Vol. 19 (2018), No. 2, pp. 823–833 DOI: 10.18514/MMN.2018.2233
APPROXIMATION BY N ¨ORLUND AND RIESZ TYPE DEFERRED CES `ARO MEANS IN THE SPACE HP.!/
U ˇGUR DE ˇGER AND HILAL BAYINDIR Received 15 February, 2017
Abstract. The deferred Ces`aro transformations which have useful properties not possessed by the Ces`aro transformation was considered by R.P. Agnew in [1]. In [9], Deˇger and K¨uc¸¨ukaslan introduced a generalization of deferred Ces`aro transformations by taking account of some well known transformations such as Woronoi-N¨orlund and Riesz, and considered the degree of ap- proximation by the generalized deferred Ces`aro means in the spaceH.˛; p/,p1,0 < ˛1 by concerning with some sequence classes. In 2014, Nayaket al.studied the rate of convergence problem of Fourier series by deferred Ces`aro mean in the spaceHp.!/introduced by Daset al.
in [7].
In this study we shall give the degree of approximation by the generalized deferred Ces`aro means in the spaceHp.!/. Therefore the results given in [17] are generalized according to the summability method.
2010Mathematics Subject Classification: 40G05; 41A25; 42A05; 42A10 Keywords: deferred Ces`aro means, N¨orlund means, generalized H¨older metric
1. INTRODUCTION
Letf be a2 periodic function andf 2LpWDLp.0; 2/forp1and let sn.fIx/D1
2a0C
n
X
kD1
.akcoskxCbksinkx/
n
X
kD0
Ak.fIx/
partial sum of the first.nC1/terms of the Fourier series off 2Lp.p1/at a point x.
The degree of approximation of the sums sn.fIx/ in different spaces has been studied by many authors. In [20], Quade investigated approximation properties of the partial sums of Fourier series inLp norms. Chandra in [5] and Leindler in [13]
generalized the results of Quade using the Woronoi-N¨orlund and the Riesz means of Fourier series for some sequence classes. A similar problems were studied for more general means in [8] and [15].
c 2018 Miskolc University Press
The space
H˛D ff 2C2 W jf .x/ f .y/j Kjx yj˛; 0 < ˛1g
whereC2 is space of all2-periodic and continuous functions defined on Rwith the supremum norm andK is a positive constant, not necessarily the same at each occurrence, is a Banach space (see Pr¨osdorff, [19]) with the normk k˛defined by
kfk˛D kfkCCsup
x¤y
˛f .x; y/ (1.1)
where
˛f .x; y/D jf .x/ f .y/j
jx yj˛ .x¤y/;
by convention0f .x; y/D0and
kfkC D sup
x2Œ ;jf .x/j:
The metric generated by the norm (1.1) onH˛is called the H¨older metric. Pr¨osdorff has studied the degree of approximation by Ces`aro means in the H¨older metric and proved the following theorem.
Theorem 1([19]). Letf 2H˛.0 < ˛1/and0ˇ < ˛1:Then kn.f / fkˇ DO.1/
nˇ ˛ ; 0 < ˛ < 1I nˇ 1lnn ; ˛D1 wheren.f /is the Ces`aro means of the Fourier series off.
The case ˇD0 in Theorem 1 is due to Alexits [2]. Leindler introduced more general classes than the H¨older classes of2-periodic continuous functions, gener- alized the results of Pr¨osdorff [11]. Chandra obtained a generalization of Theorem1 by considering the Woronoi-N¨orlund transform [4]. In [16] Mohapatra and Chandra considered the problem by a matrix means of the Fourier series off 2H˛.
Further generalizations of the H¨older metric was given in [6] and [7]. In [6], Das et al. studied the results regarding the degree of approximation by infinite matrix means involved in the deferred Ces`aro means in a generalized H¨older metric.
In [9], the authors considered the degree of approximation to functions in this space with respect to the norm in the space given in [6] by the deferred Woronoi- N¨orlund means and the deferred Riesz means of the Fourier series of the functions.
The modulus of continuity off 2C2 is defined by
!.f; ı/WD sup
0<jhjı
jf .xCh/ f .x/j: According to this, the class of functionsH!is defined by
H!WD ff 2C2 W!.f; ı/DO.!.ı//g where!.ı/is a modulus of continuity.
There are numerous papers related to the degree of approximation inH! space such as Leindler [11], Mazhar and Totik [14], Daset al.[7], Khatri and Mishra [10].
A further generalization ofH!space has been given by Das, Nath and Ray in [7].
They defined the following notations: Iff 2Lp.0; 2/,p1, then denote Hp.!/WD ff 2Lp.0; 2/; p1WA.f; !/ <1g
where!is a modulus of continuity and A.f; !/WDsup
t¤0
jjf . Ct / f ./jjp
!.jtj/
wherek:kpwill denoteLp-norm with respect toxand is defined by kfkpWD
1 2
Z 2
0 jf .xjpdx
1 p
: The norm in the spaceHp.!/is defined by
jjfjjp.!/WD jjfjjpCA.f; !/:
Given the spacesHp.!/andHp.v/, if !.t /v.t / is nondecreasing , then Hp.!/Hp.v/Lp p1
since
jjfjjp.v/max
1;!.2/
v.2/
jjfjjp.!/:
Especially, the leading studies related to degree of approximation inHp.!/ space can be found in [7] and [12].
2. DEFERREDCESARO MEAN AND ITS GENERALIZATION`
The concept of deferred Ces`aro mean has been given by Agnew as following (see [1]). LetaD.an/andbD.bn/be sequences of nonnegative integers with conditions an< bn nD1; 2; 3; ::: (2.1) and
nlim!1bnD C1: (2.2)
The deferred Ces`aro mean,Dab, determined byaandb DnDDabDSanC1CSanC2C CSbn
bn an D 1
bn an bn
X
kDanC1
Sk; where.Sk/is a sequence of real or complex numbers.
Since each Dab with conditions (2.1) and (2.2) satisfies the Silverman-Toeplitz conditions, every Dab is regular. Note that Dab involves, except in caseanD0all n, means of deferred elements of .Sn/. It is also known that Dn 1n is the identity transformation andD0nis the.C; 1/transformation. The basic properties ofDba can be found in [1]. By considering deferred Ces`aro means, Deˇger and K¨uc¸¨ukaslan gave the following notations with conditions (2.1) and (2.2) in [9].
Let.pn/be a positive sequence of real numbers. Then DbaNn.fIx/D 1
P0bn an 1
bn
X
mDanC1
pbn mSm.fIx/;
and
DabRn.fIx/D 1 Pabn
nC1 bn
X
mDanC1
pmSm.fIx/
where
P0bn an 1D
bn an 1
X
kD0
pk ¤0; Pabn
nC1D
bn
X
kDanC1
pk¤0 and
Sn.fIx/D 1
Z
f .xCt / Dn.t / dt;
in which
Dn.t /Dsin.nC12/t 2sin 2t :
These two methods are calleddeferred Woronoi-N¨orlund means,.DabN; p/, and deferred Riesz means, .DabR; p/, with respect to Sm.fIx/, respectively. In case bnDnandanD0, the methodsDabNn.fIx/andDabRn.fIx/give us the classicaly known Woronoi-N¨orlund and Riesz means, respectively. Provided thatpnD1for all n.0/, both of them yield deferred Ces`aro means
Dab.fIx/D 1 bn an
bn
X
mDanC1
Sm.fIx/
ofSm.f; x/.
In addition to this, if bnDn, anD0 andpk D1 for these two methods, then they coincide with Ces`aro method C1. In the event that anD0, .bn/ is a strictly increasing sequence of positive integers withb.0/D0andpkD1, then they give us Ces`aro submethod which is obtained by deleting a set of rows from Ces`aro matrix (see [3,8,18]).
There exist some inclusions that established relation between deferred method and Ces`aro method. Before giving this inclusions without detail, we need to give some definition. If.an/and.bn/satisfy, in addition to (2.1) and (2.2), the condition
an
bn anDO.1/ (2.3)
for alln, then (Dba) is properly deferred ; such a transformation is called a proper (Dab) (see [1]). Therefore we know that
.C; 1/D.an; bn/ if and only if .Dba/ is properly deferred, and
D.an; n/.C; 1/:
On the other hand,
D.an; n/.C; 1/ if and only if .Dab/ is properly deferred.
In2014, Nayaket al. has studied the rate of convergence problem of Fourier series by deferred Ces`aro mean in the generalized H¨older metric (Hp.w/) and gave the fol- lowing theorem.
Theorem 2 ([17]). Let v and w be moduli of continuity such that wv is non- decreasing andf 2Hp.w/; p1. Let
qnD.2jC1/pnC2j where j is a positive integer. Then
.i /kDn.Sn.fI:// f .:/kp.v/DO 1
pnC1
C O.1/
.pnC1/2 Z
.pnC1/
w.t / t3v.t /dt:
If in addition t v.t /w.t / is non-increasing then
.i i /kDn.Sn.fI:// f .:/kp.v/D O.1/
.pnC1/2 Z
.pnC1/
w.t / t3v.t /dt and a fortiori
.i i i /kDn.Sn.fI:// f .:/kp.v/DO
w.=.pnC1//
v.=.pnC1//
:
Taking into account of this theorem and the methods given in [9], we shall improve the results in [17] on the degree of approximation by the generalized deferred Ces`aro means in the spaceHp!. Therefore the results given in [17] are generalized according to the summability method.
3. THEOREMS
In this section we will give two theorems on the degree of approximation by de- ferred Woronoi-N¨orlund means and deferred Riesz means in generalized H¨older met- ric.
Theorem 3. Letvandwbe moduli of continuity such thatwv is nondecreasing and f 2Hp.w/; p1. LetbnD.2jC1/anC2j where j is a positive integer. Moreover let.pn/be a positive sequence and the conditions
.bn an/pbnDO.Pabn
nC1/ (3.1)
and
bn 1
X
mDanC1
jm.pm/j DO.jpbn panC1j/ (3.2) are satisfied wherem.pm/Dpm pmC1. Then
kDabRn.fI:/ f .:/kp.v/
DO
1Cjpbn panC1j pbn
( 1
anC1C 1 .anC1/2
Z
.anC1/
w.t / t3v.t /dt
)
Proof. By definition of the deferred Riesz means, we have DbaRn.fI:/ f .:/D 1
Pabn
nC1 bn
X
mDanC1
pm.Sm.fI:/ f .://:
By elementary methods, we get IWDDbaRn.fI:/ f .:/D 1
PabnnC1
bn an 1
X
mD0
pmCanC1.SmCanC1.fI:/ f .://:
Using Abel’s transformation, we see that I D 1
PabnnC1
bn an 2
X
mD0
ŒpmCanC1 pmCanC2Œ
m
X
kD0
.SkCanC1.fI:/ f .://
C 1 Pabn
nC1
bn an 1
X
kD0
Œ.SkCanC1.fI:/ f .://pbnDWI1CI2 (3.3) By considering the second term in right side of the above equality, we write
I2D pbn Pabn
nC1
bn an bn an
bn
X
kDanC1
.Sk.fI:/ f .://Dpbn.bn an/ Pabn
nC1
.Dab.Sn.fI:// f .://:
Hence by the condition (3.1) and Theorem2-(i), we obtain kI2kp.v/ jpbn.bn an/
Pabn
nC1
jkDba.Sn.fI:// f .:/kp.v/
DO.1/kDba.Sn.fI:// f .:/kp.v/
DO 1
anC1
C O.1/
.anC1/2 Z
.anC1/
w.t /
t3v.t /dt: (3.4) Now let us consider the first term in (3.3). Then we have
I1D 1 Pabn
nC1
bn an 2
X
mD0
Œ
m
X
kD0
.SkCanC1.fI:/ f .://ŒpmCanC1 pmCanC2
D 1 PabnnC1
bn 1
X
mDanC1
m.pm/
m an 1
X
kD0
.SkCanC1.fI:/ f .://
D 1 Pabn
nC1 bn 1
X
mDanC1
m.pm/m an
m an m
X
kDanC1
.Sk.fI:/ f .://
D 1 Pabn
nC1 bn 1
X
mDanC1
.m an/m.pm/ŒDam.Sn.fI:// f .:/ (3.5) By considering (3.5) and Theorem2-(i), we write
kI1kp.v/ 1 Pabn
nC1
.bn an/
bn 1
X
mDanC1
jm.pm/jkDam.Sn.fI:// f .:/kp.v/
D.bn an/ Pabn
nC1
( O
1 anC1
C O.1/
.anC1/2 Z
.anC1/
w.t / t3v.t /dt
) bn 1 X
mDanC1
jm.pm/j: Owing to condition (3.1) and condition (3.2) in the last term, we obtain
kI1kp.v/Djpbn panC1j pbn
( O
1 anC1
C O.1/
.anC1/2 Z
.anC1/
w.t / t3v.t /dt
)
: (3.6) Combining (3.3), (3.4) and (3.6), we get the desired result . Therefore the proof is
completed.
Corollary 1. Under conditions of Theorem3, iff 2Hp.w/forp1and t v.t /w.t / is nonincreasing then
(i)kDabRn.fI:/ f .:/kp.v/DO.1/
1CjpbnppbnanC1j
1 .anC1/2
R
.anC1/
w.t / t3v.t /dt
and furthermore
(ii)kDabRn.fI:/ f .:/kp.v/DO.1/
1CjpbnppbnanC1jw.=.a
nC1//
v.=.anC1//.
Theorem3and its corollaries are very important due to the relations between Riesz type deferred method and Riesz method. Let us recall some results establishing these relations given in [9]. Suppose that the sequences.an/and.bn/satisfy the conditions (2.1) and (2.2). If the condition
p1Cp2C Cpan
Pabn
nC1
DO.1/ (3.7)
satisfy for .pn/, then we shall say that .DabR; p/ is properly deferred and such a transformation is called a proper.DabR; p/. We see that if.pn/D1for allnwith the conditions (2.1) and (2.2), then .DabR; p/ and the condition (3.7) are reduced.D/
and the condition (2.3), respectively. In [9], we know that ”.R; p/.DabR; p/if and only if.DbaR; p/ is proper”. On the other hand, we have ”.DnaR; p/.R; p/” and
”.DanR; p/.R; p/if and only if.DnaR; p/is proper”.
Theorem 4. Letv andw be moduli of continuity such that wv is nondecreasing andf 2Hp.w/,p1. Let
bnD.2j C1/anC2j
where j is a positive integer. Moreover let.pn/be a positive sequence and the condi- tions
.bn an/pbn an 1DO.P0bn an 1/ (3.8) and
bn 1
X
mDanC1
jm.pbn m/j DO.jpbn an 1 p0j/ (3.9) are satisfied. Then
kDabNn.fI / f ./kpv
DO
p0C jpbn an 1 p0j pbn an 1
( 1
anC1C 1 .anC1/2
Z
.anC1/
w.t / t3v.t /dt
)
: (3.10) Proof. Since
JWDDabNn.fI:/ f .:/D 1 P0bn an 1
bn an 1
X
mD0
pbn m an 1.SmCanC1.fI:/ f .://;
we get that
JD 1
P0bn an 1
bn an 2
X
mD0
Œ
m
X
kD0
.SkCanC1 f .://Œpbn m an 1 pbn m an 2
C 1
P0bn an 1
bn an 1
X
kD0
Œ.SkCanC1.fI:/ f .:/p0DWJ1CJ2 (3.11) by Abel’s transformation. Let us considerJ2. Since
J2Dp0.bn an/ P0bn an 1
.Dab.Sn.fI:// f .://;
we have
kJ2kp.v/O.1/ p0
pbn an 1 ( 1
anC1C 1 .anC1/2
Z
.anC1/
w.t / t3v.t /dt
)
(3.12) by considering (3.8) and Theorem2-(i). Let us estimateJ1. By elementary methods, we know that
J1D 1 P0bn an 1
bn 1
X
mDanC1
.pbn m pbn m 1/
m an 1
X
kD0
.SkCanC1.fI:/ f .://
D 1
P0bn an 1
bn 1
X
mDanC1
m.pbn m/.m an/ŒDam.Sn.fI:// f .:/:
Taking into account of (3.8), (3.9) and Theorem2-(i), we write kJ1kp.v/O
jpbn an 1 p0j pbn an 1
( 1
anC1C 1 .anC1/2
Z
.anC1/
w.t / t3v.t /dt
)
: (3.13) Therefore, we get (3.10) by collecting of (3.11)-(3.13).
Corollary 2. Under conditions of Theorem 4, if f 2Hp.w/, p1 and t v.t /w.t / is nonincreasing then
(i)kDabNn.fI:/ f .:/kp.v/DO.1/.a 1
nC1/2
p
0Cjpbn an 1 p0j pbn an 1
R
.anC1/
w.t / t3v.t /dt and furthermore
(ii)kDabNn.fI:/ f .:/kp.v/DO.1/.w.=.av.=.anC1//
nC1///p
0Cjpbn an 1 p0j pbn an 1
.
Remark1. If we takepkD1in Theorem3, Theorem4, Corollary1and Corollary 2, then all of results coincide with Theorem2.
4. CONCLUSION
Several studies have been carried out on the degree of approach in the Hp.w/
spaces. The means used in these studies include the results obtained by approach- ing to functions in this space using the classically known Ces`aro, Woronoi-N¨orlund and Riesz means. These results depend either directly related to the degree of polyno- mials, or to the upper and lower limits of the method such as deferred Ces`aro means.
In this study we have shown that how the speed of the approach is being affected, based on the sequences that determines the Woronoi-N¨orlund and Riesz means by considering the deferred type of the Woronoi-N¨orlund and Riesz mean introduced in [9]. It is evident that the results generalized previous work on these spaces.
REFERENCES
[1] R. P. Agnew, “On deferred Ces´aro means,”Ann. Math. (2), vol. 33, no. 3, pp. 413–421, 1932, doi:
10.2307/1968524.
[2] G. Alexits, “Convergence problems of orthogonal series,” (International Series of Monographs on Pure and Applied Mathematics. Vol. 20.) New York-Oxford-London-Paris: Pergamon Press. IX, 350 p. (1961), 1961.
[3] D. H. Armitage and I. J. Maddox, “A new type of Ces`aro mean,”Analysis, vol. 9, no. 1-2, pp.
195–204, 1989.
[4] P. Chandra, “On the generalized Fejer means in the metric of H¨older space,”Math. Nachr, vol.
109, pp. 39–45, 1982, doi:10.1002/mana.19821090105.
[5] P. Chandra, “Trigonometric approximation of functions inLp-norm,”J. Math. Anal. Appl, vol.
275, no. 1, pp. 13–26, 2002, doi:10.1016/S0022-247X(02)00211-1.
[6] G. Das, T. Ghosh, and B. Ray, “Degree of approximation of functions by their Fourier series in the generalized H¨older metric,”Proc. Indian Acad. Sci., Math. Sci, vol. 106, no. 2, pp. 139–153, 1996, doi:10.1007/BF02837167.
[7] G. Das, A. Nath, and B. K. Ray, “An estimate of the rate of convergence of Fourier series in the generalized H¨older metric,” inAnalysis and applications. Outcome of the conference dedicated to Professor Tribikram Pati on the occasion of his 70th birthday, Ujjain, India, 1999. New Delhi:
Narosa Publishing House; Boca Raton, FL: Chapman and Hall/CRC, 2002, pp. 43–60.
[8] U. Deˇger and M. Kaya, “On the approximation by Ces´aro submethod,”Palest. J. Math, vol. 4, no. 1, pp. 44–56, 2015.
[9] U. Deˇger and M. K¨uc¸¨ukaslan, “A generalization of deferred Ces`aro means and some of their applications,”J. Inequal. Appl, vol. 2015, p. 16, 2015, doi:10.1186/s13660-014-0532-0.
[10] K. Khatri and V. N. Mishra, “Degree of approximation by the.TE1/means of Conjugate Series of Fourier Series in the H¨older metric,”Iran J Sci Technol Trans Sci, vol. 2017, p. 9, 2017, doi:
https://doi.org/10.1007/s40995-017-0272-3.
[11] L. Leindler, “Generalizations of Proessdorf’s theorems,”Stud. Sci. Math. Hung, vol. 14, pp. 431–
439, 1979.
[12] L. Leindler, “A relaxed estimate of the degree of approximation by Fourier series in generalized H¨older metric,”Anal. Math, vol. 35, no. 1, pp. 51–60, 2009, doi:10.1007/s10476-009-0106-6.
[13] L. Leindler, “Trigonometric approximation inLp-norm,”J. Math. Anal. Appl, vol. 302, no. 1, pp.
129–136, 2005, doi:10.1016/j.jmaa.2004.07.049.
[14] S. Mazhar and V. Totik, “Approximation of continuous functions byT-means of Fourier series,”
J. Approx. Theory, vol. 60, no. 2, pp. 174–182, 1990, doi:10.1016/0021-9045(90)90082-2.
[15] V. N. Mishra and L. N. Mishra, “Trigonometric approximation of signals (functions) inLp-norm,”
Int. J. Contemp. Math. Sci, vol. 7, no. 17-20, pp. 909–918, 2012.
[16] R. Mohapatra and P. Chandra, “Degree of approximation of functions in the H¨older metric,”Acta Math. Hung, vol. 41, pp. 67–76, 1983, doi:10.1007/BF01994063.
[17] L. Nayak, G. Das, and B. Ray, “An estimate of the rate of convergence of Fourier series in the generalized H¨older metric by deferred Ces`aro mean,”J. Math. Anal. Appl, vol. 420, no. 1, pp.
563–575, 2014, doi:10.1016/j.jmaa.2014.06.001.
[18] J. A. Osikiewicz, “Equivalence results for Ces`aro submethods,”Analysis, M¨unchen, vol. 20, no. 1, pp. 35–43, 2000, doi:10.1524/anly.2000.20.1.35.
[19] S. Pr¨ossdorf, “Zur konvergenz der fourierreihen h¨olderstetiger funktionen,” Mathematische Nachrichten, vol. 69, no. 1, pp. 7–14, 1975, doi: 10.1002/mana.19750690102. [Online].
Available:http://dx.doi.org/10.1002/mana.19750690102
[20] E. Quade, “Trigonometric approximation in the mean,”Duke Math. J, vol. 3, pp. 529–543, 1937, doi:10.1215/S0012-7094-37-00342-9.
Authors’ addresses
Uˇgur Deˇger
Mersin University, Faculty of Science and Literature, Department of Mathematics, 33343, Mersin, Turkey
E-mail address:degpar@hotmail.com(udeger@mersin.edu.tr)
Hilal Bayindir
Mersin University, Institute of Science and Literature, Department of Mathematics, 33343, Mersin, Turkey
E-mail address:hilalbayindir2@gmail.com