4
$\begingroup$

I would like to know in what generality do we have pushouts for schemes/stacks/derived stacks. More precisely, let $f : X \rightarrow Y$ be a proper flat surjective morphism of schemes of finite type over $k$ and $g: X \rightarrow Z$ be an etale map. Is there a pushout $T$ (scheme? stacks? derived scheme?) that makes the diagram commutative:

$$ \begin{matrix} X & \rightarrow & Z \\ \downarrow & & \downarrow \\ Y & \rightarrow & T \end{matrix} $$

I think that if $ f : X \rightarrow Y$ is assumed to be a closed immersion, then Ferrand proved that such a pushout exists in the category of schemes. But I am really interested when $f : X \rightarrow Y $ is a proper flat surjective morphism.

Thanks in advance.

EDIT : People gave me all sorts of nice references, but whatever the level of abstraction (schemes, stacks, derived stacks) the theorems they refer to always concern closed immersions. However, there are some "trivial" cases where the pushout exists without closed immersions coming into the picture. For instance, assume that $g : X \rightarrow Z$ is the quotient of $X$ by a finite group $G$ acting transitively without fixed points and assume that $f$ is $G$-equivariant. Then, it seems clear that the pushout exists.

If you remove the $G$-equivariant assumption on $f$, but you impose some regularity (like flatness or smoothness), I was wondering if the pushout still exists, perhaps as a very abstract object (derived stack?)?

Improve this question
$\endgroup$
3
  • $\begingroup$ It is false for $T$ a scheme with the "expected" space with $X$ finite etale over $Y$ and $Z$. Let $Z$ be a separated finite type $k$-scheme with a free action by a finite group $G$ such that the separated algebraic space $Z/G$ is not a scheme (these exist). Then $X=Z \times_{Z/G}Z$ is a scheme finite etale over $Y=Z$ and $Z$ with $Z/G$ an algebraic space pushout. If $T$ were a scheme pushout of finite type over $k$ with the "expected" underlying space then the unique $Z/G\rightarrow T$ is separated and quasi-finite, so $Z/G$ is a scheme since $T$ is. $\endgroup$
    – user36938
    Commented Jul 23, 2013 at 20:22
  • $\begingroup$ You may want to look at the appendix to version 1 of J. Hall and D. Rydh's "The Hilbert Stack". arxiv.org/abs/1011.5484v1 $\endgroup$
    – Jonathan Wise
    Commented Jul 23, 2013 at 20:24
  • $\begingroup$ Since it doesn't seem to have been mentioned yet, the title of Ferrand's paper is "Conducteur, descent, et pincement". $\endgroup$
    – Shane Kelly
    Commented May 2, 2017 at 14:28

1 Answer 1

Reset to default
1
$\begingroup$

This note of Karl Schwede gives several circumstances under which pushouts of schemes exist.

Improve this answer
$\endgroup$
4
  • 2
    $\begingroup$ It doesn't have anything in it that Ferrand didn't really already know though. $\endgroup$
    – Karl Schwede
    Commented Jul 23, 2013 at 19:01
  • $\begingroup$ I did not read Karl's paper completely, but it seems it only deals with closed immersions. I am really interested in the case where $f$ is NOT a closed immersion (but a flat proper surjective morphism). $\endgroup$
    – Libli
    Commented Jul 23, 2013 at 19:55
  • 1
    $\begingroup$ Oh well...it's a nice note though :). $\endgroup$
    – Daniel Litt
    Commented Jul 23, 2013 at 20:11
  • $\begingroup$ The link is broken.. math.utah.edu/~schwede/Papers/SchemeWithoutPoints.pdf is active now.. $\endgroup$
    – Praphulla Koushik
    Commented Apr 27, 2020 at 4:11

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .