2
$\begingroup$

In this post, we will write $M_q$ as a set of Dirichlet characters modulo $q\geq 1$.

Theorem If $\chi\in M_q$ and $\chi\neq \chi_0$, then $L(1,\chi)\neq 0$.

Corollary: If $a$, $q$ are positive integers with $(a,q) = 1$, then there are infinitely many prime numbers $p$ such that $p\equiv a\pmod q$.

The problem is about the corollary. The lecturer proved it by proving that $\sum_{p\equiv a\pmod q}\frac{1}{p^s}$ diverges as $s\to 1$. The first question is, why are they equivalent?

The lecturer has shown that, for $\Re s>1$, $$\sum_{p\equiv a\pmod q}\frac{1}{p^s}=\frac{1}{\phi(q)}\sum_{\chi\in M_q}\overline{\chi}(a)\left ( \log L(s,\chi)+\mathcal{O}(1) \right )\tag{1}$$ He didn't go further, and let us do the rest ourselves. I hope you can correct me if I am mistaken somewhere. If $\chi=\chi_0$, then $$ (s-1)L(s,\chi_0)=(s-1)\zeta(s)\prod_{p\mid q}(1-p^{-s})\xrightarrow{s\to 1^{+}} \prod_{p\mid q}(1-p^{-1}) $$ The limit is finite and is $\neq 0$. This implies that $L(s,\chi_0)\xrightarrow{s\to 1^{+}} \infty$, and so $\log L(s,\chi_0)\xrightarrow{s\to 1^{+}} \infty$. The right hand side of $(1)$ is then divergent as $s\to 1^+$, which makes the left hand side divergent. If $\chi\neq \chi_0$, then $\log L(s,\chi)\xrightarrow{s\to 1^{+}} \log L(1,\chi)$ is finite by theorem above. How do I then check if the right hand side of $(1)$ diverges? This is the second question.

  • 0
    To answer your first question rapidly, it's not so much that they're equivalent rather than that the divergence of the series implies that there are infinitely many terms in it. If you replace the primes with the squares of natural numbers then the series converges as $s\to 1$ which certainly doesn't imply that there are finitely many squares!2017-01-22

0 Answers 0