Selçuk J. Appl. Math. Selçuk Journal of Vol. 12. No. 2. pp. 101-110, 2011 Applied Mathematics
∆m-Ideal Convergence Hafize Gok Gumus, Fatih Nuray
Department of Mathematics, Faculty of Science and Arts, Afyon Kocatepe University, Afyon, Turkiye
e-mail: hafi ze_ 1409@ hotm ail.com,f_ nuray@ hotm ail.com
Received Date: April 23, 2011 Accepted Date: May 23, 2011
Abstract. Statistical convergence has several applications in different fields of Mathematics: Number theory, trigonometric series, summability theory, prob-ability theory, measure theory, optimization and approximation theory. The notion of ideal convergence corresponds to a generalization of the statistical convergence. In this paper, we define the ∆m(c
I) spaces by using generalized sequence spaces and ideal convergence. Furthermore we establish some topo-logical results and give inclusion relation between ∆m(wT
p)−convergence and ∆m−ideal convergence.
Key words: Difference Sequence Spaces, I-Convergence 2000 Mathematics Subject Classification: 40G15, 40A35. 1. Introduction
Let l∞, c and c0be the linear spaces of bounded, convergent and null sequences, respectively normed by
kxk∞= sup k
xk
where k ∈ N = {1, 2, ...} , the positive integers. In 1981 the idea of difference sequence spaces was introduced by Kızmaz.(see [10]) He defined the sequence spaces
l∞(∆) = {x = (xk) : ∆x ∈ l∞} c(∆) = {x = (xk) : ∆x ∈ c} c0(∆) = {x = (xk) : ∆x ∈ c0}
where ∆x = (∆xk) = (xk − xk+1) and showed that these spaces are Banach spaces with the norm
Then Et and Çolak generalized the concept when they defined the sequence spaces l∞(∆m) = {x = (x k) : ∆mx ∈ l∞} c(∆m) = {x = (x k) : ∆mx ∈ c} c0(∆m) = {x = (xk) : ∆mx ∈ c0} where m ∈ N and ∆mx = (∆mx k) = (∆m−1xk − ∆m−1xk+1). They showed that spaces are Banach spaces with the norm
kxk∆= m P i=1|x
i| + k∆mxk∞.
(see [4] and [5]). Recently Et and Nuray defined a general sequence spaces ∆m(X) = {x = (x
k) : (∆mxk) ∈ X} where X is any sequence space and ∆mx = (∆mx
k) = (∆m−1xk − ∆m−1xk+1) for m ∈ N. They established some inclu-sion relations and topological results. They also studied ∆m−statistical conver-gence.(see [6]).
The notion of statistical convergence of sequences of real numbers was intro-duced by H. Fast (see [8]) and it is based on the notion of asymptotic density of a set A ⊆ N.
Definition 1. For A ⊆ N and n ∈ N we define dn(A) = 1 n n P k=1 χA(k)
where χA is the characteristic function of A. The numbers d(A) = lim inf
n→∞dn(A), d(A) = lim supn→∞ dn(A)
are called, respectively, lower and upper asymptotic density of A. If we have that the limit
d(A) = lim n→∞dn(A)
exists, then d(A) is called the asymptotic density of A.(see [11])
Definition 2. A sequence x = (xk) of real numbers is said to be statistically convergent to L ∈ R provided that for each ε > 0 we have d(Aε) = 0 where
Aε= {k ∈ N : |xk− L| ≥ ε} .
The set of all statistically convergent sequences is denoted by S. (see [11]) Definition 3. A sequence x = (xk) of real numbers is said to be ∆m−statistically convergent to L ∈ R provided that for each ε > 0,
The set of all ∆m−statistically convergent sequences is denoted by ∆m(S) . (see [6])
Definition 4. A family of sets I ⊂ 2Nis an ideal in N if and only if (i) ∅ ∈ I
(ii) For each A, B ∈ I we have A ∪ B ∈ I
(iii) For each A ∈ I and each B ⊆ A we have B ∈ I
An ideal is called non-trivial if N /∈ I and a non-trivial ideal is called admissible if {n} ∈ I for each n ∈ N.(see [12])
Definition 5. A family of sets F ⊂ 2Nis a filter in N if and only if (i) ∅ /∈ F
(ii) For each A, B ∈ F we have A ∩ B ∈ F
(iii) For each A ∈ F and each B ⊇ A we have B ∈ F (see [12]) Proposition 1. I is a non-trivial ideal in N if and only if
F = F (I) = {M = N\A : A ∈ I} is a filter in N. (see [12])
In 2000, Kostyrko, ˘Salát and Wilczy´nski introduced and studied the concept of I−convergence of sequences in metric spaces where I is an ideal (see [11]). Definition 6. A real sequence x = (xk) is said to be I−convergent to L ∈ R if and only if for each ε > 0 the set
{k ∈ N : |xk− L| ≥ ε}
belongs to I. The number L is called I−limit of the sequence x. (see [11]) The set of all I−convergent sequences is denoted by cI.
2. Main Results
Definition 7. Let I ⊆ 2Nbe a non-trivial ideal in N. The sequence x = (x k) of real numbers is said to be ∆mI−convergent to L ∈ R if for each ε > 0 the set
{k ∈ N : |∆mxk− L| ≥ ε}
belongs to I. In this case we write xk → L(∆m(cI)). The space of all ∆mI−convergent sequences is denoted by ∆m(c
I).
Example 1. Define If = {A ⊆ N : A is finite} . If is an admissible ideal in N and
∆m(cIf) = ∆
Example 2. Define the non-trivial ideal Id to be Id= {A ⊆ N : d(A) = 0} . In this case
∆m(cId) = ∆
m(S).
Theorem 1. Let I be an admissible ideal, and let x = (xk) and y = (yk) be real valued sequences.
(a) If xk → L1(∆m(cI)) and yk → L2(∆m(cI)) then xk + yk → L1 + L2(∆m(cI)).
(b) If xk → L1(∆m(cI)) and λ ∈ R then λxk→ λL1(∆m(cI)). Proof. (a) Let ε > 0 be given. If xk → L1(∆m(cI)) we can write
A1= n k ∈ N : |∆mx k− L1| < ε 2 o ∈ F (I) and if yk→ L2(∆m(cI)) we can write
A2= n k ∈ N : |∆myk− L2| < ε 2 o ∈ F (I).
In this case A1∩ A2 ∈ F (I) and since F (I) is a filter, A1∩ A26= ∅.. For each k ∈ A1∩ A2, |∆m(x k+ yk) − (L1+ L2)| = |(∆mxk− L1) + (∆myk− L2)| ≤ |∆mx k− L1| + |∆myk− L2| < ε2+ε2 = ε
(b) Let xk→ L1(∆m(cI)). Then, for each k ∈ A1, |∆m(λx
k) − λL1| = |λ(∆mxk− L1)| ≤ |λ| |∆mx
k− L1| < λε2
As ε > 0 was arbitrary, it follows that {k : |∆m(λx
k) − λL1| < η} ∈ F (I) for any η > 0.
Definition 8. Let I ⊆ 2Nbe an ideal in N. If {k + 1 : k ∈ A} ∈ I
for any A ∈ I, then I is said to be a translation invariant ideal. (see [3]) Corollary 1. If I is translation invariant and (xk) ∈ cI then (xk+1) ∈ cI.
Example 3. Id is a translation invariant ideal.
Proposition 2. Suppose that I ⊆ 2N is an admissible translation invariant ideal and m ∈ N. Then
∆m−1(cI) ⊆ ∆m(cI). (see [3])
Proof. Let x ∈ ∆m−1(cI) then¡∆m−1xk¢∈ cI. Since I is translation invariant, ¡ ∆m−1x k+1 ¢ ∈ cI. We know that (∆mxk) =¡∆m−1xk− ∆m−1xk+1¢ and hence x ∈ ∆m(c I) .
Definition 9. Let I ⊆ 2Nbe a non-trivial ideal in N. If for each ε > 0 there is a number N (= N (ε)) such that
{k ∈ N : |∆mxk− ∆mxN| ≥ ε} ∈ I then x is called ∆mI−Cauchy sequence.
Theorem 2. If x = (xk) is a ∆mI−convergent sequence then x is ∆mI−Cauchy sequence.
Proof. Suppose that xk → L(∆m(cI)) and ε > 0. Then A1= n k ∈ N : |∆mxk− L| < ε 2 o
belongs to F (I). Lets choose N ∈ A1. In this case |∆mxN− L| < ε2. |∆mxk− ∆mxN| < |∆mxk− L| + |∆mxN− L|
< ε 2+
ε 2 = ε Hence x is a ∆mI−Cauchy sequence.
Theorem 3. Let I be a non-trivial ideal in N and x be a sequence. If there is a ∆mI−convergent y such that
{k ∈ N : ∆mxk6= ∆myk} ∈ I then x is also ∆mI−convergent.
Proof. Assume that {k ∈ N : ∆mx
k 6= ∆myk} ∈ I and y is ∆mI−convergent to L ∈ R. For each ε > 0
As the right-hand side of the inclusion is in I, we have that {k ∈ N : |∆mxk− L| ≥ ε} ∈ I.
