Home Mathematics A Tauberian theorem for the generalized Nörlund summability method
Article Publicly Available

A Tauberian theorem for the generalized Nörlund summability method

  • ORCID logo EMAIL logo , and ORCID logo
Published/Copyright: January 10, 2018
Become an author with De Gruyter Brill

Abstract

Let (pn) and (qn) be any two non-negative real sequences, with Rn:=k=0npkqn-k0 (n). Let k=0ak be a series of real or complex numbers with partial sums (sn), and set tnp,q:=1Rnk=0npkqn-ksk for n. In this paper, we present the necessary and sufficient conditions under which the existence of the limit limnsn=L follows from that of limntnp,q=L. These conditions are one-sided or two-sided if (sn) is a sequence of real or complex numbers, respectively.

MSC 2010: 40G15; 41A36

1 Introduction

In what follows, we give the concept of the summability method, known as the generalized Nörlund summability method (N,p,q) (see [1]). Given two non-negative sequences (pn) and (qn), the convolution (pq) is defined by

Rn:=(pq)n=k=0npkqn-k=k=0npn-kqk.

Let k=0ak be a series of real or complex numbers with partial sums (sn). When (pq)n0 for all n, the generalized Nörlund transform of the series k=0ak is the sequence (tnp,q) defined by

tnp,q=1(pq)nk=0npkqn-ksk.

We say that the series k=0ak is generalized Nörlund summable to L, determined by the sequences (pn) and (qn) or, briefly, (N,p,q) summable to L if

(1.1)limntnp,q=L.

We define the following sets:

A(n,t):={qλn-k-qn-k:k=0,1,2,,n,λ>1},
B(n,t):={qk-λn-qn-k:k=0,1,2,,tn, 0<λ<1},

where λn:=[λn] denotes the integral part of λn.

Throughout the paper we assume that the sequence q=(qn) satisfies the following conditions:

pnqn,n,qn1,n,
supnA(n,λ)<,supnB(n,λ)<.

If

(1.2)limnsn=L

implies (1.1), then the (N,p,q) method is called regular. The necessary and sufficient condition for the (N,p,q) method to be regular [6] is

pn-kqk=o(Rn)(n,k)

and

k=0n|pn-kqk|=O(Rn)(n).

Remark 1.1.

The (N,p,q) summability method reduces to the Cesàro summability method when pn=1 and qn=1 for all n, to the weighted mean method (N¯,p) when qn=1 for all n, to the method (C,α,β) (see [1]) when pn=(n+ββ), qn=(n+α-1α) and to the Nörlund method (N,q) when pn=1 for all n (see [6]).

Notice that (1.1) may imply (1.2) under a certain condition, which is called a Tauberian condition. Any theorem which states that the convergence of a series follows from its (N,p,q) summability and some Tauberian condition is said to be a Tauberian theorem for the (N,p,q) summability method. The inclusion and Tauberian type theorems are proved in the papers [7, 8], and some inclusion, Tauberian and convexity type theorems for certain families of generalized Nörlund methods are obtained in [12]. Moreover, some theorems of Abelian and Tauberian type have been recently proved for the methods of power series, (C,α) and weighted mean in [4, 3, 13, 2, 5].

In this paper, we present the necessary and sufficient conditions under which the existence of the limit limnsn=L follows from that of limntnp,q=L. These conditions are one-sided or two-sided if (sn) is a sequence of real or complex numbers, respectively. Furthermore, we have some classical Tauberian theorems for the Cesàro and weighted mean summability methods as a corollary.

2 Main results

In the following theorem we characterize the converse implication when the ordinary convergence follows from its (N,p,q) summability.

Theorem 2.1.

Let (pn) and (qn) be any two non-negative real sequences such that

(2.1)lim infnRλnRn>1for every λ>1,

where λn:=[λn] denotes the integral part of λn for every nN, and let k=0ak be a series of real numbers which is (N,p,q) summable to a finite number L. Then k=0ak is convergent to the same number L if and only if the following two conditions hold:

(2.2)limλ1+lim infn1Rλn-Rnk=n+1λnpkqλn-k(sk-sn)0,
(2.3)limλ1-lim infn1Rn-Rλnk=λn+1npkqn-k(sn-sk)0.

Remark 2.2.

Following Schmidt [11], we say that a real sequence (sn) is slowly decreasing if

(2.4)limλ1+lim infnminn+1kλn(sk-sn)0

or, equivalently,

limλ1-lim infnminλn+1kn(sn-sk)0.

