1
$\begingroup$

I'm trying to teach myself about differential forms, and my book says that functions can't be integrated on manifolds because the integral isn't coordinate independent, but if the manifold has dimension $n$, you can integrated $n$ forms in a coordinate independent way.

The example given is simple, but I'm struggling with it. when the manifold is just $(\mathbb{R},1_{i.d.})$. Let $x$ and $y=Cx$ be the coordinate, and coordinate change, and $f(x)=1$. Then $\int_a^bf(x)dx=\int_a^bdx=\int_{Ca}^{Cb}\frac{dx}{dy}dy=(b-a)\neq C(b-a)=\int_{Ca}^{Cb} f(y)dy$. So I guess you would say that the Riemann integral is not coordinate invariant. But the claim is that you can integrate one-forms on a one dimensional manifold in a coordinate independent way. But $f(x)dx$ is a one-form, so I'm a little confused.

$\endgroup$

1 Answer 1

2
$\begingroup$

When considering coordinate changes, one-forms and functions are different things, as you transform them differently when changing coordinates. The Riemann integral of a one-form does not change after a coordinate change in this context, which is what allows classical "u-substitution", but it does change if you try to transform the one-form $f(x)\,dx$ as the function $f(x)$.

Explicitly, the one-form $f(x)\,dx$ should be pulled back along the transformation $x=a(u)$ as $f(x)\,dx \mapsto f(a(u)) a'(u)\, du$, in correspondence with the chain rule, rather than simply $f(a(u)) du$.

$\endgroup$
2
  • $\begingroup$ I probably don't understand what is meant by coordinate independent. Correct me if I'm wrong, but it seems like you're saying it means if you transform $f(x)dx$ and the limits of integration to the new coordinates, nothing changes. I was thinking it meant you have an integral $\int f(x)dx$ and if you formally just substitute $x$ for $y$, you get the same value for $\int f(y)dy$ $\endgroup$
    – user153582
    Oct 13, 2015 at 6:36
  • 2
    $\begingroup$ I mean to say, $\int f(x)dx\to\int f(y)dy$ would be considered changing coordinates of the function, $\int f(y(x))y'(x)dx\to\int f(y)dy$ would be considered changng coordinates of the one form? $\endgroup$
    – user153582
    Oct 13, 2015 at 6:41

You must log in to answer this question.

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