Proposition 3. Let I ⊆ 2N be a non-trivial ideal. Then ∆m(c) ⊆ ∆m(c I). Proof. We know that c ⊆ cI and for any X and Y spaces, if X ⊆ Y then ∆m(X) ⊆ ∆m(Y ). (see [6] ). Hence it is easy to see that ∆m(c) ⊆ ∆m(c
I). Remark 1. The converse of (i) is not always true.
Example 4. Define the sequence x such that ∆mxk = ( k12 if k is a square 0 otherwise then x ∈ ∆m(c Id) but x /∈ ∆ m(c) . Example 5. Define the ideal I such that
A ∈ I ⇐⇒ A eventually only contains even natural numbers then I is a non-trivial ideal in N. When
∆mx = (∆mxk) = (1, 1, 1, 0, 1, 1, 1, 0, 0, 1, 0, 1, 0, ...) we have
Aε0 = {k ∈ N : ∆mxk 6= 0}
= {1, 2, 3, 5, 6, 7, 10, 12, 14, 16, 18, ...} and x is ∆mI−convergent to zero i.e. x
k → 0(∆m(cI)). Hence Aε0 ∈ I and ∆mx ∈ c
I. Now lets look at the statistical convergence of the sequence. limn→∞n1|Aε0| = limn→∞n1 ¯ ¯(B) +n 2 ¯ ¯ = 1 2
where B is a finite number and |Aε0| is the cardinality of Aε0. Hence ∆mx /∈ S. Definition 10. Let T = (tnk)n,k∈N be a non-negative regular matrix, A ⊆ N and χA be the characteristic function of A. For each n ∈ N let
d(n)T (A) = P∞ k=1
tnkχA(k). If we have the limit
lim n→∞d
(n)
T (A) = dT(A) then this limit is called T −density of A. The ideal
is a non-trivial admissible ideal in N. When we take tnk=
( 1
n, k ≤ n 0, k > n
then we have ∆−statistical convergence, which is the case when m = 1 in ∆m−statistical convergence.
Definition 11. Let T = (tnk)n,k∈Nbe a non-negative regular matrix and define the space ∆m(wT p) to be ∆m(wpT) = ½ x = (xk) : lim n→∞ ∞ P k=1 tnk|∆mxk− L|p= 0, p > 0 for some L ¾ .
In this case we write xk→ L(∆m(w
T
p)).
Theorem 4. Let p ∈ R, 0 < p < ∞, T = (tnk) be a non-negative regular matrix and A ⊆ N. If xk → L(∆m(wTp)) then xk → L(∆m(cIdT)). If x ∈ ∆m(l∞) and xk → L(∆m(cIdT)) then xk→ L(∆
m(wT p)). Proof. Assume that xk → L(∆m(wTp)) then
lim n→∞ ∞ P k=1 tnk|∆mxk− L|p= 0 for some L. Let ε > 0 be given and define
A∆mε= {k ∈ N : |∆mxk− L| ≥ ε} and write ∞ P k=1 tnk|∆mxk− L|p = P k∈A∆mε tnk|∆mxk− L|p+ P k /∈A∆mε tnk|∆mxk− L|p ≥ Ã P k∈A∆mε tnk ! εp Then we have xk→ L(∆m(cIdT)).
Now suppose that x ∈ ∆m(l
∞) and xk → L(∆m(cIdT)). Then there is a set A ∈ F (IdT) such that lim k∈A(∆ mx k− L) = 0 . ∞ P k=1 tnk|∆mxk− L|p = P k∈A tnk|∆mxk− L|p+ P k /∈A tnk|∆mxk− L|p = P∞ k=1 tnkχA(k) |∆mxk− L|p+ ∞ P k=1 tnkχAc(k) |∆mxk− L|p
where Ac is complement of A. If we consider regularity of T, Ac ∈ IdT and
boundedness of (∆mx
k) right side tends to zero. Hence left side also tends to zero.
3. Matrix Transformations
Kızmaz obtained some matrix transformations between c and c(∆) spaces (see [10]) and Et and Çolak generalized them for c and ∆m(c). (see [4]) Gok, studied about some matrix transformations related with summability methods and also investigated matrix transformations by using I−convergence and she obtained characterizations of matrix transformations between cI and ∆ (cI)∩∆ (l∞). (see [2] , see [3]). In this part of the paper we generalize these results for ∆m(c
I) ∩ ∆m(l
∞).
Theorem 5 Let T be a non-negative matrix, T ∈ (l∞, l∞) and I be an ad-missible ideal in N. T ∈ (cI ∩ l∞, cI) and I − lim x = I − lim(T x) if and only if (i) ½ n : ¯ ¯ ¯ ¯kP∈Atnk ¯ ¯ ¯ ¯ ≥ ε ¾
∈ Ifor all A ∈ I and ε > 0.
