I have some questions about two statements from Bosch's "Algebraic Geometry and Commutative Algebra" about algebraic varieties (page 479). Since I still don't have the permission to add images I quote the relevant excerpt:

...The notion of properness has been introduced in 9.5/4. It means that the structural morphism $p: A \to Spec(K)$ is of finite type, separated, and universally closed. For the property of smoothness see 8.5/1. It follows from 8.5/15 in conjunction with 2.4/19 that all stalks $\mathcal{O}_{A,x}$ of a smooth $K$-group scheme $A$ are integral domains. Since abelian varieties are required to be irreducible, they give rise to integral schemes. Also let us mention that for $K$-group schemes of finite type smooth is equivalent to geometrically reduced, which means that all stalks of the structure sheaf of $A×_K \bar{K}$ are reduced. In addition, let us point out that for $K$-group schemes of finite type the property irreducible can be checked after base change with $\bar{K}/K$ so that we may replace irreducible by geometrically irreducible...

We fix an abelian variety $A$ over field $K$. By definition an abelian variety over $K$ is a proper smooth $K$-group scheme that is irreducible.

Following two questions:

  1. Why is for a $K$-group scheme of finite type smooth equivalent to geometrically reduced?

  2. Why under same conditions as in 1. (so $K$-group scheme of finite type) the property irreducible is equivlaent to geometrically irreducible?

Remark: Here I previously asked this question in MSE: https://math.stackexchange.com/questions/3136827/abelian-varieties

  • $\begingroup$ I took the liberty of changing the title of your question. $\endgroup$ – Piotr Achinger Mar 7 at 19:43

Let $G/K$ be a group scheme of finite type.

  1. $G/K$ is smooth if and only if $\bar G / \bar K$ is smooth. Suppose $\bar G$ is reduced, then it has a smooth $\bar K$-point $x$ (because we are over an algebraically closed field). But $\bar G(\bar K)$ acts transitively on itself, so now every closed point of $\bar G$ is smooth, so $\bar G$ is smooth. (And of course if $\bar G$ is smooth then it is reduced.)

  2. The point is that $G$ comes with a section, the neutral element $e\in G(K)$. Suppose that $\bar G$ is reducible, then since $\bar G^{\rm red}$ is reduced and hence smooth, we see that $\bar G$ is disconnected. If $\bar G^\circ$ is the connected component of the neutral element $e$, then since the Galois group ${\rm Gal}(\bar K/K)$ acts on $\bar G$ preserving $e$, it has to preserve $\bar G^\circ$, and so $\bar G^\circ$ descends to give a component of $G$, so $G$ is disconnected.

  • $\begingroup$ Thank you for your answer. One penible question: When you talk about the ${\rm Gal}(\bar K/K)$-action on $\bar G = G \otimes \bar{K}$ do you implicitely mean the action on "points" $\bar{G}(\bar{K})= Hom(\bar{K}, \bar{G})$ via composing $\phi \mapsto \phi \circ g$ for a $g \in {\rm Gal}(\bar K/K)$ or the action via "base change" namely that $g$ induces automorphism on $\bar{G}$ via $id \otimes g$? $\endgroup$ – Karl_Peter Mar 7 at 21:06
  • $\begingroup$ @Karl_Peter I think the latter action induces the former on $\bar K$-points. Here it is enough to say that the profinite group ${\rm Gal}(\bar K/K)$ acts continuously on the underlying topological space $|\bar G|$ of $\bar G$ (or just the subspace of closed points $\bar G(\bar K)$) and the quotient space is identified with $|G|$. $\endgroup$ – Piotr Achinger Mar 7 at 22:56
  • $\begingroup$ But what is concretely the "canonical" action of ${\rm Gal}(\bar K/K)$ on the underlying topological space $|\bar G|$? $\endgroup$ – Karl_Peter Mar 7 at 23:36
  • $\begingroup$ The argument in the answer that the identity component of $G$ over the algebraic closure of $K$ is defined over $K$ applies only in characteristic zero. Otherwise the proof is more difficult. $\endgroup$ – anon Mar 9 at 7:01
  • 1
    $\begingroup$ @Karl_Peter See Section b of Chapter 1 of Milne: Algebraic Groups, The theory of group schemes of finite type over a field, Cambridge UP, 2017. $\endgroup$ – anon Mar 9 at 19:20

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.