CUBO A Mathematical Journal

Vol.10, N
o
¯ 03, (133–136). October 2008

The Fibonacci Zeta-Function is Hypertranscendental

Jörn Steuding

Department of Mathematics, Würzburg University,

Am Hubland, 97 218 Würzburg, Germany

email: steuding@mathematik.uni-wuerzburg.de

ABSTRACT

Applying a theorem of Reich on Dirichlet series satisfying difference-differential equa-

tions, we show that the Fibonacci zeta-function satisfies no algebraic differential equa-

tion.

RESUMEN

Aplicando el Teorema de Reich sobre series de Dirichlet satisfaziendo ecuaciones diferen-

ciales-diferencias, nosotros mostramos que la función zeta de Fibonacci satisfaze una

ecuación diferencial no algebraica.

Key words and phrases: Hypertranscendence, Fibonacci zeta-function.

Math. Subj. Class.: 11B39, 11M41, 34M15.



134 Jörn Steuding CUBO
10, 3 (2008)

1 Introduction

The Fibonacci numbers are recursively defined by

F0 = 0, F1 = 1 and Fn+2 = Fn+1 + Fn for n ∈ N.

In number theory one can often obtain arithmetic information by studying a generating function of

a given number theoretical object. In the case of Fibonacci numbers this is usually the correspond-

ing Lambert series; however, in the recent past also the generating Dirichlet series was studied; this

function is more interesting with respect to its analytic properties. Let s be a complex variable.

For Re s > 0 the Fibonacci zeta-function is defined by

ζF(s) =
∑

n∈N

F −sn ,

and by analytic continuation throughout the complex plane except for simple poles at s = −2k +

πi(2n + k)/ log ϕ for n ∈ Z, k ∈ N0, where ϕ is the golden ratio; this was first proved by Navas

[6] (and relies mainly on Binet’s formula). In [1], Elsner et al. obtained several results on the

algebraic independence of the values taken by ζF on the positive integers, e.g. ζF(2), ζF(4), ζF(6)

are algebraically independent.

In this note we show that the Fibonacci zeta-function ζF(s) is hypertranscendental, i.e., it

satisfies no non-trivial algebraic differential equation (that is no finite collection of derivatives of

ζF) is algebraically dependent over the field of rational functions). Actually, we shall prove a

slightly stronger statement by applying Reich’s theorem on Dirichlet series satisfying holomorphic

difference–differential equations. In order to state this result denote by D the set of all ordinary

Dirichlet series f (s) =
∑

∞

n=1
ann

−s satisfying the following two assumptions:

• the abscissa of absoulte convergence is finite: σa(f ) < ∞,

• the set of all divisors of indices n with an 6= 0 contains infinitely many prime numbers.

Furthermore, we introduce the following abbreviation: for a non-negative integer ν we write

f
[ν](s) = (f (s), f ′(s), . . . , f (ν)(s)).

Reich [9] proved the following theorem: Assume that f ∈ D. Let h0 < h1 < . . . < hm be

any real numbers, ν0, ν1, . . . , νm be any non-negative integers, and let σ0 > σa(f ) − h0. Put

k :=
∑m

j=0
(νj + 1). If Φ : C

k
→ C is continuous and the difference-differential equation

Φ(f [ν0](s + h0), f
[ν1](s + h1), . . . , f

[νm](s + hm)) = 0

holds for all s with Re s > σ0, then Φ vanishes identically. To apply this result to the Fibonacci

zeta-function it suffices to show that the set of all Fibonacci numbers Fn is not generated by finitely



CUBO
10, 3 (2008)

The Fibonacci Zeta-Function is Hypertranscendental 135

many primes. However, this follows immediately from Lucas’ theorem

gcd(Fm, Fn) = Fgcd(m,n)

(see [5]), since the right-hand side is equal to F1 = 1 for any pair m, n of relatively coprime integers.

Thus we obtain:

Theorem 1. Given any real numbers h0 < h1 < . . . < hm, any non-negative integers ν0, ν1, . . . , νm,

and any σ0 > −h0, if Φ : C
k
→ C is continuous and the difference-differential equation

Φ(ζF
[ν0](s + h0), ζF

[ν1](s + h1), . . . , ζF
[νm](s + hm)) = 0

holds for all s with Re s > σ0, then Φ vanishes identically.

Notice that the proof does not use the meromorphic continuation of ζF(s) to C, obtained by Navas.

