Suppose $M$ is a smooth $n$-manifold and $(U,\varphi=(x^i))$ a smooth chart on $M$. For each point $p\in U$ we know that $(\frac{\partial}{\partial x^i}|_p)_{i=1}^n$ is a basis for the tangent space $T_pM$. If $(\tilde U,\tilde\varphi=(\tilde x^i))$ is another smooth chart on $M$ with $U \cap \tilde U\ne \emptyset$, then for each point $p\in U \cap \tilde U$ we have the following formula for the basis - change:

$$(\frac{\partial}{\partial x^1}|_p,\dots,\frac{\partial}{\partial x^n}|_p)=(\frac{\partial}{\partial \tilde x^1}|_p,\dots,\frac{\partial}{\partial \tilde x^n}|_p) \cdot \operatorname{Jac}(\tilde\varphi\circ\varphi^{-1})_{\varphi(p)} \quad[1]$$

where $\operatorname{Jac}(\tilde\varphi\circ\varphi^{-1})_{\varphi(p)}$ is the Jacobian matrix of $(\tilde\varphi\circ\varphi^{-1})$ in $\varphi(p)$.


Suppose we have the following vector field on $\mathbb{R}^2$: $X=2x \frac{\partial}{\partial x}$. Let $f:\mathbb{R}^2 \to \mathbb{R}, (x,y)\mapsto x^2$.

We want to compute the coordinate expression for $X$ in polar coordinates (on some open subset on which they are defined) and show that it is not equal to $\frac{\partial f}{\partial r}\frac{\partial }{\partial r}+ \frac{\partial f}{\partial \theta}\frac{\partial }{\partial \theta}$.

My understanding and execution

Let $M=\{(x,y)\in \mathbb{R}^2 : x>0\}$, and

$N=(0,+\infty)\times (-\frac{\pi}{2},\frac{\pi}{2})$.

Let $F=M\to N, (x,y)\mapsto (\sqrt {x^2+y^2},\arctan\frac{y}{x})$.

Then $F$ is a diffeomorphism with inverse $F^{-1}:N \to M, (r,\theta)\mapsto (r\cos\theta, r\sin\theta)$.

Issue 1: Polar coordinates are, by definition, the component functions of the map $F$, right? So we have $F=(r, \theta)$.

Great Issue 2: I have an ambiguity on how to interpret $\frac{\partial}{\partial r}$ and $\frac{\partial}{\partial \theta}$.

(a) Should I interpret $(r,\theta)$ also as standard coordinates on $N$ and thus $\frac{\partial}{\partial r}$ and $\frac{\partial}{\partial \theta}$ are simply the ordinary partial derivatives on $N$? (Just like $\frac{\partial}{\partial x}$ and $\frac{\partial}{\partial y}$ on $M$?)

(b) I know that if $M$ is a smooth $n$-manifold and $(U,\varphi=(x^i))$ is a smooth chart on $M$, then for each point $p\in U$ we have that $\frac{\partial}{\partial x^i}|_p=d\varphi^{-1}_{\phi(p)}(\partial_i|_{\phi(p)})$ where $\partial_i$ is the standard $i$-th partial derivative on $\mathbb{R}^n$. So thinking of my $F$ as a smooth chart, I should interpret $\frac{\partial}{\partial r}|_{(x,y)}=dF^{-1}_{F(x,y)}(\frac{\partial}{\partial x}|_{F(x,y)})$ for each $(x,y)\in M$.

Which one between (a) and (b) is the correct interpretation?

Issue 3 The vector field I'm looking for, say $Y$, should be the pushforward of $X$ by $F$?

Issue 4 Applying formula [1] I have

$$(\frac{\partial}{\partial x}|_{(x,y)},\frac{\partial}{\partial y}|_{(x,y)})=(\frac{\partial}{\partial r}|_{(x,y)},\frac{\partial}{\partial \theta}|_{(x,y)}) \cdot \operatorname{Jac}(F)_{(x,y)}$$ which i can rewrite

$$(\frac{\partial}{\partial x}|_{(r\cos\theta,r\sin\theta)},\frac{\partial}{\partial y}|_{(r\cos\theta,r\sin\theta)})=(\frac{\partial}{\partial r}|_{(r\cos\theta,r\sin\theta)},\frac{\partial}{\partial \theta}|_{(r\cos\theta,r\sin\theta)}) \cdot \operatorname{Jac}(F)_{(r\cos\theta,r\sin\theta)},$$ from which I have

$$\frac{\partial}{\partial x}|_{(r\cos\theta,r\sin\theta)}=\cos\theta \frac{\partial}{\partial r}|_{(r\cos\theta,r\sin\theta)}-\frac{\sin\theta}{r}\frac{\partial}{\partial \theta}|_{(r\cos\theta,r\sin\theta)};$$ finally we have $$X=2x\frac{\partial}{\partial x}=2r\cos\theta\frac{\partial}{\partial x}|_{(r\cos\theta,r\sin\theta)}=2r\cos^2\theta \frac{\partial}{\partial r}|_{(r\cos\theta,r\sin\theta)}-2\sin\theta\cos\theta\frac{\partial}{\partial \theta}|_{(r\cos\theta,r\sin\theta)}.$$ I think there is something wrong with this and whit using formula [1] in this context. Instead, using the formula for computing the pushforward of $X$ by $F$, I got $$(F_*X)_{(r\,\theta)}=2r\cos^2\theta \frac{\partial}{\partial r}|_{(r,\theta)}-2\sin\theta\cos\theta\frac{\partial}{\partial \theta}|_{(r,\theta)}$$ which sounds correct.

So is there some "abuse of notation" which is confusing me? Am I right in saying that strictly speaking I cannot use formula [1]?

Please excuse me for the very long post, I have done the possible to make it clear.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.