1
$\begingroup$

Is it known if a functional equation of the form:

$$\zeta (s)=f(s) \zeta (s+1)$$

can exist?

If it is possible for such a functional equation to exist then I believe lots of wonderful things would happen. In particular one could solve this integral:

$$\int \log (\zeta (s)) \, ds$$

on the critical line by extending the validity of the Euler product formula to it.

$\endgroup$
  • $\begingroup$ Well you may intend some conditions on $f$. Easily $f(s)=1$ or $f(s)=1/(s+1)$ is possible to solve. $\endgroup$ – Macavity Nov 6 '17 at 13:40
  • 7
    $\begingroup$ It certainly holds for $f(s)=\frac{\zeta(s)}{\zeta(s+1)}$. Welcome to the tautology club. $\endgroup$ – Jack D'Aurizio Nov 6 '17 at 13:41
  • $\begingroup$ @JackD'Aurizio Did you know the Riemann explicit formula gives for $\Re(s) > \sigma_{RH}$ : $\frac{-\zeta'(s)}{\zeta(s)}= \sum_{n=2}^\infty (\psi(n+1/2)-\psi(n-1/2)) n^{-s} = \sum_{n=2}^\infty (1-\sum_\rho \frac{(n+1/2)^\rho-(n-1/2)^\rho}{\rho})n^{-s}$ $=(\zeta(s)-1)-\sum_\rho \frac{1}{\rho}\sum_{k \ge 0} (\zeta(s+\rho)-1) {\rho \choose k} (2^{-k}-(-2)^{-k})$ where $\rho$ are the trivial and non-trivial zeros $\endgroup$ – reuns Nov 7 '17 at 10:21
  • $\begingroup$ @reuns Shouldn't the $\zeta(s+\rho)$ in your last formula be $\zeta(s+k-\rho)$ ? $\endgroup$ – Agno Dec 2 '17 at 19:19
  • $\begingroup$ @MatsGranvik Although it is the reciprocal of your question, maybe this link is helpful mathoverflow.net/questions/23378/zetas1-zetas $\endgroup$ – Agno Dec 2 '17 at 19:22
6
+50
$\begingroup$

Substituting $s=s+1$ into the relationship $\frac{\zeta (s-1)}{\zeta (s)}=\sum _{n=1}^{\infty } \frac{\phi (n)}{n^s}$ leads to $\zeta (s)=\zeta (s+1) \sum _{n=1}^{\infty } \frac{\phi (n)}{n^{s+1}}$ which is valid for $\Re(s)>1$.

Also, $\int\log(\zeta(s))\,ds=-\sum\limits_{n=2}^\infty\frac{\Lambda(n)}{\log(n)^2}\,n^{-s}$, but again this relationship only converges for $\Re(s)>1$.

I'm not sure where you're headed with this, but I suspect you'll be interested in the paper at the following link:

DISCRETE MELLIN CONVOLUTION AND ITS EXTENSIONS, PERRON FORMULA AND EXPLICIT FORMULAE


April 3, 2018 Update:


I've attempted to derive a formula for $\int\log\zeta(s)\,ds$ that assuming the Riemann hypothesis converges for $\Re(s)>\frac{1}{2}$ starting with the following relationship.

(1) $\quad\frac{\partial\,\log\zeta(s)}{\partial s}=\frac{\zeta'(s)}{\zeta(s)}=-s\left(\frac{1}{s-1}+\int_1^N x^{-s-1}(\psi(x)-x)\,dx\right),\quad N\to\infty\land\Re(s)>\frac{1}{2}$


The following formula for $\frac{\partial\,\log\zeta(s)}{\partial s}=\frac{\zeta'(s)}{\zeta(s)}$ was derived from relationship (1) above.

(2) $\quad\frac{\partial\,\log\zeta(s)}{\partial s}=\frac{\zeta'(s)}{\zeta(s)}=\frac{s\,N^{1-s}}{1-s}+\sum\limits_{n=1}^N\Lambda(n)\left(N^{-s}-n^{-s}\right),\quad N\to\infty\land\Re(s)>\frac{1}{2}$


Integrating (2) above leads to the following where I added the $-sgn(\Im(s))\,\pi\,i$ term to adjust for the branch point of the $Ei$ function at $0$.

(3) $\quad\log\zeta(s)=-sgn(\Im(s))\pi\,i-Ei\,((1-s) \log (N))+\frac{N^{1-s}}{\log (N)}+\sum\limits_{n=2}^N\Lambda(n)\left(\frac{n^{-s}}{\log(n)}-\frac{N^{-s}}{\log (N)}\right),\\$ $\qquad\qquad\qquad\qquad\qquad\qquad\qquad N\to\infty\land\Re(s)>\frac{1}{2}$


Integrating (3) above leads to the following where this time I added the $sgn(\Im(s))\,\pi\,i$ term to adjust for the branch point of the $Ei$ function at $0$.

(4) $\quad\int\log\zeta(s)\,ds=sgn(\Im(s))\,\pi\,i+(1-s)\,Ei\,((1-s)\log(N))-\frac{(\log(N)+1)\,N^{1-s}}{\log^2(N)}+\\$ $\qquad\qquad\quad\left( \begin{array}{cc} \{ & \begin{array}{cc} -i \pi s & \Im(s)>0 \\ i \pi s & \Im(s)<0 \\ \end{array} \\ \end{array} \right)+\sum\limits_{n=2}^N\Lambda(n)\left(\frac{N^{-s}}{\log^2(N)}-\frac{n^{-s}}{\log^2(n)}\right),\quad N\to\infty\land\Re(s)>\frac{1}{2}$


The formulas for $\frac{\partial\,\log\zeta(s)}{\partial s}$, $log\,\zeta(s)$, and $\int\log\zeta(s)\,ds$ in (2), (3), and (4) above are illustrated below where all plots are evaluated along the critical line $s=\frac{1}{2}+i\,t$ using the limit $N=1000$. The reference functions are illustrated in orange and the right-side of the formulas are illustrated in blue, The red discrete portions of the plots illustrate the evaluations of the right-side of the formulas at the non-trivial zeta zeros.

Note how the diverging oscillation in the plots below seems to decrease in magnitude as successive integrals are taken first from $\frac{\partial\,\log\zeta(s)}{\partial s}$ to $log\,\zeta(s)$ and second from $log\,\zeta(s)$ to $\int\log\zeta(s)\,ds$.


The following four plots illustrate the real, imaginary, absolute, and argument of formula (2) above for $\frac{\partial\,\log\zeta(s)}{\partial s}=\frac{\zeta'(s)}{\zeta(s)}$.


Illustration of <span class=$\Re(\frac{\partial\,\log\zeta(s)}{\partial s})$ associated with Formula (2) for $s=\frac{1}{2}+i\,t$">

Figure (1): $\text{Illustration of $\Re(\frac{\partial\,\log\zeta(s)}{\partial s})$ associated with Formula (2) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$\Im(\frac{\partial\,\log\zeta(s)}{\partial s})$ associated with Formula (2) for $s=\frac{1}{2}+i\,t$">

Figure (2): $\text{Illustration of $\Im(\frac{\partial\,\log\zeta(s)}{\partial s})$ associated with Formula (2) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$Abs(\frac{\partial\,\log\zeta(s)}{\partial s})$ associated with Formula (2) for $s=\frac{1}{2}+i\,t$">

Figure (3): $\text{Illustration of $Abs(\frac{\partial\,\log\zeta(s)}{\partial s})$ associated with Formula (2) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$Arg(\frac{\partial\,\log\zeta(s)}{\partial s})$ associated with Formula (2) for $s=\frac{1}{2}+i\,t$">

Figure (4): $\text{Illustration of $Arg(\frac{\partial\,\log\zeta(s)}{\partial s})$ associated with Formula (2) for $s=\frac{1}{2}+i\,t$}$


The following four plots illustrate the real, imaginary, absolute, and argument of formula (3) above for $log\,\zeta(s)$.


