The sequence space ℓ(p) was introduced by Maddox (1967). Quite recently, the domain of the generalized difference matrix B(r,s) in the sequence space ℓp has been investigated by Kirişçi and Başar (2010). In the present paper, the sequence space ℓ(B~,p) of nonabsolute type has been studied which is the domain of the generalized difference matrix B(r~,s~) in the sequence space ℓ(p). Furthermore, the alpha-, beta-, and gamma-duals of the space ℓ(B~,p) have been determined, and the Schauder basis has been given. The classes of matrix transformations from the space ℓ(B~,p) to the spaces ℓ∞, c and c0 have been characterized. Additionally, the characterizations of some other matrix transformations from the space ℓ(B~,p) to the Euler, Riesz, difference, and so forth sequence spaces have been obtained by means of a given lemma. The last section of the paper has been devoted to conclusion.
1. Preliminaries, Background, and Notation
By w, we denote the space of all real valued sequences. Any vector subspace of w is called a sequence space. We write ℓ∞, c, and c0 for the spaces of all bounded, convergent, and null sequences, respectively. Also by bs, cs, ℓ1, and ℓp, we denote the spaces of all bounded, convergent, absolutely convergent and p-absolutely convergent series, respectively, where 1<p<∞.
A linear topological space X over the real field ℝ is said to be a paranormed space if there is a subadditive function g:X→ℝ such that g(θ)=0,g(x)=g(-x) and scalar multiplication is continuous; that is, |αn-α|→0 and g(xn-x)→0 imply g(αnxn-αx)→0 for all α’s in ℝ and all x’s in X, where θ is the zero vector in the linear space X.
Assume here and after that (pk) is a bounded sequence of strictly positive real numbers with suppk=H and M=max{1,H}. Then, the linear spaces ℓ(p) were defined by Maddox [1] (see also Simons [2] and Nakano [3]) as follows:
(1)ℓ(p)={x=(xk)∈w:∑k|xk|pk<∞},(0<pk≤H<∞)
which is the complete space paranormed by
(2)g(x)=(∑k|xk|pk)1/M.
For simplicity in notation, here and in what follows, the summation without limits runs from 0 to ∞. We assume throughout that pk-1+(pk′)-1=1 and denote the collection of all finite subsets of ℕ={0,1,2,…} by ℱ and use the convention that any term with negative subscript is equal to naught.
Let λ, μ be any two sequence spaces and let A=(ank) be an infinite matrix of real or complex numbers ank, where n,k∈ℕ. Then, we say that A defines a matrix mapping from λ into μ, and we denote it by writing A:λ→μ; if for every sequence x=(xk)∈λ the sequence Ax={(Ax)n}, the A-transform of x, is in μ, where
(3)(Ax)n=∑kankxk,foreachn∈ℕ.
By (λ:μ), we denote the class of all matrices A such that A:λ→μ. Thus, A∈(λ:μ) if and only if the series on the right side of (3) converges for each n∈ℕ and every x∈λ, and we have Ax={(Ax)n}n∈ℕ∈μ for all x∈λ. A sequence x is said to be A-summable to α if Ax converges to α which is called the A-limit of x.
The shift operator P is defined on ω by (Px)n=xn+1 for all n∈ℕ. A Banach limit L is defined on ℓ∞, as a nonnegative linear functional, such that L(Px)=L(x) and L(e)=1, where e=(1,1,1,…). A sequence x=(xk)∈ℓ∞ is said to be almost convergent to the generalized limit l if all Banach limits of x are l and is denoted by f-limxk=l. Lorentz [4] proved that
(4)f-limxk=lifflimm→∞1m+1∑k=0mxk+n=luniformlyinn.
It is well known that a convergent sequence is almost convergent such that its ordinary and generalized limits are equal. By f, we denote the space of all almost convergent sequences; that is,
(5)f:={x=(xk)∈ω:∃l∈ℂ∋limm→∞∑k=0mxn+km+1=luniformlyinn}.
Define the double sequential band matrix B(r~,s~)={bnk(rk,sk)} by
(6)bnk(rk,sk)={rk,k=n,sk,k=n-1,0,otherwise
for all k,n∈ℕ, where r~=(rk) and s~=(sk) are the convergent sequences. We should note that the double sequential band matrices were firstly used by Srivastava and Kumar [5, 6], Panigrahi and Srivastava [7], and Akhmedov and El-Shabrawy [8].
The main purpose of this paper, which is a continuation of Kirişçi and Başar [9], is to introduce the sequence space ℓ(B~,p) of nonabsolute type consisting of all sequences whose B(r~,s~)-transforms are in the space ℓ(p). Furthermore, the basis is constructed and the alpha-, beta-, and gamma-duals are computed for the space ℓ(B~,p). Moreover, the matrix transformations from the space ℓ(B~,p) to some sequence spaces are characterized. Finally, we note open problems and further suggestions.
It is clear that Δ(1) can be obtained as a special case of B(r~,s~) for r~=e and s~=-e and it is also trivial that B(r~,s~) is reduced in the special case r~=re and s~=se to the generalized difference matrix B(r,s). So, the results related to the matrix domain of the matrix B(r~,s~) are more general and more comprehensive than the corresponding consequences of the matrix domains of Δ(1) and B(r,s).
The rest of this paper is organized as follows. In Section 2, the linear sequence space ℓ(B~,p) is defined and proved that it is a complete paranormed space with a Schauder basis. Section 3 is devoted to the determination of alpha-, beta-, and gamma-duals of the space ℓ(B~,p). In Section 4, the classes (ℓ(B~,p):ℓ∞), (ℓ(B~,p):f), (ℓ(B~,p):c), and (ℓ(B~,p):c0) of infinite matrices are characterized. Additionally, the characterizations of some other classes of matrix transformations from the space ℓ(B~,p) to the Euler, Riesz, difference, and so forth sequence spaces are obtained by means of a given lemma. In the final section of the paper, open problems and further suggestions are noted.
2. The Sequence Space ℓ(B~,p) of Nonabsolute Type
In this section, we introduce the complete paranormed linear sequence space ℓ(B~,p).
The matrix domain λA of an infinite matrix A in a sequence space λ is defined by
(7)λA={x=(xk)∈ω:Ax∈λ}.
Choudhary and Mishra [10] defined the sequence space ℓ(p)¯ which consists of all sequences such that S-transforms of them are in the space ℓ(p), where S=(snk) is defined by
(8)snk={1,0≤k≤n,0,k>n,
for all k,n∈ℕ. Başar and Altay [11] have recently examined the space bs(p) which is formerly defined by Başar in [12] as the set of all series whose sequences of partial sums are in ℓ∞(p). More recently, Aydın and Başar [13] have studied the space ar(u,p) which is the domain of the matrix Ar in the sequence space ℓ(p), where the matrix Ar={ank(r)} is defined by
(9)ank(r)={1+rkn+1uk,0≤k≤n,0,k>n,
for all k,n∈ℕ, (uk) such that uk≠0 for all k∈ℕ and 0<r<1. Altay and Başar [14] have studied the sequence space rt(p) which is derived from the sequence space ℓ(p) of Maddox by the Riesz means Rt. With the notation of (7), the spaces ℓ(p)¯,bs(p),ar(u,p), and rt(p) can be redefined by
(10)ℓ(p)¯=[ℓ(p)]S,bs(p)=[ℓ∞(p)]S,ar(u,p)=[ℓ(p)]Ar,rt(p)=[ℓ(p)]Rt.
Following Choudhary and Mishra [10], Başar and Altay [11], Altay and Başar [14–17], and Aydın and Başar [13, 18], we introduce the sequence space ℓ(B~,p) as the set of all sequences whose B(r~,s~)-transforms are in the space ℓ(p); that is
(11)ℓ(B~,p):={(xk)∈w:∑k|sk-1xk-1+rkxk|pk<∞},(0<pk≤H<∞).
It is trivial that in the case pk=p for all k∈ℕ, the sequence space ℓ(B~,p) is reduced to the sequence space ℓ~p which is introduced by Kirişçi and Başar [9]. With the notation of (7), we can redefine the space ℓ(B~,p) as follows:
(12)ℓ(B~,p):=[ℓ(p)]B(r~,s~).
Define the sequence y=(yk), which will be frequently used, as the B(r~,s~)-transform of a sequence x=(xk); that is,
(13)yk={B(r~,s~)x}k=rkxk+sk-1xk-1,∀k∈ℕ.
Since the spaces ℓ(p) and ℓ(B~,p) are linearly isomorphic by Corollary 4, one can easily observe that x=(xk)∈ℓ(B~,p) if and only if y=(yk)∈ℓ(p), where the sequences x=(xk) and y=(yk) are connected with the relation (13).
Now, we may begin with the following theorem which is essential in the text.
Theorem 1.
ℓ(B~,p) is a complete linear metric space paranormed by the paranorm
(14)h(x)=(∑k|sk-1xk-1+rkxk|pk)1/M.
Proof.
It is easy to see that the space ℓ(B~,p) is linear with respect to the coordinate-wise addition and scalar multiplication. Therefore, we first show that it is a paranormed space with the paranorm h defined by (14).
It is clear that h(θ)=0 where θ=(0,0,0,…) and h(x)=h(-x) for all x∈ℓ(B~,p).
Let x,y∈ℓ(B~,p); then by Minkowski’s inequality we have
(15)h(x+y)=[∑k|sk-1(xk-1+yk-1)+rk(xk+yk)|pk]1/M={∑k[|sk-1(xk-1+yk-1)+rk(xk+yk)|pk/M]M}1/M≤(∑k|sk-1xk-1+rkxk|pk)1/M+(∑k|sk-1yk-1+rkyk|pk)1/M=h(x)+h(y).
Let (λn) be a sequence of scalars with λn→λ, as n→∞, and let (x(n))n=0∞ be a sequence of elements x(n)∈ℓ(B~,p) with h(x(n)-x)→0, as n→∞. We observe that
(16)h(λnx(n)-λx)≤h[(λn-λ)(x(n)-x)]+h[λ(x(n)-x)]+h[(λn-λ)x].
It follows from λn→λ(n→∞) that |λn-λ|<1 for all sufficiently large n; hence
(17)limn→∞h[(λn-λ)(x(n)-x)]≤limn→∞h(x(n)-x)=0.
Furthermore, we have
(18)limn⟶∞h[λ(x(n)-x)]≤max{1,|λ|M}limn⟶∞h(x(n)-x)=0.
Also, we have
(19)limn⟶∞h[(λn-λ)x]≤limn⟶∞|λn-λ|h(x)=0.
Then, we obtain from (16), (17), (18), and (19) that h(λnx(n)-λx)→0, as n→∞. This shows that h is a paranorm on ℓ(B~,p).
Furthermore, if h(x)=0, then (∑k|sk-1xk-1+rkxk|pk)1/M=0. Therefore |sk-1xk-1+rkxk|pk=0 for each k∈ℕ. If we put k=0, since s-1=0 and r0≠0, we have x0=0. For k=1, since x0=0 we have x1=0. Continuing in this way, we obtain xk=0 for all k∈ℕ. That is, x=θ. This shows that h is a total paranorm.
