Skip to main content

Section 3 Tate's Thesis (\(\GL_1\) theory)

What's the aim?

Redo Hecke's work, in an adelic setting, more canonically. I.e. obtain analytic continuation and functional equation of Hecke \(L\)-functions. This was also seemingly a bit more general than the Hecke theory. The local theory will give us terms like

\begin{equation*} (1 - \frac{\chi(p)}{p^r})\inv \end{equation*}

and the global theory will give

\begin{equation*} \prod_p (1 - \frac{\chi(p)}{p^r})\inv\text{.} \end{equation*}

Subsection 3.1 Local theory

For all of today, \(K\) is a number field and \(v\) a place of \(K\) so that

\begin{equation*} K_v = \begin{cases} \RR,\CC \amp \text{ if } v \text{ is archimidean} \\ [K_v: \QQ_v] \lt \infty \amp \text{ if } v \text{ is non-archimidean}\end{cases}\text{.} \end{equation*}

If \(\alpha \in K_v\) then

\begin{equation*} |\alpha|_v = \begin{cases} |\alpha|_\CC^2 \amp \text{ if } K_v = \CC \\ |\alpha|_\RR \amp \text{ if } K_v = \RR \\ N(\varpi_v)^{-v_v(\alpha)} \amp \text{ if } v \text{ is non-archimidean}\end{cases} \end{equation*}

where \(|N(\varpi_v) = | \ints_v / \varpi \ints_v|\text{.}\)

Subsubsection 3.1.1 Additive theory (Theory of Fourier transform)

Let \(K_v^+\) denote the additive group of \(K_v\text{.}\)

Characters of \(K_v^+\) (continuous)

Exercise.

This is not really canonical, but as far as number theorists are concerned there is a right choice of additive character to fix.

A particular non-trivial additive character

If \(K_v= \RR,\CC\) let

\begin{equation*} \lambda_0 \colon \RR \to \RR \end{equation*}
\begin{equation*} \xi \mapsto -\xi \pmod 1 \end{equation*}

then

\begin{equation*} \chi_v(\xi) = e(\lambda(\xi)) = e^{2\pi i \lambda(\xi)} \end{equation*}

where

\begin{equation*} \lambda(\xi) = \lambda_0(\tr_{K_v/\RR}(\xi))\text{.} \end{equation*}

For \(K_v\) non-archimidean let

\begin{equation*} \lambda_0 \colon \QQ_v \to \RR \end{equation*}
\begin{equation*} \xi \mapsto \xi \pmod 1 \end{equation*}

i.e. if

\begin{equation*} \xi = a_{-N} \varpi_v^{-N} + \cdots + a_{-1}\varpi_v^{-1} \underbrace{+ \cdots}_{\in \ints_v} \end{equation*}
\begin{equation*} \lambda_0 (\xi) = a_{-N} \varpi_v^{-N} + \cdots + a_{-1}\varpi_v^{-1} \end{equation*}
\begin{equation*} \chi_v(\xi) = e(\lambda(\xi)) \end{equation*}

where

\begin{equation*} \lambda(\xi) = \lambda_0(\tr_{K_v/\QQ_v}(\xi))\text{.} \end{equation*}
Haar measures
  • \(\CC\text{,}\) \(\mu_v\) is 2 times the Lebesgue measure.
  • \(\RR\text{,}\) \(\mu_v\) is the Lebesgue measure.
  • \(K_v\text{,}\) \(\mu_v(\ints_v) = N(D_v)^{-1/2}\text{.}\)

Exercise

I.e.

\begin{equation*} \int f(x) \diff \mu(x) = |\alpha|_v \int f(\alpha x) \diff \mu(x)\text{.} \end{equation*}
Definition 3.4 Fourier Transform
\begin{equation*} \hat f(\eta) = \int f(\xi) e(-\lambda_v(\eta\xi)) \diff \mu_v(\xi)\text{.} \end{equation*}

Subsubsection 3.1.2 Multiplicative theory

Definition 3.6 Unit group
\begin{equation*} U_v = \ker(x\mapsto |x|_v) \end{equation*}

it is compact, open if \(v\) is non-archimidean, e.g.

\begin{equation*} U_v = \begin{cases}S^1 \amp \text{ if }K_v = \CC \\S^1 \amp \text{ if }K_v = \RR \\ \ints_v^\times \amp\text{ if } v\text{ non-archimidean}\end{cases}\text{.} \end{equation*}
Definition 3.7

A quasicharacter is

\begin{equation*} K_v^\times \to \CC^\times \end{equation*}

A (unitary) character is

\begin{equation*} K_v^\times \to S^1\text{.} \end{equation*}

Such a map is unramified if it is trivial on \(U_v\text{.}\) E.g.

\begin{equation*} \xi \mapsto |\xi|_v^s \end{equation*}

is unramified \(s\in \CC\text{.}\)

Choose a uniformiser \(\varpi_v\) s.t.

\begin{equation*} \begin{array}{c|c|c} \alpha \amp \tilde \alpha \amp \varpi_v^{v_v(\alpha)} \\ \hline \CC^\times \amp S^1 \amp \RR_+ \\ \RR^\times \amp \{\pm 1\}\amp \RR_+ \\ K_v^\times \amp \ints_v^\times \amp \ZZ\end{array}\text{.} \end{equation*}

Exercise.

Example 3.10

Dirichlet characters mod \(p\text{.}\)

E.g. over \(\CC^\times\) then

\begin{equation*} \tilde c(\alpha) = \left(\frac{\alpha}{|\alpha|}\right)^m,\,m\in \ZZ \end{equation*}

over \(\RR^\times\) then

\begin{equation*} \tilde c(\alpha) = \left(\frac{\alpha}{|\alpha|}\right)^m,\,m\in \{0,1\} \end{equation*}

over \(\ints_v^\times\) then

\begin{equation*} \tilde c(\alpha)|_{1+\varpi^k\ints_v} \equiv 1 \end{equation*}

because \(c\) is a continuous map from a \(p\)-adic field to the complex numbers. Let

\begin{equation*} \tilde k = \min\{k\in \NN : \tilde c(\alpha)|_{1+ \varpi_v^k \ints_v} \equiv 1\} \end{equation*}

and \(\varpi_v^{\tilde k} = f_v\) is the conductor of \(\tilde c\text{.}\)

Note 3.11

Sometimes this \(\tilde k\) is called the ramification degree of \(c\text{.}\)

Note 3.12

If \(c = \tilde c | \cdot |^s\) then \(\Re(s) = \sigma\) is uniquely determined by \(c\text{.}\) It is called the exponent of \(c\text{.}\) In modern lingo this is measuring how non-tempered \(c\) is.

Multiplicative Haar measures

Let

\begin{equation*} \diff_v^\times \alpha = \begin{cases} \frac{\diff_v \alpha}{|\alpha|_v} \amp \text{ if } v \text{ archimidean}\\ \left(\frac{1}{1-1/N(\varpi_v)}\right) \frac{\diff_v \alpha}{|\alpha|_v} \amp \text{ if } v \text{ non-archimidean}\end{cases} \end{equation*}

where \(\diff_v \alpha\) is the additive Haar measure. these extra factors are really Tamagawa numbers. They make the product in the next lemma converge.

\begin{equation*} \int_{\ints_v^\times} \diff_v^\times \alpha= \int_{\ints_v^\times} \frac{\diff_v \alpha}{|\alpha|_v} (1- N(\varpi_v)\inv)\inv \end{equation*}
\begin{equation*} = \int_{\ints_v^\times} \diff_v \alpha (1- N(\varpi_v)\inv)\inv \end{equation*}
\begin{equation*} = \sum_{\beta \in (\ints_v/\varpi_v\ints_v)^\times} \int _{\beta + \varpi_v\ints_v} \diff_v \alpha(1- N(\varpi_v)\inv)\inv \end{equation*}
\begin{equation*} = |\varpi_v|_v \sum_{\beta \in (\ints_v/\varpi_v\ints_v)^\times} \int _{\ints_v} \diff_v \alpha(1- N(\varpi_v)\inv)\inv \end{equation*}
\begin{equation*} = N(D_v)^{-1/2} |\varpi_v| (N(\varpi_v) - 1) (1- N(\varpi_v)\inv)\inv = N(D_v)^{-1/2}\text{.} \end{equation*}

We are trying to set up a general machinery that will take a quasicharacter and associate a zeta function. In fact we want have

\begin{equation*} c\colon N_v^\times \to \CC^\times \leadsto \zeta(f,c) \end{equation*}

a family of \(\zeta\)-functions. We will then look at the gcd over all possible \(f\text{,}\) this will be the \(L\)-factor.

Subsubsection 3.1.3 Local \(\zeta\)-functions

Let

\begin{equation*} f \colon K_v^+ \to \CC \end{equation*}
\begin{equation*} \xi \mapsto f(\xi) \end{equation*}

restrict to

\begin{equation*} f \colon K_v\units \to \CC \end{equation*}
\begin{equation*} \alpha \mapsto f(\alpha) \end{equation*}

and such that

  1. \(f(\xi), \hat f(\xi) \in L^1(K^+)\) are continuous.
  2. \(f(\alpha) |\alpha|^\sigma\) and \(\hat f (\alpha) |\alpha|^\sigma\in L^1(K\units)\) for all \(\alpha \gt 0\text{.}\)

Call the class of such \(f\) \(S\text{.}\)

Definition 3.14

Let \(f \in S\) and \(c\) s.t. exponent of \(c\) is \(\sigma \gt 0\text{,}\) i.e. \(c= c_0 \cdot |\cdot|^s\) with \(c_0\) unitary and \(\Re(s) = \sigma \gt 0\) then we may define

\begin{equation*} \zeta(f,c) = \int_{K_v\units} f(\alpha) c(\alpha) \diff\units \alpha\text{.} \end{equation*}

In fact we have seen examples of this, for \(K_v = \RR\text{,}\) the \(\Gamma\) function, in Tate's language this is a local zeta function.

\begin{equation*} \zeta(f, |\cdot |^s) = \int_{\RR\units} e^{-|x|} |x|^s \frac{\diff x}{|x|} = 2 \Gamma(s) \end{equation*}
Remark 3.15

\(c = c_0 | \cdot |^s\) for fixed \(f\) and \(c_0\)

\begin{equation*} \implies \zeta(f,c) = \int_{K_v\units} f(\alpha) c_0(\alpha) |\alpha|_v^s\diff \units \alpha\text{.} \end{equation*}

Aim:

  1. Prove that \(\zeta(f,s)\) extends to a meromorphic function of \(s\text{.}\)
  2. Calculate this for “nice” choices of \(f\text{.}\)

Subsubsection 3.1.4 Analytic properties of \(\zeta(f,c)\)

\begin{equation*} \zeta(f,c) = \int_{K_v\units} f(\alpha) c(\alpha) \diff \units \alpha \end{equation*}

check, absolutely convergent around 0, has derivatives around 0, other points are fine.

All of the integrals are absolutely convergent.

Fubini implies the LHS is

\begin{equation*} \int\int_{(K_v\units)^2} f(\alpha) \hat g(\beta) c(\alpha) c_0\inv(\beta) |\beta|^{1-s} \diff \units \mu(\alpha, \beta) \end{equation*}
\begin{equation*} = \vdots \end{equation*}
\begin{equation*} = \zeta(\hat f, \hat c) \zeta(g ,c) \end{equation*}

change variable \(\beta \mapsto \alpha\beta\text{.}\)

Take \(f= g\) so

\begin{equation*} \rho(C) = \frac{\zeta(f,c)}{\zeta(\hat f, \hat c)} \end{equation*}

we will show \(\rho(c)\) is independent of \(f\) and that we can choose \(f\) s.t. \(\zeta(\hat f, \hat c)\) is non-zero. TBC.

\begin{equation*} \zeta(f,c) = \rho(c) \zeta(\hat f, \hat c) = \rho(c) \rho(\hat c) \zeta(\hat{\hat f}, \hat{\hat c}) \end{equation*}

as \(\hat{\hat c}\) and \(\hat{\hat f}(x) = f(-x)\) we get \(\rho(c) \rho(\hat c) c(-1) \zeta(f,c)\text{.}\)

\begin{equation*} \overline{\zeta(f,c)} = \zeta(\bar f, \bar c) = \rho(\bar c) \zeta(\hat{\bar f}, \hat{\bar c}) \end{equation*}
\begin{equation*} = \rho(\bar c) c(-1) \overline{\zeta(\hat f, \hat c)}\text{.} \end{equation*}

Remark:

\begin{equation*} \zeta(\hat{\bar f}, \hat {\bar c}) = \int \hat {\bar f}(\alpha) \hat {\bar c}(\alpha) \diff \units \alpha \end{equation*}
\begin{equation*} = \int\hat {\bar f}(\alpha) \bar c \inv(\alpha) |\alpha| \diff \units \alpha \end{equation*}
\begin{equation*} = \int \bar {\hat f}(-\alpha) \bar c\inv (\alpha) |\alpha| \diff \units \alpha \end{equation*}
\begin{equation*} = c(-1) \int \bar {\hat f}(\alpha) \bar c\inv(\alpha) |\alpha| \diff \units \alpha \end{equation*}
\begin{equation*} = c(-1)\overline{\zeta(\hat f, \hat c)} \end{equation*}

If \(\sigma= \frac12\) then \(c(\alpha) \bar c(\alpha) = |\alpha| = c(\alpha) \hat c(\alpha) = \bar c(\alpha) = \hat c(\alpha)\text{.}\) Together the previous parts give \(\rho(c)\overline {\rho(c)} = 1\text{.}\)

Subsubsection 3.1.5 Explicit \(\zeta\) functions

First let \(K_v = \RR\) then we will use the following notation: Additively \(\xi\) with \(\Lambda(\xi) = -\xi\text{,}\) \(\diff \xi\) the Lebesgue measure. Multiplicatively \(\alpha\) with \(|\alpha|_v = |\alpha|_\RR\) and \(\diff \units \alpha = \diff \alpha/|\alpha|_\RR\text{.}\) We will use characters \(|\cdot |^s\) or \(\mathrm{sgn}|\cdot|^s\text{,}\) \(f_{|\cdot|^s}= e^{-\pi \xi^2}\) and \(f_{|\cdot|^s\mathrm{sgn}} (\xi) = \xi e^{-\pi \xi^2}\text{.}\) These have fourier transforms

\begin{equation*} \hat f_{|\cdot|^s}(\xi) = f(\xi) \end{equation*}

and

\begin{equation*} \hat f_{|\cdot|^s\mathrm{sgn}}(\xi) = i f_\mathrm{sgn}(\xi)\text{.} \end{equation*}

So that

\begin{equation*} \zeta(f_{|\cdot|^s}, |\cdot|^s) = \int_{\RR\units} f(\alpha) |\alpha|^s \diff \units \alpha = \int_{\RR\units} e^{-\pi \alpha^2} | \alpha| ^s \diff\units \alpha \end{equation*}
\begin{equation*} = 2 \int_0^\infty e^{-\pi \alpha^2} \end{equation*}
\begin{equation*} = \pi^{-s/2} \Gamma\left(\frac s2\right)\text{.} \end{equation*}
\begin{equation*} \zeta(f_{|\cdot|^s\mathrm{sgn}}, |\cdot|^s\mathrm{sgn}) = \pi^{-(s+1)/2} \Gamma\left(\frac{s+1}{2}\right) \end{equation*}

now

\begin{equation*} \rho(|\cdot|^s) = 2^{1-s} \pi^{-s} \cos(\pi s /2) \Gamma(s) \end{equation*}
\begin{equation*} \rho(|\cdot|^s\mathrm{sgn}) = -i 2^{1-s} \pi^{-s} \sin(\pi s /2) \Gamma(s)\text{.} \end{equation*}

Normalizing by \(s= 1/2\) we get \(1 \) and \(- i\) respectively.

Over \(\CC\)

\begin{equation*} \begin{array}{cc} K_v^+ \amp K_v\units\\ \hline \xi = x+iy \amp \alpha = re^{i\theta}\\ \Lambda = -2x \amp |\alpha| = r^2\\ \diff \xi =2 |\diff x \diff y| \amp \diff\units \alpha =\diff \alpha /|\alpha|= 2|\diff r \diff \theta|/r\\ \end{array} \end{equation*}

characters

\begin{equation*} c \leadsto c_n,\,n\in\ZZ \end{equation*}
\begin{equation*} c_n(\alpha) = r^n e^{in \theta} \end{equation*}

equivalence class

\begin{equation*} \{c_n(\alpha) |\alpha|^s : s\in \CC \}\text{.} \end{equation*}

Functions

\begin{equation*} f_n(\xi) = \begin{cases} (x-iy)^n e^{-2\pi(x^2 + y^2)} \amp \text{ if } n \ge 0 \\ (x+iy)^{-n} e^{-2\pi(x^2 + y^2)} \amp \text{ if } n \le 0\end{cases} \end{equation*}
\begin{equation*} = \begin{cases} r^{|n|} e^{-in \theta} e^{-2\pi r^2} \amp\text{ if } n \ge 0 \\ r^{|n|} e^{in\theta} e^{-2\pi r^2}\amp \text{ if } n \le 0\end{cases} \end{equation*}

fourier transforms

Induction on \(n\text{:}\)

\(n = 0\)

\begin{equation*} f_0(\xi) = e^{-\pi (x^2 + y^2)} \end{equation*}
\begin{equation*} \hat f_0(\xi) = 2\int_{\CC} f_0(u+iv) \overbrace{e(2(ux-vy))}^{\Lambda((u+iv)(x+iy))} \diff u \diff v \end{equation*}
\begin{equation*} 2\int_{\RR} e^{-2\pi (u^2 - 2ixu)} \diff u \int_\RR e^{-2\pi(v^2 + 2ivy)} \diff v \end{equation*}

which by completing the square and Cauchy gives

\begin{equation*} e^{-2\pi(x^2 + y^2)} = \hat f_0(\xi) \end{equation*}

Assume \(\hat f_n(\xi) = i^n f_{-n}(\xi)\) for \(n \gt 0\text{.}\) Then the induction step is to use \(\diff / \diff \bar \xi\) in the integral defining \(\hat f_n(\xi)\) (exercise).

\begin{equation*} \zeta(f_n, c_n |\cdot|^s) = \int_{\CC\units} f_n(\alpha) c_n(\alpha) |\alpha|^s\diff\units\alpha \end{equation*}
\begin{equation*} = 2 \int |r|^n e^{-i n \theta} e^{-2\pi r^2} e^{in \theta} r^{2s-2} \end{equation*}
\begin{equation*} = 4\pi \int_0^\infty r^{|n| + 2s - 2} e^{-2 \pi r^2} r \diff r \end{equation*}
\begin{equation*} = (2\pi)^{1 - s - |n|/2} \Gamma\left(s+ \frac{|n|}{2}\right) \end{equation*}

now

\begin{equation*} \zeta(f_{-n}, c_{-n}|\cdot|^s) = i^{|n|} (2\pi)^{s- |n|/2} \Gamma\left(1 - s + \frac{|n|}{2}\right) \end{equation*}
\begin{equation*} \rho(c_n|\cdot |^s) = \frac{(-i)^{|n|} (2 \pi)^{k-s}}{(2\pi)^s} \frac{\Gamma(s + \frac{|n|}{2})}{\Gamma(1-s + \frac{|n|}{2})}\text{.} \end{equation*}

For \(K\) non-archimidean with \(|\varpi|_v= \frac 1q\text{.}\)

\begin{equation*} \begin{array}{cc} K^+ \amp K\units\\ \hline \xi \amp \alpha = \tilde \alpha \varpi^{v(\alpha)}\\ \Lambda (\xi) = \lambda(\tr_{K/\QQ_p}(\xi)) \amp |\alpha| = q_v^{-v(\alpha)}\\ \diff \xi \implies \int_\ints \diff \xi = N(D_v)^{-1/2} \amp \diff\units \alpha =(1-q_v\inv)\diff \alpha /|\alpha| \implies \int_{\ints\units} \diff\units \alpha = N(D_v)^{-1/2} \end{array} \end{equation*}

For \(D_v\) the different ideal.

Quasicharacters

\begin{equation*} c_n \colon \ints\units \to \CC\units \end{equation*}

of conductor \(f (\varpi^n)\ints_v\text{.}\) \(c_n(\varpi) = 1\text{.}\) Equivalence class of \(c_n\)

\begin{equation*} \{c_n |\cdot|^s : s\in \CC\}\text{.} \end{equation*}

Functions

\begin{equation*} f_n(\xi) = \begin{cases} e(\Lambda(\xi)) = e^{2\pi i \lambda(\tr(\xi))} \amp \text{ if } \xi\in D_v\inv \varpi^{-n} \\ 0 \amp\end{cases}\text{.} \end{equation*}

The fourier transforms

\begin{equation*} \hat f_n(\xi) = \int_K f_n(\eta) e(-\Lambda (\eta \xi)) \diff \eta \end{equation*}
\begin{equation*} =\int_{D_v\inv \varpi^{-n} \ints_v} e(\Lambda(\eta(1-\xi))) \diff \eta \end{equation*}
\begin{equation*} = \begin{cases} 0 \amp \text{ if } \xi \ne 1 \pmod{\varpi^n} \\ |D_v|\inv |\varpi|^{-n} \int_{\ints_v} \diff \eta \amp \end{cases} \end{equation*}
\begin{equation*} = \begin{cases} 0 \amp \text{ if } \xi \ne 1 \pmod{\varpi^n} \\ |D_v|^{-1/2} |\varpi|^{-n} \amp \end{cases} \end{equation*}

Unramified calculation:

\begin{equation*} \zeta(f_0, |\cdot |^s) = \int_{K\units} f_0(\alpha) |\alpha|^s \diff\units \alpha \end{equation*}
\begin{equation*} = \int_{D_v\inv} e(\Lambda(\alpha) ) |\alpha| ^s \diff\units \alpha \end{equation*}
\begin{equation*} = \int_{D_v\inv} |\alpha|^s \diff\units \alpha \end{equation*}
\begin{equation*} = \sum_{k=0}^\infty|D_v|^{-s} \int_{\varpi^n \ints\units}|\alpha|^s \diff\units \alpha \end{equation*}
\begin{equation*} \left(\ints = \bigsqcup_{n=0}^\infty \varpi^n \ints\units_v\right) \end{equation*}
\begin{equation*} = \sum_{k=0}^\infty |D_v|^{-s+1/2} q_v^{-ns} \end{equation*}
\begin{equation*} = |D_v|^{-s+1/2} \frac{1}{1-1/q_v^s} \end{equation*}
\begin{equation*} \zeta(\hat f_0, |\cdot |^{1-s}) = \zeta(|D_v|^{-1/2} \mathbf 1_{\ints_v}, |\cdot |^{1-s}) \end{equation*}
\begin{equation*} = N(D_v)^{1/2} \int_{\ints_v} |\alpha|^{1-s} \diff\units \alpha \end{equation*}
\begin{equation*} = \frac{1}{1-1/q_v^{1-s}} \end{equation*}

Let's recap a little, we have \(K\units \hookrightarrow K^+\) and moreover \(K\units \acts K^+\) with two orbits \(\{0\}\) and the rest. Looking at function spaces we have

\begin{equation*} \cinf_c(K\units) \hookrightarrow S(K^+)\text{,} \end{equation*}

Schwartz functions on the right. Taking duals to get spaces of distributions

\begin{equation*} 1 \to ?? \to D(K^+)_c \to \cinf_c (K\units)^\vee_c \to 0\text{.} \end{equation*}

We have done the case of unramified characters.

Case II: \(n \gt 0\)

\begin{equation*} \zeta(f_n, c_n|\cdot|^s) = \int_{K_v\units} f_n(\alpha) c_n(\alpha) |\alpha|^s \diff\units \alpha \end{equation*}
\begin{equation*} = \int_{(D_vf_v) \inv} e(\Lambda(\alpha))(c(\alpha)) |\alpha|_v^s \diff \units \alpha\text{.} \end{equation*}

Let

\begin{equation*} (D_v) = (\varpi_v^d) \end{equation*}
\begin{equation*} f= (\varpi_v^n) \end{equation*}
\begin{equation*} = \int_{(\varpi_v^{n+d})\inv} e(\Lambda(\alpha)) c_n(\alpha) |\alpha|^s \diff \units \alpha \end{equation*}
\begin{equation*} = \sum_{j = -n-d}^\infty \int_{\varpi_v^j \ints_v\units } e(\Lambda(\alpha)) c_n(\alpha) |\alpha|^s \diff \units \alpha \end{equation*}

let \(\alpha = \varpi_v^j \alpha\) which implies

\begin{equation*} = \sum_{j = -n-d}^\infty q_v^{js}\int_{\ints_v\units } e(\Lambda(\varpi^j\alpha)) c_n(\alpha) |\alpha|^s \diff \units \alpha\text{.} \end{equation*}

\(j \ge -d\) implies

\begin{equation*} \varpi_v^j \ints_v\units \subseteq D_v\inv \end{equation*}
\begin{equation*} \implies \Lambda(\varpi_v^j \ints_v\units) \subseteq\ZZ \end{equation*}
\begin{equation*} \implies e(\Lambda(\varpi_v^j \ints_v\units)) = 1 \end{equation*}

so the integral in question

\begin{equation*} = \int_{\ints_v\units} c_n(\alpha) \diff\units \alpha \end{equation*}

as \(c_n \ne 1\text{.}\)

So finitely many are non-zero.

(Note \(-n-d \lt j \lt -d \iff 0 \lt -j-d \lt n\)) Let \(\alpha = a (1 + \varpi_v^{-d-j} \alpha_1)\text{.}\) So we may write

\begin{equation*} \sum_{a \mod{\units \varpi_v^{d+j}}} \int_{\ints_v} e(\Lambda(\varpi_v^ja (1+ \varpi_v^{-d-j} \alpha_1))) c_n(a (1+ \varpi_v^{-d-j} \alpha_1)) \diff \units \alpha\text{.} \end{equation*}

(Note: \(\varpi_v^j (a(1+ \varpi_v^{-j-d} \alpha_1)) = a\varpi_v^j + a \varpi_v^{-d} \alpha_1\text{,}\) the last term is in \(D_v\inv\) so \(e(\Lambda(\varpi_v^ja (1+ \varpi_v^{-d-j} \alpha_1))) = e(\Lambda(\varpi_v^ja))\)) So the integral is

\begin{equation*} \sum_{a \mod{\units \varpi_v^{d+j}}} c_n(a) e(\Lambda(\varpi_v^ja)) \int_{\ints_v} c_n(1+ \varpi_v^{-d-j} \alpha_1) \diff \units \alpha = 0 \end{equation*}

as we are integrating over a multiplicative group.

Therefore: In the ramified case we have

\begin{equation*} \zeta(f_b, |\cdot |^s c_n) = e_v^{(n+d) s} \int_{\ints_v\units} e(\Lambda(\varpi_v^{-(n+d)} \alpha)) c_n(\alpha) \diff \units \alpha \end{equation*}
\begin{equation*} = q_v^{(n+d)s} \sum_{a \mod{\units \varpi_v^{ n }}} e(\Lambda(\varpi_v^{-(n+d)} a)) c_n(a) \int_{1 + \varpi_v^n \ints_v} \diff \units \alpha\text{.} \end{equation*}
\begin{equation*} N(D_v f_v)^s A_n G(c_n) \end{equation*}

where

\begin{equation*} G(c_n) = \sum_{a \mod{\units \varpi_v^{ n }}} e(\Lambda(a\varpi^{-d-n})) c_n(a) \end{equation*}

is a Gauss-sum for \(c_n\text{,}\) with \(|G(c_n)| = N(f)^{1/2}\text{.}\)

Exercise.

This finishes the proof of the local functional equation.

\begin{equation*} \zeta(f,c) = \int_{K_v\units} f(\alpha) c(\alpha)\diff \units \alpha\text{.} \end{equation*}

A little bit of a more modern approach that may be helpful if you want to read things since Tate.

\begin{equation*} \psi_v \text{ additive character} \end{equation*}
\begin{equation*} \mu \text{ additive measure, self-dual w.r.t } \psi_v \end{equation*}
Definition 3.26

A family of \(\zeta\)-integrals for each \(|\cdot |^s \chi_v\)

\begin{equation*} \chi_v \colon K_v\units \to \CC \units\text{.} \end{equation*}
\begin{equation*} Z(s,\chi_v, f_v, \psi_v) = \int_{K_v\units} f(\alpha) \chi_v(\alpha) |\alpha|^s \diff \units \alpha \end{equation*}
\begin{equation*} Z \colon \mathcal S(K_v\units) \to \CC\text{.} \end{equation*}
Remark 3.27

The interesting part of this integral is when \(\alpha\) is close to 0.

To study this around 0

\begin{equation*} f(\alpha) = a\phi_1( \alpha) + \phi_2(\alpha) \end{equation*}

where \(\phi_1\) is the characteristic function of a very small (depending on \(f\)) neighbourhood of 0. \(\phi_2\) is 0 around a sufficiently small neighbourhood of 0.

\begin{equation*} Z(s, \chi_v, \phi_2, \psi_v) = \int_{K_v\units} \phi_2(\alpha) \chi_v(\alpha) |\alpha|_v^s \diff\units \alpha \end{equation*}
\begin{equation*} = \sum_{j = -N_1}^{N_2} c_j \int_{\varpi_v^j \ints_v\units} \chi_v(\alpha) |\alpha|^s \diff\units \alpha \end{equation*}

This converges for all \(s \in \CC\) and lands in \(\CC\lb q_v^s, q_v^{-1}\rb\) .

\begin{equation*} Z(s, \chi_v, \phi_1, \psi_v) = \sum_{j = M }^\infty \int_{\varpi_v^j \ints_v\units} \chi_v(\alpha) |\alpha|^s \diff\units \alpha \end{equation*}
\begin{equation*} = \sum_{j = M }^\infty | \varpi_v^j|^s \int_{\ints_v\units} \chi_v(\varpi_v^j\alpha) \diff\units \alpha \end{equation*}

this is a multiple of

\begin{equation*} \begin{cases} 0 \amp \text{ if } \chi_v \text{ is ram.} \\ \frac{1}{1 - 1 /q_v^s} \amp \text{ if } \chi_v \text{ unram.}\end{cases}\text{.} \end{equation*}

Usually these observations are packed into saying

\begin{equation*} Z(s,\chi_v, f) \end{equation*}

is

  1. Rational function of \(q_v^s\) and in \(\CC\lb q_v^s, q_v^{-s}\rb\text{.}\)
  2. \(\chi_v\) is unramified if its entire.
  3. There exists \(f\in \mathcal S(K)\) s.t.
    \begin{equation*} Z(s,\chi, f) = L(s, \chi)\text{.} \end{equation*}
  4. For all \(f\)
    \begin{equation*} \frac{Z(s, \chi, f)}{L(s, \chi, f)} \end{equation*}
    is entire.

It is then said that \(L(s,\chi)\) is the GCD of the \(\zeta\) integrals \(Z(s, \chi, f)\text{.}\)

\begin{equation*} \begin{array}{c|c|c} \text{Field} \amp \chi_v \amp L_v(s, \chi_v)\\ \hline \CC \amp |\cdot|^s_\CC\chi_n \amp 2(2\pi)^{-(s+ |n|/2)} \Gamma(s + |n|/2)\\ \RR \amp |\cdot|^s_\RR \amp (\pi)^{-s/2)} \Gamma(s/2)\\ \RR \amp |\cdot|^s_\RR\mathrm{sgn} \amp \pi^{-(s+1)/2)} \Gamma((s+1)/2)\\ K_v \amp |\cdot|^s \chi_v \text{ ur} \amp (1- \chi_v( \varpi_v)/q_v^s)\inv\\ K_v \amp |\cdot|^s \chi_v \text{ ram}\amp 1\\ \end{array} \end{equation*}

Subsubsection 3.1.6 \(\epsilon\)-factors

Last time we looked at \(L\)-factors,

\begin{equation*} Z(s,f,\chi) \end{equation*}

for \(\chi\) a multiplicative character and \(f \in \mathcal S (K_v)\text{.}\)

Let \(\psi_v\colon K_v \to \CC\) be an additive character.

Definition 3.28 \(\epsilon\)-factors

The \(\epsilon\)-factor is defined to be

\begin{equation*} \frac{Z(1-s, \hat f , \chi \inv)}{L(1-s, \chi \inv)} = \epsilon(s,\chi, \psi_v) \frac{Z(s,f,\chi)}{L(s,\chi)}\text{.} \end{equation*}
Note 3.29

This is related to \(\rho(c) \inv\) in Tate's notation.

Tate's normalization of the additive character \(\psi_v\) is such that \(\psi_v( \cdots) = \Lambda(\cdots)\text{.}\)

Definition 3.30

The root number is

\begin{equation*} \frac{\epsilon(s,\chi, \psi)}{|\epsilon(s,\chi, \psi)|}\text{.} \end{equation*}

It is a complex number of norm 1.

\begin{equation*} \begin{array}{c|c|c|c|c} \text{Field} \amp \chi_v \amp \epsilon\text{-factor} \amp \text{Root number} \amp \text{Modern normalisation} \\ \hline \CC \amp |\cdot|^s_\CC\chi_n \amp i^{|n|} \amp i^{|n|} \amp i^{-|n|} \\ \RR \amp |\cdot|^s_\RR \amp 1 \amp 1 \amp 1 \\ \RR \amp |\cdot|^s_\RR\mathrm{sgn} \amp i \amp i \amp -i \\ K_v \amp |\cdot|^s \amp N(D_v)^{\frac12-\epsilon}\amp 1 \amp 1 \\ K_v \amp |\cdot|^s \chi_v \amp N(D_v)^{\frac12-\epsilon}/(G(\chi)/N(f)^{\frac12})\amp 1/(G(\chi)/N(f)^{\frac12})\amp (G(\chi)/N(f)^{\frac12})\\ \end{array} \end{equation*}

Subsection 3.2 Global theory

Subsubsection 3.2.1 Adeles, Ideles

First, some notation (restricted direct product)

\begin{equation*} \prod_v'(F_v, H_v) = \left\{ (x_v) \in \prod_v F_v: x_v \in H_v \text{ for all but finitely many } v\right\}\text{.} \end{equation*}

Then

\begin{equation*} \adeles_K = \prod_v'(K_v, \ints_v) \end{equation*}
\begin{equation*} \adeles_K\units = \prod_v'(K_v\units, \ints_v\units)\text{.} \end{equation*}

Topologies:

\begin{equation*} \adeles_{K,S} = \prod_{v\in S} K_v \times \prod_{v\not \in S}\ints_v \end{equation*}
\begin{equation*} \adeles_{K,S}\units = \prod_{v\in S} K_v\units \times \prod_{v\not \in S}\ints_v\units \end{equation*}

for \(S\) any finite set of places containing \(v | \infty\text{.}\)

Characters

Let \(U \subseteq \CC^\times\) be a neighbourhood of \(1 \) that does not contain any subgroups except for \(\{1\}\text{,}\) Let \(N \subseteq \adeles_K\) s.t. \(\psi(N)\subseteq U\text{.}\) Let \(S\) be a the set of places for which \(N_v \ne \ints_v\text{.}\) Consider

\begin{equation*} \adeles_K^S = \prod_{v\in S}1 \times \prod_{v\not \in S} \ints_v \subseteq N \end{equation*}

so that

\begin{equation*} \psi(\adeles_K^S) = 1 \end{equation*}

since the image is a subgroup \(\subseteq U\text{.}\)

From now on we will simply write \(\adeles = \adeles_K\text{.}\)

Completely analogous to \(\hat K_v \simeq K_v\) via \(\eta \mapsto \Lambda(\eta \cdot)\text{,}\) for any non-trivial \(\Lambda\text{.}\) Fix one nontrivial \(\Lambda \colon \adeles \to \CC\units\text{.}\) Rest is just \(\Lambda(\eta \cdot)\) for \(\eta \in \adeles\text{.}\)

Measures: \(\diff \xi = \prod_v \diff \xi_v\) where for almost all \(v\) we have \(\int_{\ints_v} \diff \xi_v = 1\text{.}\)

Dual measures: Normalise so that Fourier inversion holds, i.e.

\begin{equation*} \hat f(\eta) = \int f(\xi) \psi(\xi \eta) \diff \xi \end{equation*}
\begin{equation*} \hat{\hat f}(\eta) = f(- \xi) \end{equation*}

so we get \(\mu^* (\ints_v^\perp) = 1\) where

\begin{equation*} \ints_v^\perp = \{\psi_v : \psi_v|_{\ints_v} = 1\}\text{.} \end{equation*}
\begin{equation*} \mu(\alpha S) = |\alpha|_{\adeles} \mu(S) \end{equation*}

where \(|\alpha|_\adeles = \prod |\alpha|_v\text{,}\) i.e.

\begin{equation*} \int_{\alpha S} \diff \mu(x) = |\alpha| \int_S \diff \mu(x)\text{.} \end{equation*}
  1. \(\ints_K\) is a full rank lattice in \(\prod_{v|\infty} K_v = V\) (Note for \(\lb K: \QQ \rb = N\) then \(V \simeq \RR^{r_1} \times \CC^{r_2} \simeq \RR^N\)).
  2. Obvious
  3. Partial fractions.
  4. \begin{equation*} \mu(K\backslash \adeles) = \mu (K\backslash V) \mu(\pi\ints_v) = 1\text{.} \end{equation*}
  5. Since \(\adeles_K / K\) is compact, \(\widehat{\adeles_K/K}\) is discrete. Further \(K \subseteq K^\perp\text{.}\) In fact \(K^\perp /K\) is a finite vector space since
    \begin{equation*} K^\perp/K \hookrightarrow \widehat \adeles_K/K \simeq \adeles/K \end{equation*}
    which is compact, hence \(K^\perp = K\text{.}\)
Poisson summation

Let \(f, \hat f \in L^1(\adeles)\text{,}\) then \(\forall \alpha \in \adeles\units\text{,}\)

\begin{equation*} \frac{1}{|\alpha|} \sum_{\xi\in K} f\left(\frac\xi\alpha\right) = \sum_{\xi\in K} \hat f(\alpha \xi)\text{.} \end{equation*}

Tate refers to this as “Riemann-Roch”. This is due to the case where \(K = \FF_q(T)\) and \(K(\PP^1)\text{,}\) \(f = \chi_{\ints_K}\text{.}\) There is some Serre duality involved.

Ideles

Our goal is now to understand \(K\units \hookrightarrow \adeles\units_K\text{.}\)

Exercise.

Unfortunately \(K\units\backslash \adeles\units_K\) is not compact.

A non-canonical decomposition is as follows:

Fix \(v_0 \in S_\infty\) and let

\begin{equation*} \iota\colon\RR_{\gt 0} \hookrightarrow \adeles_K \end{equation*}

via

\begin{equation*} \begin{cases} (\alpha, 1,\ldots, 1) \amp K_{v_0} = \RR \\ (\sqrt\alpha, 1,\ldots, 1) \amp K_{v_0} = \CC\end{cases} \end{equation*}

Note

\begin{equation*} | \iota(x)|_{\adeles_K} = |x| = x\text{.} \end{equation*}

If \(\alpha \in \adeles\units\) then

\begin{equation*} \alpha = \underbrace{\frac{\alpha}{|\alpha|_\adeles}}_b \underbrace{|\alpha|_\adeles}_t \end{equation*}

with \(b \in \adeles^{\times,1},t\in \RR_{\gt 0}\text{.}\) Where \(\adeles^{\times,1} = \ker(|\cdot|_\adeles)\text{.}\) Then

\begin{equation*} \diff\units \alpha = \diff\units t \diff\units b \end{equation*}

so that

\begin{equation*} \int_{\adeles\units} f(\alpha) \diff\units \alpha = \int_0^\infty \int_{\adeles^{\times,1}} f(tb) \frac{\diff t}{t} \diff b\text{.} \end{equation*}
Multiplicative world

Let

\begin{equation*} \adeles^{\times, 1} = \ker(|\cdot|_\adeles \colon \adeles \units \to \CC\units) \end{equation*}
  1. \(K\units \backslash \adeles^{\times, 1}\) is compact, which leads to Dirichlet's units theorem.
  2. \(\mu(K\units \backslash \adeles^{\times, 1})\text{.}\)
  3. Analytic continuation and functional equation.
  4. Examples
1

Recall for a fixed archimidean place \(v_0\)

\begin{equation*} S_\infty = \{v|\infty\}, S_\infty' = S_\infty \smallsetminus \{v_0\} \end{equation*}

for \(v\) complex we take pairs as 1, so that \(\#S_\infty = r_1 + r_2\text{.}\)

Let \(r = r_1 + r_2 - 1 = \#S_\infty'\text{.}\)

Recall the log embedding for \(\adeles^{\times, 1}\)

\begin{equation*} l(b ) = (\log |b|_{v_1} , \ldots, \ldots, \log |b|_{v_r})_{v_i\in S_\infty'}\text{.} \end{equation*}
\begin{equation*} l \colon W \backslash \adeles^{\times, 1} \twoheadrightarrow \RR^r \end{equation*}

let \(\{\epsilon_i\}\) be a basis for the global units, \(\{l(\epsilon_i)\} \subseteq \RR^r\) span a full rank sublattice. So \(K^\times\backslash\adeles^{\times, 1}\) is compact.

2
\begin{equation*} \mu(K\units \backslash \adeles^{\times, 1}) = \frac{2^{r_1} (2\pi)^{r_2} R h_K}{\sqrt{|D_K|} W_K} \end{equation*}

Where \(r_1\) is the number of real embeddings, \(r_2\) is half the number of complex embeddings, \(R\) is the regulator of \(K\text{,}\) \(D_K\) is the discriminant of \(K\text{,}\) \(h_K\) the class number, \(W_K\) the number of roots of unity of \(K\text{.}\)

1 and 2 give us a fundamental domain.

Definition 3.38 Hecke characters

A Hecke character is a continuous map

\begin{equation*} K\units \backslash \adeles\units \to \CC \units\text{.} \end{equation*}
Note 3.39

\(|\cdot |_\adeles^s\) is a Hecke character any Hecke character \(\chi = \chi_0 | \cdot |_\adeles^s\)

\begin{equation*} \chi_0 \colon K\units \backslash \adeles^{\units,1} \to \CC\units \end{equation*}

where \(\Re(s)\) is the exponent.

Subsubsection 3.2.2 Main theorem

Global \(\zeta\)-functions \(\zeta(f,\chi)\)

\begin{equation*} \mathcal S(\adeles) = \left\{ f \colon \adeles \to \CC : f,\hat f \in L^1,\, \sum_{\xi\in K} f(a(x + \xi)) , \sum_{\xi\in K} \hat f(a(x + \xi)) \text{ both }L^1\,\forall x\in \adeles\units, x\in \adeles,\,f(\alpha) |\alpha|^\sigma, \hat f(\alpha)|\alpha|^\sigma \in L^1 \,\forall \sigma \gt 1\right\} \end{equation*}
\begin{equation*} \zeta(f,\chi) = \int_{\adeles\units} f(\alpha) \chi(\alpha) \diff \units \alpha = \prod \zeta_v(f_v,\chi_v)\text{.} \end{equation*}
  1. \begin{equation*} \zeta(f,\chi) = \int_{\adeles^\times} f(\alpha) \chi(\alpha) \diff\units \alpha \end{equation*}
    \begin{equation*} = \int_0^\infty \int_{\adeles^{\times,1}} f(tb) \chi(tb) \diff \units b \frac{\diff t}{t} \end{equation*}
    \begin{equation*} = \int_0^\infty \zeta_t(f, \chi) \frac{\diff t}{t} \end{equation*}
    where \(\zeta_t(f,\chi) = \int_{\adeles^{\units,1}} f(tb) \chi(tb) \diff \units b\text{.}\)
  2. Functional equation
    \begin{equation*} \zeta_t(f,\chi) = \int_{K\units\backslash \adeles^{\times,1}} \sum_{\alpha \in K^\times} f(t\alpha b) \overbrace{\chi(t\alpha b)}^{=\chi(tb)} \diff \units b \end{equation*}
    \begin{equation*} = \int_{K\units \backslash \adeles^{\times,1}} \left\{\sum_{\alpha\in K} f(t\alpha b) - f(0)\right\} \chi(tb) \diff\units b \end{equation*}
    \begin{equation*} = \int_{K\units \backslash \adeles^{\times,1}} \left(\sum_{\alpha\in K} f(t \alpha b )\right) \chi(tb) \diff \units b - f(0) \int_{K\units \backslash \adeles^{\units,1}} \chi(tb) \diff \units b \end{equation*}
    the rightmost term is \(\kappa\) iff \(\chi = |\cdot| ^s\) and 0 otherwise.
    \begin{equation*} \sum_{\alpha \in K} f(t\alpha b) = \frac{1}{|tb|_{\adeles}} \sum_{\alpha \in K} \hat f\left( \frac{\alpha}{tb}\right) \end{equation*}
    Remark: \(b\in \adeles^{\times, 1}\) implies \(|b|_\adeles = 1\) and \(|tb|_\adeles = |t|_\adeles = t\) via \(\RR_{\gt 0} \hookrightarrow \adeles\units\text{.}\) This implies
    \begin{equation*} \int_{K\units\backslash \adeles^{\times, 1 }} \left( \sum_{\alpha \in K} f(t \alpha b) \right) \chi(tb) \diff \units b = \frac 1t \int_{K\units\backslash \adeles^{\times, 1 }} \left( \sum_{\alpha \in K} \hat f(\frac{\alpha}{tb}) \right) \chi(tb) \diff \units b \end{equation*}
    \begin{equation*} = \frac 1t \int_{K\units\backslash \adeles^{\times, 1 }} \left( \sum_{\alpha \in K} \hat f(\frac{\alpha b}{t}) \right) \chi(tb\inv) \diff \units b \end{equation*}
    \begin{equation*} = \frac 1t \int_{K\units\backslash \adeles^{\times, 1 }} \sum_{\alpha \in K\units} \hat f(\frac{\alpha b}{t}) \chi(tb\inv) \diff \units b + \hat f(0)\int_{K\units\backslash \adeles^{\times, 1 }} \hat \chi(b/t) \diff \units b \end{equation*}
    implies
    \begin{equation*} \zeta_{1/t}(\hat f, \hat \chi) + \hat f (0 ) \int_{K\units\backslash \adeles^{\times, 1 }} \hat \chi(b/t) \diff\units b \end{equation*}
    \begin{equation*} = \zeta_t(f,\chi) + f(0) \int_{K\units\backslash \adeles^{\times, 1 }} \chi (bt) \diff\units b \end{equation*}
    take this and put it back in 1.
    \begin{equation*} \int_0^\infty \zeta_t( f,\chi) \frac{\diff t}{t} = \int_0^1 + \int_1^\infty \end{equation*}
    the rightmost term is cool, so use
    \begin{equation*} \int_0^1 \zeta(f,b) \frac{\diff t}{t} = \int_0^1\zeta_{1/t} (\hat f, \hat \chi) \frac {\diff t}{t} \end{equation*}
    \begin{equation*} + \int_0^1 \hat f(0) \int_{K\units\backslash \adeles^{\times, 1 }} \hat \chi (b/t) \diff\units b \frac{\diff t}{t} -\int_0^1 f(0) \int_{K\units\backslash \adeles^{\times, 1 }} \chi (bt) \diff\units b \frac{\diff t}{t} \end{equation*}
    \begin{equation*} = \int_1^\infty\zeta_{t} (\hat f, \hat \chi) \frac {\diff t}{t} + \hat f(0) \kappa \int_0^1 t^{s-1}\frac{\diff t}{t} - f(0)\kappa\int_0^1 t^s\frac{\diff t}{t} \end{equation*}
    the rightmost terms are present only if \(\chi_0 =1\) they come to \(\hat f(0) \kappa/(s-1) - f(0) \kappa/s\text{.}\) This is invariant under \(f\mapsto \hat f,\,\chi \mapsto \hat \chi\text{.}\) All together
    \begin{equation*} \zeta(f,\chi) = \int_1^\infty \zeta_t(f, \chi) \frac{\diff t}{t} + \int_1^\infty \zeta_t(\hat f , \hat \chi) \frac{\diff t}{t} + \left(\hat f(0) \kappa/(s-1) - f(0) \kappa/s \right)\text{.} \end{equation*}
Example 3.41 Concrete examples
  1. Let \(K\) be a number field and \(\chi = |\cdot|_\adeles^s\text{,}\) \(f = \otimes' f_v\) where
    \begin{equation*} f_v = \begin{cases} e^{-\pi x^2} \amp v = \RR\\ e^{-2\pi (x^2 + y^2)}\amp v = \CC\\ \text{char. fn. of } \ints_v \amp v\text{ non-arch.} \end{cases} \end{equation*}
    then \(\zeta(f,\chi)\) is a multiple of \(\zeta_K(s)\) the Dedekind zeta function (check this!). We have thus proved the analytic continuation of \(\zeta_K\text{.}\)
  2. For \(K = \QQ(\sqrt 2)\text{.}\) We have \(\disc(K) = 8\text{,}\) i.e. \(K\) is ramified at 2 only. \(h_K = 1\) also and \(\ints_K = \ZZ\lb \sqrt 2 \rb\text{.}\)
    \begin{equation*} \infty_1 \colon a+b\sqrt 2 \hookrightarrow a+ b\sqrt 2 \in \RR \end{equation*}
    \begin{equation*} \infty_2 \colon a+b\sqrt 2 \hookrightarrow a- b\sqrt 2 \in \RR\text{,} \end{equation*}
    the units are \(\langle -1 , 1 +\sqrt 2 \rangle\text{.}\) The Hecke characters are then as follows: Let
    \begin{equation*} \chi = \prod \chi_v = |\cdot|_{\infty_1}^{it_{\infty_1}}|\cdot|_{\infty_2}^{it_{\infty_2}}\prod_{v\nmid \infty}|\cdot|_{v}^{it_{v}} \end{equation*}
    then for \(\chi\) to be a Hecke character we need \(\chi(a) = 1\) for all \(a \in K\text{.}\) Let's check this condition on units. \(\chi(-1) = 1\) does not give us an extra condition. \(\chi(1+ \sqrt2 ) = | 1+ \sqrt 2|^{it_1} | 1- \sqrt 2|^{it_2}\) so that
    \begin{equation*} \chi( \sqrt 2 + 1 ) = 1 \iff t_{\infty_1} - t_{\infty_2} = \frac{2k \pi}{\log(\sqrt 2 - 1)} \end{equation*}
    let's fix
    \begin{equation*} t_{\infty_1} = \frac{-\pi}{\log(\sqrt 2- 1)} \end{equation*}
    \begin{equation*} t_{\infty_2} = \frac{\pi}{\log(\sqrt 2- 1)} \end{equation*}
    once we fix this we can complete uniquely to a Hecke character \(\chi\text{.}\) Let's consider the \(\zeta\)-functions now.
    \begin{equation*} \zeta(f,\chi) = \pi^{-(s+it_{\infty_1})/2}\pi^{-(s-it_{\infty_2})/2} \Gamma\left(\frac{s+it_{\infty_1}}{2}\right)\Gamma\left(\frac{s-it_{\infty_2}}{2}\right) \frac{1}{\sqrt 8} \overbrace{L(s,\chi)}^{\prod_{v\nmid \infty}(1- \chi(\varpi_v)/q^s)} \end{equation*}
    \begin{equation*} \zeta(\hat f,\hat \chi ) = \pi^{-(1-s+it_{\infty_1})/2}\pi^{-(1-s-it_{\infty_2})/2} \Gamma\left(\frac{1-s+it_{\infty_1}}{2}\right)\Gamma\left(\frac{1-s-it_{\infty_2}}{2}\right) \frac{1}{8^s} L(1-s,\chi\inv) \end{equation*}
    then Tates's implies
    \begin{equation*} \pi^{-s} \Gamma\left(\frac{s-i\pi/\log(\sqrt 2 - 1)}{2}\right)\Gamma\left(\frac{s+i\pi/\log(\sqrt 2 - 1)}{2}\right) \frac{1}{8} L(s,\chi) \end{equation*}
    \begin{equation*} = \pi^{1-s} \Gamma\left(\frac{1-s-i\pi/\log(\sqrt 2 - 1)}{2}\right)\Gamma\left(\frac{1-s+i\pi/\log(\sqrt 2 - 1)}{2}\right) \frac{1}{8^s} L(1-s,\chi\inv)\text{.} \end{equation*}