PII. S016117120100309X
http://ijmms.hindawi.com © Hindawi Publishing Corp.
GENERALIZATION OF SOME ABSOLUTE
SUMMABILITY METHODS
W. T. SULAIMAN
(Received 6 January 1999)
Abstract.Mazhar (1971) gave the characterization for the seriesannto be summable |N,pn|wheneveranis summable|C,α|k, α≥0, k≥1. Here we prove two theorems, the first concerns the sufficient conditions and the second the necessary conditions satisfied bynin order to haveannsummable|N,p¯ n|kwheneveranis summable|C,α|k, k≥1.
2000 Mathematics Subject Classification. Primary 40G.
Letanbe a given infinite series with partial sumssn.Letσnδandrnδdenote thenth Cesàro mean of orderδ (δ >−1)of the sequences{sn}and{nan}, respectively. The seriesanis said to be summable|C,δ|k, k≥1 if
∞
n=1
nk−1σδ
n−σn−δ 1k <∞, (1)
or equivalently
∞
n=1 n−1rδ
nk<∞. (2)
LetPnbe a sequence of positive real constants such that
Pn=p0+p1+···+pn, Pn → ∞asn → ∞,P−1=p−1=0. (3)
A seriesanis said to be summable|N,p¯ n|k, k≥1, (see [1]), if
∞
n=1
P n
pn k−1
tn−tn−1k<∞, (4)
where
tn= 1
Pn ∞
ν=0
Forpn=1/(n+1),the summability|N,p¯ n|kreduces to the well-known summability |R,logn,1|k.In general, the two summabilities|N,p¯ n|kand|C,α|kare independent of each other, but forpn=1 andα=1 they are equivalent. For any realαand integer
n≥0,we define
∆α n=
∞
ν=nA −α−1
ν−n ν, (6)
whenever the series is convergent.
TheoremA(see [2]). The necessary and sufficient conditions for the seriesann to be summable|N,p¯ n|wheneveranis summable|C,α|k,α≥0,k≥1are
(i) nα+1−(1/k)
∆α(n/n)∈k
,1/k+1/k=1; (ii) n−1/k
n∈k,0≤α≤1; (iii) nα−(1/k)
(pn/Pn)n∈k,α >1;
where (a) pn = O(pn+1), (b) (n+1)pn = O(Pn), and (c) Pn =O(nαpn) (α >1). Theorem A includes as special cases k=1and pn=1,the result of Mohapatra[4] and Mehdi[3], respectively.
We need the following lemmas. Lemma1(see [5]). Ifσ > δ >0,then
∞
n=ν+1
(n−ν)δ−1
nσ =O
νδ−σ. (7)
Lemma2. Ifσ > δ≥0, then
∞
n=ν+1 Aδ−1
n−ν
Aσ n =O
νδ−σ. (8)
Proof. Ifδ >0,then byLemma 1,
∞
n=ν+1 Aδ−1
n−ν
Aσ
n =O(1) ∞
n=ν+1
(n−ν)δ−1
nσ =O
νδ−σ. (9)
Ifδ=0,then byLemma 1,
∞
n=ν+1 A−1
n−ν
Aσ
n =O(1) 1 Aσ
n ∞
n=ν+1
A−1
Lemma3(see [6]). Letk 1,Pn=p0+p1+ ··· +pn→ ∞asn→ ∞, pn>0. Then for allν≥1,
1 kPk
ν−1
∞
n=ν
pn
PnPn−k 1
1
Pk ν−1
. (11)
We prove the following theorems.
Theorem4. Let{n}be monotonic nonincreasing sequence of constants. Suppose that{Pnn/n}is nonincreasing, and
npn=OPn. (12)
Then the following are sufficient conditions forannto be summable|N,p¯ n|k when-everanis summable|C,α|k,0< α≤αk≤1:
(i) n=O(npn/Pn), (ii) ∆n=O(pn/Pn).
Proof ofTheorem4. Define
Tn=
pn/Pn1/k
Pn−1
n
ν=1
Pν−1aνν, tnα=A1α n
n
ν=1 Aα−1
n−νναν,
Tn=
pn/Pn1/k
Pn−1
n
ν=1 Pν−1ν
ν
n
r=1 A−α−1
ν−r Aαrtrα
=
pn/Pn1/k
Pn−1
n
r=1 Aα
rtrα n
ν=r
Pν−1ν
ν A−α−ν−r1
=
pn/Pn1/k
Pn−1
n
r=1 Aα
rtrα n−r
ν=0
Pν−r−1ν+r
ν+r A−α−ν 1
=
pn/Pn1/k
Pn−1
n
r=1 Aα
rtrα n−r−1
ν=0
ν
µ=0 A−α−1
µ
∆νPν+r−1ν+r
ν+r
+ n−r
µ=0 A−α−1
µ Pn−1n
n
=
pn/Pn1/k
Pn−1
n
r=1 Aα
rtrα n−r−1
ν=0 A−α
ν
Pν+r−1ν+r
(ν+r )(ν+r+1)−
pν+rν+r
ν+r+1
+Pν+rν+∆rν+ν+r1
+A−α
n−rPn−n1n
,
=Tn,1+Tn,2+Tn,3+Tn,4. (13)
In order to prove sufficiency, by Minkowski’s inequality, it is sufficient to show that
m
r=1
m
n=1
Tn,1k= m n=1
pn
PnPn−k 1
n r=1 Aα
rtrα n−r−1
ν=0 A−α
ν (ν+Pν+r−r )(ν1+ν+rr+1) k ≤ m n=1 pn
PnPn−k 1
n r=1 Aα
rtrαPr−r1r n−r−1
ν=0 A−α
ν
ν+r+1 k
as Pnnn being nonincreasing
= m
n=1 pn
PnPn−k 1
n r=1 Aα
rtrαPr−r1r n
ν=r
(ν−r )−α
ν
k
=O(1)
m
n=1 pn
PnPn−k 1
n r=1 Aα
rtαrPr−r1rr−α k
≤O(1)m
n=1 pn
PnPn−k 1
n r=1
prtαrr pPr rr
k
≤O(1)m
n=1 pn
PnPn−1
n
r=1
prtαrk P
r
r pr k
rkn r=1
pr
Pn−1
k−1
=O(1)
m
r=1
prtrαk P
r
r pr k
rkm n=r
pn
PnPn−1
=O(1)
m r=1 pr Pr tα
rk P
r
r pr k
rk
=O(1)
m
r=1 1 rtrαk.
(15)
m
n=1
Tn,2k= m n=1
pn
PnPn−k 1
n r=1 Aα
rtrα n−r−1
ν=0 A−α
ν −νp+ν+rr+ν+r1 k ≤ m n=1 pn
PnPn−k 1
n r=1 Aα
rtrαprr n−r−1
ν=0 A−α
ν
ν+r+1 k
≤ m n=1
pn
PnPn−k 1
n r=1 tα
rprr k
≤O(1)m
n=1 pn
PnPn−1
n
r=1
tα
rkprrk n
r=1 pr
Pn−1
k−1
=O(1)
n
r=1
tα
rkprrk m
n=r
pn
PnPn−1
=O(1)
m r=1 pr Pr tα
rkrk
=O(1)
m
r=1 1
rtαrkrk
=O(1)
m
r=1 1
rtαrk asr=O(1)being nonincreasing.
m
n=1
Tn,3k= m n=1
pn
PnPn−k 1
n r=1 Aα
rtrα n−r−1
ν=0 A−α
ν Pν+rν+∆rν+ν+r1 k = m n=1 pn
PnPn−k 1
n r=1 Aα
rtrαPr∆r n−r−1
ν=0 A−α
ν
ν+r+1 k
≤O(1)
m
n=1 pn
PnPn−k 1
n r=1 tα
r pr∆r k
=O(1)
m
n=1 pn
PnPn−1
n
r=1
prtαrk P
r
pr k
∆rkn r=1
pr
Pn−1
k−1
=O(1)
m
r=1
prtrαk P
r
pr k
∆rkm n=r
pn
PnPn−1
=O(1)
m r=1 pr Pr tα
rk P
r
pr k
∆rk
=O(1)
m
r=1 1 rtrαk.
(17)
m
n=1
Tn,4k= m n=1
pn
PnPn−k 1
n r=1 Aα
rtrαA−αn−rPn−n1n k = m n=1 pn|n|k
Pnnk n r=1 Aα r
ktα rk
A−α n−r k n r=1 1 k−1 = m n=1 pn|n|k
Pnn n r=1 Aα r
ktα rk
A−α n−r k = m r=1 Aα rktαrk
m
n=r
pn|n|k
nPn
A−α n−rk
≤O(1)
m
r=1
Aα
rktrαkrk m n=r A−αk n−r n2
=O(1)m
r=1
Aα
rktrαkrkr−1−αk
=O(1)
m
r=1 1 rtrα
k.
(18)
Theorem5. The necessary conditions forannto be summable|N,p¯ n|kwhenever
anis summable|C,α|k, k >1, are (i) n=O{(Pn/pn)1/kν1−α−(1/k)}, (ii) ∆n=O(ν1−α−(1/k)).
Proof ofTheorem5. Fork 1,we define
A=ai:
aiis summable|C,α|k
, B=ai:
aiiis summableN,p¯ nk
These are BK-spaces, if normed by
a1=
∞
n=1 1 ntn
k1/k, a
2=
∞
n=1
Tnk1/k, (20)
respectively. Since(1/n)|tn|k<∞ ⇒|Tn|k<∞by the hypothesis ofTheorem 5, then
a1<∞ ⇒ a2<∞. (21) Consider the inclusion mapi:A→Bdefined byi(x)=x,iis continuous which follows asAandBare BK-spaces. Therefore, there exists a constantMsuch that
a2≤Ma1. (22) Applying the values oftn andTn, stated inTheorem 5, toa=eν−eν+1(whereeν is theνth coordinate vector), we have
tn=
0, ifn < ν,
ν Aα
ν, ifn=ν, ∆νAα−n−ν1
Aα
n , ifn > ν,
(23)
Tn=
0, ifn < ν,
p ν
Pν
1/k
, ifn=ν,
pn/Pn1/k
Pn−1 ∆
Pν−1ν, ifn > ν.
(24)
Equalities (20) give
a1=
1 ν
ν
Aα ν
k +
∞
n=ν+1 1 ν
∆νAα−n−ν1ν
Aα n
k1/k,
a2=
pν
Pν
νk+ ∞ n=ν+1
pn
PnPn−k 1
∆
pν−1νk 1/k
.
(25)
By inequality (22), pν
Pν
νk+ ∞ n=ν+1
pn
PnPn−k 1
∆
Pν−1νk≤Mk
1 ν
ν
Aα ν
k +
∞
n=ν+1 1 ν
∆ννAα−n−ν1
Aα n
k.
(26) Since∆νAα−1
n−ν
= −Aα−1
n−ν+ν∆νAα−n−ν1, then by Minkowski’s inequality
Mk
1 ν
ν
Aα ν
k +
∞
n=ν+1 1 ν
∆ννAα−n−ν1
Aα n
k=Ovk−αk−1+O
1 ν
∞
n=ν+1
(n−ν)αk−k
Aαk n
+O
νk
ν
∞
n=ν+1
(n−ν)αk−2k
nαk
=Oνk−αk−1+Oν−k+Oν−k,byLemma 1 =Oνk−αk−1.
As each part of the left-hand side of (26) isO(νk−αk−1), we have
ν=O P
v
pv
1/k
ν1−α−(1/k),
∆
Pν−1νk ∞
n=ν+1 pn
PnPn−k 1
=Oνk−αk−1, (28)
which, byLemma 3, implies ∆
Pν−1νk=OPν−k 1νk−αk−1. (29)
Now, as∆(Pν−1ν= −pνν+Pν∆ν,
∆ν=pPν νν+
1 Pν∆
Pν−1ν
=pPν νO
P ν
pν
1/k
ν1−α−(1/k)
+P1 νO
Pν−1ν1−α−(1/k)
=O
p ν
Pν
1−(1/k)
ν1−α−(1/k)
+P1 νO
Pν−1ν1−α−(1/k)
=Ov1−α−(1/k).
(30)
This completes the proof of the theorem.
References
[1] H. Bor,On two summability methods, Math. Proc. Cambridge Philos. Soc.97(1985), no. 1, 147–149.MR 86d:40004. Zbl 554.40008.
[2] S. M. Mazhar,On the absolute summability factors of infinite series, Tôhoku Math. J. (2)23 (1971), 433–451.MR 46#7758. Zbl 227.40005.
[3] M. R. Mehdi,Summability factors for generalized absolute summability. I, Proc. London Math. Soc. (3)10(1960), 180–200.MR 22#9760. Zbl 093.26203.
[4] R. N. Mohapatra,On absolute Riesz summability factors, J. Indian Math. Soc. (N.S.)32(1968), 113–129.MR 43#7811. Zbl 197.34404.
[5] W. T. Sulaiman,On absolute Cesàro summability of infinite series, Pure Appl. Math. Sci.31 (1990), no. 1-2, 25–30.MR 91j:40005. Zbl 713.40002.
[6] ,A study on absolute weighted mean summability method, Math. Sci. Res. Hot-Line 1(1997), no. 11, 25–29.MR 98m:40004.