Can you prove the next theorem:

Let $f$ be Dirichlet series with real, positive coefficients $(a_n>0)$. If $f$ is holomorphic on $\Re(z)\ge1$, but has one singularity at $z=1$, then

$\lim_\limits{z \to 0^+} \left[f(z)-c(z-1)^k\right]$ exists $\iff$ $\lim_\limits{x \to \infty} \frac{A(x)\ln(x)^{k+1}}{x}=c$

where $c$ and $k$ are some real constants and

$A(x)=\sum_{n=1}^x a_n$

Behind the theorem:

I'm exploring complex analysis methods in number theory, specifically for determining asymptotic growth of partial sums of multiplicative functions. I have read Newman's proof of PNT, and I tried to use these ides to find the asymptotic growth of the next sum:

$\sum_{n=1}^{x} \frac{1}{d(n)}$

where $d(n)$ is the number of divisors of $n$. It seems to be, and computer approves it, that the growth is $\frac{cx}{\sqrt{\ln(x)}}$ for some real constant $c$. Now, in order to prove it, I defined $f$ to be Dirichlet series with coefficients $a_n=\frac{1}{d(n)}$. I proved that $f(z)-c\sqrt{\zeta(z)}$ is holomorphic for $\mathfrak{R}(z)>0$. Also, I have that $f(z)-\frac{c}{\sqrt{z-1}}$ is holomorphic for $\Re(z)\ge1$ except for $1$. It is not holomorphic at the point $1$, so I can't use Newman's analytic theorem. But, if we could ignore it, we would get that,

$\int_1^\infty \frac{A(t)-ct\sqrt{\ln(t)}}{t^2}dt$

converges. (That comes out when we differentiate $f(z)-\frac{c}{\sqrt{z-1}}$). Here I used

$A(x)=\sum_{n=1}^{x} \frac{\ln(n)}{d(n)}$.

Then, using Newman's idea, we can get that $A(x)$~$cx\sqrt{\ln(x)}$. From there, using partial summation, we get that the first sum is asymptotically equal to what it is expected to be, $\frac{cx}{\sqrt{\ln(x)}}$.

Please, help me, this is going to be in my undergraduate work, I need to know these things. Help me to prove this asymptotic growth.

  • 2
    $\begingroup$ I would suggest to take a look at korevaar's tauberian theory bible and see what's known $\endgroup$ – Conrad Mar 27 at 19:17
  • $\begingroup$ @Conrad Can you please give me a link for that? $\endgroup$ – donaastor Mar 27 at 19:31
  • $\begingroup$ This is the link to the official version of the book - any decent university library should have it and of course, I am sure that unofficial versions exist too. springer.com/us/book/9783540210580 $\endgroup$ – Conrad Mar 27 at 22:28

I have found a work on arxiv that proves the very similar theorem and here is the link of it:


It is Ryo-Kato's generalization of Kable's generalization of Wiener-Ikehara Tauberian theorem. Kable states for $k$ of form $\frac{1}{n}$, and Ryo-Kato states it for all rational $k$. The real $k$ is still a mystery, though.

  • $\begingroup$ @Conrad You actually helped me by saying that I should pay attention on Tauberian Theory. Without that I wouldn't know that I should have looked for Tauberian theorems and their extensions. So, thanks $\endgroup$ – donaastor Apr 2 at 18:34

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.