}^n$, with normal bundle of degree $-1$ (namely, the zero section). On the other hand every effective compact divisor in $\mathbb{P}^{n+1}-\{x\}$ has a positive normal bundle. In particular, a hyperplane avoiding $x$ has normal bundle of degree $1$. The projection from the point identifies the complement of $x$ with the total space of this normal bundle. – Tony Pantev Nov 7 '10 at 2:20
Thank you, Tony. Maybe you know such simple description of $Tot(O(n))$ for $n>1$ also? – Klim Puhov Nov 8 '10 at 14:23
You can get $Tot(\mathcal{O}(n))$ in a similar manner by deleting a point from a weighted projective space. But this is more contrived. And is not really better than thinking of $Tot(\mathcal{O}(n))$ as a toric variety. So this probably is not what you want. – Tony Pantev Nov 9 '10 at 1:22
$Tot(\mathcal{O}_{\mathbb{P}^{n}}(k))$ is $\mathbb{P}^{n+1}(1,1,\ldots,1,k) - \{x\}$, where $x = (0:0:\ldots:0:1)$. – Tony Pantev Apr 7 at 12:13

Dear Luther King, since you ask for equations, let me add them to Tony's beautifully geometric answer.

Consider $\mathbb P^{n+1}$ with homogeneous coordinates $(z_0:z_1:\ldots:z_{n+1})$ and $\mathbb P^{n}$ enbedded as the hyperplane $z_0=0$. If $x\in\mathbb P^{n+1}$ is the point $x=(1:0:\ldots:0)$, the required total space $T=Tot \mathcal O_{\mathbb P^{n}}(1)$ is the complement $T=\mathbb P^{n+1} \setminus \{x\}$ of $x$ in $\mathbb P^{n+1}$. The fiber of our bundle $\mathcal O_{\mathbb P^{n}}(1)$ at the arbitrary point $(0:z_1:\ldots:z_{n+1}) \in \mathbb P^n$ is the set of all $(\lambda:z_1:\ldots:z_{n+1})$ with $\lambda \in k$ (base field) .

The one-dimensional vector space structure on the fiber is given by $(\lambda:z_1:\ldots:z_{n+1})+(\mu:z_1:\ldots:z_{n+1})=(\lambda +\mu:z_1:\ldots:z_{n+1})$ and similarly for products by scalars. Beware that we definitely do not have an isomorphism from our fiber to $k$ defined by $(\lambda:z_1:\ldots:z_{n+1})\mapsto \lambda$: this is not even a well-defined map. This is not surprising: after all $\mathcal O_{\mathbb P^{n}}(1)$ is not a trivial bundle!

A more canonical approach (edited) Readers of the canonical faith may suppress coordinates as follows.

Consider a $k$ - vector space $V$ of dimension $n+1$, its projectivization $\mathbb P (V)$ and its embedding $\mathbb P (V) \to \mathbb P (k \oplus V)$ sending the point $\mathbb P (l) \in \mathbb P (V) $ to the point $\mathbb P (0\oplus l) \in \mathbb P (k \oplus V)$. The vector bundle $\mathcal O _{\mathbb P (V)} (1)$ then has as total space the open subset $T\subset \mathbb P (k \oplus V)$ obtained by deleting $x=\mathbb P (k \oplus 0)$ from $\mathbb P (k \oplus V)$ i.e. $T= \mathbb P (k \oplus V)\setminus x$. The fiber over $\mathbb P (l)$ [identified with $\mathbb P (0\oplus l)$] is $\mathbb P (k \oplus l)\setminus x$, a vector space of dimension one with origin the point $\mathbb P (0 \oplus l)$ but with no prefered isomorphism to $k$.

Elementary geometry It may clarify the above to recall that given a one dimensional projective space $\mathbb P$ over $k$, if you delete a point $x$ from it you get a one dimensional affine space $\mathbb P \setminus x$ and if you choose in it an origin, you get a one dimensional vector space, but that vector space has no prefered isomorphism with the vector space $k$.

share|cite|improve this answer
Thanks for detailed answer, Georges! – Klim Puhov Nov 7 '10 at 20:22

For me the best description of $Tot(O(1))$ is the tautological --- as the relative spectrum of the sheaf of algebras $A = O \oplus O(-1) \oplus O(-2) \oplus \dots$ on $P^n$: $$ Tot(O(1)) = Spec_{P^n}(A). $$ This allows to work with $Tot(O(1))$ more effectively than any other description. For example, a coherent sheaf on $Tot(O(1))$ can be represented by a quasicoherent sheaf $F$ on $P^n$ together with a morphism $F(-1) \to F$ inducing on $F$ a structure of an $A$-module.

share|cite|improve this answer
Yes, I know about such description, but for my purposes the above one is preferable. – Klim Puhov Nov 7 '10 at 20:35

Your Answer


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.