New proper es of the Natarajan method of summability
P. N. Natarajan
Summary The (M , λ
n) method of summability was introduced in 2013 by Natarajan. In the paper some new proper es of this method are studied.
Keywords regular matrix;
the (M, λ
n) method of summability;
l-stronger;
l-equivalent;
ordered abelian semigroup;
infinite chain
MSC 2010
40C05; 40D05; 40G99 Received: 2014/05/08; Accepted: 2014/10/26
1. Introduc on and preliminaries
Throughout, the entries of matrices, sequences and series are real or complex numbers. To make the paper self-contained, we recall the following. Given an infinite matrix A ≡ (a
nk), n , k ∈ N
0∶= {0 , 1 , 2 , . . . }, and a sequence x = {x
k}, k ∈ N
0, by the A-transform of x = {x
k} we mean the sequence A (x) = {(Ax)
n},
(Ax)
n= ∑
∞k=0
a
nkx
k, n ∈ N
0,
where we assume that the series on the right converges. If lim
n→∞(Ax)
n= s, we say that x = {x
k} is A-summable or summable A to s. If lim
n→∞(Ax)
n= s whenever lim
k→∞x
k= s, we
P. N. Natarajan Old No. 2/3, New No. 3/3, Second Main Road, R.A. Puram, Chennai 600 028, India (e-mail: pinnangudinatarajan@gmail.com)
DOI 10.14708/cm.v55i1.810 © 2015 Polish Mathema cal Society
say that A is regular. The following result, which characterizes a regular matrix in terms of its entries, is well known (see, for instance, [2]).
1.1. Theorem. The infinite matrix A ≡ (a
nk) is regular if and only if
(i) sup
n⩾0
∞
∑
k=0
∣a
nk∣ < ∞ ,
(ii) lim
n→∞
a
nk= 0 , k ∈ N
0, (iii) lim
n→∞
∞
∑
k=0
a
nk= 1.
Let ℓ denote the Banach space of all sequences x = {x
k} such that ∑
∞k=0∣x
k∣ < ∞. We write A ≡ (a
nk) ∈ ( ℓ, ℓ ) if {(Ax)
n} ∈ ℓ whenever x = {x
k} ∈ ℓ . The following result is due to Knopp and Lorentz (see [3]):
1.2. Theorem. A ∈ ( ℓ, ℓ ) if and only if
sup
k⩾0
∞
∑
n=0
∣a
nk∣ < ∞ .
If A , B are infinite matrices such that lim
n→∞(Ax)
n= s implies that lim
n→∞(Bx)
n= s, then we say that A is included in B (or B includes A) and write A ⊆ B.
2. Main results
The (M , λ
n) method of summability was introduced by Natarajan in [ 4] and some of its proper es were studied in [4–6].
2.1. Defini on. Let {λ
n} be a sequence such that ∑
∞n=0∣λ
n∣ < ∞. The (M , λ
n) method is defined by the infinite matrix (a
nk), where
a
nk= ⎧⎪⎪ ⎪⎨ ⎪⎪⎪ ⎩
λ
n−k, k ⩽ n , 0 , k > n .
2.2. Remark. Note that the (M , λ
n) method reduces to the Y-method when λ
0= λ
1=
12and λ
n= 0, n ⩾ 2.
The following result is known (see [4], Theorem 2.3).
2.3. Theorem. The (M , λ
n) method is regular if and only if
∞
∑
n=0
λ
n= 1 .
Following Dafranza [1], we will study some proper es of the (M , λ
n) method. For convenience, we write (M , λ ) for (M , λ
n). Let M denote the set of all (M , λ ) methods. First note the following.
2.4. Theorem. For any method (M , λ ) ∈ M, (M , λ ) ∈ ( ℓ, ℓ ).
Let ℓ ((M , λ )) denote the set of all sequences x = {x
k} such that (M , λ )(x) ∈ ℓ . 2.5. Defini on. Given two methods (M , λ ), (M , μ ) in M, we say that
– (M , μ ) is ℓ -stronger than (M , λ ) if ℓ ((M , λ )) ⊆ ℓ ((M , μ ));
– (M , μ ) is strictly ℓ -stronger than (M , λ ) if ℓ ((M , λ )) ⫋ ℓ ((M , μ ));
– (M , λ ) and (M , μ ) are ℓ -equivalent if ℓ ((M , λ )) = ℓ ((M , μ )).
Given the methods (M , λ), (M , μ) in M, we formally define
λ (x) = ∑
∞n=0
λ
nx
n, μ (x) = ∑
∞n=0
μ
nx
nand
a (x) = λ (x) μ (x) = ∑
∞n=0
a
nx
n,
b (x) = μ (x) λ (x) =
∞
∑
n=0
b
nx
n.
Following an argument similar to the one used in [2, Theorem 18], we can prove the following results:
2.6. Theorem. If (M , λ ) ∈ M, then the series ∑
∞n=0λ
nx
nconverges for ∣x∣ < 1.
2.7. Theorem. If (M , λ ) , (M , μ ) ∈ M, then ∑
∞n=0a
nx
n, ∑
∞n=0b
nx
nhave posi ve radii of conver- gence and
(i) λ
n= a
nμ
0+ a
n−1μ
1+ ⋅ ⋅ ⋅ + a
0μ
n, (ii) μ
n= b
nλ
0+ b
n−1λ
1+ ⋅ ⋅ ⋅ + b
0λ
n.
Further, if s = {s
n} ∈ ℓ ((M , λ )), then the series s(x) = ∑
∞n=0s
nx
nhas posi ve radius of conver-
gence.
Given sequences λ = {λ
n}, μ = {μ
n}, let λ ∗ μ = {g
n} denote their Cauchy product, i.e., g
n= ∑
nk=0λ
kμ
n−k, n ∈ N
0.
2.8. Defini on. Given (M , λ ) , (M , μ ) ∈ M, we say that (M , λ ∗ μ) is the symmetric product of (M , λ ) and (M , μ ).
In the context of Defini on 2.8, ∑
∞n=0∣λ
n∣ < ∞, ∑
∞n=0∣μ
n∣ < ∞, and so ∑
∞n=0∣g
n∣ < ∞ in view of Abel’s theorem on Cauchy mul plica on of absolutely convergent series. Conse- quently, (M , λ ∗ μ) ∈ M.
We will need the following result.
2.9. Lemma. For sequences λ = {λ
n}, μ = {μ
n}, let γ = λ ∗ μ and let (M , λ ), (M , γ ) ∈ M. Then
ℓ ((M , λ )) ⊆ ℓ ((M , γ )) if and only if μ ∈ ℓ .
Proof. For any sequence x = {x
k},
((M , γ )(x))
n= γ
nx
0+ γ
n−1x
1+ ⋅ ⋅ ⋅ + γ
0x
n= (λ
nμ
0+ λ
n−1μ
1+ ⋅ ⋅ ⋅ + λ
0μ
n)x
0+ (λ
n−1μ
0+ λ
n−2μ
1+ ⋅ ⋅ ⋅ + λ
0μ
n−1)x
1+ ⋅ ⋅ ⋅ + (λ
0μ
0)x
n= (λ
nx
0+ λ
n−1x
1+ ⋅ ⋅ ⋅ + λ
0x
n)μ
0+ (λ
n−1x
0+ λ
n−2x
1+ ⋅ ⋅ ⋅ + λ
0x
n−1)μ
1+ ⋅ ⋅ ⋅ + (λ
0x
0)μ
n= ((M , λ )(x))
nμ
0+ ((M , λ )(x))
n−1μ
1+ ⋅ ⋅ ⋅ + ((M , λ )(x))
0μ
n= ∑
∞k=0
t
nk((M , λ )(x))
k,
where
t
nk= ⎧⎪⎪ ⎪⎨ ⎪⎪⎪ ⎩
μ
n−k, k ⩽ n , 0 , k > n .
In view of Theorem 1.2, ℓ ((M , λ )) ⊆ ℓ ((M , γ )) if and only if ∑
∞n=0∣μ
n∣ < ∞, i.e., μ ∈ ℓ , which completes the proof.
We deduce an inclusion result:
2.10. Theorem. Given the methods (M , λ ) , (M , μ ) ∈ M,
ℓ ((M , λ )) ⊆ ℓ ((M , μ )) if and only if b = {b
n} ∈ ℓ.
Proof. In Lemma 2.9 we replace the sequence μ by the sequence b, so γ = λ ∗ b = μ, and hence
ℓ ((M , λ ) ⊆ ℓ ((M , μ )) if and only if b = {b
n} ∈ ℓ .
2.11. Corollary. Let (M , λ ) , (M , μ ) ∈ M. Then
(i) ℓ ((M , λ )) = ℓ ((M , μ )) if and only if {a
n} ∈ ℓ and {b
n} ∈ ℓ , (ii) ℓ ((M , λ )) ⫋ ℓ ((M , μ )) if and only if {a
n} /∈ ℓ and {b
n} ∈ ℓ .
2.12. Corollary. For (M , λ ) ∈ M, let h(x) =
λ(x)1. Then ℓ ((M , λ )) = ℓ if and only if {h
n} ∈ ℓ . Proof. Let I be the iden ty mapping, so ℓ (I) = ℓ . Now, I (x) = ∑
∞n=0i
nx
n= 1, i
0= 1, i
n= 0, n ⩾ 1. Then a(x) =
λI(x)(x)= λ(x) and b(x) =
λ(x)I(x)=
λ(x)1= h(x). From Theorem 2.10 it follows that {h
n} ∈ ℓ .
