Let $f:X\rightarrow S$ be a finite type morphism of noetherian schemes and let $S$ be integral. Let $F$ be a coherent sheaf on $X$.

Let $T$ be an integral closed subscheme of $S$. Then we can apply generic flatness to $f^*F$ on $X_T\rightarrow T$. Then there is an open dense set $V\subset T$ such that $f^*F|_V$ is flat over $V$.

Can we find an open set $U$ of $S$ such that $F|_U$ is flat over $U$ and $V\subset U$?


I had another question about flat families that I am adding to this. If $W$ is the maximal subset of $S$ such that $F|_W$ is $W$-flat. Then is $W$ open in general? I think not. But what if $f$ is a faithfully flat morphism?

  • 3
    $\begingroup$ No. Let $S$ be $\text{Spec}\ k[y]$. Let $X$ equal $S$. Let $f$ be the identity map. Let $F$ be the coherent sheaf associated to the $k[y]$-module $k[y]/\langle y \rangle$. Let $T$ be the closed subscheme associated to the prime ideal $\langle y \rangle$. Then $f^*F$ is flat on $T$, which is a singleton set. Yet the maximal open subscheme $U$ of $S$ over which $F$ is flat is the open complement of $T$. Having said that, the notion of "flattening stratification" almost does what you want. $\endgroup$ – Jason Starr Mar 30 '17 at 11:56
  • $\begingroup$ @JasonStarr, thanks. Can you please elaborate on the comment -the notion of flattening stratification almost does what you want. $\endgroup$ – user349424 Mar 30 '17 at 12:55
  • $\begingroup$ When $f$ is proper, the flattening stratification of $F$ is a finite type, separated morphism $i:\Sigma \to S$ that is bijective on points and is a disjoint union of locally closed subschemes such that (i) $i^*F$ is $\Sigma$-flat on $X_\Sigma\to \Sigma$, and (ii) for every $f:T\to S$, $f^*F$ is $T$-flat on $X_T$ if and only if $f$ factors through $i$ (in which case the factorization is unique). In particular, if $T$ is integral, then there exists a unique stratum $U$ of $\Sigma$ such that $f^{-1}(U)$ contains a dense open subset $V$ of $T$. $\endgroup$ – Jason Starr Mar 30 '17 at 14:12

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.