6

Let $f$ be a non constant entire function satisfying the following conditions :

  1. $f(0)=0$
  2. for every positive real $M$, the set $\{z: \left|f(z)\right|<M\}$ is connected.

Prove that $f(x)=cz^n$ for some constant $c$ and positive integer $n$.

Let $f(z)=a_nz^n+\cdots+a_1z+a_0$ be function that satisfies the given conditions. As $f(0)=0$ we have $a_0=0$ and $f(z)=a_nz^n+\cdots+a_1z$.

As $f$ is non-constant function, its zeros are isolated. So, there exists an $r>0$ such that $f$ is non-zero on $B_r=\{z:|z|<r\}$. I was thinking of connecting this to connectedness of $\{z: \left|f(z)\right|<M\}$.

I wanted to check what goes wrong in case of $f(z)=z^2+z$. I want to check if the given set is connected for this but failed in doing so.

Daniel Fischer
  • 206,697
  • 1
    First of all, there is no immediate reason the function must be a polynomial. You must also check that, for instance, $\sin(z)$ and $e^z-1$ both fail condition 2. – Arthur Mar 02 '16 at 18:51
  • Yes... There is no reason why it should be a polynomial.. I was just checking if i can get some information if i assume it to be a polynomial.. @Arthur –  Mar 02 '16 at 18:53
  • 1
    If it was a polynomial, then think about the roots of the polynomial. What happens when $M$ is small and close to $0$? – Steven Gubkin Mar 02 '16 at 18:56
  • 1
    @StevenGubkin : As $M$ is close to $0$ then we have finitely many distinct roots in case of polynomials other than $f(z)=az^n$ which is clearly disconnected... –  Mar 02 '16 at 19:02
  • @cello, right. Actually, this shows that for any $f$ satisfying the conditions, it only has $0$ as a root. Can you write all of these $f$ in some standard form? – Steven Gubkin Mar 02 '16 at 19:05
  • @StevenGubkin : I did not understand your comment... –  Mar 02 '16 at 19:07
  • Hmm. It would help to know a bit about what background you have already covered. Do you know the Weierstrass factorization theorem? Do you know great Picard? These are maybe overkill for the problem, but that is how I thought about it. – Steven Gubkin Mar 02 '16 at 19:09
  • @StevenGubkin: I do not know any of that you have mentioned... I only know some residue theory... –  Mar 02 '16 at 19:12
  • 1
    Do you know the Casorati-Weierstraß theorem? – Daniel Fischer Mar 02 '16 at 19:51
  • @DanielFischer : Suppose $z_0$ is an essential singularity of $f(z)$. Then for every complex number $w_0$, there exists a sequence $z_n\rightarrow z_0$ such that $f(z_n)\rightarrow w_0$.. But i do not know how is this realted. There is no singularities here :O –  Mar 02 '16 at 19:53
  • We have a singularity at $\infty$. Casorati-Weierstraß shows that it can't be an essential singularity. If you already know that $f$ has a representation $f(z) = z^n\cdot e^{g(z)}$, what does that tell you about the entire function $g$? – Daniel Fischer Mar 02 '16 at 19:58
  • Ok Ok.. It says it can not be an essential singularity.. I got it –  Mar 02 '16 at 20:02
  • 1
    $w_0 = 0$ is an excellent choice. The theorem does not say $z^n e^{g(z)} \to 0$ for $z\to\infty$. It only asserts the existence of sequences $(z_k)$ with $z_k\to\infty$ and $z_k^n e^{g(z_k)} \to 0$ if $f$ is transcendental [which means it has an essential singularity at $\infty$]. – Daniel Fischer Mar 02 '16 at 20:06
  • @DanielFischer http://math.stackexchange.com/questions/531747/detailed-proof-that-no-essential-singularity-at-infinity-implies-polynomial says that as $\infty$ is not an essential singularity, $f$ has to be a polynomial... In particular $g(z)$ is constant and so $f(z)=cz^n$.. –  Mar 02 '16 at 20:10
  • Okay. Can you write up the argument to answer your question? – Daniel Fischer Mar 02 '16 at 20:18
  • @DanielFischer : I am still not quite sure about $f=z^n e^g$ part.. Once i am fine with that i will write the answer –  Mar 03 '16 at 06:59
  • 2
    Note that instead of $\lt$ you can use $\le$ by taking the closure. Then for $M=0,$ the set of zeros must be connected. – Bumblebee Aug 19 '20 at 23:35
  • M=0 is not allowed. It's given M>0. – Lawrence Mano Aug 19 '20 at 23:44
  • Agreed. It has only one zero in some closed neighborhood of 0. But it does not immediately imply that f has only one zero in the entire complex plane. – Lawrence Mano Aug 20 '20 at 00:12

4 Answers4

5

We can write $f(z) = z^kg(z)$ for some $k\in \mathbb N,$ where $g$ is entire and $g(0)\ne 0.$ Choose $r>0$ such that $g\ne 0$ in $\{|z|\le r\}.$ Then

$$m= \min_{|z|=r}r^k|g(z)|>0.$$

Now $0\in \{|f(z)| < m\},$ and this set doesn't intersect $\{|z|=r\}.$ Because $\{|f(z)| < m\}$ is given to be connected, it must lie in $\{|z|<r\}.$ Thus all zeros of $f$ lie in $\{|z|<r\}.$ It follows that $f$ has only one zero, namely the one at $0.$ Hence $g(z)$ never vanishes.

Again, $\{|f(z)| < m\}$ lies in $\{|z|<r\}.$ Thus if $|z|\ge r,$ we must have $|f(z)| \ge m.$ But an entire function that behaves this way cannot have an essential singularity at $\infty.$ Thus $f$ has at most a pole at $\infty,$ which means $f$ is a polynomial. But a polynomial with a $k$th order zero at $0$ and no other zeros, has the form $cz^k.$ That is the desired result.

zhw.
  • 105,693
2

Let $D_M=\{z:|f(z)|<M\}$ open connected set; since $f$ is non-constant $D_M$ is not the plane for any $M$; given any Jordan curve $J \subset D_M$, the interior of $J$ is contained in $D_M$ by maximum modulus, hence $D_M$ is simply connected.

Let now $B_{2r}$ a small disc of radius $2r$ around $0$ where $f$ vanishes only at $z=0$ and $2a=\min_{|z|=r}|f(z)| >0$; since then $f(\partial B_r) \cap D_a = \emptyset$ it follows that $D_a \subset B_r$ (as otherwise if there is $|w|>r, |f(z)| <a$ there is a path in $D_a$ joining $w$ and $0 \in B_r$ and that must intersect $\partial B_r$) so in particular $D_a$ bounded and $f$ vanishes only at zero.

It follows that $f$ is a polynomial since $\infty$ is not an essential singularity ($|f(z)| > a, |z|>r$ and since $f$ vanishes only at $0$ we are done!

Conrad
  • 27,433
0

We may assume that $$ f(z)=cz^n(1+c_1z+c_2z^2+\cdots)=cz^ng(z), $$ where $c>0$ (for simplicity) and $g(z)$ has no zeros in $|z|\le r_0$ for some positive $r_0$. Furthermore there is a positive $r_1<r_0$ such that \begin{align} |f(z)|>\frac{c{r_1}^n}{2} \quad \text{on}\,\;|z|=r_1,\tag{1} \end{align} since $|g(z)|\to 1$ $(\,z\to 0\,)$.

Let $M=\displaystyle\frac{c{r_1}^n}{2}$ and $A=\{z:\, |f(z)<M\}$.
If $g(z_0)=0$ for some point $z_0\,(\,|z_0|>r_0\,)$, then $A$ contains $0$ and $z_0$, and hence by the connectedness there is a curve $\gamma \subset A$ joining $0$ and $z_0$. But this is impossible by $(1)$.
Therefore $g(z)\ne 0$ and hence we can write $g(z)=e^{-h(z)}$ , $h(z)=\sum_{k=1}^\infty a_kz^k$.

Suppose that there exists $a_k\ne 0$. Let $H(r)=\max_{|z|=r} \operatorname{Re}\, h(z)$. Then as well-known $$ |a_k|r^k\le \max\{ 4H(r), 0\}-2\operatorname{Re}\, h(0)= 4H(r).$$ Hence we have \begin{align} \min_{|z|=r} |g(z)|&=\min_{|z|=r} \left|e^{-h(z)}\right|=e^{-H(r)}\le e^{-|a_k|r^k/4}. \end{align}

Thus for sufficiently large $r$ we have $$ |f(z_0)|\le cr^n\cdot e^{-|a_k|r^k/4}<M$$ at a point $z_0$ on $|z|=r$. Then $A$ contains $z_0$, but it is impossible.
Thus all $a_k=0$ and we have $f(z)=cz^n$.

ts375_zk26
  • 4,872
0

Suppose $f$ has Taylor series $$ f(z)=\sum_{n=0}^{\infty}a_nz^n \quad\text{and }\quad g(z) =f(1/z) =\sum_{n=0}^{\infty}\frac{a_n}{z^n} $$ If $f$ is not polynomial, then $\infty$ is an essential singularity of $f$ ($0$ is an essential singularity of $g$). By Casorati-Weierstrass theorem, for $0\in \Bbb{C}$, there is a sequence $z_n'\to0$ such that $\lim_{n\to\infty}g(z_n')=0$, i.e. there is $z_n=1/z_n'\to\infty$ such that $\lim_{n\to\infty}f(z_n)=0$. This means that $f$ has always $z_0$ that $f(z_0)=0$ near $\infty$.

Since zero of analytic function is isolated and $f(0)=0$, by choosing $M$ so that $z_0\in \{z: \left|f(z)\right|<M\}$ $(0\in \{z: \left|f(z)\right|<M\}$ is obvious$)$, we conclude that $\{z: \left|f(z)\right|<M\}$ can not be connected because $z_0$ and $0$ are separated. Thus $f$ must be polynomial.

If $f$ has nonzero root, i.e. $f(z_1)=0,\: z_1\ne 0$, then $\{z: \left|f(z)\right|<M\}$ including $z_1$ can not be connected for $z_1$ and $0$ are separated. So $f(z)=0$ can only has zero root. Thus we conclude that $f(z)=cz^n$.

Eugene Zhang
  • 16,805