# zeros of L_4(s) on Re(s)=1?

Discussion in 'Math Research' started by marco72, Sep 14, 2009.

1. ### marco72Guest

It is a well-known fact that the zeta function has no zeros on the
line Re(s)=1. Consider the Dirichlet L-series
associates to the non-trivial mod 4 character
L_4(s) = 1/1^s - 1/3^s + 1/5^s - 1/7^s + ...
Is it still true that this function has no zeros on the line Re(s)=1?
If so, why? Thanks.

marco72, Sep 14, 2009

2. ### JoeShipmanGuest

This is true for any Dirichlet L-series. Newman's proof is the
slickest: consider the product Zm(s) of the L-functions for the non-
trivial characters mod m. Suppose Zm(1+ia)=0. Then so is Zm(1-ia), and
the function ((Zm(s))^2)Zm(s+ia)Zm(s-ia) is both real and entire
(because the only possible pole at s=1 is balanced by the zeros at
1+ia and 1-ia). But the Dirichlet series for this function has non-
negative real coefficients (because by Euler factorization the log of
each factor has non-negative real coefficients and exponentiation
preserves this). It is a standard result that if an entire function
has a Dirichlet series with non-negative coefficients, then that
series is everywhere convergent. But it's easy to show that the
Dirichlet series diverges at 0 by looking at the subseries of terms
where n is a power of some prime.

JoeShipman, Sep 15, 2009