I came across a simple Lie group question while reading Bernard Schutz's Geometrical Methods of Mathematical Physics. I am brand new to group theory, and this is my first attempt at grasping its physical significance.

He writes (p. 93) that if $\overline{V}$ and $\overline{W}$ are any two left-invariant vector fields on a manifold $G$, then the left translation map $L_g$ (which maps tangent vectors at the identity $e$ to those at $g$) maps $[\overline{V},\overline{W}]$ at $e$ to $[\overline{V},\overline{W}]$ at $g$. Then he draws a figure of the Lie bracket, and states that the field $[\overline{V},\overline{W}]$ is also left-invariant. He says, "The reader who is not convinced by the [diagrammatic proof] is invited to use coordinates on $G$ to prove the result".

I am wondering how one uses coordinates to prove this result. The Lie bracket, of course, can be expressed $$[\overline{V},\overline{W}]=\left(V^i\frac{\partial W^j}{\partial x^i}-W^i\frac{\partial V^j}{\partial x^i}\right)\frac{\partial}{\partial x^j}$$ (summed over the repeated indices). But how would the operation $L_g$ interact with each of these component terms? I only know that $L_g(\overline{V}(e))=\overline{V}(g)$, and likewise for $\overline{W}$.


Let $M$ be a smooth manifold and $\varphi: M\to M$ a diffeomorphism. For a vector field $\xi: M\to TM$ one can define a new vector field $\varphi_*\xi$ by the relation:

\begin{equation} (\varphi_*\xi)|_q = T_p\varphi(\xi|_p) \end{equation}

where $q = \varphi(p)$. This lead to an automorphism $\varphi_*: \Gamma(TM)\to \Gamma(TM)$ of the $C^{\infty}(M)$-module of vector fields called the pushforward by $\varphi$.

The heart of your question rely on the following relation:

\begin{equation} \varphi_*[\xi_1, \xi_2] = [\varphi_*\xi_1, \varphi_*\xi_2] \end{equation}

Indeed if $M$ is a Lie group $G$, a vector field $\xi$ is left-invariant if and only if $(L_g)_*\xi = \xi$ for every $g\in G$. So using the previous relation it's obvious that the Lie bracket of two left-invariant vector fields is left-invariant.

Now since this relation works for any diffeomorphisms I don't think it's easier to prove it in the very particular case $\varphi = L_g$. Also, prooving this relation depends on how you define the Lie bracket and it's usually not done in coordinates. My own opinion is coordinates are not suitable to prove this relation; it would be tedious. For a standard proof I recommand John. M. Lee's Introduction to smooth manifolds.

  • $\begingroup$ I agree that coordinates aren't the ideal way to prove this statement, but my question is, how would one use coordinates to prove this statement. $\endgroup$ – Doubt Oct 2 '17 at 1:54
  • 1
    $\begingroup$ It's indeed possible to prove the relation I mentionned in coordinates. But it's neither usefull, nor instructive and it's extremely tedious. So I don't think it's the answer you desire. If you really want to do so, compute $\varphi_*[\xi_1, \xi_2]$ and $[\varphi_*\xi_1, \varphi_*\xi_2]$ in coordinates is a way. $\endgroup$ – paeolo Oct 3 '17 at 8:42
  • $\begingroup$ Thanks, I'm almost there! I see the last piece I'm missing: given $\xi = \xi^i\frac{\partial}{\partial x^i}$, how does one write out (using the coordinate basis) the quantity $\varphi_*\xi$? $\endgroup$ – Doubt Oct 13 '17 at 3:26
  • 1
    $\begingroup$ Use the definition $(\varphi_*\xi)|_q = T_p\varphi(\xi|_p)$ and the relation $T_p\varphi(\frac{\partial}{\partial x_i}) = \frac{\partial\varphi^j}{\partial x_i}\frac{\partial}{\partial x_j}|_{\varphi(p)}$. $\endgroup$ – paeolo Oct 14 '17 at 15:12

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.