Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Does there exist a set $S \subset \mathbb N$ such that the Dirchlet density of $S$ is well-defined and positive, the Dirchlet density of $S \cap \operatorname{PRIMES}$ is well-defined and zero, and:

$ \prod_{n \in S} \frac{1}{1-n^{-s}}$

has a meromorphic continuation to the whole complex plane? Can we construct it?

Motivation: How special is the set of primes, among sets of natural numbers of approximately the same size, for having a meromorphic continuation? I would guess that analytic continuation should be a very rare occurrence, but I don't have an intuition for how rare, or how hard it is to find an example if one exists.

Extra credit for a functional equation.

share|improve this question
I wonder what happens if $S$ is the set of numbers of the form $p+1$, $p$ being prime. –  Gerry Myerson Nov 5 '12 at 11:24
add comment

1 Answer

You're asking about how special the prime numbers are as a subset of the integers. One can equally well ask how special the sequence $a_k = 1$ is when viewed as the sequence of coefficients of a Dirichlet series. I don't have anything to offer on your original question, but have read a few things about the latter question that may be of some interest:

  1. According to Section 8 of David Farmer's article titled Basic Analytic Number Theory, if $f(s) = \displaystyle \sum_{k = 1}^{\infty} \frac{a_k}{k^s}$ where the $a_k$ are integers, then a sufficient condition for $f(s)$ to admit a meromorphic continuation to $\Re{(s)} = 0$ is that:

    (1) $a_k$ is of subpolynomial growth

    (2) $a_k$ is multiplicative

    (3) If $p$ is prime then $a_{p^m}$ is independent of $m$

    and may or may not have a natural boundary there.

  2. According to section 9.5 of Titchmarsh's The Theory of the Riemann Zeta-function, if $a_k = 0$ when $k$ is composite and $a_{k} = 1$ when $k$ is prime then $f(s)$ (provably) has a natural boundary at $\Re{(s)} = 0$

  3. I've also heard of results of the type "a Dirichlet series with $a_k$ chosen at random uniformly from $[-1, 1]$ has natural boundary $\Re{(s)} = 0$ with probability $1$," but don't know a precise statement or a reference.

share|improve this answer
What do you mean by a "natural boundary"? Unfortunately I think the second condition is never satisfied by a sequence satisfying my conditions, because it implies that $S$ consists only of prime powers, but the set of prime powers that aren't primes has Dirchlet density zero. –  Will Sawin Nov 5 '12 at 15:18
Natural Boundary: analytic continuation to that line, but no further owing to an accumulation of zeros or poles on the line forcing the function to be identically 0 or infinity under the assumption of an analytic continuation beyond the line - a contradiction. –  Jonah Sinick Nov 5 '12 at 16:37
Yes, you're right that the second condition that I list and yours are mutually exclusive - again, the things that I mention are about the cognate question that I mention, not your original one. –  Jonah Sinick Nov 5 '12 at 16:41
add comment

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.