Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I have a problems doing exercise 16 of chapter 3 (p.98 in my edition) of Spivak's book. The problem is very simple. Let $M$ be a manifold with boundary, and choose a point $p\in\delta(M)$. Now consider an element $v\in T_p M$ which is not spanned by the vectors on $T_p\delta(M)$, that is, it's last coordinate is non-zero (after good identifications). We say that $v$ is inward pointing if there is a chart $\phi: U\rightarrow \mathbb{H}^n$ ($p\in U$) such that $d_p\phi(v)=(v_1,\dots,v_n)$ where $v_n>0$.

It is asked to show that this is independent on the choice of coordinates (on the chart).

I think that Spivak's idea is to realize first that the subespace of vectors in $T_p\delta M$ is independent on the chart, which can be seen noticing that if $i:\delta (M)\rightarrow M$ then $d_pi(v_1,\dots,v_{n-1})=(v_1,\dots,v_{n-1},0)\in T_p \mathbb{H}^n$

share|improve this question
"Spivak's book" ? Spivak has writtten more than one book, and as far as I know he defines in all of them "outward pointing" by $v_n\lt 0$ ... –  Georges Elencwajg Jan 10 '13 at 19:34
The book is, as I said in the title of the question, A comprehensive introduction to differential geometry vol1. And I don't think that the sign of the last coordinate affects the core of the question, take $v_n<0$ as definition of outward pointing if you want. –  Miguel Jan 11 '13 at 2:42
It is preferred on this site that the body of the question be independent of its title. Also, although the core of a question is indeed not affected by the convention, words have an accepted meaning in mathematics and you are not allowed (here or elsewhere) to change a universally adopted convention nor to misquote an author. –  Georges Elencwajg Jan 11 '13 at 8:22
Ok, I thought this place was meant to be more informal!, anyway I will be more careful in the future, post the whole references and not to misquote things. Sorry! –  Miguel Jan 11 '13 at 12:31
No problem Miguel, and welcome to our site! –  Georges Elencwajg Jan 11 '13 at 12:36

1 Answer 1

up vote 0 down vote accepted

A change of coordinates between charts for the manifold with boundary $M$ has the form $x=(x_1, \cdots,x_n) \mapsto (\phi_1(x), \cdots,\phi_n(x))$, with $x_n, \phi_n(x)\geq0$ since $x_n,\phi_n(x)\in \mathbb H_n$.

The last line of the Jacobian $Jac_a(\phi)$ at a point $a\in \partial \mathbb H_n$ has the form $(0,\cdots , 0,\frac { \partial \phi_n}{\partial x_n}(a))$ :
Indeed, for $1\leq i\leq n-1$ we have $\frac { \partial \phi_n}{\partial x_i}(a)=0$ by the definition of partial derivatives since $\phi(\partial \mathbb H_n)\subset \partial \mathbb H_n$ and thus $\:\frac {\phi_n(a+he_i)-\phi_n(a)}{h}=\frac {0-0}{h}=0$.
Similarly $\frac { \partial \phi_n}{\partial x_n}(a)\geq 0$ because $\:\frac {\phi_n(a+he_n)-\phi_n(a)}{h}=\frac {\phi_n(a+he_n)-0}{h}\gt 0$.
Actually, we must have $\frac { \partial \phi_n}{\partial x_n}(a)\gt 0$ because the Jacobian is invertible.

The above proves that, given a tangent vector $v\in T_a(\mathbb H_n)$, its image $w=Jac_a(\phi)(v)$ satisfies $w_n=\frac { \partial \phi_n}{\partial x_n}(a)\cdot v_n$ with $\frac { \partial \phi_n}{\partial x_n}(a)\gt 0$, which shows that outward pointing vectors are preserved by the Jacobian of a change of coordinates for $M$ and thus that the notion of outward pointing vector is well-defined at a boundary point of a manifold with boundary .

share|improve this answer
Ok, I assume that $\frac{\phi(a+he_n)}{h}\geq 0$ because $\phi(\mathbb{H}^n)\subseteq\mathbb{H}^n$? Thanks for the answer! –  Miguel Jan 11 '13 at 13:24
@ Miguel: yes, exactly. Beware also that you have to assume $h\geq 0$ for $\phi (a+he_n)$ to be defined (so that the partial derivative $\frac { \partial \phi_n}{\partial x_n}(a)$ is actually a limit from the right, i.e. obtained for $h\gt 0$ tending to zero). And you are welcome! –  Georges Elencwajg Jan 11 '13 at 15:07

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.