Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm studying the "Elementary Differential Geometry" from O'Neil and he mentions what he calls the "Natural Coordinate Functions". In particular, he says that if $p = (p_1 ... p_n) \in \mathbb{R}^n$ is a point in the $n$-space then we define the $i$-th coordinate function $x_i : \mathbb{R}^n \to \mathbb{R}$ as $x_i(p) = p_i$

My question is: why it's needed to define such functions ? I mean why is it relevant to define this ? He says that in $n$-space it's very important this definition to avoid serious ambiguities, but I couldn't identify any special case when there's a ambiguity for not defining those functions.

Can someone please explain this and show an example where it causes ambiguities for not defining those functions ?

Thanks a lot in advance.

share|cite|improve this question
up vote 2 down vote accepted

The reason to introduce these functions is clear from calculus III. We don't even need to consider anything too fancy. Consider the following:

$$ \frac{\partial x}{\partial y}=0 $$

How is this shown? We need to introduce a function $f(x,y) = x$ and then calculate from the definition for partial derivative with respect to $y$ at $(a,b)$

$$ \frac{\partial x}{\partial y}(a,b) = \lim_{h \rightarrow 0} \frac{f(a,b+h)-f(a,b)}{h} = \lim_{h \rightarrow 0} \frac{a-a}{h} = \lim_{h \rightarrow 0} 0 = 0 $$

Of course this holds for all $(a,b)$ so we can drop it without ambiguity.

Fun exercise: do the same calculation for $f(\vec{x}) = e_j \cdot \vec{x} = x_j$ to prove that $\frac{\partial x_j}{\partial x_i} = \delta_{ij}$. Here the partial derivative with respect to $x_i$ would be defined by $$ \frac{\partial f}{\partial x_i}(\vec{p}) = \lim_{h \rightarrow 0}\frac{f(\vec{p}+he_j)-f(\vec{p})}{h} $$

Oneil's point is that without this notation it makes many statements relating maps on the surface to maps on $\mathbb{R}^n$ awkward. Often an embedded surface can be covered by restricting the ambient maps $x,y,z$ on $\mathbb{R}^3$. For example, a plane with $z=2$ natrually inherits the $x,y$ coordinate maps by simply restricting their domain from the natural $\mathbb{R}^3$. In short, it's about having a notation for formulas you need to write.

If coordinates are abhorrent then perhaps you should read Lang's differential geometry, I believe he makes an effort to avoid unnecessary coordinated formulas. Elementary Differential Geometry by O'neill is written at a very different level. It is meant to be a natural extension from calculus III.

share|cite|improve this answer
Hi James S. Cook, I understood what's the point now. Thanks for your aid. The book of differential geometry you speak of is Fundamentals of Differential Geometry from Serge Lang? Thanks again for your answer. – user1620696 Sep 17 '12 at 17:14
Correct, although be careful of which edition you get. There is a big difference between the 3rd and earlier editions. I just bought the third ed. and it is much bigger than the earlier ed. You can look at the reviews to decide if it's worth it... – James S. Cook Sep 17 '12 at 20:06

It is possible to reduce the use of coordinates in differential geometry.

By definition, on an $n$-manifold, for every point there is given an open neighborhood with homeomorphism to an open subset of $\mathbb R^n$. Now, the coordinates of these given homeomorphisms can be set as the coordinate functions- Its use is that differentials and integrals are then locally computable on the manifold, using these in $\mathbb R^n$.

For example, the surface of the sphere, as 2-manifold, can be given by 2 "maps", one circle of a plain stereographic-projected to the upper part of the sphere, and another circle for the lower part.

share|cite|improve this answer
Berci Pecsi, I know that idea, it's the idea of using charts to descibre a manifold right? Then, if $(U, \varphi)$ is a chart on a manifold $M$ and if $p \in M$ I'll have $\varphi(p) \in \mathbb{R}^n$. If I understood you, I'll use the coordinate functions to write $p_i = (x_i \circ \varphi)(p)$ the $i$-th coordinate of the point in the manifold? I mean, I'll use it to assign coordinates to points on the manifold using a chart? Thanks a lot for your answer. – user1620696 Sep 16 '12 at 20:55

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.