Conditions (2.2) and (2.3) are satisfied if (sn) is slowly decreasing.

Remark 2.3.

The classical one-sided Tauberian condition of Landau (see [9])

kak-C(k=1,2,)

is sufficient for (2.4) to hold.

Remark 2.4.

If we take qn=1 for all n as a special case of Theorem 2.1, we obtain the Móricz and Rhoades conditions (see [10])

limλ1+lim infn1Pλn-Pnk=n+1λnpk(sk-sn)0

and

limλ1-lim infn1Pn-Pλnk=λn+1npk(sn-sk)0,

where Pn:=(p1)n=k=0npk.

In the next result we will consider the case where k=0ak is a series of complex numbers.

Theorem 2.5.

Let condition (2.1) be satisfied and let k=0ak be a series of complex numbers which is (N,p,q) summable to a finite number L. Then k=0ak is convergent to the same number L if and only if one of the following two conditions holds:

(2.5)limλ1+lim supn|1Rλn-Rnk=n+1λnpkqλn-k(sk-sn)|=0

or

(2.6)limλ1-lim supn|1Rn-Rλnk=λn+1npkqn-k(sn-sk)|=0.

Remark 2.6.

Following Schmidt [11], we say that a real sequence (sn) is said to be slowly oscillating if

limλ1+lim infnminn+1kλn|sk-sn|=0

or, equivalently,

limλ1-lim infnminλn+1kn|sn-sk|=0.

Conditions (2.5) and (2.6) are satisfied if (sn) is slowly oscillating.

Remark 2.7.

The classical two-sided Tauberian condition

k|ak|C(k=1,2,)

is sufficient for (2.5) and (2.6) to hold.

Remark 2.8.

If we take qn=1 for all n as a special case of Theorem 2.1, we obtain the Móricz and Rhoades conditions (see [10])

limλ1+lim supn|1Pλn-Pnk=n+1λnpk(sk-sn)|=0

and

limλ1-lim supn|1Pn-Pλnk=λn+1npk(sn-sk)|=0,

where Pn:=(p1)n=k=0npk.

3 Auxiliary results

In what follows, we list some auxiliary lemmas which are needful in the sequel.

Lemma 3.1.

The condition given by relation (2.1) is equivalent to the condition

(3.1)lim infnRnRλn>1,0<λ<1.

Proof.

Suppose that relation (2.1) is valid, 0<λ<1 and m=λn=[λn], n. Then it follows that

1λ>1mλ=[λn]λn.

From the above relation and the definition of the positive real sequences (pn) and (qn), we obtain

RnRλnR[mλ]Rλnlim infnRnRλnlim infnR[mλ]Rλn>1.

Conversely, suppose that relation (3.1) is valid. Let λ>1 be a given number and let λ1 be chosen so that 1<λ1<λ. Set m=λn=[λn]. From 0<1λ<1λ1<1, it follows that

nλn-1λ1<[λn]λ1=mλ1,

provided that λ1λ-1n, which is the case when n is large enough. Under these conditions, we have

RλnRnRλnR[mλ1]lim infnRλnRnlim infnRλnR[mλ1]>1.

Lemma 3.2.

Let condition (2.1) be satisfied and let k=0ak be a series of complex numbers which is generalized Nörlund summable to L. Then

(3.2)limn1Rλn-Rnk=n+1λnpkqλn-ksk=Lfor λ>1

and

(3.3)limn1Rn-Rλnk=λn+1npkqn-ksk=Lfor 0<λ<1.

Proof.

Consider the case where λ>1. Then we obtain

1Rλn-Rnk=n+1λnpkqλn-k(sk-L)=RλnRλn-Rn1Rλnk=0λnpkqλn-k(sk-L)-RnRλn-Rn1Rnk=0npkqλn-k(sk-L)
=RλnRλn-Rn1Rλnk=0λnpkqλn-k(sk-L)-RnRλn-Rn1Rnk=0npkqn-k(sk-L)
(3.4)-RnRλn-Rn1Rnk=0npk(qλn-k-qn-k)(sk-L),

provided that Rλn>Rn.

Since, by (2.1),

(3.5)lim supnRλnRλn-Rn=1+{-1+lim infnRλnRn}-1<,

(3.2) follows from (3.4), (3.5) and the definition of the sequence (qn).

Consider the case where 0<λ<1. Then we obtain

1Rn-Rλnk=λn+1npkqn-k(sk-L)=RnRn-Rλn1Rnk=0npkqn-k(sk-L)-RλnRn-Rλn1Rλnk=0λnpkqn-k(sk-L)
=RnRn-Rλn1Rnk=0npkqn-k(sk-L)-RλnRn-Rλn1Rλnk=0λnpkqn-k(sk-L)
(3.6)-RλnRn-Rλn1Rλnk=0λnpk(qn-k-qλn-k)(sk-L),

provided that Rn>Rλn.

Since, by (2.1),

(3.7)lim supnRλnRn-Rλn={-1+lim infnRnRλn}-1<,

(3.3) follows from (3.6), (3.7) and the definition of the sequence (qn). ∎

4 Proofs of the theorems

Proof of Theorem 2.1.

Necessity. Suppose that limnsn=L, and (1.1) holds. From Lemma 3.2, we have

limn1Rλn-Rnk=n+1λnpkqλn-k(sk-sn)=limn{(1Rλn-Rnk=n+1λnpkqλn-ksk)-sn}=0

for every λ>1 . In the case where 0<λ<1, we find that

limn1Rn-Rλnk=λn+1npkqn-k(sn-sk)=limn{(1Rn-Rλnk=λn+1npkqn-ksk)-sn}=0.

Sufficiency. Assume that conditions (2.2) and (2.3) are satisfied. In what follows, we will prove that limnsn=L. Given any ϵ>0, by relation (2.2), we can choose λ1>0 so that

(4.1)lim infn1Rλn1-Rnk=n+1λn1pkqλn-k(sk-sn)-ϵ,

where λn1=[λ1n]. By the assumed summability (N,p,q) of (xn) and Lemma 3.2, we have

(4.2)lim infn1Rλn-Rnk=n+1λnpkqλn-ksk=L

for any λ>1. Taking (4.1) and (4.2) into account, we obtain

(4.3)lim supnsnL+ϵ.

On the other hand, if 0<λ<1, then, for every ϵ>0, we can choose 0<λ2<1 so that

(4.4)lim infn1Rn-Rλn2k=λn2+1npkqn-k(sn-sk)-ϵ,

where λn2=[λ2n]. By the assumed summability (N,p,q) of (xn) and Lemma 3.2, we have

(4.5)lim infn1Rn-Rλnk=λn2+1npkqn-ksk=L

for any 0<λ<1.

Taking (4.4) and (4.5) into account, we obtain

(4.6)lim infnsnL-ϵ.

Since ϵ>0 is arbitrary, combining relations (4.3) and (4.6), we obtain

limnsn=L.

Correction added on 12 September 2018 after online publication: Mistakes within the proof of Theorem 2.1, part “Sufficiency”, have been corrected.

Proof of Theorem 2.5.

Necessity. If both (1.1) and (1.2) hold, then Lemma 3.2 yields (2.5) for every λ>1 and (2.6) for every 0<λ<1.

Sufficiency. Suppose that (1.1), (2.1) and one of conditions (2.5) or (2.6) is satisfied. For any given ϵ>0, there exists some λ1>1 such that

lim supn|1Rλn1-Rnk=n+1λn1pkqλn1-k(sk-sn)|ϵ,

where λn1=[λ1n]. Taking into account the fact that k=0ak is (N,p,q) summable to L, we get the following estimation:

lim supn|L-sn|limnsup|L-1Rλn1-Rnk=n+1λn1pkqλn1-ksk|+lim supn|1Rλn1-Rnk=n+1λn1pkqλn1-k(sk-sn)|ϵ.

For any given ϵ>0, there exists some 0<λ2<1 such that

lim supn|1Rn-Rλn2k=λn2+1npkqn-k(sk-sn)|ϵ,

where λn2=[λ2n]. Taking into account the fact that k=0ak is (N,p,q) summable to L, we get the following estimation:

lim supn|L-sn|lim supn|L-1Rn-Rλn2k=λn2+1npkqn-ksk|+lim supn|1Rn-Rλn2k=λn2+1npkqn-k(sk-sn)|ϵ.

