I want to solve the following exercise from M. Spivak's

If $F$ is a differentiable vector field on $M \subseteq \mathbb{R}^n$, show that there is an open $A \supseteq M$ and a differentiable vector field $\tilde{F}$ on $A$ with $\tilde{F}(x)=F(x)$ for $x \in M$. Hint: Do this locally and use partitions of unity.

My attempt:

If $x \in M$ let $U_x$ be an open set and $h_x:U_x \to V_x$ a diffeomorphism as in the definition of a manifold. We can restrict $V_x$ to an open ball around $a=h(x)$. Since a differentiable vector field on a manifold was defined as the pushforward of a differentiable vector field $G$ in Euclidean space, we can extend $F$ locally from $U_x \cap M$ to $U_x$ in the following way:

$$\tilde{F}(x):=F(G(\pi(x)) $$

where $\pi:\mathbb{R}^n \to \mathbb{R}^n$ is the projection to the first $k$ coordinates (here $k$ is the dimension of the manifold $M$).

After extending $F$ locally, I can't see how to combine the extensions using a partition of unity. I'd appreciate any help on that.

P.S. differentiability means smoothness in this case.

Thank you!


I do assume you know what a paritition of unity is and that there exists one which is subordinate to a given cover. With that in mind look at a locally finite open cover $U_x$ of $M$ by coordinate charts as in your question and associated $F_{U_x}$ (locally extended vector fields) and $\phi_{U_x}$ (paritition of unity). Then let $F = \sum_{U_x}\phi_{U_x} F_{U_x}$

If you feel uneasy because the $F_{U_x}$ are only defined on the corresponding chart note that $\phi_{U_x} F_{U_x}$ makes sense globally since the $\phi$ are supported in the coordinate charts.

  • $\begingroup$ Thanks Thomas. What I'm having trouble understanding is why this sum gives the extensions locally. Could you please elaborate on that? $\endgroup$ – user1337 May 31 '15 at 10:12
  • $\begingroup$ Sorry, but I don't really get the question. Do you want to know why the construction agrees with the original vector field on $M$? This is because the $\phi$ are a partition of unity, i.e. $\sum \phi_{U_z} = 1$, so whenever $F_{u_k} = F $ (wich is true on $M$) then $\sum \phi_{U_z} F_{U_z}(x) = \sum \phi_{U_z} F(x)= F(x) \sum \phi_{U_z}(x) = F(x) $ $\endgroup$ – Thomas May 31 '15 at 10:59

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.