Let $\Omega$ be open and connected, and let $\{f_n\}$ be a sequence of holomorphic functions on $\Omega$ such that $f_n(\Omega)\subseteq \Omega$. If $f_n\rightarrow f$ uniformly on compact subsets $\Omega$ and $f$ is not constant, prove $f(\Omega)\subseteq \Omega$

My attempt:

Let $z_0\in \Omega$ and let $g(z)=f(z)-f(z_0)$ and $h(z)=f_n(z)-f(z_0)$. Clearly, $g(z)$ has a zero in a small neighborhood of $z_0$, which is $z_0$.

If I'm able to show that $h(z)$ also has a zero in that neighborhood, say $z_1$, this implies that $f(z_0)=f_n(z_1)\in \Omega$ and this completes the proof.

But I'm having trouble showing that $h(z)$ has a zero in the neighborhood of $z_0$. I'm trying to use Rouche's theorem by showing if $|g(z)-h(z)|<|g(z)|$ in the boundary of some neighborhood of $z_0$, $g(z)$ and $h(z)$ has the same number of roots in the neighborhood, hence, there exists some $z_1$ that satisfies my claim.

Can anyone show me how to get the inequality?


Suppose for the sake of contradiction there exists a point $z_0$ such that $z_0 \notin \Omega$, such that $f(w_0) = z_0$. WLOG $z_0 = 0$, then by the argument principle \begin{equation} \int_{\gamma} \frac{f'}{f} \geq 1 \end{equation} where $\gamma$ is a small contour around zero so that it doesn't intersect $\Omega$; this is possible since $\Omega$ is open. But observe \begin{equation} \int_{\gamma} \frac{f_n'}{f_n} = 0 \end{equation} for any $n$ since $f_n(\Omega) \subset \Omega$, so $f_n \neq 0$ for any $z \in \Omega$. So by uniform convergence we get \begin{equation} \int_{\gamma} \frac{f'}{f} = 0 \end{equation} which is our contradiction. In general uniform convergence + argument principle gives you many strong results such as if $f_n$ is $1-1$ and converges uniformly to $f$, then either $f$ is constant or $1-1$ and similarly if $f_n \neq 0$ for any $z$. (Uniform convergence allows you to pass limit to integrals, and this integral tells you how many zeros there are.)

  • $\begingroup$ So the argument principle states that if $f$ is meromorphic with roots and poles not on the contour $\gamma$, we have $\frac{1}{2\pi i}\int_{\gamma}\frac{f'}{f}=$ number of zeros in $\gamma$ - number of poles in $\gamma$. How do we know that $\int_{\gamma}\frac{f'}{f}\geq1$? I guess I just need some more explanation on your reasoning. $\endgroup$ – Ya G Dec 19 '18 at 19:22
  • $\begingroup$ As $f_n \rightarrow f$ uniformly on compact sets, $f_n$ holomorphic, we know by Morerra Theorem and uniform convergence that $f$ is holomorphic. So it has no poles, but $f(z_0) = 0$, so it has at least one zero in $\gamma$ namely $z_0$. So the integral is greater than or equal to 1. $\endgroup$ – Story123 Dec 19 '18 at 19:26
  • 1
    $\begingroup$ Got it! So you are assuming that there is a root inside the contour such that the root itself is not in $\Omega$. But since it's inside the contour, the argument principle suggests that there exists at least one zero in the contour. $\endgroup$ – Ya G Dec 19 '18 at 19:31
  • $\begingroup$ Yes, you can assume WLOG it's a root since if it isn't consider $g(z) := f(z) - w_0$, then $g'(z) = f'(z)$ and you can apply the argument principle to $g$. $\endgroup$ – Story123 Dec 19 '18 at 19:36
  • $\begingroup$ What if $z_0\in \partial \Omega?$ Then every nbhd of $z_0$ intersects $\Omega$ $\endgroup$ – Matematleta Dec 19 '18 at 22:06

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.