Skip to main content

Section 6.10 Rankin-Selberg (Aash)

Notation:

\(K\) imaginary quadratic field.

\(\mathscr A\) ideal class of \(K\)

\(D\) discriminant \(K\)

\(\epsilon (n) = \legendre{D}{n}\) associated Dirichlet character.

\(h = \#\Cl_K\text{.}\)

\(w = 2u \) twice number of units.

\(f\in S_{2k}^{new}(\Gamma _0(N))\) for \(k\ge 1\text{.}\) \((N,D) = 1\) write

\begin{equation*} f(z) = \sum_{n \ge 1} a(n) e^{2\pi i n z} \end{equation*}
\begin{equation*} L(f,s) = \sum a_n n^s \end{equation*}

define

\begin{equation*} L_{\mathscr A } (f,s) = L^{(N)} (2s-2k +1 , \epsilon ) \cdot \sum_{n=1}^\infty a(n) r_{\mathscr A} (n)n^{-s} \end{equation*}
\begin{equation*} L^{(N)} = \sum _{(n,N) = 1} \epsilon (n) n^{-2s +2k -1}\text{.} \end{equation*}

Subsection 6.10.1 Rankin's method

\begin{equation*} \theta _{\mathscr A} (z) = \sum_{n=1}^\infty r_{\mathscr A}(n) q^n \in S_1(|D|, \epsilon ) \end{equation*}
\begin{equation*} \frac{\Gamma (s+2k -1)}{(4\pi )^{s+2k-1}} \cdot \sum_{n=1}^\infty \frac{a(n) r_{\mathscr A}(n)}{n^{s+2k-1}} \end{equation*}
\begin{equation*} = \int_0^\infty \sum_{n=1}^\infty a(n)r_{\mathscr A}(n) e^{-4\pi n y} y^{s+2k-2} \diff y \end{equation*}
\begin{equation*} = \int_0^\infty \int_0^1 f(x+iy) \overline{\theta _{\mathscr A} (x+iy)} \diff x y^{s+2k -2 } \diff y \end{equation*}
\begin{equation*} = \int\int_{\Gamma _\infty \backslash \HH} f(z) \overline{\theta _{\mathscr A} (z)} \diff x \diff y/y^2 \end{equation*}
\begin{equation*} = ????? \end{equation*}

Choose \(\mathcal F\) to be a fundamental domain for \(\Gamma _0(M)\) where \(M = N|D|\) consider

\begin{equation*} \bigcup _{\gamma \in \Gamma _\infty \backslash \Gamma _0(M) } \gamma \mathcal F \end{equation*}

have

\begin{equation*} \sum_{\gamma \in \Gamma _\alpha \backslash \Gamma _0(M)} \int \in_{\gamma \mathcal F} f(z) \overline{\theta _{\mathscr A}(z)} y^{s+2k} \frac{\diff x\diff y}{y^2} \end{equation*}
\begin{equation*} = \int\int _{ \mathcal F} f(\gamma z) \overline{\theta _{\mathscr A}(\gamma z)} (\Im \gamma z)^{s+2k} \frac{\diff x\diff y}{y^2} \end{equation*}
\begin{equation*} \sum_{\gamma =\pm \begin{pmatrix} \bullet \amp \bullet \\\ c \amp d \end{pmatrix} \in \Gamma _\infty \backslash \Gamma _0(M)} \int\int_{\mathcal F} f(z) \overline{\theta _{\mathscr A}(z)} \end{equation*}

More formulae

\begin{equation*} \frac{\Gamma (s+2k-1)}{(4\pi )^{s+2k-1}} L_{\mathscr A }(f, s+2k-1) \end{equation*}
\begin{equation*} = \int\int_{\mathcal F} f(z) \overline{\theta _{\mathscr A}(z) E_s(z)} y^{2k} \frac{\diff x\diff y}{y^2} \end{equation*}
\begin{equation*} = (f, \theta _{\mathscr A} f)_{\Gamma _0(N)} \end{equation*}
\begin{equation*} E_s(z) = \frac 12 \sum_{c,d\in \ZZ,\,c \equiv 0 \pmod M} \frac{\epsilon (d)}{(cz+d)^{2k-1}} \frac{y^s}{|cz+d|^{2s}} \end{equation*}

want to take \((d,M) =1\) to \((d, N)=1\) .

we resolve this by letting

\begin{equation*} \tr_N^M \colon \widetilde M_{2k}(\Gamma _0(M)) \hookrightarrow \widetilde M_{2k}(\Gamma _0(N)) \end{equation*}
\begin{equation*} g \mapsto \sum _{\gamma \in \Gamma _0(M)\backslash \Gamma _0(N)} g|_{2k} \gamma \end{equation*}
\begin{equation*} (f,g)_{\Gamma _0(M)} = (f,\tr_N^M f)_{\Gamma _0(N)} \end{equation*}

so

\begin{equation*} (4\pi )^{-s-2k +1} \Gamma (s+2k-1) L_{\mathscr A}(f, s+2k-1)= (f, \tr_N^M \theta _{\mathscr A}E_s) \end{equation*}

Then for \(f\in S_{2k}^{new}(\Gamma _0(N))\)

\begin{equation*} (4\pi )^{-s-2k +1} N^s\Gamma (s+2k-1) L_{\mathscr A}(f, s+2k-1)= (f, \tilde \phi _s) \end{equation*}

Computing the trace

\begin{equation*} \tilde \phi _s = \epsilon _s(Nz) \theta _{\mathscr A}(z) U_{|D|} ??? \end{equation*}

where

\begin{equation*} \epsilon _s= \frac{\sum_{D= D_1\cdot D_2} \epsilon _{D_1}(N) \chi _{D_1D_2}(\mathscr A) E_s ^{(D_1)} (|D_2|z)}{\kappa (D_1) |D_1|^{s+2k-1}} \end{equation*}

\(D\) odd \(\equiv 1\pmod 4\) \(D_1,D_2\) fund disc. \(\chi _{D_1,D_2}\) genus character \(\chi_{(a)} = \epsilon _{D_1}(N(a) ) = \epsilon _{D_2}(N(a))\) for ideal prime to \(D\) with \(\kappa = 1 , D_1\gt 0\text{,}\) \(\kappa = i\) \(D_2 \lt 0\text{.}\)

\begin{equation*} E_s^{(D_1)} (z) =\frac 12 \sum_{m,n \in \ZZ,\,D_2 |m} \frac{\epsilon _1(m)\epsilon _2(n) y^s}{(mz+n)^{2k-1}{|mz+n|}^{2s}} \end{equation*}
\begin{equation*} U_n \colon f(z) \mapsto \frac 1n \sum_{j\pmod n} f\left(\frac{z+j}{n}\right) \end{equation*}

on a function \(f\) of period 1.

Fourier coefficients. Consider

\begin{equation*} \epsilon _s(z) = \sum_{n\in \ZZ} e_s(n,y) e(nx) \end{equation*}