Abstract
In the paper, an absolutely convergent Dirichlet series whose shifts approximate a wide class of analytic functions is constructed. This series is close in the mean to the Riemann zeta-function.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Let \(\zeta (s)\), \(s=\sigma +it\), be the Riemann zeta-function, i.e., for \(\sigma >1\),
The function \(\zeta (s)\) has an analytic continuation to the whole complex plane, except for a simple pole at the point \(s=1\) with residue 1. It is well known that the function \(\zeta (s)\) has a universality property discovered by Voronin [13] on the approximation of a wide class of analytic functions by shifts \(\zeta (s+i\tau )\), \(\tau \in {\mathbb {R}}\). Let \(D=\{ s\in {\mathbb {C}}: 1/2<\sigma <1\}\). Denote by \({\mathcal {K}}\) the class of compact subsets of the strip D with connected complements, and by \(H_0(K)\) with \(K\in {\mathcal {K}}\) the class of continuous non-vanishing functions on K that are analytic in the interior of K. Then the last version of the Voronin theorem, see, for example, [5], says that, for every \(K\in {\mathcal {K}}\), \(f(s)\in H_0(K)\), and \(\varepsilon >0\),
The latter inequality shows that there are infinitely many shifts \(\zeta (s+i\tau )\) that approximate uniformly on K a given function \(f(s)\in H_0(K)\) with accuracy \(\varepsilon >0\). On the other hand, no concrete such shift is known. In [7] and [9], it was obtained that “\(\liminf \)” in (1.1) can be replaced by “\(\lim \)” for all but at most countably many \(\varepsilon >0\).
Some other zeta-functions and their classes are universal in the above sense as well, see [1, 3, 6, 11] and the very informative survey paper [8].
The Riemann zeta-function \(\zeta (s)\), for \(s\in D\), cannot be written as a convergent Dirichlet series, and is defined by analytic continuation. The aim of this note is to present a certain absolutely convergent Dirichlet series in D having an approximation property similar to that of the function \(\zeta (s)\). Obviously, an absolutely convergent Dirichlet series cannot be universal. Therefore, such a series must be close in a certain sense to \(\zeta (s)\).
For \(u>0\), let \(\theta >1/2\) be a fixed number and, for \(m\in {\mathbb {N}}\),
where \(\exp \{a\}=\mathrm {e}^a\). Define the series
It will be proved below that the latter series is absolutely convergent in the half plane \(\sigma >1/2\), thus, for \(s\in D\) as well.
For the precise statement of the approximation theorem for the function \(\zeta _u(s)\), we need one particular topological group. Let \(\gamma =\{s\in {\mathbb {C}}: |s|=1\}\), \({\mathbb {P}}\) denote the set of all prime numbers, and
where \(\gamma _p=\gamma \) for all \(p\in {\mathbb {P}}\). By the Tikhonov theorem, the infinite-dimensional torus \(\Omega \) with pointwise multiplication and the product topology is a compact topological Abelian group. Therefore, on \((\Omega , {\mathcal {B}}(\Omega ))\) (\({\mathcal {B}}({\mathbb {X}})\) is the Borel \(\sigma \)-field of the space \({\mathbb {X}}\)), the probability Haar measure \(m_H\) can be defined. This leads to the probability space \((\Omega , {\mathcal {B}}(\Omega ), m_H)\). Denote by H(D) the space of analytic functions on D endowed with the topology of uniform convergence on compacta, by \(\omega (p)\) the pth component of an element \(\omega \in \Omega \), \(p\in {\mathbb {P}}\), and on the probability space \((\Omega , {\mathcal {B}}(\Omega ), m_H)\), define the H(D)-valued random element
Note that the latter infinite product converges uniformly on compact subsets of the strip D for almost all \(\omega \in \Omega \), see [1] or [5].
The main result of the paper is the following theorem.
Theorem 1.1
Suppose that \(n_T\rightarrow \infty \) and \(n_T\ll T^2\) as \(T\rightarrow \infty \). Let \(K\in {\mathcal {K}}\) and \(f(s)\in H_0(K)\). Then the limit
exists for all but at most countably many \(\varepsilon >0\).
