0

It is known that the exponential map is surjective for compact, connected Lie groups. This is generally proved through the introduction of some bi-invariant metric etc. etc. My question is related to the following: Can this be shown using only Lie theory and not deferring to other concepts like Riemannian geometry, etc.? In particular, can I use some properties of the Lie algebra of compact Lie groups? Details:

Let $G$ be a compact, connected Lie group. We know $\exp:\mathfrak{g}\to G$ is a local homeomorphism. So I can find a subset of the Lie algebra, $S$, containing $0$ such that $\exp{S}\subset G$ is open in $G$. Now, consider an open subset of $\exp{S}$, $U$, with the property that $U^{-1}:=\{g^{-1}:g\in U\}=U$ and $e\in U$. Then, with a bit of work, you can show $\exists n\in\mathbb{N}$ such that $G=U^n$ (length $n$ words in the elements of $U$). This fact is equivalent to compactness (finite subcover).

So, given $g\in G, \exists g_1,...,g_n\in U$ such that $g=g_1g_2...g_n$. By local homeomorphism property of $\exp$, $\exists x_1,...,x_n \in \exp^{-1}{(U)}\subset S\subset \mathfrak{g}$ such that $e^{x_i}=g_i \forall i$. Therefore, $g=e^{x_1}...e^{x_n}=e^{\Omega_n(x_1,...,x_n)}$ where $\Omega_n$ denotes the Baker-Campbell-Hausdorff formula of $n$ elements. In particular, $$\Omega_2(A,B):=A+B+\frac{1}{2}[A,B]+\frac{1}{12}[A,[A,B]]+...$$

However, if $\exp$ is to be surjective, $\Omega_n$ must converge for all $x_1,...,x_n$ in $\exp^{-1}{(U)}$. But in order to prove surjectivity, is there any way I can use the properties of Lie algebras of compact Lie groups to show convergence of the BCH formula in this set?

Please let me know if this question is ill-posed/ill-formed. Thanks!

Anz
  • 144
  • 1
    I don't think what you stated is true. Taking for example $G=\mathbb{S}^1\times\mathbb{Z}/2$. It's compact, and the image of $\exp$ lies in the connected components $\mathbb{S}^1\times{0}$, hence is not surjective. – anonymous67 Apr 26 '20 at 06:16
  • Yes you are right. I meant connected as well. Edited. – Anz Apr 26 '20 at 06:18
  • This might be helpful (where a Fields medalist asked the identical question): https://terrytao.wordpress.com/2011/06/25/two-small-facts-about-lie-groups/ – anonymous67 Apr 26 '20 at 06:30
  • @Learning Yes, I have come across that before. I was wondering if anyone here had any insight or thoughts. – Anz Apr 26 '20 at 07:22
  • 2
    There certainly are proves that do not need Riemannian geometry: Surjectivity of the exponential map is a simple consequence of the fact that given a maximal torus $T\subset G$ any element of $G$ is conjugate to an element of $T$. Unfortunately, I don't know a purely Lie thoretic proof of this conjugacy result. The proof I know uses mapping degree and thus de Rham cohomology, but no Reimannian geometry. – Andreas Cap Apr 26 '20 at 08:45
  • @AndreasCap Yes, I have seen the maximal torus proof. I should've mentioned it. I'm wondering if some properties of Lie algebras of compact Lie groups are useful making at least some weak claim about $\exp$ as it relates to surjectivity (at least). This seems to be a completely self-contained Lie theory result. – Anz Apr 26 '20 at 08:49
  • I understand what you are looking for. However, I think that these are fairly suble matters. Compare to the case of the "compact" proof of the Weyl character formula via integration vs. algebraic proofs, which a hugely different. – Andreas Cap Apr 26 '20 at 08:52

0 Answers0