Prime zeta function

In mathematics, the prime zeta function is an analogue of the Riemann zeta function, studied by Glaisher (1891). It is defined as the following infinite series, which converges for ⁠ℜ(s)>1⁠:

P(s)=∑p∈primes1ps=12s+13s+15s+17s+111s+… .

Properties

The Euler product for the Riemann zeta function ζ(s) implies that

log⁡ζ(s)=∑n>0P(ns)n,

which by Möbius inversion gives

P(s)=∑n>0μ(n)log⁡ζ(ns)n

When s goes to 1, we have ⁠P(s)∼log⁡ζ(s)∼log⁡(1s−1)⁠. This is used in the definition of Dirichlet density.

This gives the continuation of P(s) to ⁠ℜ(s)>0⁠, with an infinite number of logarithmic singularities at points s where ns is a pole (only ns=1 when n is a squarefree number greater than or equal to 1), or zero of the Riemann zeta function ζ(.). The line ℜ(s)=0 is a natural boundary as the singularities cluster near all points of this line.

If one defines a sequence

an=∏pk∣n1k=∏pk∣∣n1k!

then

P(s)=log⁡∑n=1∞anns.

(Exponentiation shows that this is equivalent to Lemma 2.7 by Li.)

The prime zeta function is related to Artin's constant by

ln⁡CArtin=−∑n=2∞(Ln−1)P(n)n

where Ln is the ⁠n⁠th Lucas number.

Specific values are:

sapproximate value P(s)OEIS
112+13+15+17+111+⋯→∞
20.45224 74200 41065 49850…OEIS: A085548
30.17476 26392 99443 53642…OEIS: A085541
40.07699 31397 64246 84494…OEIS: A085964
50.03575 50174 83924 25713…OEIS: A085965
60.01707 00868 50636 51295…OEIS: A085966
70.00828 38328 56133 59253…OEIS: A085967
80.00406 14053 66517 83056…OEIS: A085968
90.00200 44675 74962 45066…OEIS: A085969

Analysis

Integral

The integral over the prime zeta function is usually anchored at infinity, because the pole at s=1 prohibits defining a nice lower bound at some finite integer without entering a discussion on branch cuts in the complex plane:

∫s∞P(t)dt=∑p1pslog⁡p

The noteworthy values are again those where the sums converge slowly:

sapproximate value ∑p1/(pslog⁡p)OEIS
11.63661632…OEIS: A137245
20.50778218…OEIS: A221711
30.22120334…
40.10266547…

Derivative

The first derivative is

P′(s)≡ddsP(s)=−∑plog⁡pps

The interesting values are again those where the sums converge slowly:

sapproximate value P′(s)OEIS
2−0.493091109…OEIS: A136271
3−0.150757555…OEIS: A303493
4−0.060607633…OEIS: A303494
5−0.026838601…OEIS: A303495

Generalizations

Almost-prime zeta functions

As the Riemann zeta function is a sum of inverse powers over the integers and the prime zeta function a sum of inverse powers of the prime numbers, the k-primes (the integers that are a product of k not necessarily distinct primes) define a sort of intermediate sums:

Pk(s)≡∑n:Ω(n)=k1ns,

where Ω is the total number of prime factors.

ksapproximate value Pk(s)OEIS
220.14076043434…OEIS: A117543
230.02380603347…
320.03851619298…OEIS: A131653
330.00304936208…

Each integer in the denominator of the Riemann zeta function ζ may be classified by its value of the index ⁠k⁠, which decomposes the Riemann zeta function into an infinite sum of the ⁠Pk⁠:

ζ(s)=1+∑k=1,2,…Pk(s)

Since we know that the Dirichlet series (in some formal parameter ⁠u⁠) satisfies

PΩ(u,s):=∑n≥1uΩ(n)ns=∏p∈ℙ(1−up−s)−1,

we can use formulas for the symmetric polynomial variants with a generating function of the right-hand-side type. Namely, we have the coefficient-wise identity that Pk(s)=[uk]PΩ(u,s)=h(x1,x2,x3,…) when the sequences correspond to xj:=j−sχℙ(j) where χℙ denotes the characteristic function of the primes. Using Newton's identities, we have a general formula for these sums given by

Pn(s)=∑k1+2k2+⋯+nkn=nk1,…,kn≥0[∏i=1nP(is)kiki!⋅iki]=−[zn]log⁡(1−∑j≥1P(js)zjj).

Special cases include the following explicit expansions:

P1(s)=P(s)P2(s)=12(P(s)2+P(2s))P3(s)=16(P(s)3+3P(s)P(2s)+2P(3s))P4(s)=124(P(s)4+6P(s)2P(2s)+3P(2s)2+8P(s)P(3s)+6P(4s)).

Prime modulo zeta functions

Constructing the sum not over all primes but only over primes that are in the same modulo class introduces further types of infinite series that are a reduction of the Dirichlet L-function.

See also

References

  1. ↑ Weisstein, Eric W. "Artin's Constant". MathWorld.
  2. ↑ See Divergence of the sum of the reciprocals of the primes.