8
$\begingroup$

let $S=\operatorname{Spec}(A)$ be an affine scheme. For which ring $A$, not field is it known that $H^1(S,\mathcal{O}_S^{*})$ is trivial?

If $X\to S$ is a finite map and $H^1(S,\mathcal{O}_S^{*})$ is trivial, is it true that also $H^1(X,\mathcal{O}_X^{*})$ is trivial?

Thanks

$\endgroup$
11
$\begingroup$

Sample answers to your first question: If $S$ is Spec of a local ring, or of a UFD, then $H^1(S, \mathcal O_S^{\times})$ is trivial.

The answer to your second question is no: $X = $Spec $\mathbb C[x,y]/(y^2 - x^3 +x) \to S =$ Spec $\mathbb C[x]$ gives a counterexample of a geometric nature, and $X =$Spec $\mathbb Z[\sqrt{-5}] \to S =$ Spec $\mathbb Z$ gives a counterexample of an arithmetic nature.

$\endgroup$
  • $\begingroup$ is this true for coherent cohomology? Namely if $S$ is affine and $\mathcal{F}$ is coherent on $X$, $X\to S$ finite, is it true that $H^i(X,\mathcal{F})=0$ for $i>0$? $\endgroup$ – ullo Jul 6 '12 at 14:31
  • 3
    $\begingroup$ Yes, because $X$ is affine under your hypotheses! $\endgroup$ – Georges Elencwajg Jul 6 '12 at 15:31
  • $\begingroup$ @Matt E: can you explain how to compute $H^1(X,\mathcal{O}_X^\times)$ in your geometric example? $\endgroup$ – adrido Jun 5 '14 at 19:19
  • 2
    $\begingroup$ @adrido: Dear adrido, Recall that $H^1(X,\mathcal O_X^{\times})$ is the Picard group of $X$. In the geometric example, the curve $X$ is a once-punctured geometric curve, so its Pic is equal to Pic of an elliptic curve modulo the copy of $\mathbb Z$ generated by a single point (i.e. the point we punctured), which is naturally isomorphic to Pic${}^0$ of the elliptic curve. But this is isomorphic to the elliptic curve itself. In other words, in the geometric example, we have $H^1(X,\mathcal O_X^{\times})$ is isomorphic to an elliptic curve. Regards, $\endgroup$ – Matt E Jun 6 '14 at 5:55
  • $\begingroup$ @MattE: Dear Matt, that's brilliant, thank you! so the projective curve defined by $wy^2-x^3+wx^2$ is elliptic and once you take out a point (w=0) you get the affine curve above. I wonder if you can generalize this trick to compute the Picard group of affine varieties that are cones over a nice projective variety? e.g. V(xy-z^2)? $\endgroup$ – adrido Jun 6 '14 at 16:26

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.