1
$\begingroup$

I've posted this question Tangent space of $G \times_H M$ in MSE, but didn't get any answer. My question is the following:

Let $G$ be a Lie group and let $H$ be a Lie subgroup of $H$. Let $M$ be a smooth manifold on which $H$ acts from the left.

Let's consider the action of $H$ on $G \times M$: $$h((g,m)):= (gh,h^{-1}m), \quad h \in H , g \in G , m \in M, $$ and define the manifold $Z$ to be the quotient $G \times_H M$.

If we fix $(g,m) \in G \times M$, what is the induced equivalence relation on $T_gG \times T_mM$?

(My background is not so good in differential geometry, so please be patient with me.)

$\endgroup$
2
  • 3
    $\begingroup$ Not quite an answer to what you ask, but I think that $G\times_HM\to G/H$ is a locally trivial fibre bundle with fibre $M$, since the quotient you are considering is the associated bundle to the principal bundle $G\to G/H$. So then the tangent space will split to be isomorphic to the vector space direct sum $\mathfrak{g}/\mathfrak{h}\oplus T_mM$, where the quotient of Lie algebras is at the level of the underlying vector spaces. $\endgroup$
    – David Roberts
    Commented Jan 29, 2022 at 14:25
  • 1
    $\begingroup$ Your link intended to point to your question actually pointed to a comment, so I edited to fix the link. I hope that was correct. $\endgroup$
    – LSpice
    Commented Jan 31, 2022 at 3:18

1 Answer 1

3
$\begingroup$

When moving to the tangent space level, the general rule is to take the derivative (of everything you can). Just from this I'd expect something like the equivalence on $T_gG\oplus T_mM$ should be such that $gX - Xm = 0$ for every $X\in \mathfrak h$.

(As a side note, it may be better to write $(g,m).h$ because it's a right action. Or at least change it to $h.(g,m):=(gh^{-1},hm)$. I will stick with the way you have it for now.)

In detail, we want to explicitly describe the kernel of $\mathrm d \pi$, since this is exactly the $H$ relation you seek.

Let $h(t) = \exp(tX)$ be a path in $H$ with tangent vector $X\in \mathfrak h$, and consider the corresponding path in $G\times M$, $h(t).(g,m) = (gh(t),h(t)^{-1}m)$. The tangent vector is $gX - Xm\in T_gG\oplus T_mM$.

This whole path is projected to the same point $[(g,m)]\in Z$, so its tangent vector is in the kernel of $\mathrm d\pi$. Also since you are just quotienting by $H$ to get to $Z$, it is exactly these types of paths which make up the whole kernel. You can also convince yourself by counting dimensions. You should have exactly $\dim(H)$ conditions for the equivalence.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .