1
$\begingroup$

Let $\mu(n)$ be the Möbius function and let $a(n)$ be the Dirichlet inverse of the Euler totient function:

$$a(n) = \sum\limits_{d|n} d \cdot \mu(d)$$

Let the matrix $T$ be defined as:

$$T(n,k)=a(GCD(n,k))$$

where GCD = the Greatest Common Divisor.

This matrix $T$ starts:

$$T = \left( \begin{array}{ccccccc} +1&+1&+1&+1&+1&+1&+1&\cdots \\ +1&-1&+1&-1&+1&-1&+1 \\ +1&+1&-2&+1&+1&-2&+1 \\ +1&-1&+1&-1&+1&-1&+1 \\ +1&+1&+1&+1&-4&+1&+1 \\ +1&-1&-2&-1&+1&+2&+1 \\ +1&+1&+1&+1&+1&+1&-6 \\ \vdots&&&&&&&\ddots \end{array} \right)$$

There is then the symmetry that the von Mangoldt function can be computed as: $$\Lambda(n) = \sum\limits_{k=1}^{k=\infty}\frac{T(n,k)}{k}$$

$$\Lambda(k) = \sum\limits_{n=1}^{n=\infty}\frac{T(n,k)}{n}$$

In the language of Dirichlet series this is equivalent to: $$\Lambda(n)=\lim\limits_{s \rightarrow 1} \zeta(s)\sum\limits_{d|n} \frac{\mu(d)}{d^{(s-1)}}$$

Can we then say that the Dirichlet series generating function for the whole matrix $T$ is:
$$\sum\limits_{k=1}^{\infty}\sum\limits_{n=1}^{\infty} \frac{T(n,k)}{n^c \cdot k^z} = \sum\limits_{n=1}^{\infty} \frac{\lim\limits_{s \rightarrow z} \zeta(s)\sum\limits_{d|n} \frac{\mu(d)}{d^{(s-1)}}}{n^c} = \frac{\zeta(z) \cdot \zeta(c)}{\zeta(c + z - 1)}$$?

For $c$ a value close to one such as $c=1+\frac{1}{30}$ we have this plot for $\frac{\zeta (c) \zeta (\frac{1}{2}+i t)}{\zeta (c+\frac{1}{2}+i t-1)}$ on the critical line in the complex plane:

Dirichlet series spectra

where the downward spikes are at Riemann zeta zeros.

