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 an exercise of Lie group as follows: "Let $G,H$ be closed connected subgroup of $GL_n(\mathbb{R})$, and $H$ be subgoup of $G$. Suppose that $Lie(H)$ is an ideal of $Lie(G)$. Prove that $H$ is a normal subgroup of $G$." I get stuck to solve this problem. Also I have no idea to use the connectedness of $G$ and $H$. Some one can help me? Thanks a lot!

share|improve this question

This question has an open bounty worth +50 reputation from Tim kinsella ending in 3 days.

The current answers do not contain enough detail.

This has been asked at least twice on this site, so I don't want to post it again. There doesn't seem to be a full proof in either post of this extremely natural question.

2 Answers 2

This is essentially an application of the Lie subalgebra-subgroup correspondence:

Let $G$ be a Lie group with Lie algebra $\mathfrak{g}$. Suppose that $\mathfrak{h}$ is a Lie subalgebra of $\mathfrak{g}$. Then there is a unique connected immersed Lie subgroup $H\subseteq G$ whose Lie algebra corresponds to $\mathfrak{h}$.

You are given $H$ a closed connected Lie subgroup of the Lie group $G$. Choose $g\in G$, and let $H' = gHg^{-1}$. Then $H'$ is a Lie group with corresponding Lie algebra $Lie(H')\subseteq Lie(G)$. However, the assumption that $Lie(H)$ is an ideal of $Lie(G)$ says exactly that $Lie(H') = Lie(H)$. You then have that $H$ and $H'$ are two connected Lie subgroups of $G$ with the same Lie algebra. By the uniqueness in the above theorem, it follows that $H = H'$, and hence that $H$ is normal.

share|improve this answer
Thank you so much. But I don't know about the theorem you said. Can you tell me the name of the theorem? By chance, if the Lie group $G$ is linear, i.e a closed subgroup of $GL_n{\mathbb{R}}$, do we have simple proof for this theorem? –  mapping Nov 3 '12 at 15:11
@mapping: I don't know if the theorem has a name. It is essentially the Frobenius Theorem‌​. The idea is as follows. Take a basis $X_1,\ldots, X_r$ for $\mathfrak{h}$. These are invariant vector fields. Since $\mathfrak{h}$ is a subalgebra, these vector fields involutive. Frobenius' Theorem says there is a unique connected integral submanifold $H\subseteq G$ through the identity element. This is the subgroup $H$. Of course, you have to prove it is a Lie subgroup. –  froggie Nov 3 '12 at 16:44
Thanks a lot! I will try to prove by this way. However I have trouble to prove $Lie(H')=Lie(H)$: If I take $X\in Lie{H'}$ then we have $\exp(tX)\in H'\forall t$. Then I can deduce that $g^{-1}Xg=Z\in Lie(H)$. So I need to prove $gZg^{-1}\in Lie(H)$. Using the condition $Lie(H)$ is an ideal of $Lie(G)$, we obtain $gZ-Zg\in Lie(H)$, hence $[gZ-Zg,g^{-1}]\in Lie(H)$. Finally I get $gZg^{-1}+g^{-1}Zg\in Lie(H)$. Until here, I don't know how to continue. Can you help me? –  mapping Nov 3 '12 at 17:53

There seems to be a full proof contained in lemma 0.1 here http://math.berkeley.edu/~ianagol/261A.F09/Simplegroups.pdf

Here is the argument:

Claim: For $X \in \frak{g}$, $Y \in \frak{h}$, we have $e^Xe^Ye^{-X}\in H$.

Proof of claim: Denote by $exp$ the exponential map $End(\frak{g}) \rightarrow $ $Aut(\frak{g}) $ (where Aut and End are the vector space automorphisms and endomorphisms) . Since $\frak{h}$ is an ideal, $ad_X^n$ preserves $\frak{h}$ for all $n \in \mathbb{N}$, and therefore so does $exp(ad_X)$. We therefore have $$e^Xe^Ye^{-X}=e^{Ad_{e^X}(Y)}=e^{exp(ad_X)Y} \in e^{\frak{h}} $$

This establishes the claim.

Now the fact that $G$ normalizes $H$ follows from the fact that $$H= \bigcup_{n\in \mathbb{N}} (e^\frak{h})^n$$ $$G= \bigcup_{n \in \mathbb{N}}(e^\frak{g})^n$$ Since $G$ and $H$ are connected.

In particular, this doesn't seem to use the fact that $G$ is a closed subgroup of $GL(n, \mathbb{R})$, although this is a hypothesis of lemma 0.1. Nor does it require even that $H$ be closed in $G$.

share|improve this answer

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.