Theorem 1.1 implies that there exists \(T_0=T_0(f,K, \varepsilon )>0\) such that, for every \(T\ge T_0\), there are infinitely many shifts \(\zeta _{n_T}(s+i\tau )\) approximating a given function f(s).
2 The function \(\zeta _u(s)\)
Denote by \(\Gamma (s)\) the Euler gamma-function, and define
where \(\theta >1/2\) is from the definition of \(v_u(m)\). For convenience, recall some properties of \(\Gamma (s)\).
Lemma 2.1
For arbitrary \(\sigma _1<\sigma _2\), there exists \(c>0\) such that the estimate
holds uniformly in \(\sigma \), \(\sigma _1\le \sigma \le \sigma _2\).
Proof of the lemma can be found, for example, in [4].
Lemma 2.2
For positive a and b,
The equality of the lemma is the classical Mellin formula, for the proof, see, for example, [12].
Lemma 2.3
The series for \(\zeta _u(s)\) is absolutely convergent for \(\sigma >1/2\). Moreover, the equality
is valid.
Proof
In view of Lemma 2.2,
Therefore, by Lemma 2.1,
Since \(\theta > 1/2\), this shows the absolute convergence in the half-plane \(\sigma >1/2\) of the series for \(\zeta _u(s)\).
Now, using (2.1) and the definition of \(l_u(s)\), we find
\(\square \)
3 Mean distance between \(\zeta _{n_T}(s)\) and \(\zeta (s)\)
For \(n_T\) sufficiently large, the coefficients of the series for \(\zeta _{n_T}(s)\) are close to 1. This suggests that the function \(\zeta _{n_T}(s)\) is close to \(\zeta (s)\) even in the strip D. Actually, those two functions are close in the mean.
Lemma 3.1
Suppose that \(n_T\rightarrow \infty \) and \(n_T\ll T^2\) as \(T\rightarrow \infty \). Then, for every compact set \(K\subset D\),
Proof
By Lemma 2.3, for \(s\in D\), we have
Let \(0<{\hat{\theta }}<2\theta \). Then (3.1), properties of the function \(\zeta (s)\), and the residue theorem imply
Denote the points of the set K by \(s=\sigma +iv\). Then there exists \(\varepsilon >0\) such that \(1/2+2\varepsilon \le \sigma \le 1-\varepsilon \) for all \(s\in K\). Taking
we obtain from (3.2) that, for all \(s\in K\),
Hence, putting t in place of \(v+t\) gives
Therefore, in view of Lemma 2.1,
where
and
It is well known that, for fixed \(\sigma \), \(1/2<\sigma <1\),
Therefore, for all \(t\in {\mathbb {R}}\) and \(T\ge 1\),
Moreover, taking into account Lemma 2.1, we find that, for all \(s\in K\),
This and (3.4) show that
Similarly, we obtain that, for \(s\in K\),
Therefore,
Since \(n_T\ll T^2\), this, (3.3), and (3.5) prove the lemma. \(\square \)
4 Limit theorem
The inequality (1.1) and its modification with “\(\lim \)” are derived in [1, 5, 7] from a probabilistic limit theorem for measures in the space of analytic functions H(D). For \(A\in {\mathcal {B}}(H(D))\), define
and denote by \(P_\zeta \) the distribution of the H(D)-valued random element \(\zeta (s, \omega )\), i.e.,
Then the following statement is true [1, 5].
Lemma 4.1
\(P_T\) converges weakly to \(P_\zeta \) as \(T\rightarrow \infty \). Moreover, the support of the measure \(P_\zeta \) is the set \(S=\{g\in H(D): g(s)\ne 0\; or\; g(s)\equiv 0\}\).
