The Dedekind zeta function for a number field $K$ is defined as $\zeta_K(s)=\sum_{I\subset O_K} (N_{K/\mathbb{Q}}(I))^{-s}$.

By attaching a Hecke character $\psi$, we can define $L(s,\psi)=\sum_{I\subset O_K} \psi(I)(N_{K/\mathbb{Q}}(I))^{-s}$.

Is there any way to calculate the special value of such $L(s,\psi)$ at $1$ when $\psi$ is nontrivial? Trying to do it numerically in Sage but do not have any idea where to start.

I am particularly interested in cases when $K$ is totally real, $\psi$ is a product of sign functions of the embeddings. Two examples:

1)$K=\mathbb{Q}(\sqrt{3})$ and $\psi(x)=sgn(Norm(x))$

2)$K=\mathbb{Q}(\alpha)$ the cubic extension given by $\alpha^3 + 2\alpha^2 - 3\alpha - 1=0$. $\alpha$ has 3 real embeddings, two negative ones called $\sigma_1$ and $\sigma_2$ and one positive $\sigma_3$. Now let $\psi(x)=sgn(\sigma_1(x)\sigma_2(x))$ (One can verify it is well-defined on the class group b/c it is 1 on the units)

  • $\begingroup$ By "modified Dedekind zeta-function" do you mean a Hecke $L$-function, i.e., is $\psi$ a Hecke character or do you have something else in mind? $\endgroup$ – KConrad Dec 11 '14 at 5:57
  • $\begingroup$ @KConrad Yes, I should say Hecke L-function and $\psi$ is a Hecke character. Modifying my original thread. $\endgroup$ – Ted Mao Dec 11 '14 at 6:04
  • $\begingroup$ I don't think Hecke characters are in Sage to any great degree (yet). Watkins has a paper where he gives the algorithms for this, and there is a Magma implementation. He also mentions a Pari package of Radziejewski for Hecke characters. see magma.maths.usyd.edu.au/~watkins/papers/hecke.pdf $\endgroup$ – NAME_IN_CAPS Dec 11 '14 at 6:38

I think the following Magma code gives you what you want:

> K:=QuadraticField(3);
> G:=HeckeCharacterGroup(1*Integers(K),[1,2]);
> psi:=G.1;
> L:=LSeries(psi);
> Evaluate(L,1);

Note that in fact $L(\psi)=L(\chi_{-3})\cdot L(\chi_{-4})$.

> K:=NumberField(Polynomial([-1,-3,2,1]));
> [Sign(Real(e)) : e in Conjugates(K.1)]; // [1,-1,-1]
> G:=HeckeCharacterGroup(1*Integers(K),[2,3]);
> psi:=G.1;
> L:=LSeries(psi);
> Evaluate(L,1);

The field is nonabelian here, but you still might be able to write $L(\psi)$ in terms of Artin representations (which are 1-dim, thus Dirichlet chars, in the previous case). Indeed, the Conductor of $L(\psi)$ is 257, and it is valuable to search the LMFDB for dimension 3 Artin representations of this conductor.


The LMFDB searching mechanism lists two Artin representations of this dimension and conductor, but they have the same $L$-function (compare the computed zeros).

| cite | improve this answer | |
  • $\begingroup$ Thank you for the reference. The codes are self-explanatory and very helpful. $\endgroup$ – Ted Mao Dec 11 '14 at 7:27

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.