Is there a simple proof that every holomorphic function $M\to\mathbb{C}$ on a compact complex manifold $M$ is constant?


You mean assuming that $M$ is also connected. Yes, a simple proof exists if you can use the maximum modulus principle, by showing the set $$ \{ x \in M \mid f(x)=f(x_0) \} $$ is all of $M$ (where $x_0$ is a point where the maximum modulus is achieved).


Non-constant holomorphic functions on connected complex manifolds are open maps.
So, if $M$ were compact and $f:M\to \mathbb C$ were non-constant, its image would be an open, compact non-empty subset $f(M)\subset\mathbb C$. Such a beast does not exist.

  • $\begingroup$ Does the same argument hold if I have a holomorphic map $f: \mathbb{P}^n \to \mathbb{C}^{n+1}$? $\endgroup$ – Kristina Apr 18 '17 at 11:53
  • 2
    $\begingroup$ @Kristina: the datum of a holomorphic map $f:\mathbb P^n \to \mathbb C^{n+1}$ is equivalent to the datum of $n+1$ holomorphic functions $f_j=pr_j\circ f:\mathbb P^n \to \mathbb C^{n+1} \to \mathbb C$ . Since the $f_j$ are constant, so is $f$. $\endgroup$ – Georges Elencwajg Apr 18 '17 at 22:02

It depends on what you mean by simple but if you take the maximum principle for holomorphic functions on $\mathbb{C}$ for granted (a non-constant holomorphic function doesn't admit local maxima), the statement follows easily: Suppose $f : M \to \mathbb{C}$ is a holomorphic function from a compact connected Riemann surface $M$. Then by compactness of $M$, the function $|f| : M \to \mathbb{R}$ attains a maximum at some point $p \in M$. If $(U,\varphi)$ is a holomorphic coordinate patch around $p$, say $\varphi : \mathbb{C} \supset V \overset{\sim}\to U \subset M$, then this gives a holomorphic map $f \circ \varphi : V \to \mathbb{C}$ with a maximum at $0$, hence $f \circ \varphi \equiv C$ is constant on $V$, so $f \equiv C$ is constant on $U$. Since $U$ is open and $M$ is connected, analytic continuation implies that $f \equiv C$ on all of $M$.

Edit: I just realised you asked for complex manifolds in general but the proof is the same.

  • $\begingroup$ However, this works only for $\dim_{\mathbb{C}} M=1$, not? $\endgroup$ – Peter Franek Jul 29 '14 at 17:17
  • 1
    $\begingroup$ You can use the maximum modulus principle linked in Jonas Meyer's answer with the same proof in the genral case. $\endgroup$ – jef808 Jul 29 '14 at 17:19
  • $\begingroup$ Yes, I got it. Thanks a lot (it was hard to decide which answer to accept, both are ok)! $\endgroup$ – Peter Franek Jul 29 '14 at 17:20

proof. ‎$‎\left| f \right|:X \to R‎$‎ is a continuous function, ‎$X‎‎$ ‎is ‎compact, ‎‎$‎‎\Rightarrow‎ \left\{ {\left| {f\left( x \right)} \right|:x \in X} \right\}‎$‎ is compact, hence bounded. Thus, there is an ‎$‎x_0 \in X$‎ such that ‎$‎‎\left| {f(x_0 )} \right| = M$ ‎is ‎maximal. ‎Let ‎‎$‎‎a = f(x_0 ) \in \mathbb{C}‎$ .‎ ‎‎Obviously, ‎$‎‎f^{ - 1} (a)$‎ is closed in ‎$X$‎ (it is a pre-image of a point); if we are able to show that $‎‎f^{ - 1} (a)$ is open, too, then $‎‎f^{ - 1} (a) = X$, that implies ‎$‎‎f(x) = a~~~~\forall x \in X$‎. Let $x \in ‎‎f^{ - 1} (a)$ and ‎$ (U,z) $ be a chart with ‎$ z(x)=0 $‎. Then ‎$ F:foz^{ - 1} :z(U) \to ‎\mathbb{C}‎ $‎ is holomorphic on the open subset ‎$ z(U) \subseteq C^n ;F(0)=f(x) = a $ ‎and‎ ‎$ \left| F \right| $‎ has a maximum in ‎$ z=0 $.‎ Let‎ ‎$ \varepsilon > 0 $ ‎such ‎that‎ ‎$ B_\varepsilon : = \left\{ {y \in C^n :\left\| y \right\| < \varepsilon } \right\} \subseteq z(U) $. ‎For‎ ‎$ y \in B_\varepsilon $‎, the function ‎$ g(t): = F(ty) $‎ is holomorphic on ‎$ \left\{ {t \in C:\left\| {ty} \right\| < \varepsilon } \right\} $ ‎and‎ ‎$ \left| g \right| $‎ takes its maximum in ‎$ t=0 $.‎ By the "maximum principle", ‎$ g $‎ is constant ‎$ \Rightarrow a = g(0) = g(1) = F(y) $‎, that means ‎$ F(y) = a~~~ \forall y \in B_‎\varepsilon‎ $‎. Hence ‎$ f \equiv a $ ‎on‎ ‎$ z^{ - 1} (B_\varepsilon ) $‎, an open subset of ‎$ X $‎ containing ‎$ x $‎. Hence ‎$ f^{ - 1} (a) $‎ is open.


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.