I have a very basic question regarding Chern classes. Let $X$ be a smooth projective variety and $\mathcal{E}$ a vector bundle on it. Let $\pi:\mathbb{P}(\mathcal{E})\to X$ denote the projective space bundle over $X$ associated to $\mathcal{E}$.

How are the Chern classes $c_i$ of the vector bundle $\mathcal{E}$ on $X$ related to the Chern classes $c_i'$ of the projective variety $\mathbb{P}(\mathcal{E})$, i.e. the Chern classes of the tangent bundle of $\mathbb{P}(\mathcal{E})$? My naive hope would be that we simply have $c_i'=\pi^* c_i$. Is that true? Is there a good reference?

  • 1
    $\begingroup$ The answer is correct, but one can say more. There is a formula for the tangent bundle of a projective bundle that relates it to the pullback of the tangent bundle from below AND the relative tangent bundle. So there is a formula, but the formula makes it clear that what you are asking will almost never hold. $\endgroup$
    – meh
    Apr 13, 2019 at 20:10
  • $\begingroup$ Hi @aginensky ! That sounds interesting, can you maybe state this formula in a new answer or point me to a reference? $\endgroup$
    – Hans
    Apr 15, 2019 at 7:55

1 Answer 1


The nice relation that you hope for is not true unfortunately. Here is a counter-example: consider the trivial rank 2 vector bundle $E= \mathcal{O} \oplus \mathcal{O}$ over $\mathbb{CP}^{1}$. Then $$\mathbb{P}(E) \cong \mathbb{CP}^{1} \times \mathbb{CP}^{1}.$$

Now, note that $c_{2}(E)=0$ since $E$ is a trivial bundle, but $\int_{\mathbb{P}(E)} c_{2}(T\mathbb{P}(E)) = 4$, since this is equal to the topological Euler characteristic of $\mathbb{P}(E)$ (it is a standard result that the integral of the top Chern class of a complex manifold is equal to the topological Euler characteristic).

In general, I don't think that there will be a nice relation (although I could be wrong) since when $E$ has different Chern numbers, the topological type of $\mathbb{P}(E)$ will be different in general, so we are not even talking about Chern classes on the same complex manifold. However, you can actually compute some topological invariants of $\mathbb{P}(E)$ from the chern classes of $E$: see Proposition 15 of Okonek and Van de Ven's "Cubic forms and complex 3-folds".


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .