0

Question is :

Are there any non-constant entire functions $f$ that satisfy an inequality of the form $|f(z)|\leq A+B\log |z|$ for all $z$ with $|z|\geq 1$, where $A,B$ are positive constants.

Let $z_0\in \mathbb{C}$. As $f$ is entire and $z_0\in \mathbb{C}$, given any $R>0$ and the circle $D_R$ at $z_0$ of radius $R$, we have
$$f^{(1)}(z_0)=\frac{1}{2\pi i}\int_{\partial D_R} \frac{f(z)}{(z-z_0)^2}dz.$$

For simplicity, take $z_0=0$. Parametrizing the circle we have $z(\theta)=Re^{i\theta}$ then $dz=iRe^{i\theta}d\theta$.

So $$f^{(1)}(0)=\frac{1}{2\pi i}\int_0^{2\pi} \frac{f(Re^{i\theta})}{R^2e^{i2\theta}}dz.$$ Considering modulus, we have $$\left|f^{(1)}(0)\right| \leq\frac{1}{2\pi }\int_0^{2\pi} \frac{\left|f(Re^{i\theta})\right|}{R}d\theta.$$ Now, we have $|f(z)|\leq A+B \log |z|$ for all $z$ with $|z|\geq 1$. We have $|Re^{i\theta}|=R$. So, for $R>1$ we have $$|f^{(1)}(0)| \leq\frac{1}{2\pi }\int_0^{2\pi} \frac{|f(Re^{i\theta})|}{R}d\theta \leq \frac{1}{2\pi }\int_0^{2\pi} \frac{A+B\log R}{R}d\theta=\frac{A+B\log R}{R}.$$ As $R\rightarrow \infty$ we have $\frac{A+B\log R}{R}\rightarrow 0$. So, $|f^{(1)}(0)|=0$.

Everything till now is fine. Now, i wanted to do this for arbitrary $z_0$ other than $0$. Everything would be same except that this $R\geq 1$ may not work.

We want $R$ such that $\left|z_0+Re^{i\theta}\right|\geq 1$. I do not see how to get this. Using the inequality $$\left|z_0+Re^{i\theta}\right|\geq |z_0|-|Re^{i\theta}|=|z_0|-R$$ I thought considerng $R$ such that $-R\geq 1-|z_0|$ equivalently $R\leq |z_0|-1$ may work.. But then this would restrict $R$ and we can not think of $R$ going to infinity and it may happen that $|z_0|<1$ so $R$ would turn out to be negative which is not acceptable.

Seeing all this I think we only have $\left|f'(0)\right|=0$ and for arbitrary $z_0$ it may not be true. Then i have to give one counterexample $f$.. Could not think of any such.

  • an entire function bounded by a "polynomial" of degree $a$ is itself a polynomial of degree (at most) $\lfloor a \rfloor$, hence an entire function bounded by a polynomial in $\ln z$ has to be constant. – reuns Apr 07 '16 at 12:21
  • @user1952009 : Please provide proof for that –  Apr 07 '16 at 12:22
  • for the bounded by a polynomial part, you can look at http://math.stackexchange.com/questions/1665885/growth-estimate-of-an-entire-function/1665954#1665954 or search for "Liouville polynomial" – reuns Apr 07 '16 at 12:23
  • @user1952009 : I know that it is true for bounded polynomial.. I am looking for a proof of logarithm result –  Apr 07 '16 at 12:23
  • $\ln z = \mathcal{O}(z^{1/2})$ and $\lfloor 1/2 \rfloor = 0$ – reuns Apr 07 '16 at 12:24
  • see the steps I use in the link I gave : $f(z) = \mathcal{O}(z^a)$ and is entire $ \implies$ $f(z)$ is a polynomial of degree $\le \lceil a \rceil$, hence comparing the growth rate, it is a polynomial of degree at most $\lfloor a \rfloor$ – reuns Apr 07 '16 at 12:27
  • Almost (but not quite) exact duplicates: http://math.stackexchange.com/questions/171610 http://math.stackexchange.com/questions/458071 – mrf Apr 07 '16 at 12:27
  • It seems to me you may estimate the Taylor coefficients from the Cauchy Integral Formula (Theorem 23.4 in http://stat.math.uregina.ca/~kozdron/Teaching/Regina/312Fall12/Handouts/312_lecture_notes_F12_Part2.pdf for example), over an origin centered circle with radius r which can be made arbitrarily large. You should get that all the coefficients are 0 apart from the 0-th coefficient. –  Apr 07 '16 at 12:29
  • @JohnDonn : yes that it is the usual proof of the Liouville theorem and pf its extended version (when bounded by a polynomial) – reuns Apr 07 '16 at 12:30
  • @user1952009 : I do not know about the floor functions and that notation $\mathcal{O}$ –  Apr 07 '16 at 12:31
  • @mrf : That was useful. –  Apr 07 '16 at 12:31
  • @JohnDonn : I think that is what i have done.. Please see the question –  Apr 07 '16 at 12:33
  • If you show that all the Taylor coefficients at z=0 of your entire function are zero except perhaps for the constant term, you have proved that your function is a constant. Or maybe I did not quite understand your problem. –  Apr 07 '16 at 12:47
  • It would have been nice if some one has commented something about my work. Just saying this is a duplicate of some other question which have not showed any effort would only demotivate the user... –  Apr 07 '16 at 13:01
  • @JohnDonn : If i prove that at $z=0$ all taylor coefficients are zero then it does not say the function is constant.. See http://math.stackexchange.com/questions/240026/all-derivatives-zero-at-a-point-implies-constant-function –  Apr 07 '16 at 13:05
  • but you know that your function is not just smooth, it is entire.. –  Apr 07 '16 at 13:16
  • @JohnDonn : I was careless... Got it... Thanks a lot.. –  Apr 07 '16 at 13:20

0 Answers0