5
$\begingroup$

Let $G$ be a compact Lie group that acts on a smooth, finite dimensional, oriented manifold $M$, and suppose that such action preserves orientation, i.e., for each $g\in G$, the diffeomorphism $\mu_g$ of $M$ induced by the action preserves orientation. Consider the stratification $$M=\bigcup M_j$$ by orbit types, i.e., $M_j$ is the set of points that have the same isotropy, up to conjugation.

My question is, are there "simple" conditions under which the quotient manifold $M_j/G$ is orientable? Is it automatic true with the orientation hypothesis on $M$ and the action?

$\endgroup$
3
$\begingroup$

In your question, you may as well take $M=M_j=$ a ($G$-invariant) tubular neighborhood of a single orbit $G/S$. Near the point $S/S$, the space looks like $\mathfrak g/\mathfrak s \times N$, where $N$ is the normal space. Indeed, using a $G$-invariant metric, we can make $S$ act on $N$ and make this identification $S$-equivariant. By your condition that the stabilizers are all conjugate, the $S$-action on $N$ should be trivial.

You say that $\mathfrak g/\mathfrak s \times N$ is oriented, and ask whether $N$ (the tangent space in the quotient) is naturally oriented. This is therefore equivalent to $G/S$ being oriented. (Which is easy to violate, e.g. $SO(3)$ acting on $\mathbb{RP}^2$.) If we assume $G$ to be simply connected (which we don't get for free, since taking the universal cover might spoil its compactness), then for $G/S$ to be nonorientable you'd need $S$ to be disconnected. (Here's it's $S(O(2)\times O(1))$.)

For a counterexample, we want a $G/S$-bundle over a non-orientable manifold $M$. These are easy to cook up but I haven't yet seen how to guarantee that the total space is oriented.

$\endgroup$

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.