7
$\begingroup$

Suppose that $X$ and $Y$ are smooth varieties over a field $k$ (not necessarily algebraically closed), of dimension $m$ and $n$. Suppose we are given a morphism $\pi:X\rightarrow Y$.

We know that if $\pi$ is smooth, it is flat (by definition) and all the fibers are smooth of dimension $m-n$.

Under these assumptions:

  • if all the (geometric) fibers are smooth of dimension $m-n$, does it follow that $\pi$ is flat (and hence smooth, e.g. by Vakil, Theorem 25.2.2)
  • more generally, can we say something about flatness of $\pi$ even if the fibers are not smooth, but still of constant dimension?

The first part is exercise 25.2.F from Vakil.

We can readily perform a number of reductions to simplify the question: the question is local on the target and on the source, so locally we can replace $Y$ by $\mathbb{A}^n_k$, by choosing an étale surjective map to $\mathbb{A}^n_k$. We can also replace $X$ by a smooth affine variety. It should also be sufficient to check this at closed points. Hence we have a morphism of rings $B \colon= k[y_1,\dots,y_n]\rightarrow A$, where A is a finitely generated $k$-algebra of dimension $m$ with $\Omega_{\mathrm{Spec }A/k}$ locally free of rank $m$, such that all the tensor products with maximal ideals of $B$ are (possibly smooth) of dimension $m-n$. Does it follow that $A$ is a flat $B$-algebra?

I must say that I don't really know how to approach this. I can't seem to relate the flatness and smoothness conditions, and the answer to this question is clearly false if $Y$ is not smooth, think of the normalization of the node, where all fibers are smooth of dimension $0$ (and probably also if $X$ isn't, although I don't have a counterexample in my head).

$\endgroup$
  • $\begingroup$ I misread your question, answered. Reread your question incorrectly again, answered again. Finally I think I've fixed it. I believe below is what you want. If you know that $X$ and $Y$ are both smooth over $k$, then the constant fibral dimension implies flatness, and so implies smoothness. Let me know if I have (for a third time!) misinterpreted you. :) $\endgroup$ – Alex Youcis Nov 22 '14 at 12:07
  • $\begingroup$ This time it's correct. Thank you! $\endgroup$ – Jakob Oesinghaus Nov 22 '14 at 12:08
  • $\begingroup$ No problem. Sorry for the (initial) confusion. :) $\endgroup$ – Alex Youcis Nov 22 '14 at 12:09
5
$\begingroup$

If you know that both $X$ and $Y$ are regular (even Cohen-Macauly), than constant fiber dimension implies flatness. See this for example. I believe this answers your question.

$\endgroup$
0
$\begingroup$

I guess that one should add some condition - that $X$ should be connected. Consider $(\mathbb{A}^1 - 0) \coprod pt \to \mathbb{A}^1$; Source and target are smooth of dimension 1, fibers are smooth of dimension 0, target and fibers are all connected, but the morphism is not flat.

$\endgroup$
  • $\begingroup$ Sorry, this is not correct. Scheme-theoretic fibers are not smooth. $\endgroup$ – Sasha Nov 24 at 1:56

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.