Different approaches to the distribution of primes (Q627053)

The author gives both a survey of the classic history, for the study of prime numbers, and a glimpse into a very recent approach to the distribution of primes, namely, a more elementary study based on the concept of ``pretentiousness'', among multiplicative functions (namely, functions \(f\) such that \(f(mn)=f(m)f(n)\), for all coprime natural numbers \(m,n\)). This new approach allows to avoid use of the Riemann \(\zeta\)-function \[ \zeta(s):=\sum_{n=1}^{\infty}n^{-s}, \quad \forall s\in \mathbb C, \; \Re(s)>1. \] The paper gives a brief account of the major historical issues about primes and this function, starting from the greatest pearl regarding our knowledge about prime numbers, the celebrated Prime Number Theorem, discovered by a (very!) young Gauss, namely (in modern notation) \[ \sum_{p\leq x}\log p \sim x, \quad x\to \infty. \tag{\text{PNT}} \] We don't have room, here, to report the whole story that the author tells (not a comprehensive one, of course, but quite a good historical report). However, from the section names it transpires the author's underlying idea that, after (well after!) Riemann's memoir (his famous \(1860\) only eight pages, celebrated here for the \(150\)th year), the quest for an elementary approach started, with (see section \(2\)) the elementary proof of PNT by Erdős and Selberg (independently!); this was based on the Selberg formula: the elementary ``core'', so to speak, of the 1949 PNT proof. This formula may also prove PNT in arithmetic progressions \[ \sum_{{p\leq x}\atop {p\equiv a\pmod q}}\log p \sim {x\over {\varphi(q)}}, \quad x\to \infty, \tag{\text{PNT-AP}} \] where \((a,q)=1\) (indicating with \((r,s)\) the \(gcd\) of \(r\) and \(s\), i.e., the progression \(qn+a=q+a,2q+a,\dots\) is non-trivial) and \(\varphi(q):=|\{ 1\leq n\leq q, (n,q)=1\}|\) is Euler's function. This is equivalent to \[ \pi(x;q,a)\sim {{\pi(x)}\over {\varphi(q)}}, \] where \(\pi(x):=|\{ p\leq x\}|\) and \(\pi(x;q,a):=|\{p\leq x, p\equiv a\pmod q\}|\) (again, the progression being non-trivial). The author (in Section 3) acquaints the reader with the other new approach to the distribution of primes in arithmetic progressions; the classic approach to prove PNT-AP dates back to Dirichlet, in fact, through His \(L\)-functions, \[ L(s,\chi):=\sum_{n=1}^{\infty}\chi(n)n^{-s} \] where \(\chi\) is a Dirichlet character \(\bmod q\), i.e. \(\chi(n+q)=\chi(n)\) and \(\chi(mn)=\chi(m)\chi(n)\) (i.e., they are periodic mod \(q\) and completely multiplicative). Already Selberg's formula avoids the application of these \(L\)-functions properties. In \(1970\) Gallagher gave a ``classic'', so to speak, proof (i.e., based on Dirichlet \(L\)-functions) that \[ \pi(x;q,a)\sim {{\pi(x)}\over {\varphi(q)}}+{{\chi(a)}\over {\varphi(q)}}\sum_{p\leq x}\overline{\chi(p)}, \] for an ``exceptional'', say, character \(\chi\), that (if exists at all!) is real, otherwise may be removed and this approximation becomes like the one above. (Hiding here the definition of a ``Siegel zero'', see 1970 paper.) Then, Elliott, in the \(2002\), proved the same in an elementary fashion, under a suitable technical hypothesis (a lower bound for the class-number of \(\mathbb Q(\sqrt{-q})\) that may be even proved, knowing \(L\)-functions zeros location). See that the sum over \(p\) is large exactly when \(\chi\) ``pretends'', say, to be the Möbius function, \(\mu\), defined in fact to be \(\mu(p):=-1\) on primes, multiplicative and \(0\) out of square-free numbers (i.e., \(\mu(n)=0\) whenever has at least a prime-power divisor). This is the origin of the name of ``pretentious'', say a multiplicative function \(f:\mathbb N\rightarrow \mathbb C\) with \(|f(n)|\leq 1\), \(\forall n\): \[ \sum_{{n\leq x}\atop {n\equiv a\pmod q}}f(n)\sim {{\chi(a)}\over {\varphi(q)}}\sum_{n\leq x}f(n)\overline{\chi(n)}, \] for an exceptional character \(\chi\) (see the paper for the exact statement), that renders the second sum large (because pretends to be this character \(\chi\)!). This was only a name in the hands of Elliott (who used that in the context of the Large Sieve techniques), but the author, together with Soundararajan, has started to develop an elementary theory, in the millennium turning.