Let $G$ be a (finite-dimensional) complex Lie group, and suppose $f : G \to \mathbb{C}$ is holomorphic. Let $X$ be a left-invariant vector field on $G$. Must $Xf$ be holomorphic?

I think I have a proof, but I feel that I may have missed some subtlety, or overcomplicated matters. I'm still learning this area and I don't think my intuition is completely calibrated yet. I would also be happy to see alternate proofs.

Let $\mathfrak{g}$ be the Lie algebra of $G$, considered as the tangent space of $G$ at the identity $e$. Then $\mathfrak{g}$ has a complex structure; call it $J$. Without loss of generality we can assume $X$ is a real vector field, so let $\xi = X(e) \in \mathfrak{g}$.

By left-invariance it is sufficient to show $Xf$ is holomorphic at $e$. Fix a holomorphic system of coordinates $(z^1, \dots, z^n)$ in a neighborhood of $e$. I claim $\displaystyle \frac{\partial}{\partial \bar{z}_j} f = 0$ at $e$. Let $\bar{Z}$ be a left-invariant complex vector field with $\bar{Z} = \displaystyle \frac{\partial}{\partial \bar{z}_j} $ at $e$. Then $\bar{Z}$ is of type (0,1) so we can write $\bar{Z}(e) = \eta + i J \eta$ for some $\eta \in \mathfrak{g}$.

Now we have $(\bar{Z}Xf)(e) = (X \bar{Z} f)(e) + ([\bar{Z}, X]f)(e)$. Since $f$ is holomorphic and $\bar{Z}$ is of type (0,1), $\bar{Z} f = 0$ so the first term vanishes. The second term equals $[\eta + iJ\eta, \xi]f = ([\eta, \xi] + i J [\eta, \xi]) f$ which also vanishes since $[\eta, \xi] + i J [\eta, \xi]$ is also of type (0,1).


An easier way to see it is as follows.

As before, assume without loss of generality that $X$ is real, let $\xi = X(e) \in \mathfrak{g}$ and let $\exp : \mathfrak{g} \to G$ be the exponential map. Let $f_t(g) = f(g \exp(t \xi))$ and recall that $Xf(g) = \frac{d}{dt}\Big|_{t=0} f_t(g)$. Since the group operation is holomorphic, $f_t$ is a one-parameter family of holomorphic functions, hence its derivative at $t=0$ is also holomorphic. (For instance, we can interchange derivatives with respect to $t$ and $\bar{z}_j$ to see that $Xf$ satisfies the Cauchy-Riemann equations.)


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.