8
$\begingroup$

Let $X\subset \mathbb P^3$ be a smooth variety and $D \subset X$ be an effective, reduced, irreducible divisor. My question is the following.

If I know the defining equations of $X$ and $D$ then is there any software that can compute $H^2(X, T_X(-\log D))$?

$\endgroup$

1 Answer 1

3
$\begingroup$

Computing cohomology of the logarithmic tangent sheaf $T_X(-\log D)$ is a usually easier once you know the cohomology of the sheaves $T_X, \quad T_X(-D)$ and $N_D$, the normal bundle of $D$ in $X$. To get $H^i(X,T_X(-\log D))$ one can use the exact sequences $$ 0 \to T_X(-D) \to T_X(-\log D) \to T_D \to 0 $$or $$ 0 \to T_X(-\log D) \to TX\to N_{D}\to 0. $$These sequences make sense even when $D$ is singular (see Sernesi's book on deformation theory). These sheaves and their cohomology groups can be handeled in, say, Macaulay2, if you have the explicit equations.

EDIT: Let me try to explain the above in more detail: If one is in the lucky position that some of the cohomology groups of, say $N_D$, vanish (e.g., a line $L$ on a cubic surface has $h^0(L,N_L)=h^1(L,N_L)=0$, you can calculate the cohomology almost immediately. In the general case, you will have to write down the maps and calculate the cohomology using say, Cech cohomology. Macaulay2 have built-in routines for this, see for example Francesco Polizzi's answer to this question.

$\endgroup$
2
  • $\begingroup$ I have been trying to define such exact sequence in Macaulay2. Can you give me some reference of commands or a little example. Thanks $\endgroup$
    – SAG1
    Feb 18, 2011 at 2:52
  • $\begingroup$ The best reference for computing in M2 is Eisenbud/Sturmfels book on Macaulay2, I think you'll find this online. As for this question, you can define $T_X$ and $N_D$ as sheafification of modules in Macaulay2, see Polizzi's answer (and Eisenbud/Sturmfels). Similarily, you can define the map between them, $T_X \to N_D$ and let $T_X(-\log D)$ as the kernel of this map. This gives you a concrete presentation of $F=T_X(-\log D)$ and you can compute $h^2$ using 'HH^2(F)'. $\endgroup$
    – J.C. Ottem
    Feb 18, 2011 at 7:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.