# Pullbacks of cohomology classes along embeddings vs classes of intersections

Let $$X,Y$$ be a subvarieties of a smooth complex projective variety $$Z$$ and let $$i:X\hookrightarrow Z$$ denote the embedding. Lets further assume that the intersection $$X\cap Y$$ is transverse in $$Z$$.

Is it true that the cohomology classes $$i^*([Y])$$ and $$[X\cap Y]$$ in $$H^*(X)$$ are equal?

Here we are viewing $$X\cap Y$$ as a subvariety of $$X$$.

I've been told that if $$X$$ is smooth, this is likely true. However if we don't assume that $$X$$ is smooth or even irreducible and anything be said about the relationship between these two classes?

• See Dold's "Lectures on algebraic topology", proposition 11.18 and section 13 in chapter VIII. – Igor Belegradek Nov 8 at 13:44

This is true in the smooth case, assuming $$j:Y\hookrightarrow Z$$ is proper (inverse images of compact sets are compact; this holds for example if $$Y$$ is compact). It is a consequence of a more general fact, that pushforwards (or Gysin, or Umkehr homomorphisms) for proper maps commute with pullbacks. That is, given a transverse pullback square such as
$$\require{AMScd} \begin{CD} X\cap Y @>{f}>> Y\\ @V{g}VV @VV{j}V \\ X @>{i}>> Z \end{CD}$$ where $$j$$ is proper, and a cohomology class $$a\in H^*(Y)$$, we have $$i^*j_!(a) = g_!f^*(a) \in H^{*+\operatorname{dim}(Z)-\operatorname{dim}(Y)}(X).$$ At least, this is true under some orientability assumptions (such that the Gysin maps exist in whatever cohomology theory you are using) which should be satisfied for complex subvarieties and integral cohomology. Then your statement follows by taking $$a=1\in H^0(Y)$$, since $$[Y]=j_!(1)$$.