Efficient prime counting and the Chebyshev primes

Abstract: The function

e p s i l o n (x) = m b o x l i (x) - p i (x)

is known to be positive up to the (very large) Skewes' number. Besides, according to Robin's work, the functions

e p s i l o n_{h e t a} (x) = m b o x l i [h e t a (x)] - p i (x)

and

e p s i l o n_{p s i} (x) = m b o x l i [p s i (x)] - p i (x)

are positive if and only if Riemann hypothesis (RH) holds (the first and the second Chebyshev function are

h e t a (x) = s u m_{p l e x} l o g p

and

p s i (x) = s u m_{n = 1}^{x} L a m b d a (n)

, respectively,

m b o x l i (x)

is the logarithmic integral,

m u (n)

and

L a m b d a (n)

are the M"obius and the Von Mangoldt functions). Negative jumps in the above functions

e p s i l o n

,

e p s i l o n_{h e t a}

and

e p s i l o n_{p s i}

may potentially occur only at

x + 1 i n m a t h c a l P

(the set of primes). One denotes

j_{p} = m b o x l i (p) - m b o x l i (p - 1)

and one investigates the jumps

j_{p}

,

j_{h e t a (p)}

and

j_{p s i (p)}

. In particular,

j_{p} < 1

, and

j_{h e t a (p)} > 1

for

p < 1 0^{11}

. Besides,

j_{p s i (p)} < 1

for any odd

p i n m a t h c a l m b o x C h

, an infinite set of so-called {it Chebyshev primes } with partial list

109, 113, 139, 181, 197, 199, 241, 271, 281, 283, 293, 313, 317, 443, 449, 461, 463, l d o t s

. We establish a few properties of the set

m a t h c a l m b o x C h

, give accurate approximations of the jump

j_{p s i (p)}

and relate the derivation of

m b o x C h

to the explicit Mangoldt formula for

p s i (x)

. In the context of RH, we introduce the so-called {it Riemann primes} as champions of the function

p s i (p_{n}^{l}) - p_{n}^{l}

(or of the function

h e t a (p_{n}^{l}) - p_{n}^{l}

). Finally, we find a {it good} prime counting function

S_{N} (x) = s u m_{n = 1}^{N} f r a c m u (n) n m b o x l i [p s i (x)^{1 / n}]

, that is found to be much better than the standard Riemann prime counting function.

Review text

The authors study the distribution of primes, applying very classical tools of Analytic Number Theory; esp., \[ \pi(x):=|\{ p\leq x \}|\,, \] i.e. the number of primes up to \(x\) (\(\to \infty\)), where \(p\) always denotes a prime; the von Mangoldt function \(\Lambda(n)\) which is (a weighted counting function of primes and) defined as \(\log p\) for all \(n=p^r\) (\(r>0\) an integer), zero otherwise, contained (implicitly/explicitly) in the two sums \[ \theta(x):=\sum_{p\leq x}\log p, \quad \psi(x):=\sum_{n\leq x}\Lambda(n), \] i.e., the Chebyshev functions; the Riemann zeta function, namely \[ \zeta(s):=\sum_{n=1}^{\infty}n^{-s}, \quad \text{Re}(s)>1 \] (and then extended by analytic continuation, etc.: see the 8-pages memoir of \textit{B. Riemann} himself [Mitt. Königl. Preuss. Akad. Wiss. Berlin 1859, 671--680 (1860), an English translation can be found as appendix to {\textit{H. M. Edwards}' book ``Riemann's zeta-function. London: Academic Press (1974; Zbl 0315.10035)] and, then, \textit{A. Ivić}'s book on it [``The Riemann Zeta Function. New York etc.: John Wiley (1985; Zbl 0556.10026)]). The most famous Riemann Hypothesis (RH) is the (still unproven) statement: all complex (non-real) zeros of \(\zeta\) have real part \(1/2\). First of all, starting from Gauss and the Prime Number Theorem (one form is: \(\psi(x)\sim x\), as \(x\to \infty\)): \[ \text{li}(x):=\int_{2}^{x}{1\over {\log t}}\,dt \] i.e., the Logarithmic Integral (sometimes written \(\text{Li}(x)\), instead: there's a typo in the paper), has been used as a good approximation to \(\pi(x)\), even after the discovery that say \(\varepsilon(x):=\text{li}(x)-\pi(x)\) is oscillating in sign (after the very big Skewes number, see his paper [\textit{S. Skewes}, J. Lond. Math. Soc. 8, 277--283 (1933; Zbl 0007.34003, JFM 59.0370.02)]). By the way, \textit{G. Robin} [Ann. Fac. Sci. Toulouse, V. Sér., Math. 6, 257--268 (1984; Zbl 0543.10032)] has found the equivalence between RH and the positivity of \(\varepsilon_{\theta(x)}:=\text{li}(\theta(x))-\pi(x)\); and the authors use this to prove RH is equivalent to \(\varepsilon_{\psi(x)}:=\text{li}(\psi(x))-\pi(x)>0\) (see Theorem 1 \& Corollary 2 in the paper). Many variations on the theme of jumps at primes, for different functions (Chebyshev ones, but not only) is then the main theme of the paper, with a wealth of Conjectures and ``experiments, for the heuristic side of this very interesting subject. Last but not least, a more efficient formula for prime-counting (namely, a good approximation to \(\pi(x)\) function) with respect to the Riemann original one (see the paper, consult the classic memoir: it's linked to Riemann-von Mangoldt explicit formula, i.e. (3) in the paper) is derived by the authors who dedicate to the ``experimental, so to speak, tests (see the tables of primes completing the paper).}

Recommendations

Cites work

Cited in

(5)

Describes a project that uses

Uses Software

OEIS

This page was built for publication: Efficient prime counting and the Chebyshev primes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2249928)