I'm following Both-Tu book about differential topology and I wanted to understand the construction of the Thom Class of an (oriented) $n$-vector bundle $\xi\colon E \to B$ via forms. I already know the algebraic-topology way of construct if, I want to understand what's going on when working with forms.

The authors construct a global form $\psi$ on $E^0$ called the global angular form which is identified by the property that the restriction of the fibre defines a generator of $H^{n-1}(\Bbb R^n \setminus 0)$. Now via a bump form $d\rho=\rho'(r)dr$ one can claim that $d\rho \wedge \psi$ is a generator of $H^n_c(\Bbb R^n)$. Then the authors observe (page 71):

Once we have such angular form $\psi$, it is then easy to check that $\Psi=d(\rho\wedge \psi)$ is the Thom class

What I don't understand is that it seems to me that they are claiming that the Thom class is exact and therefore $0$ is cohomology, something clearly untrue. My algebraic topology background suggests me that such formula should therefore hold in $E^0$, where it is true that the Thom class is zero ($\Psi\in H^n(E,E^0)$), but the point of the bump form was to extend it to the entire $E$.

In fact at page 74 it's claimed: Although $\psi$ is defined only outside the zero section of $E$, the form $\Psi$ is a global form on $E$ since $d\rho \cong 0$ near the zero section.

I need some guidance in understanding the aforementioned equality, because it causing me a lot of confusion!


First, there's no need to put a wedge in your formula, since $\rho$ is a $0$-form. Second, $\rho\psi$ is not a globally-defined form on $E$ (even though it does have compact vertical support), although $d(\rho\psi)$ is a globally defined $n$-form. [This is akin to our writing $d\theta$ on the unit circle. $\theta$ is not globally defined, but $d\theta$ is globally defined. Similarly, $d\theta$ actually is a generator of $H^1(S^1)$, despite the fact that the notation suggests it's the zero class.]

  • $\begingroup$ Dear Ted, thank you. I see your point. $\endgroup$ – Luigi M May 8 '17 at 20:38

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.