(*Mathematica 8 start*)
Clear[s, z, c]
z = 9
c = 7
N[Zeta[z]*Zeta[c]/Zeta[c + z - 1], 20]
N[Sum[Limit[
    Zeta[s] Total[1/Divisors[n]^(s - 1)*MoebiusMu[Divisors[n]]], 
    s -> z]/n^c, {n, 1, 64}], 20]
Clear[c, ss, t]
ss = 30
c = 1 + 1/ss;
sigma = 1/2;
Plot[Re[N[Zeta[sigma + I*t]*Zeta[c]/Zeta[c + sigma + I*t - 1], 
   20]], {t, 0, 60}, PlotRange -> {0, ss + 2}]
(*Mathematica 8 end*)
$\endgroup$
  • 1
    $\begingroup$ I think you should look at $\sum_{k \ge 1} k^{-s} T(n,k) = \sum_{k \ge 1} k^{-s} \sum_{d | gcd(n,k)} \mu(d) d = \sum_{d | n} \mu(d) d \sum_{k \ge 1, d | k} k^{-s} = \sum_{d | n} \mu(d) d \sum_{m \ge 1} (md)^{-s} = \sum_{d | n} \mu(d) d^{1-s} \zeta(s) $ $\endgroup$ – reuns Jun 19 '16 at 17:22
  • 1
    $\begingroup$ so that $$\sum_k \sum_n n^{-z} k^{-c} T(n,k) = \zeta(c) \sum_n n^{-z} \sum_{d | n} \mu(d) d^{1-c} $$ on the other hand $$\frac{\zeta(z)}{\zeta(z+c-1)} = \sum_n n^{-z} \sum_k \mu(k) k^{-z-c+1} =\sum_m m^{-z} \sum_{d | m} \mu(d) d^{1-c}$$ $\endgroup$ – reuns Jun 19 '16 at 17:28
  • $\begingroup$ Mathematica tells me: $$-\frac{\zeta '(s)}{\zeta (s)}=\lim_{c\to 1} \, \left(\frac{\zeta (c) \zeta (s)}{\zeta (c+s-1)}-\zeta (c)\right)$$ $\endgroup$ – Mats Granvik Jun 26 '16 at 14:25
  • 1
    $\begingroup$ mathematica told me... as $c \to 1$ : $\zeta(c) = \frac{1}{c-1}+ \mathcal{O}(1)$ and with $f(c) = \frac{1}{\zeta(c+s-1)}$, $\ f(c) =f(1)+(c-1)f'(1) + o(c-1)$ where $f(1) = \frac{1}{\zeta(s)}$ and $f'(1) = -\frac{\zeta'(s)}{\zeta(s)^2}$ hence $$\lim_{c \to 1} \zeta(c) \left(\frac{\zeta(s)}{\zeta(c+s-1)}-1\right) =\lim_{c \to 1} (\frac{1}{c-1}+\mathcal{O}(1))\left(\zeta(s)(f(1)+(c-1)f'(1) + o(c-1))-1\right) $$ $$= \lim_{c \to 1} (\frac{1}{c-1}+ \mathcal{O}(1))\left(-(c-1)\frac{\zeta'(s)}{\zeta(s)}+ o(c-1)\right) = \ldots$$ $\endgroup$ – reuns Jun 26 '16 at 18:52
0
$\begingroup$

joriki proved here that for $n>1$ the von Mangoldt function is:

$$\Lambda(n)=\displaystyle \sum\limits_{k=1}^{\infty}\frac{T(n,k)}{k}$$

so if $$\frac{\zeta (c) \zeta (s)}{\zeta (c+s-1)}$$ is to be the generating function associated with the double sum:

$$\sum\limits_{k=1}^{\infty}\sum\limits_{n=1}^{\infty} \frac{T(n,k)}{n^c \cdot k^s} = \frac{\zeta(s) \cdot \zeta(c)}{\zeta(c + s - 1)}$$

(explicitly valid for $c>1$ and $s>1$), then when subtracting $\zeta(c)$, which is the first column in the matrix $T$, we should get the Dirichlet generating function for the von Mangoldt function $\Lambda(n)$:

$$\sum _{n=1}^{\infty } \frac{\Lambda (n)}{n^s}=-\frac{\zeta '(s)}{\zeta (s)}=\lim_{c\to 1} \, \left(\frac{\zeta (c) \zeta (s)}{\zeta (c+s-1)}-\zeta (c)\right)$$

which we do according to Mathematica.

$\endgroup$
  • $\begingroup$ as $c \to 1$ : $\zeta(c) = \frac{1}{c-1}+\gamma+\mathcal{O}(c-1)$. Do you get that (this is the Taylor expansion of the analytic function $\zeta(c) (c-1)$) ? And again from the Taylor expansion $\zeta(s+c-1)= \zeta(s)+ (c-1)\zeta'(s)+\mathcal{O}((c-1)^2)$ so that $$\lim_{c \to 1}\frac{\zeta(c)\zeta(s)}{\zeta(c+s-1)}-\zeta(c) = \lim_{c \to 1}\frac{1}{c-1}(\frac{\zeta(s)}{\zeta(s)+(c-1)\zeta'(s)+\mathcal{O}((c-1)^2)}-1)=\lim_{c \to 1}\frac{1}{c-1}\frac{-(c-1)\zeta'(s)}{\zeta(s)+(c-1)\zeta'(s)}=\frac{-\zeta'(s)}{\zeta(s)}$$ $\endgroup$ – reuns Nov 28 '16 at 1:54
  • $\begingroup$ This is detailed in any real analysis course $\endgroup$ – reuns Nov 28 '16 at 1:54

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.