Illustration of <span class=$\Re(\log\zeta(s))$ associated with Formula (3) for $s=\frac{1}{2}+i\,t$">

Figure (5): $\text{Illustration of $\Re(\log\zeta(s))$ associated with Formula (3) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$\Im(\log\zeta(s))$ associated with Formula (3) for $s=\frac{1}{2}+i\,t$">

Figure (6): $\text{Illustration of $\Im(\log\zeta(s))$ associated with Formula (3) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$Abs(\log\zeta(s))$ associated with Formula (3) for $s=\frac{1}{2}+i\,t$">

Figure (7): $\text{Illustration of $Abs(\log\zeta(s))$ associated with Formula (3) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$Arg(\log\zeta(s))$ associated with Formula (3) for $s=\frac{1}{2}+i\,t$">

Figure (8): $\text{Illustration of $Arg(\log\zeta(s))$ associated with Formula (3) for $s=\frac{1}{2}+i\,t$}$


The following four plots illustrate the real, imaginary, absolute, and argument of formula (4) above for $\int\log\zeta(s)\,ds$.


Illustration of <span class=$\Re(\int\log\zeta(s)\,ds)$ associated with Formula (4) for $s=\frac{1}{2}+i\,t$]">

Figure (9): $\text{Illustration of $\Re(\int\log\zeta(s)\,ds)$ associated with Formula (4) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$\Im(\int\log\zeta(s)\,ds)$ associated with Formula (4) for $s=\frac{1}{2}+i\,t$">

Figure (10): $\text{Illustration of $\Im(\int\log\zeta(s)\,ds)$ associated with Formula (4) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$Abs(\int\log\zeta(s)\,ds)$ associated with Formula (4) for $s=\frac{1}{2}+i\,t$">

Figure (11): $\text{Illustration of $Abs(\int\log\zeta(s)\,ds)$ associated with Formula (4) for $s=\frac{1}{2}+i\,t$}$


Illustration of <span class=$Arg(\int\log\zeta(s)\,ds)$ associated with Formula (4) for $s=\frac{1}{2}+i\,t$">

Figure (12): $\text{Illustration of $Arg(\int\log\zeta(s)\,ds)$ associated with Formula (4) for $s=\frac{1}{2}+i\,t$}$


12/04/2018 Update:


Formula (7) below for $\zeta(s)=f(s)\,\zeta(s+1)$ was derived from the relationships illustrated in (5) and (6) below. Formula (7) below assumes the Riemann Hypothesis (RH).


(5) $\quad\zeta(s)=\frac{2\,\pi^{-\frac{1}{2}}\,\Gamma\left(\frac{s+3}{2}\right)}{(s-1)\,\Gamma\left(\frac{s}{2}\right)}\frac{\xi(s)}{\xi(s+1)}\zeta(s+1)$

(6) $\quad \xi(s)=\xi(0)\prod\limits_{k=1}^\infty\left(1-\frac{s}{\rho_k}\right)\left(1-\frac{s}{\rho_{-k}}\right)$

(7) $\quad \zeta(s)=\frac{2\,\pi^{-\frac{1}{2}}\Gamma\left(\frac{s+3}{2}\right)}{(s-1)\Gamma\left(\frac{s}{2}\right)}\left(\prod\limits_{k=1}^\infty\left(1-\frac{8\,s}{4\,\Im\left(\rho_k\right){}^2+(2\,s+1)^2}\right)\right)\zeta(s+1)\qquad\text{(assuming RH)}$


The following four figures illustrate formula (7) for $\zeta(s)$ above evaluated along the critical line $s=\frac{1}{2}+i\,t$ where formula (7) is evaluated over the first $200$ non-trivial zeta zeros in the upper half plane. Formula (7) is illustrated in orange, and $\zeta(s)$ is illustrated in blue as a reference.


Illustration of formula (7) for Abs(zeta(1/2+i t))

Figure (13): Illustration of formula (7) for $\left|\zeta(\frac{1}{2}+i\,t)\right|$


Illustration of formula (7) for Re(zeta(1/2+i t))

Figure (14): Illustration of formula (7) for $\Re\left(\zeta(\frac{1}{2}+i\,t)\right)$


Illustration of formula (7) for Im(\zeta(1/2+i t))

Figure (15): Illustration of formula (7) for $\Im\left(\zeta(\frac{1}{2}+i\,t)\right)$


Illustration of formula (7) for Arg(zeta(1/2+i t))

Figure (16): Illustration of formula (7) for $Arg\left(\zeta(\frac{1}{2}+i\,t)\right)$


$\endgroup$
  • $\begingroup$ What I am asking for is probably an impossibility. What you posted here is probably as close as one can get. $\endgroup$ – Mats Granvik Feb 20 '18 at 18:16

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.