Section 4Selberg orthogonality and strong multiplicity one for $\GL(n)$
The proof of Theorem 2.1.2 used only standard techniques
from analytic number theory. Utilizing recent
results concerning the Selberg orthonormality conjecture,
and restrictng to the case of $L$-functions of
cuspidal automorphic representations of $\GL(n)$,
one obtains
the following theorem, which is stronger than (3.1.1).
Theorem4.1
Suppose that $\pi$, $\pi'$ are (unitary) cuspidal automorphic representations of $\GL(n,\A_F)$, and suppose
\begin{equation}
\sum_{p\le X} \frac{1}{p}\left| \tr A(\pi_p) - \tr A(\pi_p')\right|^2
\le (2-\epsilon) \log\log(X)\tag{4.1}
\end{equation}
for some $\epsilon>0$ as $X\to\infty$. If $n\le 4$, or if
Hypothesis H holds for both $L_\mathrm{fin}(s,\pi)$ and $L_\mathrm{fin}(s,\pi')$
(in particular if the partial Ramanujan conjecture $\theta < \frac14$ is true
for $\pi$ and $\pi'$), then $\pi=\pi'$.
Using the fact that $1.4^2 < 2$ and the consequence of the
prime number theorem
\begin{equation}
\sum_{p\le X}\frac{1}{p} \sim \log\log(X),\tag{4.2}
\end{equation}
we see that condition (4.1) holds
if $\left| \tr A(\pi_p) - \tr A(\pi_p')\right| < 1.4$ for all but finitely
many $p$. Thus, the strong multiplicity one theorem only requires considering
the traces of $\pi_p$, and futhermore those traces can differ at every
prime, and by an amount which is bounded below.
For $GL(2,\A_\Q)$, the Ramanujan bound along with (4.2)
implies a version of a result of Ramakrishnan [Ram]:
if $ \tr A(\pi_p) = \tr A(\pi_p')$ for $\frac78+\varepsilon$ of all
primes $p$, then $\pi=\pi'$. This result was extended by Rajan [Raj].
The proof of Theorem 4.1 is a straightforward application of
recent results toward the Selberg orthonormality conjecture [LWY, AvSm], which make use of progress on Rudnick and Sarnak's "Hypothesis H" [RudSar, K]. Suppose
\begin{equation}L_1(s)=\sum\frac{a(n)}{n^s},\qquad L_2(s)=\sum\frac{b(n)}{n^s}\tag{4.3}
\end{equation}
are $L$-functions, meaning that they have a functional equation and
Euler product as described in Section 2.1.
The point of the strong multiplicity one theorem is that
two $L$-functions must either be equal, or else they must be
far apart. The essential idea was elegantly described by
Selberg; see [Sel]. Recall
that an $L$-function is primitive if it cannot be written
nontrivially as a product of $L$-functions.
Conjecture4.2(Selberg Orthonormality Conjecture)
Suppose that $L_1$ and $L_2$ are primitive $L$-functions with Dirichlet
coefficients $a(p)$ and $b(p)$. Then
\begin{equation}
\sum_{p\le X} \frac{a(p)\overline{b(p)}}{p}=\delta({L_1, L_2}) \log\log(X)+O(1),\tag{4.4}
\end{equation}
where $\delta({L_1, L_2}) = 1$ if $L_1=L_2$, and $0$ otherwise.
Proof of Theorem 4.1. Rudnick and Sarnak's Hypothesis H is the assertion
\begin{equation}
\sum_p \frac{a(p^k)^2\log^2(p)}{p^k} < \infty
\end{equation}
for all $k\ge 2$. For a given $k$, this follows from
a partial Ramanujan bound $\theta < \frac12 - \frac{1}{2k}$.
Since $k\ge 2$,
Hypothesis H follows from the partial Ramanujan bound
$\theta < \frac14$.
For the standard $L$-functions of cuspidal
automorphic representations on $\GL(n)$,
Rudnick and Sarnak [RudSar] proved Selberg's
orthonormality conjecture under the assumption of Hypothesis H,
and they proved Hypothesis H for
$n=2$, $3$.
The case of $n=4$ for Hypothesis H was proven by Kim [K].
Thus, under the conditions in Theorem 4.1,
the Selberg orthonormality conjecture is true.
Since $\pi$ and $\pi'$ are cuspidal automorphic
representations of $\GL(n,\A_F)$, the $L$-functions $L_1(s) = L_\mathrm{fin}(s,\pi)$
and $L_2(s) = L_\mathrm{fin}(s, \pi')$ are primitive $L$-functions. Hence,
by (4.4)
\begin{align}
\sum_{p\le X} \frac{1}{p} |a(p)-b(p)|^2
=\mathstrut&\sum_{p\le X} \frac{1}{p} \bigl( |a(p|^2 + |b(p)|^2 - 2 \Re(a(p)\overline{b(p)})\bigr) \cr
=\mathstrut& 2\log\log(X) - 2 \delta_{L_1, L_2} \log\log(X)+ O(1) \cr
=& \begin{cases}
O(1) & \text{ if } L_1= L_2 \cr
2\log\log(X) + O(1) & \text{ if } L_1\not = L_2. \cr
\end{cases}\tag{4.5}
\end{align}
We have $\sum_{p\le X} \frac{1}{p} |a(p)-b(p)|^2 \le (2-\epsilon)\log\log(X)$ for some $\epsilon > 0$. This implies that
$\epsilon \log\log(X)$ is unbounded, and hence (4.5) implies that $L_1(s) = L_2(s)$. This gives us $\pi = \pi'$. \qed
\vspace{2ex}
Recently, the transfer of full level Siegel modular forms to $\GL(4)$
was obtained in [PSS]. Hence, we can apply Theorem 4.1
to the transfer to $\GL(4)$ of a Siegel modular form of full level
and thus obtain a stronger version of Theorem 3.2.1.
Theorem4.3
Suppose $F_j$, for $j=1,2$, are Siegel Hecke eigenforms of weight $k_j$ for $\Sp(4,\Z)$, with Hecke eigenvalues $\mu_j(n)$. If
\begin{equation}
\sum_{p\le X} \frac 1p \left|p^{3/2-k_1}\mu_1(p)-p^{3/2-k_2}\mu_2(p)\right|^2 \le (2-\epsilon) \log\log(X)\tag{4.6}
\end{equation}
for some $\epsilon > 0$, as $X\to\infty$, then $k_1=k_2$ and $F_1$ and $F_2$ have the same eigenvalues for the Hecke operator $T(n)$ for all $n$.
\proof[Acknowledgements] We thank Farrell Brumley and Abhishek Saha for carefully reading an
earlier version of this paper and for providing useful feedback on it.