Now, we show that ℓ(B~,p) is complete. Let {xn} be any Cauchy sequence in ℓ(B~,p) where xn={x0(n),x1(n),x2(n),…}. Here and after, for short we write B~ instead of B(r~,s~). Then for a given ε>0, there exists a positive integer n0(ε) such that h(xn-xm)<ε for all n,m>n0(ε). Since for each fixed k∈ℕ(20)|(B~xn)k-(B~xm)k|≤[∑k|(B~xn)k-(B~xm)k|pk]1/M=h(xn-xm)<ε
for every n,m>n0(ε), {(B~x0)k,(B~x1)k,(B~x2)k,…} is a Cauchy sequence of real numbers for every fixed k∈ℕ. Since ℝ is complete, it converges, say (B~xn)k→(B~x)k as n→∞. Using these infinitely many limits (B~x)0,(B~x)1,(B~x)2,… we define the sequence {(B~x)0,(B~x)1,(B~x)2,…}. For each K∈ℕ and n,m>n0(ε)(21)[∑k=0K|(B~xn)k-(B~xm)k|pk]1/M≤h(xn-xm)<ε.
By letting m,K→∞, we have for n>n0(ε) that
(22)h(xn-x)=[∑k|(B~xn)k-(B~x)k|pk]1/M<ε.
This shows us xn-x∈ℓ(B~,p). Since ℓ(B~,p) is a linear space, we conclude that x∈ℓ(B~,p); It follows that xn→x, as n→∞ in ℓ(B~,p), thus we have shown that ℓ(B~,p) is complete.
Therefore, one can easily check that the absolute property does not hold on the space ℓ(B~,p); that is, g1(x)≠g1(|x|), where |x|=(|xk|). This says that ℓ(B~,p) is the sequence space of nonabsolute type.
Theorem 2.
Convergence in ℓ(B~,p) is stronger than coordinate-wise convergence.
Proof.
First we show that h(xn-x)→0, as n→∞ implies xkn→xk; as n→∞ for every k∈ℕ. We fix k, then we have
(23)limn→∞|sk-1xk-1(n)+rkxk(n)-sk-1xk-1-rkxk|pk≤limn→∞∑k|sk-1xk-1(n)+rkxk(n)-sk-1xk-1-rkxk|pk=limn→∞[h(xn-x)]M=0.
Hence, we have for k=0 that
(24)limn→∞|s-1x-1(n)+r0x0(n)-s-1x-1-r0x0|=0,
which gives the fact that |x0(n)-x0|→0, as n→∞. Similarly, for each k∈ℕ, we have |xk(n)-xk|→0, as n→∞.
A sequence space λ with a linear topology is called a K-space provided each of the maps pi:λ→ℂ defined by pi(x)=xi is continuous for all i∈ℕ, where ℂ denotes the complex field. A K-space λ is called an FK-space provided λ is complete linear metric space. An FK-space whose topology is normable is called a BK-space. Given a BK-space λ⊃ϕ, we denote the nthsection of a sequence x=(xk)∈λ by x[n]:=∑k=0nxke(k), and we say that x=(xk) has the property AK if limn→∞∥x-x[n]∥λ=0. If AK property holds for every x∈λ, then we say that the space λ is called AK-space (cf. [19]). Now, we may give the following.
Theorem 3.
(ℓp)B~ is the linear space under the coordinatewise addition and scalar multiplication which is the BK-space with the norm
(25)∥x∥:=(∑k|sk-1xk-1+rkxk|p)1/p,where1≤p<∞.
Proof.
Because the first part of the theorem is a routine verification, we omit the detail. Since ℓp is the BK-space with respect to its usual norm (see [20, pages 217-218]) and B(r~,s~) is a normal matrix, Theorem 4.3.2 of Wilansky [21, page 61] gives the fact that (ℓp)B~ is the BK-space, where 1≤p<∞.
Let us suppose that 1<pk≤sk for all k∈ℕ. Then, it is known that ℓ(p)⊂ℓ(s) which leads us to the immediate consequence that ℓ(B~,p)⊂ℓ(B~,s).
With the notation of (13), define the transformation T from ℓ(B~,p) to ℓ(p) by x↦y=Tx. Since T is linear and bijection, we have the following.
Corollary 4.
The sequence space ℓ(B~,p) of nonabsolute type is linearly paranorm isomorphic to the space ℓ(p), where 0<pk≤H<∞ for all k∈ℕ.
Theorem 5.
The space ℓ(B~,p) has AK.
Proof.
For each x=(xk)∈ℓ(B~,p), we put
(26)x〈m〉=∑k=0mxke(k),∀m∈{1,2,…}.
Let ε>0 and x∈ℓ(B~,p) be given. Then, there is N=N(ε)∈ℕ such that
(27)∑k=N∞|sk-1xk-1+rkxk|pk<εM.
Then we have for all m≥N,
(28)h(x-x〈m〉)=h(x-∑k=1mxke(k))=(∑k=m+1∞|sk-1xk-1+rkxk|pk)1/M≤(∑k=N∞|sk-1xk-1+rkxk|pk)1/M<ε.
This shows that x=∑kxke(k).
Now we have to show that this representation is unique. We assume that x=∑kλke(k). Then for each k,
(29)(|sk-1λk-1+rkλk-sk-1xk-1-rkxk|pk)1/M≤(∑k|sk-1λk-1+rkλk-sk-1xk-1-rkxk|pk)1/M=h(x-x)=0.
Hence, sk-1λk-1+rkλk=sk-1xk-1+rkxk for each k.
For k=0, r0λ0=r0x0. Since r0≠0, we have λ0=x0.
For k=1, s0λ0+r1λ1=s0x0+r1x1. Since r1≠0, we also have λ1=x1.
Continuing in this way, we obtain λk=xk for each k. Therefore, the representation is unique.
We firstly define the concept of the Schauder basis for a paranormed sequence space and next give the basis of the sequence space ℓ(B~,p).
Let (X,g) be a paranormed space. A sequence (bk) of the elements of X is called a basis for X if and only if, for each x∈X, there exists a unique sequence (αk) of scalars such that
(30)limn→∞g(x-∑k=0nαkbk)=0.
The series ∑kαkbk which has the sum x is then called the expansion of x with respect to (bn) and written as x=∑kαkbk. Since it is known that the matrix domain λA of a sequence space λ has a basis if and only if λ has a basis whenever A=(ank) is a triangle (cf. [22, Remark 2.4]), we have the following.
Corollary 6.
Let 0<pk≤H<∞ and αk=(B~x)k for all k∈ℕ. Define the sequence b(k)={bn(k)}n∈ℕ of the elements of the space ℓ(B~,p) by
(31)bn(k):={(-1)n-krn∏j=kn-1sjrj,0≤k≤n,0,k>n,
for every fixed k∈ℕ. Then, the sequence {b(k)}k∈ℕ given by (31) is a basis for the space ℓ(B~,p) and any x∈ℓ(B~,p) has a unique representation of the form x:=∑kαkb(k).
3. The Alpha-, Beta-, and Gamma-Duals of the Space ℓ(B~,p)
In this section, we state and prove the theorems determining the alpha-, beta-, and gamma-duals of the sequence space ℓ(B~,p) of nonabsolute type.
For the sequence spaces λ and μ, the set S(λ,μ) defined by
(32)S(λ,μ):={z=(zk)∈ω:xz=(xkzk)∈μ∀x=(xk)∈λ}
is called the multiplier space of the spaces λ and μ. With the notation of (32), the alpha-, beta-, and gamma-duals of a sequence space λ, which are, respectively, denoted by λα, λβ, and λγ, are defined by
(33)λα:=S(λ,ℓ1),λβ:=S(λ,cs),λγ:=S(λ,bs).
Since the case 0<pk≤1 may be established in similar way to the proof of the case 1<pk≤H<∞, we omit the detail of that case and give the proof only for the case 1<pk≤H<∞ in Theorems 10–12 below.
We begin with quoting three lemmas which are needed in proving Theorems 10–12.
Lemma 7 ([23, (i) and (ii) of Theorem 1]).
Let A=(ank) be an infinite matrix. Then, the following statements hold.
Let 0<pk≤1 for all k∈ℕ. Then, A∈(ℓ(p):ℓ∞) if and only if
(34)supn,k∈ℕ|ank|pk<∞.
Let 1<pk≤H<∞ for all k∈ℕ. Then, A∈(ℓ(p):ℓ∞) if and only if there exists an integer M>1 such that
(35)supn∈ℕ∑k|ankM-1|pk′<∞.
Lemma 8 ([23, Corollary for Theorem 1]).
Let 0<pk≤H<∞ for all k∈ℕ. Then, A=(ank)∈(ℓ(p):c) if and only if (34) and (35) hold, and
(36)limn→∞ank=βk,∀k∈ℕ.
Lemma 9 ([24, Theorem 5.1.0]).
Let A=(ank) be an infinite matrix. Then, the following statements hold
Let 0<pk≤1 for all k∈ℕ. Then, A∈(ℓ(p):ℓ1) if and only if
(37)supN∈ℱsupk∈ℕ|∑n∈Nank|pk<∞.
Let 1<pk≤H<∞ for all k∈ℕ. Then, A∈(ℓ(p):ℓ1) if and only if there exists an integer M>1 such that
(38)supN∈ℱ∑k|∑n∈NankM-1|pk′<∞.
Theorem 10.
Define the sets S1(p) and S2(p) by
(39)S1(p)=⋃M>1{|∑n∈N(-1)n-krn∏j=kn-1sjrjanM-1|pk′a=(ak)∈ω:⋃B>1supN∈ℱ∑k|∑n∈N(-1)n-krn∏j=kn-1sjrjanM-1|pk′<∞|∑n∈N(-1)n-krn∏j=kn-1sjrjanM-1|pk′},S2(p)={a=(ak)∈ω:supN∈ℱsupk∈ℕ|∑n∈N(-1)n-krn∏j=kn-1sjrjan|pk<∞}.
Then,
(40){ℓ(B~,p)}α={S1(p),1<pk≤H<∞,∀k∈ℕ,S2(p),0<pk≤1,∀k∈ℕ.
Proof.
Let us take any a=(an)∈ω. By using (13) we obtain that
(41)xn=∑k=0n(-1)n-krn∏j=kn-1sjrjyk
holds for all n∈ℕ which leads us to
(42)anxn=∑k=0n(-1)n-krn∏j=kn-1sjrjanyk=(Cy)n,(n∈ℕ),
where C=(cnk) is defined by
(43)cnk={(-1)n-krn∏j=kn-1sjrjan,0≤k≤n,0,k>n
for all k,n∈ℕ. Thus, we observe by combining (42) with the condition (37) of Part (i) of Lemma 9 that ax=(anxn)∈ℓ1 whenever x=(xk)∈ℓ(B~,p) if and only if Cy∈ℓ1 whenever y=(yk)∈ℓ(p). That means {ℓ(B~,p)}α=S1(p).
Theorem 11.
Define the sets S3(p), S4(p), and S5(p) by
(44)S3(p)=⋃M>1{a=(ak)∈ω:supn∈ℕ∑k|∑i=kn(-1)i-kri∏j=ki-1sjrjaiM-1|pk′<∞},S4(p)={a=(ak)∈ω:∑i=k∞(-1)i-kri∏j=ki-1sjrjai<∞},S5(p)={a=(ak)∈ω:supn,k∈ℕ|∑i=kn(-1)i-kri∏j=ki-1sjrjai|pk<∞}.
