I'd like to know to what point is it possible to generalize this method for obtaining the functional equation for the Dedekind zeta function $\zeta_K(s)$ of a number field ?

Let $\mathfrak{C}$ be the ideal class group $$\zeta_K(s) =\sum_{I \subset \mathcal{O}_K} N(I)^{-s} = \sum_{C_j \in \mathfrak{C}} \ \ \sum_{I \subset \mathcal{O}_K,I \,\sim \, C_j} N(I)^{-s}$$ Assuming $\mathcal{O}_K^\times$ is finite, each ideal $C_j \in \mathfrak{C}$ being a rank $n=[K:\mathbb{Q}]$ free $\mathbb{Z}$-module with basis $b_{j,1}, \ldots, b_{j,n}$, letting $C_j C_j^{-1} = (d_j)$ : $$\zeta_{K,C_j}(s)=\sum_{I \subset \mathcal{O}_K,I \,\sim \, C_j} N(I)^{-s} = \frac{1}{|O_K^\times|} \sum_{a \in C_j^{-1}} N(\frac{a}{d_j} C_j)^{-s}$$ $$=\frac{1}{|O_K^\times|}\frac{N(C_j)^{-s}}{N(d_j)^{-s}}\sum_{m \in \mathbb{Z}^n \setminus (0)} N(\sum_{l} m_l b_{j,l})^{-s}$$ $$\Gamma(s) \zeta_{K,C_j}(s) = \frac{1}{|O_K^\times|} \frac{N(C_j)^{-s}}{N(d_j)^{-s}}\int_0^\infty x^{s-1} \sum_{m \in \mathbb{Z}^n \setminus (0)} e^{-x\ N(\sum_{l} m_l b_{j,l})}dx$$ Therefore it reduces to finding a Poisson summation formula for $\displaystyle\Theta_j(x) = \sum_{m \in \mathbb{Z}^n} e^{-x \, N(\sum_{l} m_l b_{j,l})}$.

  • I did it for $\mathbb{Q}(\sqrt{-5})$ whose ideal class group has two elements, obtaining that $$\Lambda(s)= 5^{s/2}\pi^{-s}\Gamma(s)4 \zeta_{\mathbb{Q}(\sqrt{-5})}(s)=\Lambda(1-s) $$ I guess this method works at least for any imaginary quadratic field ?

  • If $K$ is imaginary quadratic, $\mathcal{O}_K = \mathbb{Z}[w]$ and any ideal has the form $(k,l+ow)$ then due to the property of $2\times 2$ matrices (see how $\scriptstyle\begin{pmatrix} 2 & 1 \\ 0 & \sqrt{5}\end{pmatrix}$ is treated in the link), I'll get a functional equation $\Lambda_{K,C_j}(s) = \lambda_j^{s/2} \pi^{-s}\Gamma(s)\zeta_{K,C_k}(s)=\Lambda_{K,C_j}(1-s)$, but I don't know how to show $\lambda_j $ is the same for each ideal class $C_j$.

  • What about the other cases ($\mathcal{O}_K^\times$ infinite, cubic extension, non-monogenic field...) ?

  • 2
    $\begingroup$ Didn't Hecke do this for any number field? Isn't it in his book Algebraic Number Theory? $\endgroup$ – Stopple Apr 4 '17 at 16:03
  • $\begingroup$ @Stopple : He shows the Poisson summation formula of the theta function of an ideal in the Minkowski space p.215 eq.(180) and (185) $\theta(x) = \sum_{a \in I} e^{-\pi (\sum_{m=1}^n x_m |\sigma_m(a)|^2)}$ where $\{\sigma_m\}_{m=1}^n$ is the Galois group. But he doesn't mention the functional equation for $\zeta_K(s)$ $\endgroup$ – reuns Apr 4 '17 at 17:11
  • 1
    $\begingroup$ I'm writing exactly this proof up for the notes in my ongoing algebraic number theory course and will post a link when done. $\endgroup$ – Lior Silberman Apr 4 '17 at 21:33
  • $\begingroup$ There are also versions in Lang's and Neukirch's ANT books and I think also in Weil's "basic number theory". $\endgroup$ – Lior Silberman Apr 4 '17 at 21:40
  • $\begingroup$ @LiorSilberman Neukrich does it p.218, but his notation is horrible. $\endgroup$ – reuns Apr 4 '17 at 21:56

In any event, this is essentially Hecke's original argument, the only other truly independent known proof being that of Tate. The original paper (in german) is, I believe,

  • Erich Hecke "Über die Zetafunktion beliebiger algebraischer Zahlkörper" (1917)

The method works for arbitrary number fields and, with some extra work, for any twist by a Hecke character.

English expositions of the proof can be found at least in Neukirch and Lang's textbooks.

| cite | improve this answer | |
  • $\begingroup$ In fact, Hecke's paper is clearer than the account in Lang, I think. $\endgroup$ – paul garrett Apr 8 '17 at 13:09

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.