Suppose we have a manifold $M$ of dimension $n$ and its cotangent bundle $T^*M$. The Liouville form $\lambda$ on $T^*M$ is defined as $\lambda_{\omega_p} = \pi^*(\omega_p)$ where $\pi$ is the standard projection map from $T^*M$ to $M$. The problem is to prove that $d\lambda$ is a nondegenerate 2-form on $T^*M$. I was able to solve this problem by simply expressing the form with the aid of local coordinates and deriving the result by straightforward computation. However, I also have a bogus proof where I can't identify the incorrect step. Exponents are repeated wedge products.

$d\lambda_{\omega_p} = d\pi^*(\omega_p)$

$(d\lambda_{\omega_p})^n = (d\pi^*(\omega_p))^n = \pi^*(d\omega_p)^n$

(Because pullbacks commute with exterior differentiation and wedge products).

However, $\pi^*(d\omega_p)^n = 0$ because it is the image of a $2n$ form on a $n$ dimensional manifold.

Where am I going wrong? I'm guessing I'm missing some hypothesis in order to use the commutative fact but I am not sure. Thank you.


1 Answer 1


You can't differentiate a pointwise expression. In particular, $\lambda$ is not the pullback of a form on $M$.

  • $\begingroup$ I think I understand what you're saying but just to make it clear I want to see if what I'm saying in my own words is correct. You can't differentiate a specific covector i.e. $d\omega_p$ makes no sense. However, if $\omega$ is a form on $M$ we do have $(d\omega)_p$. $\endgroup$
    – Memeozuki
    Aug 20, 2014 at 0:48
  • 1
    $\begingroup$ Right! Similarly, a calculus student must take the derivative and then evaluate, not vice-versa. $\endgroup$ Aug 20, 2014 at 1:59

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.