1
$\begingroup$

Let $X$ be a smooth quasi-projective toric variety of dimension $n$ over $\mathbb C$. Take it to be non-compact, so its fan is not complete. (A good example to keep in mind is a toric Calabi-Yau.)

If we denote by $D_i$ for $i=1,\ldots,d$ the irreducible subvarieties of codimension one that are stable under the torus action (corresponding to edges of the fan), what can be said about the kernel $K$ in the exact sequence $$ 0 \to K \to \oplus_{i=1}^d \mathbb Z D_i \to A_{n-1}(X) \to 0 $$ in this case?

I know that if $D_i$ is also compact, then its Poincare dual $PD[D_i]$ defines a class in compact support cohomology $H^2_c (X)$, but I'm confused how this is related with usual $H^2(X)$, hence my question above.

$\endgroup$
4
  • 1
    $\begingroup$ The kernel $K$ is just the monomial lattice $M$, and the map $M\to \bigoplus \mathbf{Z}D_i$ sends $m$ to $\operatorname{div}(m) = \sum \langle m, \rho_i\rangle \cdot D_i$ where $\rho_i\in N = M^\vee$ is the generator of the ray of the fan corresponding to $D_i$. $\endgroup$ Apr 30 at 20:40
  • 2
    $\begingroup$ Strictly speaking, the kernel is the image of this map, but it is injective in most cases. A necessary and sufficient condition is that the fan is not contained in a hyperplane of $N_\mathbf{R}$. $\endgroup$ Apr 30 at 20:42
  • $\begingroup$ @PiotrAchinger Thanks. Since $X$ is non-compact, the fan is contained in a hyperplane, right? if so, it is meaningful to ask for a characterization of the kernel, or more generally how to best deal with divisors in the non-compact setting. $\endgroup$
    – jj_p
    May 1 at 13:33
  • 1
    $\begingroup$ No, $X$ is compact if and only if the fan is complete i.e. it covers $N_\mathbf{R}$. There is plenty of fans which are not complete and not contained in a hyperplane, e.g. the fan corresponding to $\mathbf{A}^n$. $\endgroup$ May 1 at 15:17
4
$\begingroup$

As Piotr mentioned in the comments, you don't need completeness of the fan to be sure that the sequence $$0 \to M \to \oplus \mathbb Z D_i \to Cl(X_{\Sigma}) \to 0$$ is exact. The only condition that you actually need is that fan $\Sigma \subset N_{\mathbb R}$ is not contained inside (real) linear subspace of codimension 1. See Cox Little Schenck Theorem 4.1.3.

I believe you wanted to say "since $X$ is Calabi-Yau (necessary non-compact) the fan is contained in hyperplane" (which is true). In this case you can always write $$0 \to M/(\operatorname{ker}(M \to \bigoplus \mathbb Z D_i)) \to \bigoplus \mathbb Z D_i \to Cl(X_{\Sigma}) \to 0$$ This sequence is exact for any toric variety $X_{\Sigma}$ associated with fan $\Sigma$. And $\operatorname{ker}(M \to \bigoplus \mathbb Z D_i)$ is exactly the covectors which are ortohonal to the fan generators, so you can write $$0 \to M/L^{\bot} \to \bigoplus \mathbb Z D_i \to Cl(X_{\Sigma}) \to 0$$ where $L = \operatorname{span}_{\mathbb Z} \Sigma(1)^{gen} \subset N$ is $\mathbb Z$-submodule inside $N$ generated by ray generators of fan $\Sigma$ and $L^{\bot} = \{\phi \in M | \forall l \in L : \phi(l) = 0\}$ the annihilator.

$\endgroup$

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.