The Riemann Xi-Function is defined as $$ \xi(s) = \tfrac{1}{2} s(s-1) \pi^{-s/2} \Gamma\left(\tfrac{1}{2} s\right) \zeta(s) $$ and it satisfies the reflection formula $$ \xi(s) = \xi(1-s). $$ But the area $A$ of a $s$-dimensional sphere is $$ A(s) = \frac{2 \pi^{s/2}}{\Gamma\left(\tfrac{1}{2} s\right)} $$ so that we can write the Xi-Function like $$ \xi(s) = s(s-1) \frac{\zeta(s)}{A(s)}. $$ If we insert this into the reflection formula we get the following relation between the area of a n-sphere and the Riemann zeta function $$ \frac{\zeta(s)}{\zeta(1-s)} = \frac{A(s)}{A(1-s)} $$ Has this relation been noted in the literature? Why should there be such a connection between the area of a n-sphere and the zeta function? Can this relation be explained geometrically?

UPDATE: I agree with the comments that negative dimensional spheres are difficult to interpret geometrically. However if we insert the formula for the area $A$ into Eulers reflection formula $$ \Gamma(z)\Gamma(1-z)= \frac{\pi}{\sin(\pi z)} $$ we get $$ A(1-s) = \frac{4\sin\left(\pi \frac{s+1}{2}\right)}{A(s+1)} $$ Inserting this into our relation with the Riemann zeta function yields $$ \frac{\zeta(s)}{\zeta(1-s)} = \frac{A(s)A(s+1)}{4\sin\left(\pi \frac{s+1}{2}\right)} $$ which avoids the negative dimensions for $s>0$. Can this formula be interpreted geometrically?

  • 4
    $\begingroup$ Why should there be such a connection between the area of a n-sphere and the zeta function ? - What is a sphere ? A bounded sum of powers: $x^2+y^2+z^2=r^2.~$ And what is the $\zeta$ function ? A bounded sum of powers: recall its famous infinite series expression for $\Re(s)>1.~$ And by expanding the power of a sum into a sum of powers through Newton's binomial series we get binomial coefficients, aka beta functions, which can be expressed in terms of factorials, aka $\Gamma$ functions. In general, all $\Gamma$ functions of argument $\frac1n$ are connected to superellipses $x^n+y^n=r^n$. $\endgroup$ – Lucian May 20 '16 at 10:14
  • $\begingroup$ I can't follow you. You are saying e.g. $\zeta(3) = \frac{1}{1^3} + \frac{1}{2^3} + \frac{1}{3^3} + ...$ is a bounded sum of powers? It is a bounded sum, but it is a sum with infinite terms to the power 3. Whereas a 3d-sphere is a bounded sum of three terms to the power two. Can you elaborate again on what kind of connection you see between these sum of powers? $\endgroup$ – asmaier May 21 '16 at 20:49
  • $\begingroup$ The superellipse $x^n+y^n=r^n$ has an area which is directly expressible in terms of beta$/\Gamma$ functions, whose arguments are rational numbers of denominator n. At the same time, $\zeta(n)=\displaystyle\sum_{k=1}^\infty\frac1{k^n}.~$ Both are bounded sums of powers. For instance, the fact that $\Gamma\Big(\tfrac12\Big)=\sqrt\pi$ and $\zeta(2n)=a_n\cdot\pi^n$ are both related to the the same fundamental constant is not a coincidence. Since $A(s)$ is expressible in terms of $\Gamma(s),$ the conclusion follows. $\endgroup$ – Lucian May 27 '16 at 7:43
  • 4
    $\begingroup$ What geometrical meaning do you give to $A(-\frac 1 2)$ and $A(\textrm i)$? $\endgroup$ – Alex M. Jun 5 '16 at 17:05
  • $\begingroup$ I don't think you can hope for a simple geometrical interpretation of this. Saying that $A(s)$ is the area of a sphere only makes sense when $s$ is an integer (the dimension of the space). But if $s$ is an integer then $1-s$ is negative (or zero) so there are no choices where both $A(s)$ and $A(1-s)$ can be given a classical geometrical interpretation. $\endgroup$ – Winther Jun 9 '16 at 2:51

The relation $$\frac{\zeta(s)}{A(s)}=\frac{\zeta(1-s)}{A(1-s)}$$ stems from the fact that $$\frac{\zeta(s)}{A(s)}=\frac{1}{2}\mathcal{M}\left(\psi\right)\left(\frac{s}{2}\right),$$ where $\mathcal{M}$ denotes the Mellin transform and $$\psi(x)=\sum_{n=1}^\infty e^{-\pi n^2 x}.$$ This function is symmetric due to the functional equation for $\psi(x)$, and that yields the functional equation for the zeta function. The reason for the connection to the surface area appears in my previous answer https://math.stackexchange.com/a/1494471/6075, which provides a geometric explanation for integer dimensions - the case where we can actually interpret the surface area. In that answer, I showed that the factor of $A(s)$ appears due to the spherical symmetry of a naturally arising $s$-dimensional integral expression for $\zeta(s)$.

Remark 1: This gives an excellent way to remember the precise form of the functional equation!

Remark 2: Note that the surface area here, $A(s)$, corresponds to the $s-1$-dimensional sphere, that is the surface of the $s$-dimensional ball, whereas in the linked answer, $A_{k-1}$ is the surface area of the $k-1$-dimensional sphere, that is the surface of the $k$ dimensional ball.


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.