(ii) I − lim P∞ k=1
tnk= 1
(see [2] and [3])
Theorem 6. Let T = (tnk) be a non-negative regular matrix, T ∈ (l∞, l∞), and I be an admissible translation invariant ideal in N. Then T ∈ (∆ (cI) ∩ ∆ (l∞) , cI) if and only if (i) ( n ∈ N : ¯ ¯ ¯ ¯ ¯ P k∈A à ∞ P j=k tnj !¯¯ ¯ ¯ ¯≥ ε )
∈ I for all A ∈ I and ε > 0.
(ii) I − limn µP∞ k=1 ktnk ¶ = 1 (see [2] and [3])
Theorem 7. Let T = (tnk) be a nonnegative regular matrix, T ∈ (l∞, l∞) and I be an admissible translation invariant ideal in N. Then, for each m ∈ N, T ∈ (∆m(c I) ∩ ∆m(l∞) , cI) if and only if (i) ( n ∈ N : ¯ ¯ ¯ ¯ ¯ P k∈A à ∞ P j=k (j−(k−1))(j−(k−2))...(j−(k−(m−1))) (m−1)! tnk !¯¯¯ ¯ ¯≥ ε ) ∈ I for all A ∈ I and ε > 0. (ii) I − limn µ∞ P k=1 k(k+1)(k+2)...(k+(m−1) m! ¶ = 1
Poof. Suppose that T ∈ (∆m(cI) ∩ ∆m(l∞) , cI) and define the matrix Smby Sm= ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ 1 0 0 0 0 ... m 1 0 0 0 ... m(m+1) 2! m 1 0 0 ... m(m+1)(m+2) 3! m(m+1) 2! m 1 0 ... m(m+1)(m+2)(m+3) 4! m(m+1)(m+2) 3! m(m+1) 2! m 1 ... .. . ... ... ... ... ... ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦
Then for each x ∈ cI ∩ l∞ we have that ∆m(Smx) =
(
−xn+m , if m is odd xn+m , if m is even By the translation invariance of the ideal ∆m(Smx) ∈ c
I ∩ l∞ and hence (Smx)n ∈ cI(∆m)∩l∞(∆m). By the definition of (Sm)−1, for each y ∈ (∆m(cI)∩ ∆m(l
∞)) we have (Sm)−1(y) ∈ cI∩l∞. Then the matrix R = T Smmaps cI∩l∞ into cI. If we apply theorem 5 to this matrix, then we have that T satisfies (i) and (ii). In order to prove the converse, note that (i) and (ii) yield that R is a matrix transformation between cI∩ l∞and cI. For each x ∈ ∆m(cI) ∩ ∆m(l∞)
³
R¡S−1¢m´=³(T Sm)¡S−1¢m´(x) = T³Sm¡S−1¢m´(x) = T (x) and, as x is arbitrary, we have T ∈ (∆m(c
I) ∩ ∆m(l∞) , cI). References
1. Bekta¸s,Ç.A., Et, M.and Çolak R., 2004, Generalized difference sequence spaces and their dual spaces, J.Math.Anal.Appl. 292, 423-432
2. Gok, H., 2005, Some summability methods and matrix transformations, Master’ Thesis, Afyon Kocatepe University, Science Institute, Afyonkarahisar
3. Gok Gumus, H., I−convergence and asymptotic I−equivalence of difference se-quences, PhD. thesis, Afyon Kocatepe University, Science Institute, Afyonkarahisar, to appear
4. Et, M., 1993, On Some Difference Sequence Spaces, Doga Tr. J. of Mathematics 17, pp. 18-24
5. Et, M.and Çolak, R., 1995, On Some Generalized Difference Sequence Spaces, Soochow Journal of Mathematics, Volume 21,No.4, pp. 377-386
6. Et, M and Nuray, F., 2001, ∆m−Statistical Convergence, Indian J. pure appl.
Math., 32 (6), 961-969
7. Et, M.and Çolak, R., 1997, On some generalized difference sequence spaces and related matrix transformations, Hokkaido Mathematical Journal Vol.26, 483-492 8. Fast, H., 1951, Sur la convergence statistique, Colloq. Math. 2, 241—244.
9. Fridy, J. A., 1985, On statistical convergence, Analysis, 5, 301-313
10. Kızmaz, H., 1981, On Certain Sequence spaces, Canad. Math. Bull. Vol. 24(2) 11. Kostyrko. P., ˘Salát, T. and Wilezy´nski, 2000, W., I-Convergence, Real Analysis Exchange Vol. 26(2), pp. 669-686
12. Kostyrko, P., Maˇcaj, M., Šalát, T. and Sleziak, M., 2005, I-Convergence and extremal I-limit points, Math. Slovaca, 55, 443—464