Abstract:
Background and Objective: This study comes from the notion of I-convergence which is the generalization of statistical convergence. The idea of I-convergent sequence spaces was motivated by the statistical convergence for double sequence. The I-convergence for double sequence in real line and general metric space. Methodology: This has motivated us to study the ideal convergence for double sequences using compact operator and define these spaces
INTRODUCTION
Fast1 and Steinhaus2 were the first who introduced the notion of the statistical convergence independently. Later on it was further investigated from a sequence space point of view and linked with suitability theory by Fridy3, Salat4 and Tripath5. The above concept is extended to double sequences by using the idea of a two dimensional analogue of natural density6. Kostyrko et al.7,8 defined I-convergence for single sequences which is a natural generalization of statistical convergence.
The idea of I-convergence is based on the notion of the ideal I of subsets of ℕ, the set of positive integers. Salat et al.9 studied the concept of I-convergence and I-Cauchy for sequences and proved some properties. Tripathy and Hazarika10,11 introduced some spaces of I-convergence of single sequences and proved some properties related to the solidity, symmetrically, completeness and denseness. Das et al.12 defined the concept of I-convergence for double sequences.
Let us denote 2ω for the space of all real or complex double sequences x = (xi,j), where i, j ∈ ℕ. A double sequence x = (xi,j) ∈ 2ω is said to be Pringsheims convergent to some number j ∈ ℂ (or P-convergent) if for given ε>0, there exists an integer N such that:
(1) |
It will be written as:
where, i and j tending to infinity and independent of each other13.
Definition 1: A normal linear space X is said to be a Banach space if it is complete, that is if every Cauchy sequence in X is convergent in X14.
Definition 2: An operator T is said to be linear operator if the domain D(T) of T is a vector space and the rang R(T) lies in a vector space over the same field15:
T(αx+βy) = αT(x)+βT(y)∀α, β∈K, for all x, y∈D(T)
Definition 3: Let X and Y be two normed linear spaces and T: D(T)→Y be a linear operator, where D(T)⊂X. Then, the operator T is said to be bounded, if there exists a positive real k>0 such that:
||Tx||<k||x||, for all x∈D(T)
the set of all bounded linear operators B(X, Y) is a linear space normed by
Definition 4: A linear space T: X→Y is said to be a compact linear operator (or completely continuous linear operator) if T maps every bounded sequence xk in X onto a sequence (Txk) in Y which has a convergence subsequence15. The set of all compact linear operators C(X,Y) is closed subspace of B(X, Y) and it is Banach space if Y is Banach space.
Throughout the paper, it was denoted 2l∞, 2c and 2c0 as the Banach spaces of bounded, convergent and null double sequences of reals respectively with norm:
Following Basar and Altay16 and Sengonul17, it was introduced the double sequence spaces 2S and 2S0 with help of compact operator T on ℝ as follows:
2S = {x = (xi, j) ∈ 2l∞: Tx ∈ 2c},
2S0 = {x = (xi, j) ∈ 2l∞: Tx ∈ 2c0}
Definition 5: A sequence x = (xk) ∈ ω is said to be statistically convergent to limit L if for every ε>0, the density of the set {k ∈ ℕ: |xk-L|>ε} equal zero.
Definition 6: Let ℕ×ℕ be a non-empty set. A family of sets I⊆2ℕ×ℕ is said to be an ideal if:
(i) | φ ∈ I |
(ii) | A, B ∈ I⇒A∪B ∈ I, (additivity) |
(iii) | A ∈ I, B⊆A⇒B ∈ I, (hereditary) |
• | An ideal I⊆2ℕ×ℕ is said to be non-trivial if I≠2ℕ×ℕ |
• | A non-trivial ideal I⊆2ℕ×ℕ is said to be admissible if I⊇{{x}: x ∈ ℕ×ℕ} |
• | A non-trivial ideal I⊆2ℕ×ℕ is said to be maximal if there cannot exist any non-trivial ideal J≠I containing I as a subset18 |
Definition 7: Let ℕ×ℕ be a non-empty set. Then a family of sets F⊆2ℕ×ℕ is said to be a filter on ℕ×ℕ if and only if18:
(i) | φ ∉ F |
(ii) | A, B ∈ F⇒A∩B ∈ F |
(iii) | A ∈ F with A⊆B⇒B ∈ F |
Remark 1: For each ideal I there is a filter F(I) which correspponding to I (filter associate with ideal I), that is18:
(2) |
Definition 8: A double sequence (xi,j) ∈ 2ω is said to be I-convergent to a number L ∈ ℝ if, for every ε>0, the set:
(3) |
And write I-lim(i,j)xi,j = L. In case L = 0 then (xi,j) ∈ 2ω is said to be I-null.
Definition 9: A double sequence (xi,j) ∈ 2ω is said to be I-Cauchy if, for each ε>0, there exists a numbers s = s(ε) and t = t(ε) such that the set:
{(i, j) ∈ ℕ×ℕ: |(xi,j)-(xs,t)|>ε} ∈ I
Definition 10: A double sequence (xi,j) ∈ 2ω is said to be I-bounded if there exists M>0,such that, the set:
{(i, j) ∈ ℕ×ℕ: |(xi,j)|>M} ∈ I
Definition 11: Let x = (xi,j) and y = (yi,j) be two double sequences. It can say that xi,j = yi,j for almost all i and j relative to I (in short a.a.i and j.r.I) if the set:
{(i, j) ∈ ℕ×ℕ: xi,j≠yi,j} ∈ I
Definition 12: Let x = (xi,j) be a double sequence and I be an ideal in ℕ×ℕ. A subset D of ℂ, the field of complex numbers, is said to contain xi,j for a.a.i and j.r.I if the set:
{(i,j) ∈ ℕ×ℕ: xi,j ∉ D} ∈ I
Definition 13: A double sequence space E is said to be solid or normal, if (αi,jxi,j) ∈ E whenever (xi,j) ∈ E and for any double sequence of scalars (αi,j) with |(αi,j) |<1, for all (i,j) ∈ ℕ×ℕ.
Definition 14: A double sequence space E is said to be symmetric, if (xπ(i,j)) ∈ E whenever (xi,j) ∈ E, where π(i,j) is a permutation on ℕ×ℕ.
Definition 15: A double sequence space E is said to be sequence algebra, if (xi,j)×(yi,j) = (xi,j.yi,j) ∈ E whenever xi,j, yi,j ∈ E.
Definition 16: A double sequence space E is said to be convergent free, if (yi,j) ∈ E whenever (xi,j) ∈ E and xi,j = 0 implies yi,j = 0, for all (i,j) ∈ ℕ×ℕ.
Definition 17: Let K = {(in, jn) ∈ ℕ×ℕ: i1<i2< and j1<j2< } ⊆ N×N and let E be a double sequence space. A K-step space of E is a sequence space:
A canonical pre-image of a step space is a set of canonical pre-images of all elements in
Definition 18: A double sequence space E is said to be monotone, if it is contains the canonical pre-images of it is step space.
Definition 19: A map h: D⊂X→ℝ is said to satisfy Lipschitz condition if:
|h(x)-h(y)|<k|x-y|
where, k is known as Lipschitz constant.
Definition 20: The I-convergence can be considered as a summability methods that (In case of admissibility of I are regular). Denoted by F(I) the convergence field of I-convergence, that is:
F(I) = {x = xk) ∈ l∞: there exists I-lim x ∈ ℝ}
The convergence field F(I) is a closed linear subspace of l∞ with respect to the supremum norm:
The function h: F(I)→ℝ defined by h(x) = I-lim x, for all x ∈ F(I) is said to be Lipschitz function17.
Following lemmas were used to establish some results of this article:
(i) | Every solid is monotone17 |
(ii) | Let K ∈ F(I) and M ⊆ ℕ. If M ∉ I, then M ∩ K ∉ I |
(iii) | If I ⊂ 2ℕ and M ⊆ ℕ. If M ∉ I, then M ∩ ℕ ∉ I18 |
Throughout this article, T is considered as a compact operator on the space ℝ.
RESULTS
In this article the following classes of double sequences are studied:
(4) |
(5) |
(6) |
(7) |
Theorem 1: The classes of double sequences
Proof: Let x = (xi,j), y = (yi,j) be two arbitrary elements of the space 2S1 and α, β are scalars. Now, since (xi,j), (yi,j) ∈ 2S1 then for given ε>0, there exist L1, L2 ∈ ℂ, such that:
and:
Now, let:
be such that
(8) |
Thus, the sets on right hand side of Eq. 8 belongs to F(I). By definition of filter associate with ideal, the complement of the set on left hand side of (Eq. 8) belongs to I. This implies that α(xi,j)+β(yi,j) ∈ 2S1. Hence 2S1 is linear space.
Theorem 2: The spaces 2S1 and are 2S1 normed spaces normed by:
Proof: The proof of the result is easy in view of existing techniques and hence omitted.
Theorem 3: A double sequence = (xi,j) ∈ 2l∞ is I-converges if and only if for every ε>0, there exists s = s(ε), t = t(ε) ∈ ℕ×ℕ, such that:
{(s, t) ∈ ℕ×ℕ: |T(xs,t)-L|<ε} ∈ F (I)
Proof: Suppose that the double sequence x = (xi,j) ∈ 2l∞ is I-convergent to some number L ∈ ℂ, then for a given ε>0, the set:
Fix an integers s = s(ε), t = t(ε) ∈ Bε. Then:
which holds for all (i,j) ∈ Bε. Hence the set:
{(i,j) ∈ ℕ×ℕ: |T(xi,j)-T(xs,t)|<ε} ∈ F(I)
Conversely, suppose that:
{(i,j) ∈ ℕ×ℕ: |T(xi,j)-T(xs,t)|<ε} ∈ F(I)
Then the set:
Cε = {(i,j) ∈ ℕ×ℕ: T(xi,j) ∈ [T(xi,j)-ε, T(xi,j)+ε]} ∈ F(I)
for all ε>0. Let Jε = [T(xi,j) ∈ [T(xi,j)-ε, T(xi,j)+ε]. If fix ε>0, then it was Cε ∈ F(I)as well as
That is the set:
{(i,j) ∈ ℕ×ℕ: T(xi,j)} ∈ F(I)
That is:
diam J< diam Jε
where, the diam J denote the length of interval J. Proceeding in this way, it will have a sequence of closed intervals:
Jε = I0 ⊇ I1 ⊇ ⊇ Ik ⊇
with the property that:
diam Ik < diam Lk+1, for k = (1,2,3 )
and:
{(i,j) ∈ ℕ×ℕ: T(xi,j) ∈ Ik} ∈ F(I), for k = (1, 2, 3 )
Then there exists a number L ∈∩ Ik where k ∈ ℕ such that L = 1-lim T(xi,j). Hence the result holds.
Theorem 4: Let I be an admissible ideal. Then the following are equivalent:
(i) | (Xi,j) ∈ 2S1 |
(ii) | There exists (yi,j) ∈ 2S1 such that xi,j = yi,j, for a, a, i and j, r, I |
(iii) | There exists (yi,j) ∈ 2S1 and (zi,j) ∈ |
(i,j) ∈ ℕ×ℕ and {(i,j) ∈ ℕ×ℕ: |T(xi,j)-L|>ε} ∈ I
There exists a subset K = {(is, jt): s, t ∈ ℕ, i1<i2<i3<⋯ and j1<j2<j3<⋯} of ℕ×ℕ, such that K ∈ F(I) and limi,j|T (xi,j)-L| = 0.
Proof: (i) Implies (ii): Let xi,j ∈ 2SI, then for any ε > 0, there exists a number L ∈ ℂ such that the set:
{(i,j) ∈ ℕ×ℕ: |T(xi,j)-L|>ε} ∈ I
Let (ms,t) be an increasing double sequence with ms,t ∈ ℕ×ℕ such that:
{(i,j)<ms,t: |T(xi,j)-L|>(st)‾1} ∈ I
Define a double sequence (yi,j) as (yi,j) = (xi,j) for all (i,j)<ms,t. For ms,t<(i,j)<ms+1,t+1, (s,t)∈ℕ×ℕ. That is:
Then yi,j ∈ 2SI and from the following inclusion:
{(i,j)< ms,t: xi,j≠yi,j}⊆{(i,j) ∈ ℕ×ℕ: |T(xi,j)-L|> ε} ∈ I
It get for xi,j = yi,j for a, a, I and j, r, I.
(ii) Implies(iii): For xi,j ∈ 2SI there exists yi,j ∈ 2SI such that xi,j = yi,j for a, a, I and j, r, I.
Let K = {(i,j) ∈ ℕ×ℕ: xi,j≠yi,j}, then K ∈ I. Define a double sequence (zi,j) as:
Then zi,j ∈ 2SI and yi,j ∈ 2SI.
(iii) Implies (iv): Let Pc = {(i,j) ∈ ℕ×ℕ: |T(zi,j)| > ε} ∈ I and K = Pc = {(is,jt) ∈ N×ℕ: s, t ∈ ℕ, i1<i2<i3<⋯ and j1<j2<j3<⋯} ∈ F(I).
Then
(iv) Implies (i): Let {(is,jt)∈N×ℕ: s, t ∈ ℕ, i1<i2<i3<⋯ and j1<j2<j3<⋯} ∈ F(I). And let lims,t |T(xis, jt)-L| = 0. Then for any ε>0 and by lemma (II) have:
{(i,j) ∈ ℕ×ℕ: |T(xi,j)-L|>ε}⊆Kc∪{(i,j) ∈ K: |T(xi,j)-L|>ε}
Thus (xi,j) ∈ 2SI.
Theorem 5: The function h: 2SI→ℝ defined by h(x) = I-lim x, for all x ∈ 2SI is a lipschitz function and hence uniformly continuous.
Proof: Clearly the function is well defined. Let x = (xi,j), y = (yi,j) ∈ 2SI, x≠y, then the sets:
Ax = {(i,j) ∈ ℕ×ℕ: |T(x)-h(x)|>∥x-y∥} ∈ I
Ay = {(i,j) ∈ ℕ×ℕ: |T(y)-h(y)|>∥x-y∥} ∈ I
where, ∥x-y∥ = supi,j |T(xi,j-yi,j)|. Thus the sets:
Bx = {(i,j) ∈ ℕ×ℕ: |T(x)-h(x)|<∥x-y∥} ∈ F(I)
By = {(i,j) ∈ ℕ×ℕ: |T(y)-h(y)|<∥x-y∥} ∈ F(I)
Hence, B = Bx ∩ By ∈ F(I), so that B is non-empty set, can choose (i,j) ∈ B, therefore:
|h(x)-h(y)| <|h(x)-T(x)|+|T(x)-T(y)|+|T(y)-h(y)|< 3∥x-y∥
Thus, h is lipschitz function and hence uniformly continuous.
Theorem 6: If T is an identity operator and h: 2SI→ℝ is a function defined by h(x) = I-lim x, for all x ∈ 2SI and if x = (xi,j), y = (yi,j)x ∈ 2SI then (x.y) ∈ 2SI and h(x.y) = h(x).h(y).
Proof: For ε>0, the sets:
(9) |
(10) |
where, ∥x-y∥ = ε. Now since T is identity operator:
(11) |
As 2SI ⊆ 2l∞, there exists an M ∈ ℝ such that |xi,j|<M and |h(y)|<M.
Therefore, from the Eq. 9, 10 and 11:
|T(x.y)-h(x) h(y)| = |T(xi,j yi,j)-h(x) h(y)|
< M ∈+M ∈ = 2M ∈
for all (i,j) ∈ Bx ∩ By ∈ F(I). Hence, (x.y) ∈ 2SI and h(x.y) = h(x).h(y).
Theorem 7: The space 2SI is solid and monotone.
Proof: Let (xi,j) ∈
(12) |
Let (αi,j) be a double sequence of scalars with |αi,j|<1 for all (i, j) ∈ ℕ×ℕ. Therefore:
|T (αi,j, xi,j)| = |αi,j T(xi,j)|
< |αi,j| |T (xi,j)| < |T (xi,j)|, for all (i, j) ∈ ℕ×ℕ
Thus, from the above inequality and Eq. 12:
{(i, j) ∈ ℕ×ℕ: |T (αi,j xi,j)|>ε}⊆{(i, j) ∈ ℕΧℕ: |T(xi,j)|>ε} ∈ I
Implies that:
{(i,j) ∈ ℕ×ℕ: |T(αi,j xi,j)|>ε} ∈ I
Therefore,
Theorem 8: The inclusions
Proof: Let (xi, j) ∈ 2SI. Then there exists a number L ∈ ℝ such that:
That is, the set:
{(i,j) ∈ ℕ×ℕ:|T(xi,j)-L|>ε} ∈ I
Where:
|T(xi,j)| = |T(xi,j)-L+L|<|T(xi,j))-L|+|L|
Taking the supremum over i and j on both sides, it get
The inclusion:
It was showed that ap,q converse
Theorem 9: The set is 2SI closed subspace of 2l∞.
Proof: Let
I-lim x = a
Since
That is for a given ε>0, it have:
Now, it have sets:
and:
Then,
B = {(i, j) ∈ ℕ×ℕ: |ap,q-as,t)|<ε}, then Bc ∈ I
Consider n0 ∈ BC. Then for each p, q, s, t>n0 it have:
Thus, (ap, q) is a Cauchy sequence of scalars in ℂ, so there exists a scalar a ∈ ℂ such that (ap, q)→a as p, q→∞.
For this step, let 0<δ<1 be given. Then it showed that if:
U = {(i,j) ∈ ℕ×ℕ: |T(x)-a|<δ} then Uc ∈ I
Since
(13) |
Which implies that PC ∈ I. The numbers p0, q0 can be chosen together with Eq. 13, it have:
Which implies that QC ∈ I. Since:
Then it have a subset of such that SC ∈ I. Where:
Let UC = PC∪QC∪SC, where:
U = {(i, j)∈ℕ×ℕ: |T(x)-a|<δ}
Therefore, for (i, j)∈UC each it have:
(14) |
The sets on the right hand side of Eq. 14 are belongs to F(I). Therefore, the set on the left hand side of Eq. 14, it is also belongs to F(I). Hence its complement belongs to I. Thus, I-Limx = α.
DISCUSSION
The notion of I-convergence, which is a generalization of of statistical convergence, was introduced by Kostyrko et al.8. The notion of statistical convergence of double sequences x = xij has been defined and studied by Mursaleen and Edely19. Motivated by this definition, Das et al.12 studied the notion of I and I*-convergence of double sequences in R. Recently, a compact operator was used to define the single sequence spaces with the help of the notion I-convergence in more general settings by Khan and Ebadullah20 and Khan et al.21-23. It keep the same direction up, in order to prove that a compact operator can be also used to define spaces of ideal convergence for double sequences. Indeed, most of our results in this paper are just minor adaptations of results in Khan et al.23-25, through which it was seen that the compact operator has preserved some topological and algebraic properties for these spaces. As no contradictions are yet found in this study. Regarding this case, it was mentioned some different attempts that have been taken place in some previous works which used different operators and the compact operator as well26. The opportunity is still open for other researchers to study some other different properities by the use of compact operator in an extended way to fugure out whether there is any discrepancy or not.
CONCLUSION
In this study, the compact linear operator is used to define spaces of ideal convergent double sequence
SIGNIFICANT STATEMENT
This study discovers the idea of ideal convergence double sequence space defined by compact operator that can be beneficial for scholars working in this field. This study will help the researcher to uncover the critical areas of ideal convergence by using compact operator that many researchers were not able to explore. Thus a new theory on ideal convergence double sequence spaces may be arrived at.