Let $P$ be a normal, $\mathbb{Q}$-Gorestein variety with terminal singularities. Let $X \subseteq P$ be a normal, irreducible Weil divisor such that $X \sim_{\mathbb{Q}} - K_P$, that is $\mathbb{Q}$-linearly equivalent to the anticanonical divisor of $P$ (which is $\mathbb{Q}$-Cartier) .

My question is, does $X$ have $\mathbb{Q}$-trivial canonical divisor? In the adjuction formula for Weil divosr, it seems there is a $Diff$ term appears. I was wondering if this term is zero here?

  • $\begingroup$ You probably mean $\mathbb{Q}$-trivial? $\endgroup$ – abx Jan 28 '15 at 14:55
  • $\begingroup$ Yes, that is want I meant, thank you! $\endgroup$ – Li Yutong Jan 28 '15 at 16:01
  • 1
    $\begingroup$ Spelling? "Gorenstein"? $\endgroup$ – paul garrett Jan 29 '15 at 15:42

How about this: As you assumed, $P$ has terminal singularities, which means that the singular locus of $P$ has codimension at least 3 (see Corollary 5.18 in Kollár-Mori's book). Hence, the singular locus has codimension at most 2 in $X$. We can consider everything outside the singular locus because we work on the level of divisors of $X$. Hence $P$ is just smooth and $X$ is Cartier and everything is true for you.

EDIT: @LiYutong, of course I did not mean that $K_P$ is Cartier (which is not true in general). I mean that there exists a codimension 3 subset $Z$ of $P$ such that $U=P-Z$ is smooth and hence $X|_U$ is Cartier divisor on $U$. Hence $K_{X|_U}=(K_U+X|_U)|_{X|_U}$ is $\mathbb{Q}$-trivial by adjunction. Note that $X-X|_U$ has codimension at least 2 in $X$, so the Weil divisors class $\rm{Cl}(X)$ on $X$ coincides with the Weil divisors class $\rm{Cl}(X|_U)$ on $X|_U$ by restriction. So $K_X\sim_\mathbb{Q}0$.

| cite | improve this answer | |
  • $\begingroup$ Dear Chen Jiang, thank you for your answer! Since I assume $X$ is normal, $X$ is already smooth in codimensional 1. Do you want to argue: when $P$ smooth in codimensional 2, then the canonical divisor $K_P$ is Cartier (which I don't think is true by considering some toric example)? $\endgroup$ – Li Yutong Jan 29 '15 at 13:59
  • $\begingroup$ @LiYutong, I edited. Please check additional explanation. $\endgroup$ – Chen Jiang Jan 29 '15 at 14:44
  • $\begingroup$ Yes, this makes sense for me, thank you again!! $\endgroup$ – Li Yutong Jan 29 '15 at 15:07

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.