Suppose I have a Cartier divisor D in a smooth variety X, and suppose I have a subvariety Y in X. Is it always possible to talk about restrict D to Y even when Y might be contained in D? I think it is true from a bundle point of view, because when I look at the example for standard blow up of Cn at the origin and restrict the exceptional divisor to itself, it makes sense. So my question is: what if I restrict D to a singular subvariety Y contained in D.

  1. Does it make sense. How to understand it?
  2. If the restriction is meaningful, is the restriction still Cartier in Y
  3. If D is only a Weil divisor, in general, is it meaningful to restrict it to a normal subvariety contained in D?

I really want to make it clear.

  • 4
    $\begingroup$ You essentially already answered your questions 1 and 2. For 3: if X is smooth, every Weil divisor is Cartier. If it isn't, then restriction doesn't always work. $\endgroup$ May 30, 2017 at 19:08
  • 2
    $\begingroup$ I would personally say that you cannot restrict the divisor, but you can restrict the divisor class, at least in the case of a Cartier divisor. For Weil divisors, even that might be out of reach. For an explicit example, see Example 11.4.6 in Fulton's Intersection Theory. $\endgroup$ May 30, 2017 at 22:30

1 Answer 1

  1. Chow's Moving lemma says given algebraic cycles Y, Z on a nonsingular quasi-projective variety X, there is another algebraic cycle Z' on X such that Z' is rationally equivalent to Z and Y and Z' intersect properly. That means there exists a Weil divisor D' (since X is smooth Cartier, Weil, Line bundles all are equivalent) which is linearly equivalent to Weil divisor D and intersect Y along a codimension 1 sub variety of Y. This may not be a Weil divisor(Since Y may not be locally factorial etc.). Also this restriction here is not unique in any case (as R. van Dobben de Bruyn was saying).

  2. It is a Cartier divisor if you assume Y is Projective or Integral. Because for projective ( or Integral) varieties Cartier and line bundles are equivalent (Hartshorne Remark 6.14.1).

  3. same as 1.


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.