17

Here is a graph of the function $y=-1/x$:

enter image description here

If we add infinitely many similar functions with a shift of $\pi/2$ each in both directions, we get $\tan x$. But if we do the same only in one direction, we get "incomplete tangent":

http://storage7.static.itmages.ru/i/14/0910/h_1410326921_7988832_91f3fd7d7d.png

The yellow one is $\operatorname{pg}(x)=\frac 1\pi \psi (\frac x\pi)$, the blue one is $\operatorname{cpg}(x)=-\frac 1\pi \psi (1-\frac x\pi)$. They obey $\operatorname{cpg}(x)+\operatorname{pg}(x)=-\cot(x)$.

Now if we differentiate cpg(x) we get:

$$(\operatorname{cpg}(x))^{(s-1)}=\pi^{-s}\Gamma(s)\zeta(s,1-\frac x\pi)$$

At $x=0$ it would be

$$\operatorname{cpg}^{(s-1)}(0)=\pi^{-s}\Gamma(s)\zeta(s)$$

Compare it with this formula involving Riemann Xi-function:

$$\xi(2s) = \pi^{-s}\Gamma\left(s\right)\zeta(2s)$$

I wonder is the similarity between these two formulas just a coincidence or Riemann Xi function serves as a consecutive derivative (generating function) of some simple or notable function?

Glorfindel
  • 3,955
Anixx
  • 9,119
  • I don't see what $\tan(x)$ or these other two functions have to do with $1/x$. – 2'5 9'2 Nov 19 '14 at 06:42
  • @alex.jordan: There is a formula expressing the $($co$)$tangent function in terms of digamma functions; there is another formula, expressing digamma functions in terms of harmonic numbers; and then there's yet another formula, expressing harmonic numbers in terms of $\dfrac1x$. This graphic seems to be a geometric and very visually intuitive embodiment of the afore-mentioned algebraic formulas. – Lucian Nov 19 '14 at 07:18
  • I prefer considering such identities in complex. We have some meromorphic function. If we can reproduce the poles such that the difference is bounded (or at least it does not increase too rapidly) then we can apply Liouville.

    For example the function $\cot z=\frac{\sin'}{\sin}(x)$ has simple poles at $k\pi$ with residue $1$. The sum $f(z)=\frac1z+\sum_{k=1}^\infty\left(\frac1{z+k\pi}+\frac1{z-k\pi}\right)$ has the same property. It can be verified that difference is bounded and both functions are odd, so they are equal.

    You can try the same for $\frac1{\sin z}$ or $\frac{\Gamma'}{\Gamma}$.

    – G.Kós Nov 19 '14 at 07:47
  • 1
    @Lucian I just do not understand what "If we add infinitely many similar functions with a shift of pi/2 each in both directions, we get tanx." means. Do you? – 2'5 9'2 Nov 19 '14 at 08:24
  • Oh I see. $\sum\frac{1}{x+n\pi}$ converges to $\cot x$. I wasn't aware of that. – 2'5 9'2 Nov 19 '14 at 08:29
  • I still don't see how the "blue" and "yellow" curves are defined. If you only sum $\sum\frac{1}{x+n\pi}$ in one direction, then for any $x$ all but finitely many terms are of the same sign, and grow harmonically, giving a divergent series. – 2'5 9'2 Nov 19 '14 at 08:35
  • @alex.jordan: Using either Desmos, or GeoGebra, or some other software of your choosing, plot $\dfrac1x$ and $\dfrac1{x\pm1}$, as well as their sum, onto the same graphic. Notice the similarity with the $($co$)$tangent's graph. Continue adding symmetrical pairs of the form $\dfrac1{x\pm n}$, and plotting their sum. Notice what happens. Then plot $\cot(\pi x)$ onto the same graphic. – Lucian Nov 19 '14 at 09:16
  • 3
    @Lucian I get that; what I don't get is what OP means by "if we do the same only in one direction". $\sum_{n\geq n_0}\frac{1}{x+n}$ is divergent. So is $\sum_{n\leq n_0}\frac{1}{x+n}$. – 2'5 9'2 Nov 19 '14 at 13:58
  • @alex.jordan digamma is indefinite sum of 1/x if u prefer http://en.wikipedia.org/wiki/Indefinite_sum#Antidifferences_of_rational_functions – Anixx Nov 19 '14 at 14:01
  • @Anixx Where did you get these graphs of "incomplete tangent" functions? They should not look like that if they are divergent sums ... – Zubin Mukerjee Jan 06 '15 at 19:23
  • @ZubinMukerjee Just calculate the darn sums anyways so we can get on with the question at hand. – Simply Beautiful Art Jun 17 '16 at 11:49
  • They're not divergent if you only add the positive translations, the asymptotes keep it 'balanced' in a sense. – Jacob Claassen May 02 '17 at 03:32
  • the Herglotz trick, see https://math.stackexchange.com/questions/581162/how-does-the-herglotz-trick-work – graveolensa Oct 04 '19 at 03:31

0 Answers0