Let $f$ be analytic on $H=\{z:\Re(z)\ge 0\}$. Suppose that:

  • There exists $M>0$ such that if $\Re(z)=0$ then $|f(z)|\le M$.
  • There are $z_1,...,z_n$ such that $\Re(z_k)>0$ and $f(z_k)=0$.

Prove that for all $z\in H$: $$|f(z)|\le M\dfrac{\prod_{k=1}^n|z-z_k|}{\prod_{k=1}^n|z+\overline{z_k}|}.$$

This is suppose to be solved with the Maximum Modulus Theorem. But I don't know exactly which one. Since $f$ shouldn't be neccesary constant, the theorem must be this:

M. M. Theorem: Let $U\subseteq\mathbb{C}$ a bounded domain. Let $f$ be continuous on $\overline{U}$ and analytic on $U$. Then: $$\max\{|f(z)|:z\in\overline{U}\}=\max\{|f(z)|:z\in\partial U\}.$$

Since we need a bounded domain, let $U$ be the unit disk and consider the Möbius transformation $T(z)=\dfrac{-z-1}{z-1}$, so $T$ sends $U$ onto the right half plane.

By the M. M. Theorem, the maximum value of $|f\circ T|$ on $\overline{U}$ occurrs on the unit circle $|z|=1$. But if $|z|=1$, by the first point we have $|f(T(z))|\le M$. Therefore $|f(T(z))|\le M$ for all $z\in\overline{U}$.

But isn't it true that $T(\overline{U})=H$ and then $|f(z)|\le M$ for all $z\in H$? This is not the desired inequality. What am I doing wrong?

Thank you.

  • 1
    $\begingroup$ This is not true. The function $e^z$ is a counterexample with $n=0$ (if you don't like $n=0$, the function $(1-z)e^z/(1+z)$ is a counterexample with $n=1$.) You're leaving out a crucial hypothesis (or the author left it out). $\endgroup$ – David C. Ullrich Jan 16 '16 at 21:44
  • $\begingroup$ @DavidC.Ullrich Oh, thank you. It seems the autor left it out. If anyone knows what a crucial hypothesis could it be, please let me know. $\endgroup$ – JonSK Jan 16 '16 at 21:54
  • $\begingroup$ Hard to believe. Where is this from? Look again - what you wrote is exactly what it says? $\endgroup$ – David C. Ullrich Jan 16 '16 at 22:31
  • $\begingroup$ It is from here. It is in spanish though posgradomatematicas.unam.mx/contenidoEstatico/archivo/files/pdf/… $\endgroup$ – JonSK Jan 16 '16 at 22:37
  • $\begingroup$ Author's typo. It should be: There exists $M>0$ such that if $\Re(z)\ge 0 $ then $|f(z)|\le M$ $\endgroup$ – ts375_zk26 Jan 16 '16 at 23:37

The result as stated is not true. The function $e^z$ is a counterexample with $n=0$; this can easily be modified to give a counterexample for $n>0$.

The result becomes true if we assume in addition that $f$ is bounded. We need the following version of MMT:

Theorem. Suppose $f$ is analytic and bounded in the region $\Re z\ge0$. If $|f(z)|\le M$ for all $z$ with $\Re z=0$ then $|f(z)|\le M$ for all $z$ with $\Re z > 0$.

Proof. The simplest proof may be a Phragmen-Lindelofish argument. For $\epsilon>0$ let $$g_\epsilon(z)=f(z)/(1+\epsilon z).$$Then $|g_\epsilon(z)|\le M$ for $\Re z=0$. And also $g_\epsilon\to0$ at infinity, since $f$ is bounded. So the lim sup of $g_\epsilon$ is less than or equal to $M$ at every boundary point in the extended plane, hence a suitable version of MMT shows that $|g_\epsilon|\le M$ in $\Re z>0$. Now let $\epsilon\to0$. QED.

Now to do the problem. Given $f$ as in the problem, assuming $f$ is also bounded, let $g=$???. Applying the theorem to $g$ shows that $|g|\le M$ in the right half plane, and that shows that $f$ satisfies the conclusion. (Saying $|g|\le M$ implies what you need should be a good hint what $g$ should be...)

The Point: When you figure out what function $g$ you need to apply the theorem to to do the problem, you see that it's not obvious that $|g|\le M$ in the entire region, even if we begin by assuming thet $|f|\le M$ in the entire region. But it is clear that $g$ is bounded (by something, we don't know what) in the entire region, and that $|g|\le M$ on the imaginary axis. Hence the theorem shows that $|g|\le M$ everywhere. (In other words, even if we assume $|f|\le M$ everywhere to start we still need that theorem.)

| cite | improve this answer | |
  • $\begingroup$ Some questions. From what you wrote it is pretty obvious that $g=f/h$ where $h(z)=\frac{\prod_{k=1}^n|z-z_k|}{\prod_{k=1}^n|z+\overline{z_k}|}$. Then $g$ is analytic because $h$ has the zeros $z_k's$ and they are simple zeros. But I don't see why is it $g$ bounded? The theorem you wrote says $g$ must be bounded in order to apply it. Second question is, why $|g|\le M$ in $\Re(z)=0$? Perhaps it is $|h(z)|=1$ for all $\Re (z)=0$? $\endgroup$ – JonSK Jan 17 '16 at 1:25
  • 1
    $\begingroup$ First, $h$ is not quite what you say. We want $h$ to be holomorphic; leave out the absolute value signs. Yes, $|h(z)|=1$ for $\Re z=0$; this follows from the fact that $|\overline\alpha|=|\alpha|$. Why is $f/h$ bounded? Well, it's bounded in a neighborhood of each $z_k$ because it's continuous. And if $z$ satisfies $|z-z_k|>\delta>0$ for every $k$ you can show directly that $|1/h(z)|$ is less than or equal to something (something independent of $z$). $\endgroup$ – David C. Ullrich Jan 17 '16 at 2:11
  • $\begingroup$ Thank you! I forgot to wipe out the absolute values, but I understood it now. Thanks again. $\endgroup$ – JonSK Jan 17 '16 at 3:31

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.