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 know two definitions of an orientation of a smooth n-manifold $M$:

1) A continuous pointwise orientation for $M$.

2) A continuous choice of generators for the groups $H_n(M,M-\{x\})=\mathbb{Z}$.

Why are these two definitions equivalent? In other words, why is a choice of basis of $\mathbb{R}^n$ equivalent to a choice of generator of $H_n(\mathbb{R}^n,\mathbb{R}^n-\{0\})=\mathbb{Z}$?

See comments for precise definitions.


share|cite|improve this question
The first definition you have given is called a parallelization, not an orientation. The two concepts are not equivalent. See – Jim Belk Jun 7 '11 at 1:29
@Jim Belk, I disagree, a pointwise choice of basis for each tangent space is continuous if every point of $M$ is in the domain of an oriented local frame. If there is a global frame then the manifold is parallelizable. – Manuel Jun 7 '11 at 1:46
@Manuel I do not agree. What is your precise definition of a continuous pointwise choice of basis for the tangent space? – Jim Belk Jun 7 '11 at 2:14
A pointwise orientation on $M$ is a choice of orientation of each tangent space. A local frame over a nieghborhood $U$ in $M$ is an ordered n-tuple $(e_1,...,e_n)$ of vector fields over $U$ such that $(e_1(p),...,e_n(p))$ is a basis for $T_pM$ for each $p\in U$. A local frame over $U$ is oriented if $(e_1(p),...,e_n(p))$ is positively oriented for each $p\in U$. As I said above, a pointwise orientation on $M$ is continuous if every point of $M$ is in the domain of an oriented local frame. I will edit my question so there is no ambiguity. – Manuel Jun 7 '11 at 2:28
See this answer to a very similar MO question:… – Matt E Jun 7 '11 at 2:41
up vote 2 down vote accepted

Observe that in (1) there is no difference between using tangent and cotangent bundle, and in (2) one can use $H^n$ instead of $H_n$.

Now, the equivalence becomes especially clear if in (2) one uses de Rham cohomology (instead of, say, singular).

Indeed, (1) is just existence of a (non-vanishing) section $\omega$ for $\Lambda^{top} T^*M$. So $\omega$ is a differential form, and for any $x\in U$ one can take a function $f_U$ that is 1 near $x$ and 0 outside of $U$ — and $\omega\cdot f_U$ is a generator of $H^n_{dR,c}(U)=H^n(M,M-\{x\})$. And using partitions of unit it's not hard to go in the opposite direction (i.e. reconstruct $\omega$ from local orientations).

share|cite|improve this answer
I am bit confused with the dimensions of the cohomology/homology groups. Do you mean $H^n_{dR}(U)=H_n(M,M-\{x\})$ where the first group is the n-th de rham compactly supported cohomology group? If so, what is this isomorphism? – Manuel Jun 7 '11 at 23:33
@Manuel Yes, exactly. It's an isomorphism of Poincare duality, if you will (but it's Poincare duality just for $S^n$: $H^n_{dR,c}(U)=\tilde H^n(S^n)=\tilde H_n(S^n)=H_n(M,M-\{x\})$). – Grigory M Jun 8 '11 at 6:50
Wouldn't Poincare duality give an iso $H^n(S^n)=H_0(S^n)$ sending a cell to its dual cell? – Manuel Jun 8 '11 at 12:27
Well, right, one shouldn't call it Poincare duality — it's just duality by universal coefficients. And the statement implicit in my answer is that one can use $H^n$ (but not $H^0$!) instead of $H_n$ in the definition (2). – Grigory M Jun 8 '11 at 12:34
Right, you mean the isomorphism given by $H^n(S^n)=\text{Hom}(H_n(S^n),\mathbb{Z})$ – Manuel Jun 8 '11 at 16:08

Recall that an element of $H_n(M,M-\{x\})$ is an equivalence class of singular $n$-chains, where the boundary of any chain in the class lies entirely in $M-\{x\}$. In particular, any generator of $H_n(M,M-\{x\})$ has a representative consisting of a single singular $n$-simplex $\sigma\colon \Delta^n\to M$, whose boundary lies in $M-\{x\}$. Moreover, the map $\sigma$ can be chosen to be a differentiable embedding. (Think of $\sigma$ as an oriented simplex in $M$ that contains $x$.)

Now, the domain $\Delta^n$ of $\sigma$ is the standard $n$-simplex, which has a canonical orientation as a subspace of $\mathbb{R^n}$. Since $\sigma$ is differentiable, we can push this orientation forward via the derivative of $\sigma$ onto the image of $\sigma$ in $M$. This gives a pointwise orientation on a neighborhood of $x$.

share|cite|improve this answer
Will the converse follow from the existence of solutions for ODE's? Given an oriented local frame on a neigborhood $U\subset M$ of $x$ we want to find a smooth non zero cycle $\sigma: \Delta^n \rightarrow M$ such that $\partial \sigma$ lies in $M-\{x\}$ the pushforward of the orientation of $\Delta^n$ is the one determined by the frame... – Manuel Jun 7 '11 at 19:27

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.