Then,
(45){ℓ(B~,p)}β={S3(p)∩S4(p),1<pk≤H<∞∀k∈ℕ,S4(p)∩S5(p),0<pk≤1∀k∈ℕ.
Proof.
Take any a=(ai)∈ω and consider the equation obtained with (13) that
(46)∑i=0naixi=∑i=0n[∑k=0i(-1)i-kri∏j=ki-1sjrjyk]ai=∑k=0n[∑i=kn(-1)i-kri∏j=ki-1sjrjai]yk=(Dy)n,
where D=(dnk) is defined by
(47)dnk={∑i=kn(-1)i-kri∏j=ki-1sjrjai,0≤k≤n,0,k>n
for all k,n∈ℕ. Thus, we deduce from Lemma 8 with (46) that ax=(aixi)∈cs whenever x=(xi)∈ℓ(B~,p) if and only if Dy∈c whenever y=(yk)∈ℓ(p). Therefore, we derive from (35) and (36) that
(48)supn∈ℕ∑k|∑i=kn(-1)i-kri∏j=ki-1sjrjaiM-1|pk′<∞,∑i=k∞(-1)i-kri∏j=ki-1sjrjai<∞.
This shows that {ℓ(B~,p)}β=S3(p)∩S4(p).
From Lemma 7 and (46), we obtain that ax=(aixi)∈bs whenever x=(xi)∈ℓ(B~,p) if and only if Dy∈ℓ∞ whenever y=(yk)∈ℓ(p), where D=(dnk) is defined by (47). Therefore, we obtain from (34) and (35) that {ℓ(B~,p)}γ=S3(p) for 1<pk, {ℓ(B~,p)}γ=S5(p) for pk≤1.
4. Matrix Transformations on the Sequence Space ℓ(B~,p)
In this section, we characterize some matrix transformations on the space ℓ(B~,p). Theorem 13 gives the exact conditions of the general case 0<pk≤H<∞ by combining the cases 0<pk≤1 and 1<pk≤H<∞. We consider only the case 1<pk≤H<∞ and leave the case 0<pk≤1 to the reader because it can be proved in similar way.
Theorem 13.
Let A=(ank) be an infinite matrix. Then, the following statements hold.
Let 1<pk≤H<∞ for all k∈ℕ. Then, A∈(ℓ(B~,p):ℓ∞) if and only if there exists an integer M>1 such that
(50)supn∈ℕ∑k|∑i=kn(-1)i-kri∏j=ki-1sjrjaniM-1|pk′<∞,(51)∑i=k∞(-1)i-kri∏j=ki-1sjrjani<∞.
Let 0<pk≤1 for all k∈ℕ. Then, A∈(ℓ(B~,p):ℓ∞) if and only if the condition (51) holds, and
(52)supn,k∈ℕ|∑i=kn(-1)i-kri∏j=ki-1sjrjani|pk<∞.
Proof.
Suppose that the conditions (50) and (51) hold, and x∈ℓ(B~,p). In this situation, since {ank}k∈ℕ∈{ℓ(B~,p)}β for every fixed n∈ℕ, the A-transform of x exists. Consider the following equality obtained by using the relation (13) that
(53)∑k=0mankxk=∑k=0m∑i=km(-1)i-kri∏j=ki-1sjrjaniyk
for all m,n∈ℕ. Taking into account the hypothesis we derive from (53) as m→∞ that
(54)∑kankxk=∑k∑i=k∞(-1)i-kri∏j=ki-1sjrjaniyk,foreachn∈ℕ.
Now, by combining (54) with the following inequality (see [23]) which holds for any M>0 and any a,b∈ℂ(55)|ab|≤M(|aM-1|p′+|b|p),
where p>1 and p-1+p′-1=1, one can easily see that
(56)supn∈ℕ|∑kankxk|≤supn∈ℕ∑k|∑i=k∞(-1)i-kri∏j=ki-1sjrjani||yk|≤supn∈ℕ∑kM(|∑i=k∞1ri∏j=ki-1sjrjaniM-1|pk′+|yk|pk)≤M(supn∈ℕ∑k|∑i=k∞1ri∏j=ki-1sjrjaniM-1|pk′+∑k|yk|pk)<∞.
Conversely, suppose that A∈(ℓ(B~,p):ℓ∞) and 1<pk≤H<∞ for all k∈ℕ. Then Ax exists for every x∈ℓ(B~,p) and this implies that {ank}k∈ℕ∈{ℓ(B~,p)}β for all n∈ℕ. Now, the necessity of (51) is immediate. Besides, we have from (54) that the matrix B=(bnk) defined by bnk=∑i=k∞((-1)i-k/ri)∏j=ki-1(sj/rj)ani for all n,k∈ℕ, is in the class (ℓ(p):ℓ∞). Then, B satisfies the condition (35) which is equivalent to (50).
This completes the proof.
Lemma 14 ([25, Theorem 1]).
A=(ank)∈(ℓ(p):f) if and only if (34) and (35) hold, and
(57)∃αk∈ℂ∋f-limank=αkforeveryfixedk∈ℕ.
Theorem 15.
Let the entries of the matrices E=(enk) and F=(fnk) be connected with the relation
(58)enk:=sk-1fn,k-1+rkfnkorfnk:=∑i=k∞(-1)iri∏j=ki-1sjrjeni
for all k,n∈ℕ. Then, E∈(ℓ(B~,p):f) if and only if F∈(ℓ(p):f) and
(59)Fn∈(ℓ(p):c)
for every fixed n∈ℕ, where Fn=(fmk(n)) with
(60)fmk(n):={∑mi=k(-1)iri∏j=ki-1sjrjeni,0≤k≤m,0,k>m,
for all m,k∈ℕ.
Proof.
Let E=(enk)∈(ℓ(B~,p):f) and take x∈ℓ(B~,p). Then, we obtain the equality
(61)∑k=0menkxk=∑k=0menk[∑i=0k(-1)k-iri∏j=ik-1sjrjyi]=∑k=0m[∑i=km(-1)iri∏j=ki-1sjrj]yk=∑k=0mfmk(n)yk
for all m,n∈ℕ. Since Ex exists, Fn∈(ℓ(p):c). Letting m→∞ in the equality (61) we have Ex=Fy. Since Ex∈f, then Fy∈f. That is F∈(ℓ(p):f).
Conversely, let F∈(ℓ(p):f), and Fn∈(ℓ(p):c), and take x∈ℓ(B~,p). Then, since (fnk)k∈ℕ∈{ℓ(p)}β and F∈(ℓ(p):f) we have (enk)k∈ℕ∈{ℓ(B~,p)}β for all n∈ℕ. So, Ex exists. Therefore we obtain from equality (61) as m→∞ that Ex=Fy, that is E∈(ℓ(B~,p):f).
Theorem 16.
Let 0<pk≤H<∞ for all k∈ℕ. Then, A∈(ℓ(B~,p):c) if and only if (50)–(52) hold and
(62)limn→∞∑k=i∞(-1)k-irk∏j=ik-1sjrjank=αk,foreveryfixedk∈ℕ.
Proof.
Let A∈(ℓ(B~,p):c) and 1<pk≤H<∞ for all k∈ℕ. Then, since the inclusion c⊂ℓ∞ holds, the necessities of (50) and (51) are immediately obtained from part (i) of Theorem 13.
To prove the necessity of (62), consider the sequence b(k) defined by (31) which is in the space ℓ(B~,p) for every fixed k∈ℕ. Because the A-transform of every x∈ℓ(B~,p) exists and is in c by the hypothesis,
(63)Ab(k)={∑k=i∞(-1)k-irk∏j=ik-1sjrjank}n∈ℕ∈c
for every fixed k∈ℕ which shows the necessity of (62).
Conversely suppose that conditions (50), (51), and (62) hold, and take any x=(xk) in the space ℓ(B~,p). Then, Ax exists. We observe for all m,n∈ℕ that
(64)∑k=0m|∑k=im(-1)k-irk∏j=ik-1sjrjankM-1|pk′≤supn∈ℕ∑km|∑k=im(-1)k-irk∏j=ik-1sjrjankM-1|pk′<∞,
which gives the fact that by letting m,n→∞ with (50) and (62) that
(65)limm,n⟶∞∑k=0m|∑k=im(-1)k-irk∏j=ik-1sjrjankM-1|pk′≤supn∈ℕ∑km|∑k=im(-1)k-irk∏j=ik-1sjrjankM-1|pk′<∞.
This shows that ∑k|αkM-1|pk′<∞ and so (αk)k∈ℕ∈{ℓ(B~,p)}β which implies that the series ∑kαkxk converges for every x∈ℓ(B~,p).
Let us now consider the equality obtained from (54) with ank-αk instead of ank(66)∑k(ank-αk)xk=∑i∑k=i(-1)k-irk∏j=ik-1sjrj(ank-αk)yi=∑kcniyi,∀n∈ℕ,
where C=(cni) defined by cni=∑k=i((-1)k-i/rk)∏j=ik-1(sj/rj)(ank-αk) for all n,i∈ℕ. Therefore, we have at this stage from Lemma 8 that the matrix C belongs to the class (ℓ(p):c0) of infinite matrices. Thus, we see by (66) that
(67)limn→∞∑k(ank-αk)xk=0.
Equation (67) means that Ax∈c whenever x∈ℓ(B~,p) and this is what we wished to prove.
Therefore, we have the following
Corollary 17.
Let 0<pk≤H<∞ for all k∈ℕ. Then, A∈(ℓ(B~,p):c0) if and only if (50)–(52) hold, and (62) also holds with αk=0 for all k∈ℕ.
Now, we give the following lemma given by Başar and Altay [26] which is useful for deriving the characterizations of the certain matrix classes via Theorems 13, 15, and 16 and Corollary 17.
Lemma 18 ([26, Lemma 5.3]).
Let λ, μ be any two sequence spaces, let A be an infinite matrix, and let B also be a triangle matrix. Then, A∈(λ:μB) if and only if BA∈(λ:μ).
It is trivial that Lemma 18 has several consequences. Indeed, combining Lemma 18 with Theorems 13, 15, and 16 and Corollary 17, one can derive the following results.
Corollary 19.
Let A=(ank) be an infinite matrix and define the matrix C=(cnk) by
(68)cnk=∑j=0n(nj)(1-t)n-jtjajk,∀n,k∈ℕ.
Then, the necessary and sufficient conditions in order to A belongs to anyone of the classes (ℓ(B~,p):e∞t), (ℓ(B~,p):ect) and (ℓ(B~,p):e0t) are obtained from the respective ones in Theorems 13, 16 and Corollary 17 by replacing the entries of the matrix A by those of the matrix C; where 0<t<1, e∞t and ect, e0t, respectively, denote the spaces of all sequences whose Et-transforms are in the spaces ℓ∞ and c, c0 and are recently studied by Altay et al. [27] and Altay and Başar [28], where Et denotes the Euler mean of order t.
Corollary 20.
Let A=(ank) be an infinite matrix and define the matrix C=(cnk) by
(69)cnk=san-1,k+rank,∀n,k∈ℕ.Then, the necessary and sufficient conditions in order to A belongs to the class (ℓ(B~,p):f^) is obtained from Theorem 15 by replacing the entries of the matrix A by those of the matrix C; where r,s∈ℝ∖{0} and f^ denotes the space of all sequences whose B(r,s)-transforms are in the space f and is recently studied by Başar and Kirişçi [29].
Corollary 21.
Let A=(ank) be an infinite matrix and define the matrix C=(cnk) by
(70)cnk=tan-2,k+san-1,k+rank,∀n,k∈ℕ.Then, the necessary and sufficient conditions in order to A belongs to the class (ℓ(B~,p):f(B)) is obtained from Theorem 15 by replacing the entries of the matrix A by those of the matrix C; where r,s,t∈ℝ∖{0} and f(B) denotes the space of all sequences whose B(r,s,t)-transforms are in the space f and is recently studied by Sönmez [30].
Corollary 22.
Let A=(ank) be an infinite matrix and define the matrix C=(cnk) by
(71)cnk=1n+1∑j=0najk,∀n,k∈ℕ.Then, the necessary and sufficient conditions in order to A belongs to the class (ℓ(B~,p):f~) is obtained from Theorem 15 by replacing the entries of the matrix A by those of the matrix C, where f~ denotes the space of all sequences whose C1-transforms are in the space f and is recently studied by Kayaduman and Şengönül [31].
Corollary 23.
Let A=(ank) be an infinite matrix and let t=(tk) be a sequence of positive numbers and define the matrix C=(cnk) by
(72)cnk=1Tn∑j=0ntjajk,∀n,k∈ℕ,
where Tn=∑k=0ntk for all n∈ℕ. Then, the necessary and sufficient conditions in order to A belongs to anyone of the classes (ℓ(B~,p):r∞t), (ℓ(B~,p):rct) and (ℓ(B~,p):r0t) are obtained from the respective ones in Theorems 13, 16 and Corollary 17 by replacing the entries of the matrix A by those of the matrix C, where r∞t, rct, and r0t are defined by Altay and Başar in [32] as the spaces of all sequences whose Rt-transforms are, respectively, in the spaces ℓ∞, c, and c0, and are derived from the paranormed spaces r∞t(p), rct(p) and r0t(p) in the case pk=p for all k∈ℕ.
Since the spaces r∞t, rct, and r0t reduce in the case t=e to the Cesàro sequence spaces X∞, c~, and c~0 of nonabsolute type, respectively, Corollary 23 also includes the characterizations of the classes (ℓ(B~,p):X∞), (ℓ(B~,p):c~), and (ℓ(B~,p):c~0), as a special case, where X∞ and c~, c~0 are the Cesàro spaces of the sequences consisting of C1-transforms are in the spaces ℓ∞ and c, c0 and studied by Ng and Lee [33] and Şengönül and Başar [34], respectively, where C1 denotes the Cesàro mean of order 1.
Corollary 24.
Let A=(ank) be an infinite matrix and define the matrix C=(cnk) by cnk=ank-an+1,kfor all n,k∈ℕ. Then, the necessary and sufficient conditions in order to A belongs to anyone of the classes (ℓ(B~,p):ℓ∞(Δ)), (ℓ(B~,p):c(Δ)) and (ℓ(B~,p):c0(Δ)) are obtained from the respective ones in Theorems 13 and 16 and Corollary 17 by replacing the entries of the matrix A by those of the matrix C, where ℓ∞(Δ), c(Δ), c0(Δ) denote the difference spaces of all bounded, convergent, and null sequences and are introduced by Kızmaz [35].
Corollary 25.
Let A=(ank) be an infinite matrix and define the matrix C=(cnk) by cnk=∑j=0najk for all n,k∈ℕ. Then the necessary and sufficient conditions in order to A belongs to anyone of the classes (ℓ(B~,p):bs), (ℓ(B~,p):cs) and (ℓ(B~,p):cs0) are obtained from the respective ones in Theorems 13, 16 and Corollary 17 by replacing the entries of the matrix A by those of the matrix C, where cs0 denotes the set of those series converging to zero.
5. Conclusion
The difference spaces ℓ∞(Δ), c(Δ), and c0(Δ) were introduced by Kızmaz [35]. Since we essentially employ the infinite matrices which is more different than Kızmaz and the other authors following him, and use the technique of obtaining a new sequence space by the matrix domain of a triangle limitation method. Following this way, the domain of some triangle matrices in the sequence space ℓ(p) was recently studied and were obtained certain topological and geometric results by Altay and Başar [14, 16], Choudhary and Mishra [10], Başar et al. [36], and Aydın and Başar [13]. Although bv(e,p)=[ℓ(p)]Δ is investigated, since B(1,-1)≡Δ, our results are more general than those of Başar et al. [36]. Also in case pk=p for all k∈ℕ the results of the present study are reduced to the corresponding results of the recent paper of Kirişçi and Başar [9]. We should note that the difference spaces Δc0(p), Δc(p) and Δℓ∞(p) of Maddox’s spaces c0(p), c(p), and ℓ∞(p) were studied by Ahmad and Mursaleen [37]. Of course, a natural continuation of the present paper is to study the sequence spaces [c0(p)]B(r~,s~), [c(p)]B(r~,s~) and [ℓ∞(p)]B(r~,s~) to generalize the main results of Ahmad and Mursaleen [37] which fills up a gap in the existing literature.
It is clear that Δ(1) can be obtained as a special case of B(r~,s~) for r~=e and s~=-e and it is also trivial that B(r~,s~) is reduced in the special case r~=re and s~=se to the generalized difference matrix B(r,s). So, the results related to the domain of the matrix B(r~,s~) are much more general and more comprehensive than the corresponding consequences of the domain of the matrix B(r,s). We should note from now that the main results of the present paper are given as an extended abstract without proof by Nergiz and Başar [38], and our next paper will be devoted to some geometric and topological properties of the space ℓ(B~,p).
Acknowledgments
The authors would like to thank Professor Bilâl Altay, Department of Mathematical Education, Faculty of Education, İnönü University, 44280 Malatya, Turkey, for his careful reading and constructive criticism of an earlier version of this paper which improved the presentation and its readability. The main results of this paper were presented in part at the conference First International Conference on Analysis and Applied Mathematics (ICAAM 2012) to be held October 18–21, 2012, in Gümüşhane, Turkey, at the University of Gümüşhane.
MaddoxI. J.Spaces of strongly summable sequences196718345355MR022114310.1093/qmath/18.1.345ZBL0156.06602SimonsS.The sequence spaces ℓ(pv) and m(pv)196515422436MR017632510.1112/plms/s3-15.1.422ZBL0128.33805NakanoH.Modulared sequence spaces195127508512MR004792910.3792/pja/1195571225ZBL0044.11302LorentzG. G.A contribution to the theory of divergent sequences194880167190MR002786810.1007/BF02393648ZBL0031.29501SrivastavaP. D.KumarS.Fine spectrum of the generalized difference operator Δv on sequence space ℓ1201082221233MR2763685ZBL1236.47029SrivastavaP. D.KumarS.Fine spectrum of the generalized difference operator Δuv on sequence space ℓ12012218116407641410.1016/j.amc.2011.12.010MR2879121PanigrahiB. L.SrivastavaP. D.Spectrum and fine spectrum of generalized second order difference operator Δuv2 on sequence space c02011915774MR2833753AkhmedovA. M.El-ShabrawyS. R.On the fine spectrum of the operator Δa,b over the sequence space c201161102994300210.1016/j.camwa.2011.03.085MR2799825ZBL1222.40002KirişçiM.BaşarF.Some new sequence spaces derived by the domain of generalized difference matrix20106051299130910.1016/j.camwa.2010.06.010MR2672929ZBL1201.40001ChoudharyB.MishraS. K.On Köthe-Toeplitz duals of certain sequence spaces and their matrix transformations1993245291301MR1223217ZBL0805.46008BaşarF.AltayB.Matrix mappings on the space bs(p) and its α-, β- and γ-duals20022117991MR1956161BaşarF.Infinite matrices and almost boundedness199263395402MR1196134ZBL0867.47021AydınC.BaşarF.Some generalizations of the sequence space apr2006302175190MR2397564AltayB.BaşarF.On the paranormed Riesz sequence spaces of non-absolute type2003265701715MR2045105ZBL1058.46002AltayB.BaşarF.Some paranormed sequence spaces of non-absolute type derived by weighted mean2006319249450810.1016/j.jmaa.2005.06.055MR2227919ZBL1105.46005AltayB.BaşarF.Generalization of the sequence space ℓ(p) derived by weighted mean2007330117418510.1016/j.jmaa.2006.07.050MR2302914ZBL1116.46003AltayB.BaşarF.Certain topological properties and duals of the domain of a triangle matrix in a sequence space2007336163264510.1016/j.jmaa.2007.03.007MR2348531ZBL1152.46003AydınC.BaşarF.Some new paranormed sequence spaces20041601–4274010.1016/j.ins.2003.07.009MR2041857ZBL1049.46002Grosse-ErdmannK.-G.On ℓ1-invariant sequence spaces2001262111213210.1006/jmaa.2001.7546MR1857218MaddoxI. J.19882ndCambridge, UKCambridge University PressMR984352WilanskyA.198485Amsterdam, The NetherlandsNorth-Holland PublishingNorth-Holland Mathematics StudiesMR738632JarrahA.MalkowskyE.BK spaces, bases and linear operators199052177191LascaridesC. G.MaddoxI. J.Matrix transformations between some classes of sequences19706899104MR0261212ZBL0193.41102Grosse-ErdmannK.-G.Matrix transformations between the sequence spaces of Maddox1993180122323810.1006/jmaa.1993.1398MR1250293ZBL0791.47029NandaS.Matrix transformations and almost convergence1976133305312MR0440245ZBL0358.46010BaşarF.AltayB.On the space of sequences of p-bounded variation and related matrix mappings200355113614710.1023/A:1025080820961MR2034909AltayB.BaşarF.MursaleenM.On the Euler sequence spaces which include the spaces ℓp and ℓ∞. I2006176101450146210.1016/j.ins.2005.05.008MR2207938AltayB.BaşarF.On some Euler sequence spaces of nonabsolute type200557111710.1007/s11253-005-0168-9MR2190951ZBL1096.46011BaşarF.KirişçiM.Almost convergence and generalized difference matrix201161360261110.1016/j.camwa.2010.12.006MR2764055ZBL1217.40001SönmezA.Almost convergence and triple band matrix201210.1016/j.mcm.2011.11.079KayadumanK.ŞengönülM.The spaces of Cesàro almost convergent sequences and core theorems20123262265227810.1016/S0252-9602(12)60176-3AltayB.BaşarF.Some paranormed Riesz sequence spaces of non-absolute type2006304591608MR2257553ZBL1118.46009NgP. N.LeeP. Y.Cesàro sequence spaces of non-absolute type1978202429433MR519380ŞengönülM.BaşarF.Some new Cesàro sequence spaces of non-absolute type which include the spaces c0 and c2005311107119MR2130500ZBL1085.46500KızmazH.On certain sequence spaces198124216917610.4153/CMB-1981-027-5MR619442ZBL0454.46010BaşarF.AltayB.MursaleenM.Some generalizations of the space bvp of p-bounded variation sequences200868227328710.1016/j.na.2006.10.047MR2369898AhmadZ. U.MursaleenKöthe-Toeplitz duals of some new sequence spaces and their matrix maps1987425761MR937453ZBL0647.46006NergizH.BaşarF.Some topological and geometric properties of the domain of the double sequential band matrix B(r˜,s˜) in the sequence space ℓ(p)20121470163168