Let $M$ be connected and let $\pi:M\times N \rightarrow N$ be the natural projection. Prove that $p-$form $\omega$ on $M\times N$ is $\delta \pi(\alpha)$ for some $p-$form $\alpha$ on $N$ if and only if $i(X)\omega=0$ and $L_X\omega=0$ for every vector field $X$ on $M\times N$ for which $d\pi(X(m,n))=0$ at each point $(m,n)\in M\times N$

Here, $i(X)\omega$ is interior multiplication, $L_x\omega$ denotes the lie derivative of $\omega$ and $\delta \pi$ is the pullback of $\pi$.

I am not at all comfortable with the notion of interior multiplication. In my book, interior multiplication is defined as the transpose of the exterior product. That is,

Let $u\in \Lambda(V)$ and defined the endomorphism $\epsilon(u)$ by $\epsilon(u)(v)=u\wedge v$. Denote the transpose of $\epsilon(u)$ by $i(u)$. Then $i(u)$ is called interior multiplication by $u$ and it satisfies $(i(u)v^*,w)=(v^*,u\wedge w)$, where $v^*\in \Lambda(V^*)$ and $w\in \Lambda(V)$.

We define interior multiplcation by vector fields pointwise. That is,


The lie derivative of a differential form with respect to a vector field is defined as follows:

$(L_X\omega))_m=\frac{d}{dt}|_{t=0}(\delta X_t(\omega_{X_t(m)})),$ where $\delta X_t$ is the pullback.

Inner multiplication and lie derivatives are related by the following identity:

$L_X=i(X)\circ d+d\circ i(X)$

I think I don't understand how interior multiplication and lie derivatives work. Here is my attempt at the problem.

Suppose $\omega=\delta \pi(\alpha)$ and let $X$ be a vector field on $M\times N$ for which $d\pi(X)=0$ for all $(m,n) \in M\times N$. If we can show that $i(X)\omega=0$, then the identity will show that $L_x(\omega)=0$. Fix a point $p=(m,n)\in M\times N$ and $w \in \Lambda(M\times N)$. Then $(i(X_p)\omega_p,w)=(\omega_p,X_p\wedge w)=(\delta \pi(\alpha)|_p,X_p\wedge w)=(\alpha(d\pi),X_p \wedge w).$ I need to show that this is zero.

I have basically no idea about the other direction. Any help would be much appreciated.


(1) $M,\ N$ has charts $x,\ y$ Then if $\omega$ is a pull back then $\omega= (f\circ \pi) (x,y) dy_{1}\cdots dy_p$ where $f: N\rightarrow {\bf R}$ (For convenience, we can write it) So $$ \partial_{x_i} (f\circ \pi )= df\ d\pi \partial_{x_i} =0 $$

If $d\pi X=0$, then $X=g_i(x,y)\partial_{x_i}$ So $i_X\omega =0$ In further since $L_X=i_Xd +di_X$ so that $$L_X\omega = i_Xd\omega = i_X \bigg( \partial_{y_i}(f\circ \pi) dy_i dy_{1}\cdots dy_p \bigg) =0 $$

(2) $\omega=gdx_1\cdots dx_k dy_{k+1}\cdots dy_p+f dy_{1}\cdots dy_p$ where $1\leq k\leq p$ Since $i_X\omega =0$, then for $d\pi\ X=0$, $g=0$

Now we must prove that $f$ is independent of $x$ : If $X=\partial_{x_i}$, then $0=L_X\omega =i_Xdf dy_{1}\cdots dy_p $ That is $\partial_{x_i} f=0$

| cite | improve this answer | |
  • 1
    $\begingroup$ This doesn't seem correct. In (2) you need to consider terms with at least one $dx_i$. $\endgroup$ – Ted Shifrin Apr 19 '16 at 7:00
  • $\begingroup$ You are right I see I will correct it $\endgroup$ – HK Lee Apr 19 '16 at 7:11
  • $\begingroup$ @HKLee how is $\omega=(f\circ \pi)(x,y)dy_1\ldots dy_p$? $\endgroup$ – user210552 Apr 19 '16 at 7:47
  • 1
    $\begingroup$ Note that $(x,y)$ is chart for $M\times N$ so that $\omega$ is a combination of wedge products of $dx_i,\ dy_j$ But $\pi^\ast dy_i=dy_i$ so that $dx_i$ can not appear $\endgroup$ – HK Lee Apr 19 '16 at 7:56
  • $\begingroup$ @HKLee ah alright...also could you explain why $i_X\omega=0$? $\endgroup$ – user210552 Apr 19 '16 at 7:58

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.