1. Let $X$ be a smooth irreducible $\Bbb C$-variety, on which the algebraic $\Bbb C$-group $G={\bf G}_{a,{\Bbb C}}$ (the additive group) acts freely on the right: $$ X\times _{\Bbb C} G\to X,\quad (x,g)\mapsto x\cdot g.$$ Assume that there exists a surjective morphism onto a smooth $\Bbb C$-variety $Y$ \begin{equation}\label{e:*} \varphi\colon X\to Y \tag{$*$} \end{equation} whose fibres are the orbits of $G$ in $X$. Then the morphism $\varphi$ is smooth, from which one can deduce that $\varphi$ induces a locally trivial fibre bundle (in the usual topology) of $C^\infty$-manifolds $$\varphi\colon X(\Bbb C)\to Y(\Bbb C).$$

Question 1. Does it follow that $(*)$ is locally trivial in the flat topology, that is, a $Y$-torsor under $G$? In other words, is the morphism $$ X\times_Y G\to X\times_Y X,\quad (x,g)\mapsto (x,x\cdot g) $$ an isomorphism of $\Bbb C$-varieties?

2. Assume that $(*)$ is a torsor. Since $H^1({\Bbb C}(Y),G)=\{1\}$, we know that $(*)$ admits a rational section.

Question 2. Does $(*)$ admit a regular section? In other words, does there exist a regular map (morphism) $s\colon Y\to X$ such that $\varphi\circ s={\rm id}_Y\,$?


For 1, yes. In fact, any smooth morphism of varieties admits a section locally in the etale topology everywhere.

Proof: A generic hypersurface section is smooth of dimension one lower over any particular point. Repeat until the relative dimension is zero.

For 2, no. Any variety $Y$ with $H^1 (Y, \mathcal O_Y) \neq 0$ is a counterexample, as every class gives a torsor $X$.

  • $\begingroup$ Thank you, Will, for a prompt answer! Could you please add details to the proof for 1? A generic hypersurface section of what? $\endgroup$ Apr 29 at 16:51
  • 2
    $\begingroup$ @MikhailBorovoi of $X$. For example, take $y\in Y$, $x \in X$ with $\varphi(x)=y$, and choose any function on (a neighborhood of $x$ in) $X$ vanishing at $x$ and whose first derivative is nonzero on the relative tangent space of $X$ over $Y$. The vanishing locus of this function is still smooth over $Y$ at the point $x$, thus smooth over $Y$ in a neighborhood of $x$. $\endgroup$
    – Will Sawin
    Apr 29 at 17:28
  • $\begingroup$ Does it follow that a morphism $\varphi$ as in my question induces a locally trivial fibre bundle of complex analytic manifolds (with the usual topology)? $\endgroup$ May 1 at 18:05
  • 1
    $\begingroup$ @MikhailBorovoi Yes. Locally trivializable in the etale topology implies locally trivializable in the flat, complex analytic, and several other topologies. In fact, the comparison between etale and Zariski cohomology for a quasicoherent sheaf implies that it is locally trivializable in the Zariski topology as well. The only subtlety I can think of, which applies equally to all these topologies, is that you want the schematic fibers to be orbits, i.e. you want the fibers to be reduced. $\endgroup$
    – Will Sawin
    May 1 at 18:14
  • $\begingroup$ Thank you! Is your argument for "In fact, any smooth morphism of varieties admits a section locally in the etale topology everywhere" written somewhere, so that I could refer? $\endgroup$ May 1 at 18:30

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.