Journal de Théorie des Nombres de Bordeaux

Joshua Harrington

Andrew Vincent

Daniel White

The factorization of $f(x)x^n+g(x)$ with $f(x)$ monic and of degree $\le 2$.

Clemens Heuberger

Daniel Krenn

Optimality of the Width-$w$ Non-adjacent Form: General Characterisation and the Case of Imaginary Quadratic Bases

Gabriele Nebe

Elisabeth Nossek

Boris Venkov

Low dimensional strongly perfect lattices.  II: Dual strongly perfect lattices of dimension 13 and 15.

Shahed Sharif

A descent map for curves with totally degenerate semi-stable reduction

David Masser

Umberto Zannier

Bicyclotomic polynomials and impossible intersections

Christoph Aistleitner

On the limit distribution of the well-distribution measure of random binary sequences

Noam D. Elkies

Daniel M. Kane

Scott Duke Kominers

Minimal $\mathcal{S}$-universality criteria may vary in size

Ghassan Sarkis

Joel Specter

Galois extensions of height-one commuting dynamical systems

Karin Halupczok

Goldbach’s problem with primes in arithmetic progressions and in short intervals

Jörn Steuding

Jörn STEUDING
Ergodic Universality Theorems for the Riemann Zeta-Function and other
L-Functions
Tome 25, no 2 (2013), p. 471-476.
<http://jtnb.cedram.org/item?id=JTNB_2013__25_2_471_0>
© Société Arithmétique de Bordeaux, 2013, tous droits réservés.
L’accès aux articles de la revue « Journal de Théorie des Nom-
bres de Bordeaux » (http://jtnb.cedram.org/), implique l’accord
avec les conditions générales d’utilisation (http://jtnb.cedram.
org/legal/). Toute reproduction en tout ou partie de cet article sous
quelque forme que ce soit pour tout usage autre que l’utilisation à
fin strictement personnelle du copiste est constitutive d’une infrac-
tion pénale. Toute copie ou impression de ce fichier doit contenir la
présente mention de copyright.
cedram
Article mis en ligne dans le cadre du
Centre de diffusion des revues académiques de mathématiques
http://www.cedram.org/
Journal de Théorie des Nombres
de Bordeaux 25 (2013), 471–476
Ergodic Universality Theorems for the Riemann
Zeta-Function and other L-Functions
par Jörn STEUDING
Résumé. Nous prouvons un nouveau type de théorème d’univer-
salité pour la fonction zêta de Riemann et d’autres fonctions L (qui
sont universelles au sens du théorème de Voronin). Contrairement
aux théorèmes d’universalité précédents pour la fonction zêta ou
ses généralisations diverses, ici les approximations sont obtenues
à partir de l’orbite d’une transformation ergodique sur la droite
réelle.
Abstract. We prove a new type of universality theorem for
the Riemann zeta-function and other L-functions (which are uni-
versal in the sense of Voronin’s theorem). In contrast to previous
universality theorems for the zeta-function or its various general-
izations, here the approximating shifts are taken from the orbit of
an ergodic transformation on the real line.
1. Motivation and Statement of The Main Theorem
In 1975, Voronin [7] proved that the Riemann zeta-function ζ(s) pos-
sesses the following remarkable approximation property: Let 0 < r < 14
and suppose that f(s) is a continuous function on the disk |s| ≤ r which is
analytic and non-vanishing in the interior of the disk. Then, for any ε > 0,
there exists a real number τ such that
max
|s|≤r
|ζ(s+ 34 + iτ)− f(s)| < ε;
moreover, his proof showed that the set of such τ has positive lower density
with respect to Lebesgue measure. This so-called universality theorem has
been extended and generalized in various ways. For instance, Voronin [8]
obtained a simultaneous approximation theorem for Dirichlet L-functions
