1. ## Circle is parallelizable.

it is clear (geometrically) that ${\mathbb S}^1$ is parallelizable because we can consider a vector field of unit tangent vectors at each point (say, in anti-clockwise direction).

I am having trouble understanding how to write down the above argument mathematically ....we need to come up with a smooth vector field $X: {\mathbb S}^1 -> T{\mathbb S}^1$, so that $X(p) = X^1(x) d/dx$.

what is the choice for a smooth function $X^1(x)$ to have the vector field as above??

can we consider the following (coordinate) vector field $X(p) = d/dx$??

2. Originally Posted by Different
it is clear (geometrically) that ${\mathbb S}^1$ is parallelizable because we can consider a vector field of unit tangent vectors at each point (say, in anti-clockwise direction).

I am having trouble understanding how to write down the above argument mathematically ....we need to come up with a smooth vector field $X: {\mathbb S}^1 -> T{\mathbb S}^1$, so that $X(p) = X^1(x) d/dx$.

what is the choice for a smooth function $X^1(x)$ to have the vector field as above??

can we consider the following (coordinate) vector field $X(p) = d/dx$??
I'd say yes... $X^1(x)=1$ is non-zero and continuous.

(note: you should write [ math ]...[ /math ] (without spaces) instead of $$, so that your math can be displayed properly) 3. Originally Posted by Laurent I'd say yes... $X^1(x)=1$ is non-zero and continuous. (note: you should write [ math ]...[ /math ] (without spaces) instead of$$, so that your math can be displayed properly)
Thanks a lot.

....I think we have to be carefull here, we need $X^1(x)=1$ to be smooth in any chart. will it be the case here?

For example, if we consider ${\mathbb S}^2$ then I don't think the field $X(p) = d/dx + d/dy$ will do the same trick...even though it's coordinate functions are constant=1 in some chart....but it is known that any smoth (or even continuous) field on a 2-sphere has at least one point where the tangent is zero.

4. Originally Posted by Different
Thanks a lot.

....I think we have to be carefull here, we need $X^1(x)=1$ to be smooth in any chart. will it be the case here?

For example, if we consider ${\mathbb S}^2$ then I don't think the field $X(p) = d/dx + d/dy$ will do the same trick...even though it's coordinate functions are constant=1 in some chart....but it is known that any smoth (or even continuous) field on a 2-sphere has at least one point where the tangent is zero.
The key fact is that $\frac{d}{dx}$ can be defined on the whole circle in a canonical way, while I don't know what you would mean by $X(p) = d/dx + d/dy$ on the sphere. Locally, sure it can be defined, but not globally. That's why the argument doesn't translate to the 2-d sphere.

For an explicit proof, you can say that the unit vector field defined by $X({\cos\theta\choose\sin\theta})={-\sin\theta\choose\cos\theta}$ is continuous.

5. Originally Posted by Laurent
The key fact is that $\frac{d}{dx}$ can be defined on the whole circle in a canonical way, while I don't know what you would mean by $X(p) = d/dx + d/dy$ on the sphere. Locally, sure it can be defined, but not globally. That's why the argument doesn't translate to the 2-d sphere.

For an explicit proof, you can say that the unit vector field defined by $X({\cos\theta\choose\sin\theta})={-\sin\theta\choose\cos\theta}$ is continuous.
yes, I meant locally.

now, why $\frac{d}{dx}$ defines a smooth vector field on the whole circle?

Also, regarding your last example of a vector field $X({\cos\theta\choose\sin\theta})={-\sin\theta\choose\cos\theta}$, ...I think I have a problem understanding how a vector field on the ambient space (in this case ${\mathbb R}^2$) translates to a vector field of a submanifold.

I am thinking of tangent vectors as derivations (or germs, in case of complex/analytic manifolds).

Now, can I argue as follows to show that the above vector field defines a smooth vector field on ${\mathbb S}^1$ :

Consider the following (smooth) vector field on ${\mathbb R}^2$ defined as $X = -y \frac{d}{dx} + x \frac{d}{dy}$.
I want to show that this defines a smooth vector field on ${\mathbb S}^1$. In other words, I need to show that this defines a derivation on ${\mathbb S}^1$.
So, let $f: U -> {\mathbb R}$ be a smooth function on some open set $U \subset {\mathbb S}^1$, so that $f(x , y) = f(\cos\theta , \sin\theta)$.

Therefore $X(f) = -\sin\theta \frac{df}{dx} + \cos\theta \frac{df}{dy} =$ (by the chain rule) $= \frac{df}{d\theta}$.

So that $X$ is a tangent vector to ${\mathbb S}^1$.

Now, why can I just, instead of the above formal argument, say that this defines a tangent vector to ${\mathbb S}^1$ simply because $(\cos\theta, \sin\theta)$ is orthogonal to $(-\sin\theta, \cos\theta)$?

Also, is there a general result that will guarantee that this restricted smooth vector field (restricted to a submanifold) will be smooth again?

Sorry about too many questions, but I feel that there is a gap in my understanding of the subject...

Thanks a lot!

6. Hi,

I won't have time for a lengthy answer; I hope the following will be of some interest nevertheless:

- the notation $\tfrac{d}{dx}$ is to be taken with care: the definition of $\tfrac{d}{dx}$ at $x\in M$ depends on the choice of a chart at $x$; $\tfrac{d}{d\theta}$ for instance refers implicitly to a parametrization of $\mathbb{S}^1(\subset\mathbb{C})$ by $\theta\mapsto e^{i\theta}$ (or with cos/sin) on the open subset $\mathbb{S}^1\setminus\{1\}$ or on $\mathbb{S}^1\setminus\{-1\}$. These two charts cover the circle, and the map is the "same" in both charts (we need two charts nevertheless because of the lack of injectivity), so that the derivation $\tfrac{d}{d\theta}$, defined locally (on each chart), is in fact global (coherent definition for both charts).

- there is a (nontrivial) equivalence between vector fields (i.e. indefinitely differentiable functions $X:x\in M\mapsto X(x)\in T_xM\subset TM$) and derivations.

- in the case of submanifolds of $\mathbb{R}^n$, the tangent subspace can be thought in various equivalent ways: either it is defined intrinsically, like for a general manifold (no need of the surrounding manifold), or it can be viewed as a subspace of $\mathbb{R}^n$ tangent to the submanifold (no wonder it is called "tangent" space). For the circle, as a submanifold of $\mathbb{R}^2$, the tangent space at a point can be identified with the line that is tangent to the circle at this point.

- if $N$ is a submanifold of $M$, and if $X$ is a vector field of $M$ such that, for $x\in N$, $X(x)\in T_xN$, then $X$ (restricted to $N$) is a vector field on $N$.

7. fantastic! everything makes more sense now. thanks a lot.