2
$\begingroup$

consider a proper, flat family of schemes $X\rightarrow S$, with $S$ affine. I would like to know under which condition on the family the functor $Sch/S \rightarrow Ab$

$ T \rightarrow \Gamma(X_T,\mathcal{O}_T^{*}) $

is smooth.

Namely if $Spec(B)\rightarrow Spec(A)$ is a nilpotent immersion between artinian $S$-rings then the restriction

$ \Gamma(X_{Spec(A)},\mathcal{O}_{X_{Spec(A)}}^{*})\rightarrow\Gamma(X_{Spec(B)},\mathcal{O}_{X_{Spec(B)}}^{*}) $ is surjective. Is this at least true for a family of curves over a discrete valuation ring, with generic fiber smooth and special fiber nodal?

Improve this question
$\endgroup$
2
$\begingroup$

Suppose $S$ is noetherian (otherwise suppose $X\to S$ is finitely presented). It is known that the functor $$ T \mapsto \Gamma(X_T, O_{X_T})$$ is represented by an $S$-scheme $V$ which is the Spec of the symetric algebra of some finitely generated module $M$ over $R=O(S)$ (see Bosch-Lütkebohmert-Raynaud: Néron models, §8.1, Corollary 8), and your functor is represented by an open subscheme $V^*$ of $V$ (op. cit., 8.1, Lemma 10). By construction, the fibers of $V\to S$ are vector spaces and $V^*$ is fiberwise dense in $V$.

If $V^*\to S$ is smooth, then $V$ is smooth because $V\to S$ is a group scheme. The converse is obviously true because $V^*$ is open in $V$. So $V^*\to S$ is smooth if and only if $V\to S$ is smooth.

Now $V\to S$ is smooth if and only if $X\to S$ is cohomologically flat in dimension $0$ (op. cit., 8.1, Cor.8), i.e. $\Gamma(X_T, O_{X_T})=\Gamma(X, O_X)\otimes_R O(T)$ for any affine $T$ over $S$. This is also equivalent to $$\Gamma(X_s, O_{X_s})=\Gamma(X, O_X)\otimes_R k(s), \quad \forall s\in S.$$ This is true if the fibers of $X\to S$ are geometrically reduced (EGA III, 7.8.6). In particular, if the fibers of $X\to S$ are nodal curves, then your functor is smooth.

Improve this answer
$\endgroup$
  • $\begingroup$ There are 2 questions: 1)the smoothness is weaker than the cohomological flatness in dimension zero. Namely we only need lifting over nilpotent extensions, do you know when this happens? 2) If I change $\mathcal{O}_X^{*}$ with a locally(\'etale topology) constant sheaf of finitely generated abelian group $G$ on $X$ do you know under which conditions on the family the functor $T\rightarrow f_{T,\'et,*}(G_T)$ is formally smooth? $\endgroup$ – unky Nov 6 '12 at 18:48
  • $\begingroup$ There are 2 questions: 1)the smoothness is weaker than the cohomological flatness in dimension zero. Namely we only need lifting over nilpotent extensions, do you know when this happens? 2) If I change $\mathcal{O}_X^{*}$ with a locally(\'etaletopology)constant sheaf of finitely generated abelian group $G$ on $X$ do you know under which conditions on the family the functor $T\rightarrow f_{T,\'et,}(G_T)$ is formally smooth ? $\endgroup$ – unky Nov 6 '12 at 18:53
  • $\begingroup$ @unky:the lifting property for nilpotent extension is equivalent to the smoothness for scheme of finite type. I explained that the smoothness is equivalent to coh. flatness in dimension $0$. Anyway you said you are satisfied with semi-stable curve over DVR. They are coh. flat. in dimension $0$. For (2), please post a different message, preferably at mathoverflow rather than here. $\endgroup$ – user18119 Nov 6 '12 at 20:53

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.