associated with non-equivalent characters; here two characters are said to be
non-equivalent if they are not induced by the same character. The precise
formulation of this so-called joint universality theorem is as follows: Let
χ1, . . . , χm be pairwise non-equivalent Dirichlet characters, K1, . . . ,Km be
compact subsets of the strip 12 < Re s < 1 with connected complements.
Further, for each j = 1, . . . ,m, let fj(s) be a continuous function on Kj
472 Jörn Steuding
which is analytic and non-vanishing in the interior of Kj. Then, for any
ε > 0,
lim inf
T→∞
1
T
meas
{
τ ∈ [0, T ] : max
1≤j≤m
max
s∈Kj
|L(s+ iτ, χj)− fj(s)| < ε
}
> 0.
Further examples of universal zeta- and L-functions are Dedekind zeta-
functions ζK(s) to a number field K, L-functions associated with modular
forms, and certain Hurwitz zeta-functions, to mention just a few exam-
ples. Reich [5] succeeded in proving so-called discrete universality theorems
where the approximating shifts are taken from an arithmetic progression,
e.g.,
lim inf
N→∞
1
N
]
{
n ∈ N ∩ [1, N ] : max
s∈K
|ζK(s+ i∆n)− f(s)| < ε
}
> 0
for suitable target functions f defined on suitable sets K; here ∆ is an arbi-
trary fixed real number different from zero and ]M denotes the cardinality
of the discrete setM. For more information we refer to [5, 6].
Given a domain D we call a family of analytic functions L1, . . . ,Lm a
jointly universal family with respect to D if, for any collection of compact sub-
sets K1, . . . ,Km of D with connected complements, any family of continuous
functions fj defined on Kj , each of which is analytic and non-vanishing in
the interior of Kj , and any ε > 0, there exists some real number τ > 0 such
that
(1.1) max
1≤j≤m
max
s∈Kj
|Lj(s+ iτ)− fj(s)| < ε.
We shall say that such a family is strongly jointly universal with respect to
D if the assumption on the non-vanishing of the target functions can be
dropped (as, for example, for certain Hurwitz zeta-functions; see [3]).
In this note we shall investigate the phenomenon of universality on orbits
of certain ergodic transformations. In order to state the main result we first
recall some basic notions from ergodic theory.
Given a probability space (X,F ,P) consisting of a non-empty space X,
an associated σ-algebra F , and a probability measure P on F , a measurable
transformation T : X → X is said to be measure preserving with respect to
P if P(T−1A) = P(A) for all A ∈ F ; it is said to be ergodic with respect to
P if for any measurable set A with T−1A = A either P(A) = 0 or P(A) = 1
holds; in this case (X,F ,P,T) is called an ergodic dynamical system. For
our purpose we shall consider probability spaces with X = R and F as the
Borel σ-algebra.
We introduce some further notion to abbreviate the formulation of the
main results. Given a domain D, we call a family of analytic functions
L1, . . . ,Lm a jointly ergodic universal family with respect to D if, for any
Ergodic Universality Theorems for the Riemann Zeta-Function 473
collection of compact subsets K1, . . . ,Km of D with connected comple-
ments, any family of continuous functions fj defined on Kj which is ana-
lytic and non-vanishing in the interior of Kj , any ergodic dynamical system
(R,F ,P,T), almost all real numbers x, and any positive ε, there exists a
positive integer n such that
(1.2) max
1≤j≤m
max
s∈Kj
|Lj(s+ iTnx)− fj(s)| < ε
holds. Here and in the sequel we use the abbreviation Tx for T(x) and Tnx
is recursively defined by T ◦ Tn−1x and T0x = x; the notion for almost all
x is an abbreviation for all real numbers except a set of P-measure zero.
We call a family of analytic functions L1, . . . ,Lm a strongly jointly ergodic
universal family with respect to D if the assumption on the non-vanishing of
the target functions fj can be dropped.
Theorem 1.1. Let (R,F ,P,T) be an ergodic dynamical system where P
is a probability measure with a positive density function. Then a family of
L-functions is jointly (strongly) universal with respect to some domain D
if, and only if, it is jointly (strongly) ergodic universal with respect to D;
in this case,
lim inf
N→∞
1
N
]
{
n ∈ N ∩ [1, N ] : max
1≤j≤m
max
s∈Kj
|Lj(s+ iTnx)− fj(s)| < ε
}
> 0.
It may be noticed that although we do not assume that the approxima-
tion problem (1.1) is solvable on a set of τ of positive lower density (as in
Voronin’s theorem) already a positive proportion of the points of the er-
godic orbit fulfill (1.2). This is related to the reccurence property of ergodic
orbits due to Poincaré’s theorem (see [4]); it also indicates that the question
whether (1.1) holds for a set of positive lower density is not directly related
to (1.2).
We give an explicit example of an ergodic dynamical system (R,F ,P,T)
which satisfies the conditions of the theorem. The underlying ergodic trans-
formation on R has been found by Lind (cf. [4]). Consider the mapping
T : R → R given by T0 := 0 and Tx := 12(x −
1
x) for x 6= 0. It is easy
to see that T is an ergodic transformation with respect to the probability
measure P defined by
(1.3) P((α, β)) = 1
π
∫
(α,β)
dτ
1 + τ2 ;
for details we refer to [4], Example 2.9.
2. Proof of Theorem 1.1
Assume we are given a domain D and a jointly universal family of ana-
lytic functions L1, . . . ,Lm with respect to D; i.e., for any compact subsets
474 Jörn Steuding
K1, . . . ,Km of D with connected complements, any family of continuous
functions fj defined on Kj , each of which is analytic and non-vanishing in
the interior of Kj , and any ε > 0, there exists some τ > 0 satisfying
(2.1) max
1≤j≤m
max
s∈Kj
|Lj(s+ iτ)− fj(s)| < ε/2.
Since T is ergodic, for any positive δ and for almost all real x, there exists
some positive integer n such that
(2.2) |Tnx− τ | < δ.
In order to see this we define the indicator function 1A of the measurable
set A by
1A(ξ) =
{
1 if ξ ∈ A,
0 otherwise.
Now recall the Birkhoff-Khintchine Ergodic Theorem: Let T be an ergodic
transformation on a probability space (X,F ,P); if f is integrable with re-
spect to (X,F ,P), then, for almost all x ∈ X,
lim
N→∞
1
N
∑
0≤n<N
f(Tnx) =
∫
X
f dP;
i.e., the limit exists and equals the right-hand side (for a proof see [4])).
Applying this with the indicator function of the open interval A = (τ −
δ, τ + δ) yields
(2.3) lim
N→∞
1
N
∑
0≤n<N
1A(Tnx) =
∫
R
1A dP = P(A)
for almost all x. By assumption, P possesses a positive density function, g
say; hence,
P(A) =
∫ τ+δ
τ−δ
g(t) dt
is positive, which proves the existence of a positive integer n satisfying
(2.2).
Continuous functions on compact sets are in particular uniformly contin-
uous. In our situation the compact set is given by Kj + iτ . Hence, it follows
that
max
1≤j≤m
max
s∈Kj
|Lj(s+ iTnx)− Lj(s+ iτ)| < ε/2
if in (2.2) δ is chosen sufficiently small. The latter inequality in combination
with (2.1) shows
(2.4) max
1≤j≤m
max
s∈Kj
|Lj(s+ iTnx)− fj(s)| < ε.
Ergodic Universality Theorems for the Riemann Zeta-Function 475
It remains to prove that this approximation holds on a set of n of positive
density. However, this follows already from (2.3) and the positivity of P.
More precisely, we note that the ergodic lower density is bounded below by
lim inf
N→∞
1
N
∑
0≤n<N
1A(Tnx) ≥ P(A) > 0,
where A = {τ ∈ R+ : Inequality (2.4) is true }, as follows from inserting
this set in place of the open interval in (2.3).
The converse implication is trivial (by setting τ = Tnx for some appro-
priate n). Hence, Theorem 1.1 is proved. 
3. Concluding Remarks
Reviewing the proof of the theorem, it appears that the reasoning for the
first implication applies to a family of L-functions which satisfies a discrete
joint universality theorem too; the first example for this type of universality
was given by Bagchi [1]. The converse implication, however, does not need
to be true in general.
Neither Voronin’s universality theorem nor any of its generalizations is
effective in the sense that an approximating shift τ can be bounded above or
below by some non-trivial explicit quantity (apart from a remarkable varia-
tion of Garunkštis [2] which applies to a restricted class of target functions).
It may be noticed, however, that the expectation that a sequence Tnx vis-
its a given subset (as, for example, the set A at the end of the proof of
Theorem 1.1) can be computed explicitly from Kac’s theorem (which is a
quantitative version of Poincaré’s recurrence theorem; see [4]).
It is expected that universality itself is an ergodic phenomenon. Unfor-
tunately, our results do not shed any new light on this conjecture. Our
reasoning is just based on continuity and basic results from ergodic the-
ory, nevertheless, it leads to a rather astonishing result. We interpret this
as another incident for the deepness and beauty of Voronin’s universality
theorem and its generalizations.
References
[1] B. Bagchi, A joint universality theorem for Dirichlet L-functions. Math. Z. 181 (1982),
319–334.
[2] R. Garunkštis, The effective universality theorem for the Riemann zeta-function, in: ‘Spe-
cial activity in Analytic Number Theory and Diophantine equations’. Proceedings of a work-
shop at the Max Planck-Institut Bonn 2002, R.B. Heath-Brown and B. Moroz (eds.), Bonner
math. Schriften 360 (2003).
[3] R. Garunkštis, A. Laurinčikas, The Lerch zeta-function. Kluwer Academic Publishers,
Dordrecht 2002.
[4] Geon Ho Choe, Computational Ergodic Theory. Springer 2005.
[5] A. Reich, Wertverteilung von Zetafunktionen. Arch. Math. 34 (1980), 440–451.
[6] J. Steuding, Value-Distribution of L-Functions. Lecture Notes in Mathematics, Vol. 1877,
Springer, 2007.
476 Jörn Steuding
[7] S.M. Voronin, Theorem on the ’universality’ of the Riemann zeta-function. Izv. Akad.
Nauk SSSR, Ser. Matem. 39 (1975), 475–486 (in Russian); engl. translation in Math. USSR
Izv. 9 (1975), 443–453.
[8] S.M. Voronin, On the functional independence of Dirichlet L-functions. Acta Arith. 27
(1975), 493–503 (in Russian).
Jörn Steuding
University of Würzburg
Emil-Fischer-Str. 40
97074 Würzburg, Germany
E-mail: steuding@mathematik.uni-wuerzburg.de


Ergodic Universality Theorems for the Riemann Zeta-Function and other $L$-Functions

Not a member yet

The factorization of $f(x)x^n+g(x)$ with $f(x)$ monic and of degree $\le 2$ .

Optimality of the Width- $w$ Non-adjacent Form: General Characterisation and the Case of Imaginary Quadratic Bases

Low dimensional strongly perfect lattices. II: Dual strongly perfect lattices of dimension 13 and 15.

A descent map for curves with totally degenerate semi-stable reduction

Bicyclotomic polynomials and impossible intersections

On the limit distribution of the well-distribution measure of random binary sequences

Minimal $\mathcal{S}$ -universality criteria may vary in size

Galois extensions of height-one commuting dynamical systems

Goldbach’s problem with primes in arithmetic progressions and in short intervals

Ergodic Universality Theorems for the Riemann Zeta-Function and other $L$ -Functions

1,235

1,351