Several equivalents of weak convergence of probability measures are known, see, for example, [2]. For us, the equivalent in terms of continuity sets is useful. Recall that a set \(A\in {\mathcal {B}}({\mathbb {X}})\) is called a continuity set of a measure P on \(({\mathbb {X}}, {\mathcal {B}}({\mathbb {X}}))\) if \(P(\partial A)=0\), where \(\partial A\) denotes the boundary of the set A.
Lemma 4.2
Suppose that \(P_n\), \(n\in {\mathbb {N}}\), and P are probability measures on \(({\mathbb {X}}, {\mathcal {B}}({\mathbb {X}}))\). Then \(P_n\) converges weakly to P as \(n\rightarrow \infty \) if and only if, for every continuity set A of P,
Proof of the lemma can be found, for example, in [2].
5 Proof of Theorem 1.1
We will apply a method of characteristic functions which is used in the theory of weak convergence of probability measures on \(({\mathbb {R}}, {\mathcal {B}}({\mathbb {R}}))\). We recall that every left continuous non-decreasing function F(x) on \({\mathbb {R}}\) such that \(F(+\infty )=1\) and \(F(-\infty )=0\) coincides with a certain distribution function. Note that the left continuity of distribution functions can be replaced by right continuity, however, for our aims the left continuity is more convenient. We say that the distribution function \(F_n(x)\), \(n\in {\mathbb {N}}\), converges weakly to a distribution function F as \(n\rightarrow \infty \) if \(\lim _{n\rightarrow \infty } F_n(x)= F(x)\) for every continuity point x of F(x).
Every distribution function F(x) is uniquely defined by its characteristic function
Moreover, the following classical continuity theorem is valid.
Lemma 5.1
Suppose that \(F_n(x)\), \(n\in {\mathbb {N}}\), and F(x) are distribution functions, and \(g_n(u)\) and g(u) are the corresponding characteristic functions. If \(F_n(x)\), as \(n\rightarrow \infty \), converges weakly to F(x), then \(\lim _{n\rightarrow \infty } g_n(u)= g(u)\), \(u\in {\mathbb {R}}\). This convergence is uniform in every finite interval. If \(\lim _{n\rightarrow \infty }g_n(u)=g(u)\), \(u\in {\mathbb {R}}\), where g(u) is a continuous function at \(u=0\), then there exists a distribution function F(x) such that \(F_n(x)\) converges weakly to F(x) as \(n\rightarrow \infty \). In this case, g(u) is the characteristic function of F(x).
For the proof of Theorem 1.1, the Mergelyan theorem on the approximation of analytic functions by polynomials [10] is also needed, see the next lemma.
Lemma 5.2
Suppose that \(K\subset {\mathbb {C}}\) is a compact set with connected complement, and g(s) is a continuous function in K and analytic in the interior of K. Then, for every \(\varepsilon >0\), there exists a polynomial p(s) such that
Proof of Theorem 1.1
By Lemma 4.1, \(P_T\) converges weakly to \(P_\zeta \) as \(T\rightarrow \infty \). Define the set
The boundaries \(\partial A_{\varepsilon _1}\) and \(\partial A_{\varepsilon _2}\) do not intersect for different positive \(\varepsilon _1\) and \(\varepsilon _2\). Therefore, the set \(A_\varepsilon \) is a continuity set of the measure \(P_\zeta \) for all but at most countably many \(\varepsilon >0\). This, the weak convergence of \(P_T\), and Lemma 4.2 imply the relation
for all but at most countably many \(\varepsilon >0\). Moreover, since by Lemma 4.1, the support of \(P_\zeta \) is the set S, the inequality
where p(s) is a polynomial and
is true. Lemma 5.2 ensures the choice of the polynomial p(s) satisfying
The latter inequality together with the definitions of the sets \(A_\varepsilon \) and \({\hat{A}}_\varepsilon \) implies the inclusion \({\hat{A}}_\varepsilon \subset A_\varepsilon \). Therefore, in view of (5.2),
By the definitions of \(P_T\), \(P_\zeta \), and \(A_\varepsilon \),
and
Thus, \(F_T(\varepsilon )\mathop {=}\limits ^{def} P_T(A_\varepsilon )\) and \(F_\zeta (\varepsilon )\mathop {=}\limits ^{def} P_\zeta (A_\varepsilon )\) with respect to \(\varepsilon \) are distribution functions. Moreover,
where
Thus,
This shows that \(F_\zeta (\varepsilon )\) is right continuous, thus continuous, only if the set \(A_\varepsilon \) is a continuity set of the measure \(P_\zeta \). Therefore, in view of (5.1), the distribution function \(F_T\) converges weakly to \(F_\zeta \) as \(T\rightarrow \infty \). Hence, by the first part of Lemma 5.1,
uniformly in u in every finite interval, where \(g_T(u)\) and \(g_\zeta (u)\) are the characteristic functions of the distribution functions \(F_T\) and \(F_\zeta \), respectively.
