This is the set up of my problem:

Let $M$ be a manifold with local coordinates $x^1,\dots, x^n$. Let $x^1,\dots,x^n,\xi_1,\dots,\xi_n$ denote the induced coordinates on $T^\ast M$. Let $f:M\to M$ be a diffeomorphism and consider the induced map

$$F:T^\ast M \to T^\ast M \ , \ (p,\xi)\mapsto (f(p),(f^\ast)^{-1}(\xi)).$$ This is the inverse of the pullback map. We get a new local coordinate system on $T^\ast M$ given by $\widetilde x^i=f(x^i)$ and $\widetilde\xi_i=(f^\ast)^{-1}_x(\xi_i)$

I know the chain rule gives $$\frac{\partial}{\partial x_i}=\frac{\partial \widetilde x^j}{\partial x^i}\frac{\partial}{\partial\widetilde x^j}.$$For simplicity let $x=(x^1,\dots,x^n)$ and $\xi=(\xi_1,\dots,\xi_n)$. I know that if the $\xi$'s were coordinates on the tangent bundle we would have $$\xi_i=\frac{\partial\widetilde x^j}{\partial x^i}\widetilde\xi_j$$ but I am not sure if this is true as coordinates on the cotangent bundle.

Anyway, I was told that since $(f^\ast)^{-1}(x,\xi)=\widetilde\xi$ it follows that $$\widetilde\xi_i=\frac{\partial x^j}{\partial \widetilde x^i}\xi_j$$I am hoping that someone can explain it to me. I know it has something to do with the inverse transpose of the change of basis matrix. That is, working with induced coordinates $x^1,\dots,x^n,v^1,\dots,v^n$ on the tangent bundle we get the formula $$v^j=\frac{\partial\widetilde x^i}{\partial x^j}\widetilde v^i$$so that $$\frac{\partial \widetilde v^j}{\partial v^i}=\frac{\partial \widetilde x^j}{\partial x^i}$$and we are flipping the roles of $i,j$ and the tildes.


Let $f:\mathbb{R}^n\to\mathbb{R}^n$ be a smooth function. Let $\xi\in T_{f(0)}^*\mathbb{R}^n$ a cotangent vector, and say $\xi=(\xi_1,\ldots,\xi_n)$, when the coordinates are with respect to the dual of the standard basis i.e. $\xi=\sum \xi_idx^i$. By definition of the pullback, $$f^*\xi=\xi\circ df,$$ and if we put $f^*\xi=u\in T_0^*\mathbb{R}^n$ we have $$\xi\circ df=u,$$ and so $$\xi=u\circ df^{-1}.$$ If we want to express the coordinates $\xi_1,\ldots,\xi_n$, we let $J$ denote the derivative matrix of $f$ and get $$(\xi_1,\ldots,\xi_n)=(u_1,\ldots,u_n)J^{-1},$$ or rather $$\left(\begin{array}{c}\xi_1\\ \vdots\\ \xi_n\end{array}\right)={J^{-1}}^T\left(\begin{array}{c}u_1\\\vdots\\u_n\end{array}\right).$$


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.