Setting. Let $M$ be a Riemann surface and $\Gamma$ a discrete group that acts properly discontinuously on $M$ by holomorphic maps. It is well known that each $x \in M$ has a finite stabilizer, that the points with nontrivial stabilizer form a discrete set $R \subset M$ (ramification points), and that the projection $M \overset \pi \to M/ \Gamma$ is a (holomorphic) covering map outside of $R$.

Moreover, it is a consequence of the uniformization theorem that $M / \Gamma$ has a unique holomorphic structure for which $\pi$ is holomorphic. In particular, $M / \Gamma$ is a smooth manifold. (Note that this is not true in the case of, say the proper group action $\{\pm 1 \} \curvearrowright \mathbb R$ by multiplication: the quotient $\mathbb R / \pm 1 \cong [0, + \infty)$ is not a smooth manifold.)

We have that if $f : M \to \mathbb C$ is $\Gamma$-invariant and holomorphic, it descends to $M / \Gamma$ as a holomorphic function.

Question. What other smoothness properties are preserved under $\pi$? Ex:

  1. If $f : M \to \mathbb C$ is smooth and $\Gamma$-invariant, does it descend to $M / \Gamma$ as a smooth function? (Compare with this MO post)
  2. If $g$ is a Riemannian metric on $M$ and $\Gamma$ acts by isometries, does $g$ descends to $M / \Gamma$ as a smooth metric?

Is there a good reference on such questions? Perhaps in the particular case where $M$ is the complex upper half plane?

  • 1
    $\begingroup$ Part 2 fails even for the flat metric on ${\mathbb C}$ when $\Gamma$ is generated by the antipodal involution $z\mapsto -z$. $\endgroup$ – Moishe Kohan Feb 25 at 4:33
  • $\begingroup$ @MoisheCohen You're right, that's interesting. $\endgroup$ – punctured dusk Feb 25 at 8:23

Partial answer:

Part 1 is true, and follows from the answer to part 3 of this MO post: https://mathoverflow.net/questions/65264/integral-representation-of-higher-order-derivatives. For smoothness of $f$ at a branch point $\pi(p)$, the uniformization theorem allows us to reduce to the case of $\pi : B(0, 1) \to B(0, 1)$ given by $z \mapsto z^p$ and $\Gamma = \mathbb Z/p$ with the action given by multiplication by $p$th roots of unity. A special case of a theorem of G. Schwartz (Smooth functions invariant under the action of a compact lie group, Topology 14, 1975) implies that the $\mathbb Z/p$-invariant smooth functions $B(0, 1) \to \mathbb C$ are given by smooth functions in the invariant polynomials for this action. In this case, they are $Re(z^p)$ and $Im(z^p)$. Hence $f$ is a smooth function of $z^p$.

For part 2, we can reduce in the same way to a Riemannian metric on $B(0, 1)$ and $\pi(z) = z^n$. We may look at the metric as a smooth map $g : B(0, 1) \to SPD_2(\mathbb R)$ that takes values in (wlog) positive definite symmetric matrices. The invariance under $\Gamma$ means that for any $k \in \mathbb Z$ and $z \in B(0, 1)$, $$\zeta_p^{-k} g(\zeta_p z) \zeta_p^{k} = g(z) \tag{1}$$ where we identify a complex number $x+iy$ with its differential, $$\begin{pmatrix} x & -y \\ y & x \end{pmatrix}$$ The question is now equivalent to whether equation $(1)$ implies that $g(z)$ is a smooth function of $z^p$. The same result of Schwartz tells us that the trace and determinant of $g(z)$ are smooth functions of $z^p$, but this is not enough to conclude.


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.