1. ## Analytic continuation

This seems to be a hard problem for me. Can someone help?
Suppose $F(z)=\sum_{n=1}^{\infty} d(n)z^n, for\mid{z}\mid <1$ $d(n)$ is the number of divisors of n. The radius of convergence of this series is 1. Show that $\sum_{n=1}^{\infty} d(n)z^n=\sum_{n=1}^{\infty}\frac{z^n}{1-z^n}$. Then use this to show that if $z=r$ with $0, then $\mid{F(r)}\mid\geq c\frac{1}{1-r}log(1/(1-r))$ as $r\rightarrow 1$.
Smilarly, if $\theta=2\pi p/q$ $p,q$ are positive integers and $z=re^{i\theta}$, then $\mid{F(re^{i\theta})}\mid\geq c_{p/q}\frac{1}{1-r}log(1/(1-r))$ as $r\rightarrow 1$. Conclude that there can not be any regular point on the unit disc.

2. I can perhaps start it:

$\frac{q^n}{1-q^n}=\sum_{k=1}^{\infty} \left(q^{n}\right)^k$

So that

$\sum_{n=1}^{\infty}\frac{q^n}{1-q^n}=\sum_{n=1}^{\infty}\sum_{k=1}^{\infty}\left(q ^n\right)^k$

$=q+q^2+q^3+q^4+\cdots$
$+q^2+q^4+q^6+q^8+\cdots$
$+q^3+q^6+q^9+q^{12}+\cdots$
$+q^4+q^8+q^{12}+q^{16}+\cdots$
$+q^5+q^{10}+q^{15}+q^{20}+\cdots$

$+\vdots$

Grouping like-powers, it becomes obvious that the number of terms for each power is the divisor function $d(nk)$ so that that sum can be written as:

$\sum_{n=1}^{\infty}\frac{q^n}{1-q^n}=\sum_{n=1}^{\infty}d(n)q^n$

The rest I assume may be approached from the general theory of Lacunary series and the Hadamard Gap Theorem although I wouldn't know how.

3. Originally Posted by shawsend
I can perhaps start it:

$\frac{q^n}{1-q^n}=\sum_{k=1}^{\infty} \left(q^{n}\right)^k$

So that

$\sum_{n=1}^{\infty}\frac{q^n}{1-q^n}=\sum_{n=1}^{\infty}\sum_{k=1}^{\infty}\left(q ^n\right)^k$

$=q+q^2+q^3+q^4+\cdots$
$+q^2+q^4+q^6+q^8+\cdots$
$+q^3+q^6+q^9+q^{12}+\cdots$
$+q^4+q^8+q^{12}+q^{16}+\cdots$
$+q^5+q^{10}+q^{15}+q^{20}+\cdots$

$+\vdots$

Grouping like-powers, it becomes obvious that the number of terms for each power is the divisor function $d(nk)$ so that that sum can be written as:

$\sum_{n=1}^{\infty}\frac{q^n}{1-q^n}=\sum_{n=1}^{\infty}d(n)q^n$

The rest I assume may be approached from the general theory of Lacunary series and the Hadamard Gap Theorem although I wouldn't know how.
Thanks shawsend. I don't think I have learned either of the theorems you mentioned yet.

4. Ok. Then I probably misunderstand the problem. Don't want to lead you down the wrong path.

I think I was confusing convergence of the series on the disc of convergence and the ability to analytically continue the series past the boundary. These two I think are not related. My understanding of the problem then is that there are no point on the boundary that the series can be analytically continued to a disc containing a point on the boundary. Or no?

5. You can get the result for z=r by an elementary calculation:

\begin{aligned}F(r) = \frac r{1-r} + \frac{r^2}{1-r^2} + \frac{r^3}{1-r^3} + \ldots &= \frac1{1-r}\left(r + \frac {r^2}{1+r} + \frac{r^3}{1+r+r^2} + \ldots\right)\\ &\geqslant \frac1{1-r}\left(r + \frac {r^2}2 + \frac{r^3}3 + \ldots\right)\\ &= \frac1{1-r}\ln(1/(1-r)).\end{aligned}

But the result for $z = re^{2\pi pi/q}$ doesn't look as though it will come out easily from that method.