4
$\begingroup$

I think I need some hints about a proof I am currently reading in order to understand it. This question is similar to the construction used in Lemma 17.9 in the book "Introduction to smooth manifolds" by John Lee. Unfortunately, I am currently not at the level to understand his notation, although it would probably answer my question. Let $\omega: M \times [0,1]$ be a differential 2-form ( $M$ is a smooth manifold), then I want to transform it into a 1-form on $M$ by a linear map $I : \Omega^2(M \times [0,1]) \rightarrow \Omega^1(M).$

So the idea is somehow to integrate with an operator $I$ over the $[0,1]$ part in order to get the right differential form.

First, I notice that each $2-$ form on $M \times[0,1]$ is necessarily of the form $\omega = \omega_1 \wedge dt + \omega_2,$ where $\omega_1$ is a 1-form depending only on coordinates of the manifold, $t$ is the coordinate of the interval here and $\omega_2$ is a 2-form depending only on coordinates of $M$. Clearly, by skew symmetry of the wedge product, there cannot be a 2-form with $dt \wedge dt.$

Now, the big question is: How do I define $I(\omega)$ to get the following results; cause I am currently reading some lecture notes (not online unfortunately) where exactly such a construction is used, and it is said that its integration with respect to the $[0,1]$ part, but it is not explicitely mentioned how it is defined, so what this operator explicitly does to a form on $M \times [0,1]$. Altogether, I am looking for an explanation of what this operator $I$ exactly does. Let me provide you with the calculation given in the notes: So $\omega_1 = \sum_{i=1}^{n} a_i(x,t) dx_i,$ where $x_1,...,x_n$ are coordinates of the manifold $M$ and $\omega_2 = \sum_{i<j} b_{i,j}(x,t)dx_i \wedge dx_j$

Now, it is said that $I(\omega ) = \int_0^1 \omega_{p-1} dt.$ So somehow this operator does not affect $\omega_p$ if I get this correctly(maybe because there is no $dt$). Furthermore, for $d \omega = d \omega_{p-1} \wedge dt + d \omega_{p}$ which can be written in coordinates as $$ d\omega_{p-1} \wedge dt= \sum_{i=1}^{n} \sum_{j=1}^{n} \partial_{x_j} a_i(x,t) dx_j\wedge dx_i \wedge dt$$

and $$d \omega_p = \sum_{i<j} \left(\sum_{k=1}^{n} \partial_{x_k}b_{i,j}(x,t) dx_k \wedge dx_i \wedge dx_j + \partial_t b_{i,j}(x,t)dt\wedge dx_i \wedge dx_j \right),$$ we get by linearity $I(d\omega) = I( d \omega_p) + I (d\omega_{p-1}\wedge dt)$ which shall altogether give us: $d(I(\omega))-I(d\omega) = (-1) (\sum_{i<j} b_{i,j}(x,1)-b_{i,j}(x,0)dx_i\wedge dx_j).$

This is what the lecture notes contain, can we tell from this how $I$ exactly acts on forms on $M \times[0,1]$?

Especially, why this operator apparently gives zero, if the form does not depend on $t$ via $dt$, although $a,b$ depend on $t$ in general?

If anything is unclear, please let me know.

$\endgroup$
2
$\begingroup$

You have things sort of confused. For a $2$-form $\omega$, represented as $\omega = dt\wedge\omega_1 + \omega_2$ (I prefer to put the $dt$ first to spare sign problems), we define $$I(\omega) = I(dt\wedge\omega_1) = \sum_{i=1}^n\big(\int_0^1 a_i(x,t)dt\big)dx_i.$$ However, to prove your formula, we need to define the operator $I$, in general, on $p$-forms, and we do this analogously. In local coordinates $(x_1,\dots,x_n)$, write a general $p$-form $\omega$, similarly to the above, as $\omega=dt\wedge \omega_1 + \omega_2$, where $\omega_1$ is a $(p-1)$-form involving only $dx_i$'s, and $\omega_2$ is a $p$-form involving only $dx_i$'s. Write $\omega_1 = \sum\limits_I a_I(x,t)dx_I$, where these are multi-indices of length $p-1$. Then we set, once again, $$I(\omega)= I(dt\wedge\omega_1) = \sum_I\big(\int_0^1 a_I(x,t)dt\big)dx_I.$$

Now it will follow that for any $p$-form $\omega$ on $M\times [0,1]$, $d(I(\omega))+I(d\omega) = \iota_1^*\omega - \iota_0^*\omega$, where $\iota_t(x)=(x,t) \in M\times [0,1]$.

$\endgroup$
  • $\begingroup$ thank you. that was really illuminating $\endgroup$ – RealAnalysis May 18 '15 at 21:32
  • $\begingroup$ You're most welcome. $\endgroup$ – Ted Shifrin May 18 '15 at 23:48

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.