Consider the $2$-form $e^{xy}dx \wedge dy$ on $\mathbb{R}^2$.

Determine the $1$-form that it induces on $S^1$, viewed as the boundary of $B_2$.

Check if the obtained $1$-form respects or not the induced orientation.

Honestly I don't know where to begin, some help would be appreciated.

Somehow the answer for the first part is $\frac{e^{xy}}{y}dx$ or $-\frac{e^{xy}}{x}dy$.

The induced orientation of $S^1$ is given by $dx$ (this I know).

Thanks in advance.

up vote 1 down vote accepted

Assuming Euclidean metric on $\mathbb R^2$ and given a weighted "volume form" $\omega=e^{xy} dx \wedge dy$ which is not the natural volume form for this metric, we're asked to find its induced "volume form" $i^{*}(\iota_\nu \omega)$ on $S^1$, where we first take the interior product with the unit outer normal $\nu=x\partial_x+y\partial_y$ and then pull back to $S^1$ along the inclusion map.

Using the linearity of the interior product we arrive directly at @James's result: $\iota_\nu \omega=(-ydx+xdy)e^{xy}$. One could transform it further to either $\frac{e^{xy}}{y}dx$ or $-\frac{e^{xy}}{x}dy$ by differentiating and using again the $x^2+y^2=1$ and then for some reason changing the sign as @Ted noted.

I guess what I write here is wrong. But, I'm not sure exactly why, so I post it here in the hope someone points out the error of my ways. The two-form $\omega = e^{xy}dx \wedge dy$ may be written as $\omega = e^{r^2 \sin \theta \cos \theta} r \, dr \wedge d\theta$. Then, in polar coordinates, the unit circle is $r=1$ and $\partial / \partial r$ is an outward-pointing vector field on the circle. If I understand the induced orientation of $\omega$ correctly, then all I need to do is to take the inner product (sometimes called "hook") of $\omega$ with $\partial/\partial r$ and pull that back to the circle. Calculate $$ \omega(\partial/\partial r, \cdot) = e^{r^2 \sin \theta \cos \theta} r dr(\partial/r)d\theta = e^{r^2 \sin \theta \cos \theta} r d\theta$$ but, this is to be pulled-back to the circle. In Cartesians, $$ d\theta = \frac{-ydx+xdy}{x^2+y^2} $$ thus, using a hybrid notation momentarily, $$ \omega(\partial/\partial r, \cdot) = e^{r^2 \sin \theta \cos \theta} r \frac{-ydx+xdy}{x^2+y^2} = \frac{-ydx+xdy}{\sqrt{x^2+y^2}}e^{xy}$$ But, I would think the pull-back of this form to the curve $x^2+y^2=1$ simply gives us: $$ \phi = (-ydx+xdy)e^{xy}$$ Given all this, I'd think $\phi$ is consistent with $dx$ when $y<0$ but, opposite $dx$ when $y>0$. I'm not sure how to get the quoted answers in the stated question. Roughly I follow Lee's Smooth Manifolds, which is also discussed in Inducing orientations on boundary manifolds .

  • 1
    James and Leafar: The $e^{xy}$ is a complete red herring, inasmuch as it is always positive. The $2$-form $dx\wedge dy$ gives the "usual" orientation on $\Bbb R^2$, and the boundary orientation on $S^1=\partial D^2$ is, indeed, given by the $1$-form obtain by contracting against the outward normal ($\partial/\partial r$), so the induced orientation is given by $d\theta = -y\,dx+x\,dy$. – Ted Shifrin Jun 9 '15 at 14:32
  • 1
    @TedShifrin Thanks for the comment, do you have any idea where they get $(e^{xy}/y) dx$ or $-(e^{xy}/x) dy$ as the induced orientations? – James S. Cook Jun 9 '15 at 15:00
  • 1
    No, in fact, it's off by a sign. As you said, the convention for boundary orientation is that the outward-pointing normal goes in the first slot. – Ted Shifrin Jun 9 '15 at 15:06

Your Answer


By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.