The statement of the theorem can easily be extended to other Dirichlet series built from linear

recursive sequences. Here we only need that such sequences are divisible by infinitely many prime

numbers which is true except for degenerate cases when the characteristic polynomial has two roots

whose quotient is a root of unity; since roots are counted with multiplicities, this also includes the

case of repeated roots. This was first shown by Pólya [8] and has been rediscovered by several

mathematicians (see [2, 10] for some history).

We conclude with a few historical remarks on hypertranscendence and an interesting question.

In 1887, Hölder [4] proved that the Gamma-function is hypertranscendental. In his challenging

lecture at the International Congress for Mathematicians in Paris 1900, Hilbert [3] asked in problem

18 for a description of classes of functions definable by differential equations. In this context

Hilbert stated that the Riemann zeta-function ζ(s) is hypertranscendental; the first published proof

was written down by Stadigh in his dissertation (cf. Ostrowski [7]). The idea is to deduce the

hypertranscendence of ζ(s) from Hölder’s theorem and the fact that the Gamma-function appears

in the functional equation for zeta. Besides, Hilbert [3] asked for a proof of the hypertranscendence

for the more general series
∑

∞

n=1
xnn−s. This problem was solved by Ostrowski [7] as a particular

case of a more general theorem which also applies to the case when there is no functional equation

at hand; his argument relies on a comparison of the differential independence with the linear

independence of its frequencies. Reich’s theorem [9], which we have used to prove Theorem 1,

may be regarded as the most general and powerful extension of this method. A different way for

proving hypertranscendence was found by Voronin. In [11], he developped a new technique to

study the joint value distribution of Dirichlet L-functions to pairwise inequivalent characters and

their derivatives; in [12], he extended the method in order to prove his famous universality theorem

for the Riemann zeta-function: Let 0 < r < 1
4

and suppose that g(s) is a non-vanishing continuous

function on the disk |s| ≤ r which is analytic in the interior. Then, for any ǫ > 0,

lim inf
T →∞

1

T
meas

{

τ ∈ [0, T ] : max
|s|≤r

∣

∣ζ
(

s + 3
4

+ iτ
)

− g(s)
∣

∣ < ǫ

}

> 0.



136 Jörn Steuding CUBO
10, 3 (2008)

Voronin’s results imply the hypertranscendence of these Dirichlet series. It is natural to ask whether

the Fibonacci zeta-function shares this or some other universality property: is it true or false that

any (suitable) analytic function g(s) can be uniformly approximated by certain shifts of ζF(s)?

Acknowledgements. The author wants to thank the anonymous referee for her or his remarks

and references to clarify to which extent the statement of the theorem can be generalized.

Received: April 2008. Revised: June 2008.

References

[1] C. Elsner, S. Shimomura and I. Shiokawa, Algebraic relations for reciprocal sums of
Fibonacci numbers, Acta Arith., 130 (2007), 37–60.

[2] G. Everest, A. van der Poorten, I. Shparlinski and T. Ward, Recurrence sequences,
AMS Mathematical Surveys and Monographs, 104 (2003).

[3] D. Hilbert, Mathematische Probleme, Archiv f. Math. u. Physik, 1 (1901), 44–63, 213–317.

[4] O. Hölder, Über die Eigenschaft der Gammafunktion keiner algebraischen Differentialgle-
ichung zu genügen, Math. Ann., 28 (1887), 1–13.

[5] T. Koshy, Fibonacci and Lucas Numbers with applications, John Wiley & Sons, New York
2001.

[6] L. Navas, Analytic continuation of the Fibonacci Dirichlet series, Fibonacci Q., 39 (2001),
409–418.

[7] A. Ostrowski, Über Dirichletsche Reihen und algebraische Differentialgleichungen, Math.
Z., 8 (1920), 241–298.

[8] G. Pólya, Arithmetische Eigenschaften der Reihenentwicklungen rationaler Funktionen, J.
Reine Angew. Math., 151 (1921), 1–31.

[9] A. Reich, Über Dirichletsche Reihen und holomorphe Differentialgleichungen, Analysis, 4
(1984), 27–44.

[10] H. Roskam, Prime divisors of linear recurrences and Artin’s primitive root conjecture for
number fields, J. Théo. Nombres Bordeaux, 13 (2001), 303–314.

[11] S.M. Voronin, On the functional independence of Dirichlet L-functions, Acta Arith., 27
(1975), 493–503 (in Russian).

[12] S.M. Voronin, Theorem on the ‘universality’ of the Riemann zeta-function, Izv. Akad. Nauk
SSSR, Ser. Matem., 39 (1975) (in Russian); engl. translation in Math. USSR Izv., 9 (1975),
443–445.


	N11