## 110.34 Morphisms

An important question is, given a morphism $\pi : X \to S$, whether the morphism has a section or a rational section. Here are some example exercises.

Exercise 110.34.1. Consider the morphism of schemes

$\pi : X = \mathop{\mathrm{Spec}}(\mathbf{C}[x, t, 1/xt]) \longrightarrow S = \mathop{\mathrm{Spec}}(\mathbf{C}[t]).$

1. Show there does not exist a morphism $\sigma : S \to X$ such that $\pi \circ \sigma = \text{id}_ S$.

2. Show there does exist a nonempty open $U \subset S$ and a morphism $\sigma : U \to X$ such that $\pi \circ \sigma = \text{id}_ U$.

Exercise 110.34.2. Consider the morphism of schemes

$\pi : X = \mathop{\mathrm{Spec}}(\mathbf{C}[x, t]/(x^2 + t)) \longrightarrow S = \mathop{\mathrm{Spec}}(\mathbf{C}[t]).$

Show there does not exist a nonempty open $U \subset S$ and a morphism $\sigma : U \to X$ such that $\pi \circ \sigma = \text{id}_ U$.

Exercise 110.34.3. Let $A, B, C \in \mathbf{C}[t]$ be nonzero polynomials. Consider the morphism of schemes

$\pi : X = \mathop{\mathrm{Spec}}(\mathbf{C}[x, y, t]/(A + Bx^2 + Cy^2)) \longrightarrow S = \mathop{\mathrm{Spec}}(\mathbf{C}[t]).$

Show there does exist a nonempty open $U \subset S$ and a morphism $\sigma : U \to X$ such that $\pi \circ \sigma = \text{id}_ U$. (Hint: Symbolically, write $x = X/Z$, $y = Y/Z$ for some $X, Y, Z \in \mathbf{C}[t]$ of degree $\leq d$ for some $d$, and work out the condition that this solves the equation. Then show, using dimension theory, that if $d >> 0$ you can find nonzero $X, Y, Z$ solving the equation.)

Remark 110.34.4. Exercise 110.34.3 is a special case of “Tsen's theorem”. Exercise 110.34.5 shows that the method is limited to low degree equations (conics when the base and fibre have dimension 1).

Exercise 110.34.5. Consider the morphism of schemes

$\pi : X = \mathop{\mathrm{Spec}}(\mathbf{C}[x, y, t] /(1 + t x^3 + t^2 y^3)) \longrightarrow S = \mathop{\mathrm{Spec}}(\mathbf{C}[t])$

Show there does not exist a nonempty open $U \subset S$ and a morphism $\sigma : U \to X$ such that $\pi \circ \sigma = \text{id}_ U$.

Exercise 110.34.6. Consider the schemes

$X = \mathop{\mathrm{Spec}}(\mathbf{C}[\{ x_ i\} _{i = 1}^{8}, s, t] /(1 + s x_1^3 + s^2 x_2^3 + t x_3^3 + st x_4^3 + s^2t x_5^3 + t^2 x_6^3 + st^2 x_7^3 + s^2t^2 x_8^3))$

and

$S = \mathop{\mathrm{Spec}}(\mathbf{C}[s, t])$

and the morphism of schemes

$\pi : X \longrightarrow S$

Show there does not exist a nonempty open $U \subset S$ and a morphism $\sigma : U \to X$ such that $\pi \circ \sigma = \text{id}_ U$.

Exercise 110.34.7. (For the number theorists.) Give an example of a closed subscheme

$Z \subset \mathop{\mathrm{Spec}}\left({\mathbf Z}[x, \frac{1 }{ x(x-1)(2x-1)}]\right)$

such that the morphism $Z \to \mathop{\mathrm{Spec}}({\mathbf Z})$ is finite and surjective.

Exercise 110.34.8. If you do not like number theory, you can try the variant where you look at

$\mathop{\mathrm{Spec}}\left({\mathbf F}_ p[t, x, \frac{1 }{ x(x-t)(tx-1)}]\right) \longrightarrow \mathop{\mathrm{Spec}}({\mathbf F}_ p[t])$

and you try to find a closed subscheme of the top scheme which maps finite surjectively to the bottom one. (There is a theoretical reason for having a finite ground field here; although it may not be necessary in this particular case.)

Remark 110.34.9. The interpretation of the results of Exercise 110.34.7 and 110.34.8 is that given the morphism $X \to S$ all of whose fibres are nonempty, there exists a finite surjective morphism $S' \to S$ such that the base change $X_{S'} \to S'$ does have a section. This is not a general fact, but it holds if the base is the spectrum of a dedekind ring with finite residue fields at closed points, and the morphism $X \to S$ is flat with geometrically irreducible generic fibre. See Exercise 110.34.10 below for an example where it doesn't work.

Exercise 110.34.10. Prove there exist a $f \in \mathbf{C}[x, t]$ which is not divisible by $t - \alpha$ for any $\alpha \in \mathbf{C}$ such that there does not exist any $Z \subset \mathop{\mathrm{Spec}}(\mathbf{C}[x, t, 1/f])$ which maps finite surjectively to $\mathop{\mathrm{Spec}}(\mathbf{C}[t])$. (I think that $f(x, t) = (xt - 2)(x - t + 3)$ works. To show any candidate has the required property is not so easy I think.)

Exercise 110.34.11. Let $A \to B$ be a finite type ring map. Suppose that $\mathop{\mathrm{Spec}}(B) \to \mathop{\mathrm{Spec}}(A)$ factors through a closed immersion $\mathop{\mathrm{Spec}}(B) \to \mathbf{P}^ n_ A$ for some $n$. Prove that $A \to B$ is a finite ring map, i.e., that $B$ is finite as an $A$-module. Hint: if $A$ is Noetherian (please just assume this) you can argue using that $H^ i(Z, \mathcal{O}_ Z)$ for $i \in \mathbf{Z}$ is a finite $A$-module for every closed subscheme $Z \subset \mathbf{P}^ n_ A$.

Exercise 110.34.12. Let $k$ be an algebraically closed field. Let $f : X \to Y$ be a morphism of projective varieties such that $f^{-1}(\{ y\} )$ is finite for every closed point $y \in Y$. Prove that $f$ is finite as a morphism of schemes. Hints: (a) being finite is a local property, (b) try to reduce to Exercise 110.34.11, and (c) use a closed immersion $X \to \mathbf{P}^ n_ k$ to get a closed immersion $X \to \mathbf{P}^ n_ Y$ over $Y$.

Comment #2364 by Marco Castronovo on

$\pi\circ\sigma = \operatorname{id}_S$ in part (1) of Exercise 92.33.1 right?

In your comment you can use Markdown and LaTeX style mathematics (enclose it like $\pi$). A preview option is available if you wish to see how it works out (just click on the eye in the toolbar).