MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

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

Let $S$ be a base-scheme, for our purposes we can assume it is the spectrum of a field $k$. Given a smooth, proper and connected $S$-scheme $X$ and an integer $n\geq 1$, is then the fibre-product $X\times_{S}\mathbb{P}^{n}_{S}$ still connected? If not, are their any assumptions on $X$ respectively the base scheme $S$ that would guarantee connectedness?

share|cite|improve this question
The fibered product is just P^n_X, so it should be connected if X is. – Mattia Talpo Aug 25 '10 at 13:49

This question is taking ridiculously many attempts to get a clean, correct answer.

First: As inkspot points out, a scheme is connected iff it has no idempotent global sections other than zero and one. Thus, whether a scheme is connected is determined purely by its ring of global sections.

Second: As Mattia Talpo and Charles Siegel both observe, $X \times_S \mathbb{P}^n_S$ is naturally isomorphic to $\mathbb{P}^n_X$.

Third: The global section ring of $\mathbb{P}^n_X$ is isomorphic to the global section ring of $X$. To see this, first observe that if $f \colon \mathbb{P}^n_X \to X$ is the obvious morphism, then $$\Gamma(\mathbb{P}^n_X, \mathcal{O}_{\mathbb{P}^n_X}) \cong \Gamma(X, f_{*}\mathcal{O}_{\mathbb{P}^n_X}).$$ This follows from the definition of $f_{*}$. Thus, it suffices to show that $f_* \mathcal{O}_{\mathbb{P}^n_X} = \mathcal{O}_X$. For this, we can work locally on $X$. If $U \subset X$ is an open affine, then our desired statement $f_* \mathcal{O}_{\mathbb{P}^n_U} = \mathcal{O}_U$ is true by standard facts about projective space over an affine scheme.

Conclusion: $X \times_S \mathbb{P}^n_S$ is connected iff $X$ is. This statement requires no hypotheses on either $X$ or $S$ (not even local Noetherianness).

share|cite|improve this answer
BCnrd's answer was perfectly clean and correct. – Angelo Aug 26 '10 at 4:09
I think I wrote that in large part out of a sense of frustration that all of the actual answers so far have required corrections in the comments. – Charles Staats Aug 26 '10 at 15:43

A scheme $Z$ is connected if and only if the only idempotent $f\in H^0(Z,\mathcal O_Z)$ are $f=0,1$. Since $H^0(S,\mathcal O_S)=H^0(S\times Y,\mathcal O_{S\times Y})$ for any proper $k$-variety $Y$ such that $k$ is algebraically closed in $H^0(Y,\mathcal O_Y)$, it follows that $S$ is connected if and only if $S\times Y$ is so.

share|cite|improve this answer
Your statement $H^0(S, \mathcal{O}_S) = H^0(S \times Y, \mathcal{O}_{S \times Y})$ for any proper $k$-variety $Y$ is not true if $k$ is not algebraically closed. For instance, there are proper (even projective) varieties over $\mathbb{R}$ having global sections $\mathbb{C}$. However, the statement should hold if $Y$ is, specifically, a projective space. – Charles Staats Aug 25 '10 at 18:22
Charles, you're quite right, thank you. I've added an appropriate hypothesis. – inkspot Aug 25 '10 at 18:29
Charles, this is a good illustration of why "$k$-variety" should never be used with clarifying the intended meaning: no one knows if (in addition to "finite type and separated") it means "irreducible and reduced", or "geometrically integral over $k$", etc. If one takes the definition in Hartshorne, namely to require geometric integrality (with any $k$) then what inkspot originally wrote is correct. Inkspot's new condition that $k$ be algebraically closed in the $k$-finite domain ${\rm{H}}^0(Y,O_Y)$ is the same as the more concrete statement ${\rm{H}}^0(Y,O_Y) = k$ (via the canonical map). – BCnrd Aug 26 '10 at 1:22
BCnrd: Your comment took me a bit to parse, so let me see if I understand it. Let's define a variety over an algebraically closed field to be a separated integral scheme of finite type over $k$. I have never actually seen a scheme-theoretic definition of variety over a non-algebraically closed field, but in my comment, I was assuming this still means "separated integral scheme of finite type over $k$." You point out that if one instead requires that the pullback to $\bar{k}$ be a variety over $\bar{k}$, then inkspot's original statement was correct. – Charles Staats Aug 26 '10 at 15:55

The product is just $\mathbb{P}^n_X$ (because it is $X\times_S (S\times\mathbb{P}^n_\mathbb{Z})$. The fibers then of $\mathbb{P}^n_X\to X$ are projective spaces, and thus connected, so all you need is $X$ is connected.

share|cite|improve this answer
Not every continuous surjective map with connected fibers onto a connected base has connected source; must use that the map to the base is either open (since flat and locally of finite presentation; concretely in this case it is Zariski-locally an affine space, for which life is easy) or closed (because of properness). – BCnrd Aug 25 '10 at 14:24
A natural place where this happens is for instance the case on the smooth stratification of the union $X$ of the two axes in the affine plane. Obviously the scheme $X$ is connected; on the other hand, we can look at the stratification of $X$ into smooth strata, and in this case we find the complement of the origin in $X$ and the origin. Clearly there is a surjective morphism from the disjoint union of the two strata to $X$, each fiber is connected (and consists of a single point), and the target is connected. Clearly again, the whole space is not connected! – damiano Aug 25 '10 at 14:50
Wouldn't noting that the map $\mathbb{P}^n_X \to X$ admits a section suffice? – Jack Huizenga Aug 25 '10 at 19:45

To second BCnrd's remark about the need to have assumptions on the surjection, this is a purely topological issue that can occur whether you're looking at schemes or not. For instance, the projection of $$ [0,1) \times \{0\}\ \cup\ [1,2]\times \{1\} $$ on the first factor has both a connected base and connected source (but is not a proper map).

share|cite|improve this answer

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.