7
$\begingroup$

Let $D(s)$ be a Dirichlet series with abscissa of convergence $\sigma_c=\sigma_a$. Does it follow that the Dirichlet series defined by $P(s)=D(s)\bar D(\bar s)$ has the same abscissa of convergence?

In light of Noam D. Elkies answer below, I would like to know about conditions on the coefficients that also lead to the divergence of the square? What if the coefficients are completely multiplicative?

Improve this question
$\endgroup$
1
  • $\begingroup$ It's certainly true that $\sigma_a(P) \le \sigma_a(D)$ -- this is immediate from the fact that the divisor function $d(n)$ is $O(n^{\epsilon})$ for any $\epsilon > 0$. $\endgroup$
    – David Loeffler
    Commented Sep 8, 2013 at 15:11

1 Answer 1

Reset to default
11
$\begingroup$

That's not true even for power series with real coefficients. Let $$ D(s) = \sqrt{1-2^{-s}} = 1 - \frac12 2^{-s} - \frac18 4^{-s} - \frac1{16} 8^{-s} - \frac5{128} 16^{-s} - \cdots . $$ Then $P(s) = 1 - 2^{-s}$, so $\sigma_c(D) = \sigma_a(D) = 0$ but $\sigma_c(P) = \sigma_a(P) = -\infty$.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Great example, it's simplicity is striking. Thank you. Now, do you think there are further hypotheses that would work? The case of interest to me has completely multiplicative coefficients. $\endgroup$
    – Kevin Smith
    Commented Sep 9, 2013 at 14:14

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .