0
$\begingroup$

I have the following question: Let $f$ be analytic in a neighborhood of the unit disk such that $|f(z)|<1$ for $|z|=1$. Show that $f$ has exactly one fixed point $w$ in the unit disk and also show that $|f'(w)|<1$.

I tried to define a new function to apply Schwarz lemma and maximum modulus principle but couldn't get the statement. Any help would be great.

$\endgroup$
0
$\begingroup$

Let $D$ be the open unit ball in $\mathbb C$.

First of all, the maximum modulus principle tells you that: $$\forall z\in \overline D, \quad |f(z)|<1 \quad (\star).$$ Thus, $f$ induices a continuous map from the closed unit ball $\overline D$ into it self and due to the Brouwer fixed point theorem, it must have at least one fixed point in $\overline D$.

However, considering $(\star)$, any fixed point has to belong in $D$.

Let $\omega\in D$ be a fixed point of $f$ and $\varphi:D\rightarrow D$ the holomorphic map given by $$\varphi(z)=\dfrac{z-\omega}{1-\overline\omega z}.$$ Actually, $\varphi$ in an automorphism of the unit ball : $\varphi$ is holomorphic, bijective and so is its inverse. Furthermore, one has :$$\varphi(\omega)=0.$$ So the holomorphic map $g=\varphi\circ f\circ \varphi^{-1}$ is well defined on $D$ and satisfies :

  1. $g(0)=0$,
  2. $|g(z)|<1$,
  3. $g'(0)=f'(\omega)$,
  4. $g$ and $f$ have the same fixed point on $D$.

Now, applying the Schwarz lemma to $g$ we get that $$\forall z\in D, \quad |g(z)|\leq |z| \quad \text{ and } \quad |g'(0)|\leq 1$$ and moreover if $|g'(0)|=1$ then $g$ is of the form $z\mapsto az$ with $|a|=1$.

But we can't have both the assumption $(\star)$ and $g(z)=az$ on $D$ with $|a|=1$. Indeed, if $g(z)=az$ then $z\mapsto \varphi^{-1}\circ g\circ \varphi(z)$ extends to a holomorphic function on a neighborhood of $\overline D$ that coincides with $f$ on $\overline D$. But you can easily check that $\varphi^{-1}$, $g$ and $\varphi$ send the unit circle to the unit circle so $f$ has to do the same: this is in contradiction to $(\star)$.

So $$\forall z\in D, \quad |g(z)|\leq |z| \quad \text{ and }\quad |f'(\omega)|=|g'(0)|<1.$$

Finally, since $$|[g(z)-z]+z|=|g(z)|\leq |-z|$$ the Rouché theorem applied to $z\mapsto -z$ and $z\mapsto g(z)-z$ on $D$ says that $g$ (and thus $f$) has exactly one fixed point on $D$.

Edit: An other proof of the unicity of the fixed point is given in the comment.

$\endgroup$
2
  • $\begingroup$ Second part is identical with the one I wrote. Yesterday. $\endgroup$ – Gary. Aug 8 '15 at 2:32
  • $\begingroup$ Yes, but in order to apply Rouché's theorem, I need the schwarz lemma to get $|f(z)-z+z|\leq |z|$ and it wasn't clear to me how you processed in your solution. Actually, one could also prove that $g$ has only one fixed point with the Schwarz lemma again: if $f$ has another fixed point than $\omega$ then there exists $z\in D$ such that $|g(z)|=|z|$ and so $g$ has to be of the form $z\mapsto az$ with $|a|=1$. This is a contradiction with the hypothesis $(\star)$. $\endgroup$ – Bebop Aug 8 '15 at 13:15
0
$\begingroup$

For the first part, I think this should work: Using Rouche's theorem with $h(z)=f(z)-z$ and $g(z)=-z$ on $|z|<1$ :

$|h(z)-g(z)|=|f(z)-z-(-z) |=|f(z)|<|f(z)-z|+|z|$ Since $|f(z)|<1$ by assumption and $|z|=1$ on $\partial D$, so that $|f(z)-z|$ and $|-z|$ , both have the same number of zeros in $|z|<1$.

Let me try the second part.

EDIT: We can use Schwarz -Pick lemma http://mathworld.wolfram.com/Schwarz-PickLemma.html to show $|f'(z)| \leq 1$: We are given that $|f((z)| \leq 1$ we can just take $f(w)=w$ , then Schwarz-Pick tells us that $$f'(z) \leq \frac {1-w^2}{1-w^2}=1 $$.

So we can conclude $|f'(z)| \leq 1$ . Let me see if we can sharpen it to $|f'(z)|<1$.

$\endgroup$
8
  • $\begingroup$ Thanks for help. The rest is as follows: Let $F(z)=\frac{f(z)}{z-w}$ if $z \neq w$ and $f'(w)$ if $z=w$. Then we can apply maximum modulus to get $f'(w)<1$. $\endgroup$ – delueze Aug 6 '15 at 0:58
  • $\begingroup$ @delueze: Thanks for the follow up. Wouldn't that be $\frac {f(z)-w}{z-w}$ ? $\endgroup$ – Gary. Aug 6 '15 at 1:29
  • $\begingroup$ Yes, thanks for correction. But now I think my solution is not working, I will try again. $\endgroup$ – delueze Aug 6 '15 at 1:36
  • $\begingroup$ @delueze: I think Schwrz-Pick lemma may answer the second question. $\endgroup$ – Gary. Aug 6 '15 at 12:37
  • $\begingroup$ @Bebop: Why we couldnt have both the condition 2. and g(z)=az $\endgroup$ – delueze Aug 8 '15 at 0:28

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.