Since ϵ>0 is arbitrary, in either case, we get limnsn=L. ∎

References

[1] D. Borwein, On products of sequences, J. Lond. Math. Soc. 33 (1958), 352–357. 10.1112/jlms/s1-33.3.352Search in Google Scholar

[2] I. Çanak and Y. Erdem, On Tauberian theorems for (A)(C,α) summability method, Appl. Math. Comput. 218 (2011), no. 6, 2829–2836. 10.1016/j.amc.2011.08.026Search in Google Scholar

[3] I. Çanak and U. Totur, Extended Tauberian theorem for the weighted mean method of summability, Ukrainian Math. J. 65 (2013), no. 7, 1032–1041. 10.1007/s11253-013-0839-xSearch in Google Scholar

[4] I. Çanak and U. Totur, A theorem for the (J,p) summability method, Acta Math. Hungar. 145 (2015), no. 1, 220–228. 10.1007/s10474-014-0452-ySearch in Google Scholar

[5] Y. Erdem, I. Çanak and B. P. Allahverdiev, Two theorems on the product of Abel and Cesàro summability methods, C. R. Acad. Bulgare Sci. 68 (2015), no. 3, 287–294. Search in Google Scholar

[6] G. H. Hardy, Divergent Series, Clarendon Press, Oxford, 1949. Search in Google Scholar

[7] R. Kiesel, General Nörlund transforms and power series methods, Math. Z. 214 (1993), no. 2, 273–286. 10.1007/BF02572404Search in Google Scholar

[8] R. Kiesel and U. Stadtmüller, Tauberian- and convexity theorems for certain (N,p,q)-means, Canad. J. Math. 46 (1994), no. 5, 982–994. 10.4153/CJM-1994-056-6Search in Google Scholar

[9] E. Landau, Über die Bedeutung einiger neuen Grenzwertsätze der Herren Hardy und Axel, Prac. Mat.-Fiz. 21 (1910), 97–177. Search in Google Scholar

[10] F. Móricz and B. E. Rhoades, Necessary and sufficient Tauberian conditions for certain weighted mean methods of summability, Acta Math. Hungar. 66 (1995), no. 1–2, 105–111. 10.1007/BF01874356Search in Google Scholar

[11] R. Schmidt, Über divergente Folgen und lineare Mittelbildungen, Math. Z. 22 (1925), no. 1, 89–152. 10.1007/BF01479600Search in Google Scholar

[12] U. Stadtmüller and A. Tali, On certain families of generalized Nörlund methods and power series methods, J. Math. Anal. Appl. 238 (1999), no. 1, 44–66. 10.1006/jmaa.1999.6503Search in Google Scholar

[13] U. Totur and I. Çanak, Some general Tauberian conditions for the weighted mean summability method, Comput. Math. Appl. 63 (2012), no. 5, 999–1006. 10.1016/j.camwa.2011.12.005Search in Google Scholar

Received: 2015-07-16
Accepted: 2016-05-23
Published Online: 2018-01-10
Published in Print: 2020-03-01

© 2020 Walter de Gruyter GmbH, Berlin/Boston

Articles in the same Issue

  1. Frontmatter
  2. An improvement of the constant in Videnskiĭ’s inequality for Bernstein polynomials
  3. On generalized α-ψ-Geraghty contractions on b-metric spaces
  4. New approximate solutions to electrostatic differential equations obtained by using numerical and analytical methods
  5. A Tauberian theorem for the generalized Nörlund summability method
  6. A multilinear reverse Hölder inequality with applications to multilinear weighted norm inequalities
  7. The Robin function and conformal welding – A new proof of the existence
  8. Effects of the initial moment and several delays perturbations in the variation formulas for a solution of a functional differential equation with the continuous initial condition
  9. The well-posedness of a nonlocal multipoint problem for a differential operator equation of second order
  10. Wavelets method for solving nonlinear stochastic Itô–Volterra integral equations
  11. On an approximate solution of a class of surface singular integral equations of the first kind
  12. On Φ-Dedekind, Φ-Prüfer and Φ-Bezout modules
  13. On the geometrical properties of hypercomplex four-dimensional Lie groups
  14. On sets of singular rotations for translation invariant differentiation bases formed by intervals
  15. Certain commutativity criteria for rings with involution involving generalized derivations
  16. The ℳ-projective curvature tensor field on generalized (κ,μ)-paracontact metric manifolds
  17. Ripplet transform and its extension to Boehmians
  18. Variable exponent fractional integrals in the limiting case α(x)p(n) ≡ n on quasimetric measure spaces
Downloaded on 22.3.2026 from https://www.degruyterbrill.com/document/doi/10.1515/gmj-2017-0062/html
Scroll to top button