2.13. Corollary. Let (M , λ ) , (M , μ ) ∈ M and γ = λ ∗ μ. Then ℓ ((M , λ )) ⊆ ℓ ((M , γ )).
2.14. Theorem. Let (M , λ )) ∈ M and let μ = {λ
′0, λ
1, λ
2, . . . }, λ
′0≠ λ
0. Then (M , μ ) ∈ M and ℓ ((M , λ) ∩ ℓ ((M , μ)) = ℓ.
Proof. Since {λ
n} ∈ ℓ , it follows that {μ
n} ∈ ℓ , hence the method (M , μ ) ∈ M. For any sequence x = {x
k},
((M , μ )(x))
n= μ
0x
n+ μ
1x
n−1+ ⋅ ⋅ ⋅ + μ
nx
0= λ
′0x
n+ λ
1x
n−1+ ⋅ ⋅ ⋅ + λ
nx
0= (λ
0x
n+ λ
1x
n−1+ ⋅ ⋅ ⋅ + λ
nx
0) + (λ
′0− λ
0)x
n= ((M , λ )(x))
n+ (λ
′0− λ
0)x
n. (1) Let {x
n} ∈ ℓ . Since (M , λ ) , (M , μ ) ∈ ( ℓ, ℓ ), hence by Theorem 2.4, {((M , λ )(x))
n} ∈ ℓ and {((M , μ )(x))
n} ∈ ℓ , so {x
n} ∈ ℓ ((M , λ )) ∩ ℓ ((M , μ )). Consequently, ℓ ⊆ ℓ ((M , λ )) ∩ ℓ ((M , μ )). Let {x
n} ∈ ℓ ((M , λ )) ∩ ℓ ((M , μ )). Then {((M , λ )(x))
n} ∈ ℓ and {((M , μ )(x))
n} ∈ ℓ . Now, from (1) it follows that {x
n} ∈ ℓ . Thus
ℓ ((M , λ ) ∩ ℓ ((M , μ )) ⊆ ℓ,
and so ℓ ((M , λ )) ∩ ℓ ((M , μ )) = ℓ , which completes the proof of the theorem.
3. M as an ordered abelian semigroup
We recall that M denotes the set of all (M , λ ) methods. In this sec on, we prove that M is an ordered abelian semigroup, the order rela on being the set inclusion between summability fields of type ℓ ((M , λ )), and the binary opera on being the symmetric product ∗.
3.1. Lemma. If (M , λ ) , (M , μ ) ∈ M and γ = λ ∗ μ, then (M , γ ) ∈ M and ℓ ((M , λ )) ∪ ℓ ((M , μ )) ⊆ ℓ ((M , γ )) .
Proof. Since ∑
∞n=0∣λ
n∣ < ∞ and ∑
∞n=0∣μ
n∣ < ∞, from Abel’s theorem we have ∑
∞n=0∣γ
n∣ < ∞, so (M , γ ) ∈ M. By Corollary 2.13,
ℓ ((M , λ )) ∪ ℓ ((M , μ )) ⊆ ℓ ((M , γ )) .
3.2. Lemma. Let (M , λ ) , (M , μ ) , (M , γ ) ∈ M. Let θ = λ ∗ γ and ϕ = μ ∗ γ.
(i) If ℓ ((M , λ )) ⊆ ℓ ((M , μ )), then ℓ ((M , θ )) ⊆ ℓ ((M , ϕ )).
(ii) If ℓ ((M , λ )) ⫋ ℓ ((M , μ )), then ℓ ((M , θ )) ⫋ ℓ ((M , ϕ )).
Proof. Let b (x) =
μλ(x)(x)and c (x) =
ϕθ(x)(x). By Theorem 2.10, {b
n} ∈ ℓ . We claim that {c
n} ∈ ℓ . First we note that
θ (x) = λ(x)γ(x) and ϕ(x) = μ(x)γ(x) , so
∞
∑
n=0
c
nx
n= c(x) = ϕ (x)
θ(x) = μ (x)γ(x) λ(x)γ(x) = μ (x)
λ(x) = ∑
∞n=0
b
nx
n.
Thus c
n= b
n, n ∈ N
0, and hence {c
n} ∈ ℓ . Using Theorem 2.10 again, we have
ℓ ((M , θ )) ⊆ ℓ ((M , ϕ )) . The second part of the theorem follows by Corollary 2.11(ii).
Therefore, we can state the main result of this sec on:
3.3. Theorem. With “strictly ℓ -weaker than” as the order rela on and the symmetric product ∗ as the binary opera on, M is an ordered abelian semigroup.
3.4. Lemma. Let (M , λ ) ∈ M. Define λ
−1= 0, μ
n=
λn−12+λn, n ⩾ 0. Then (M , μ ) ∈ M and
ℓ ((M , λ )) ⫋ ℓ ((M , μ )).
Proof. Since {λ
n} ∈ ℓ , {μ
n} ∈ ℓ , hence (M , μ ) ∈ M. Now,
μ (x) = ∑
∞n=0
μ
nx
n= ∑
∞n=0
λ
n−1+ λ
n2 x
n= 1 + x 2 λ (x) . It is clear that
ℓ ((M , λ )) ⫋ ℓ ((M , μ )) in view of Corollary 2.11(ii).
We conclude the paper with the following interes ng result:
3.5. Theorem. There are infinite chains of (M , λ ) methods in M.
Proof. Let {λ
(1)n} be a sequence such that ∑
∞n=0∣λ
(1)n∣ < ∞. Let λ
(1)(x) = ∑
∞n=0λ
(1)nx
n. Then (M , λ
(1)) ∈ M. For n ⩾ 2, define
λ
(n)(x) = ( 1 + x 2 )
n−1 ∞
∑
k=0