Denote by \({\hat{g}}_T(u)\) the characteristic function of the distribution function
Using the inequality \(|\mathrm {e}^{iu}-1|\le |u|\), \(u\in {\mathbb {R}}\), and the triangle inequality
we obtain by (5.3) that
as \(T\rightarrow \infty \). Therefore, in view of Lemma 3.1, we have
as \(T\rightarrow \infty \), uniformly in u in every finite interval. The function \(g_\zeta (u)\), \(|u|\le C\) with every \(C>0\), as a characteristic function is continuous at the point \(u=0\). This and the second part of Lemma 5.1 show that \({\hat{F}}_T\), as \(F_T\), converges weakly to \(F_\zeta \) as \(T\rightarrow \infty \), and we have
for all but at most countably many \(\varepsilon >0\). \(\square \)
References
Bagchi, B.: The statistical behaviour and universality properties of the Riemann zeta-function and other allied Dirichlet series. PhD Thesis, Indian Statistical Institute, Calcutta (1981)
Billingsley, P.: Convergence of Probability Measures. Wiley, New York (1968)
Gonek, S.M.: Analytic properties of zeta and \(L\)-functions. PhD Thesis, University of Michigan, Ann Arbor (1975)
Ivič, A.: The Riemann Zeta-Function. The Theory of the Riemann Zeta-Function with Applications. Wiley, New York (1985)
Laurinčikas, A.: Limit Theorems for the Riemann Zeta-Function. Kluwer Academic Publishers, Dordrecht (1996)
Laurinčikas, A., Garunkštis, R.: The Lerch Zeta-Function. Kluwer, Dordrecht (2002)
Laurinčikas, A., Meška, L.: Sharpening of the universality inequality. Math. Notes 96(5–6), 971–976 (2014)
Matsumoto, K.: A survey on the theory of universality for zeta and \(L\)-functions. In: Number Theory, pp. 95–144. Ser. Number Theory Appl., vol. 11. World Sci. Publ., Hackensack, New York (2015)
Mauclaire, J.-L.: Universality of the Riemann zeta-function: two remarks. Ann. Univ. Sci. Budapest. Sect. Comput. 39, 311–319 (2013)
Mergelyan, S.N.: Uniform approximations to functions of a complex variable. Amer. Math. Soc. Translation 1954(101), 99 pp. (1954)
Steuding, J.: Value-Distribution of \(L\)-Functions. Lecture Notes Math., vol. 1877. Springer, Berlin (2007)
Titchmarsh, E.C.: The Theory of Functions, 2nd edn. Oxford University Press, Oxford (1952)
Voronin, S.M.: Theorem on the “universality” of the Riemann zeta-function. Math. USSR Izv. 9(3), 443–453 (1975)
Acknowledgements
The author thanks the referee for useful remarks.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The research is funded by the European Social Fund (Project No. 09.3.3-LMT-K-712-01-0037) under grant agreement with the Research Council of Lithuania (LMT LT).
Rights and permissions
About this article
Cite this article
Laurinčikas, A. Approximation of analytic functions by an absolutely convergent Dirichlet series. Arch. Math. 117, 53–63 (2021). https://doi.org/10.1007/s00013-021-01616-x
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00013-021-01616-x