• Nem Talált Eredményt

26A24 Keywords: Rieszp-variation, bounded variation spaces 1

N/A
N/A
Protected

Academic year: 2022

Ossza meg "26A24 Keywords: Rieszp-variation, bounded variation spaces 1"

Copied!
14
0
0

Teljes szövegt

(1)

Vol. 19 (2018), No. 1, pp. 157–170 DOI: 10.18514/MMN.2018.1988

A NOTE ON MERENTES–RIESZ BOUNDED VARIATION SPACES

REN ´E ERL´IN CASTILLO, HUMBERTO RAFEIRO, AND EDUARD TROUSSELOT Received 03 May, 2016

Abstract. In this paper we introduce a function space with some generalization of bounded vari- ation and study some of its properties, like embeddings, decompositions and others.

2010Mathematics Subject Classification: 26A45; 26B30; 26A16; 26A24 Keywords: Rieszp-variation, bounded variation spaces

1. INTRODUCTION

Around two centuries ago C. Jordan introduced the notion of a function of bounded variation and established the relation between these functions and monotonic ones when he was studying convergence of Fourier series. Later on the concept of bounded variation was generalized in various directions by many mathematicians, such has, L.

Ambrosio, R. Caccioppoli, L. Cesari, E. Conway, G. Dal Maso, E. de Giorgi, S. Hud- jaev, J. Musielak, O. Oleinik, W. Orlicz, F. Riesz, J. Smoller, L. Tonelli, A. Vol’pert, N. Wiener, among many others. It is noteworthy to mention that many of these gen- eralizations where motivated by problems in such areas as calculus of variations, convergence of Fourier series, geometric measure theory, mathematical physics, etc.

For many applications of functions of bounded variation in mathematical physics see the monograph [6]. For a thorough exposition regarding bounded variation spaces and related topics, see the recent monograph [1].

In 1992 N. Merentes [4] generalized the concept of boundedp-variation in the sense of Riesz defining the notion of bounded.p; 2/-variation. We say that a function f WŒa; b !Rhas bounded.p; 2/-variation inŒa; bif the number

Vp;2R .f /DVp;2R .f; Œa; b/

Dsup

˘ n 1

X

jD1

ˇ ˇ ˇ ˇ

f .bj/ f .dj/ bj dj

f .cj/ f .aj/ cj aj

ˇ ˇ ˇ ˇ

p 1

.bj aj/p 1

is finite, wherep1and the supremum is taken on the set of all block partitions of Œa; b.

c 2018 Miskolc University Press

(2)

The set of all bounded .p; 2/-variation functions is denoted by RV.p;2/.Œa; b/

which has an algebra structure. Moreover, it was shown that all functions that have bounded.p; 2/-variation also has bounded second-variation. In this work we gener- alize this concept to obtain the bounded.p; 2; ˛/-variation functions in the sense of Riesz and obtain some characterizations of this new space.

2. PRELIMINARIES

Before introducing the bounded variation space we will need some auxiliary res- ults.

Definition 1. A functionf WŒa; b !Ris said to be an˛-Lipschitz function if there exists a constantM > 0such that

jf .x/ f .y/j Mj˛.x/ ˛.y/j

for allx; y2.a; b/with x¤y. We define the space ˛-LipŒa; bas the space of all

˛-Lipschitz functions. This space is normable, via the norm

kfk˛-LipWD jf .a/j Csup

x¤y

jf .x/ f .y/j j˛.x/ ˛.y/j:

We now introduce the concepts of ˛ absolutely continuous function and ˛- derivative.

Definition 2. A functionf WŒa; b!Ris said to beabsolutely continuous with respect to˛if, for every " > 0, there exists some ı > 0 such that if f.aj; bj/gjnD1

are disjoint open subintervals of Œa; b; then

n

X

jD1

j˛.bj/ ˛.aj/j< ı implies

n

X

jD1

jf .bj/ f .aj/j< ":

All functions in˛-ACŒa; bare bounded and form an algebra of functions under point- wise defined standard operations.

Definition 3. Supposef and˛are real-valued functions defined on the same open intervalI and letx02I. We say thatf is˛-differentiable atx0if the following limit exists

xlim!x0

f .x/ f .x0/

˛.x/ ˛.x0/:

If the limit exists we denote its value byf˛0.x0/, which we call the˛-derivative off atx0.

(3)

3. RV.p;2;˛/Œa; bIS A NORMED SPACE

We want to recall the so-called Popoviciu variation (introduced in 1933 by T.

Popoviciu in [5]) for a partition ˘ D faDx1< x2< < xmDbgand a function f WŒa; b !Ris given by

Vark;1.f; ˘; Œa; b/D

m kC1

X

jD1

jf Œxj; : : : ; xjCk 1 f Œxj 1; : : : ; xjCk 2j wheref Œ; : : : ;is defined recursively in the following way:

f Œx0WDf .x0/;

f Œx0; x1WDf Œx1 f Œx0 x1 x0

f Œx0; x1; x2WDf Œx1; x2 f Œx0; x1 x2 x0

f Œx0; x1; : : : ; xkWDf Œx1; x2; : : : ; xk f Œx0; x1; : : : ; xk 1 xk x0

:

In the following we will consider a block partition˘ of the intervalŒa; b. It will be taken in the following way

˘ D faDx1;1< x1;26x1;3< x1;4Dx2;1< x2;26x2;3< x2;4

Dx3;1< < xn 1;4Dxn;1< xn;26xn;3< xn;4Dbg; (3.1) in place of the regular partition.

Definition 4. Let f be a real-valued function defined on Œa; b, ˘ be a block partition ofŒa; band˛be an increasing function. Let

.p;2;˛/R .f; ˘ /D

n

X

jD1

ˇˇf˛Œxj;4; xj;3 f˛Œxj;2; xj;1ˇ ˇ

p

j˛.xj;4/ ˛.xj;1/jp 1 where

f˛Œa; bD f .b/ f .a/

˛.b/ ˛.a/

and

VR.p;2;˛/.f; Œa; b/DVR.p;2;˛/.f /Dsup

˘

.p;2;˛/R .f; ˘ /;

where the supremum is taken on all block partitions of Œa; b. To the number VR.p;2;˛/.f; Œa; b/ we call theRiesz.p; 2; ˛/-variation of the functionf inŒa; b. If VR.p;2;˛/.f; Œa; b/ <1, then we say that f hasbounded Riesz .p; 2; ˛/-variation.

The set of all functions is denoted byRV.p;2;˛/.Œa; b/.

(4)

Remark1. WhenpD1we observe thatBV.2;˛/Œ.a; b/DRV.1;2;˛/.Œa; b/.

In the following result we will show that iff has Riesz bounded.p; 2; ˛/-variation, thenf has bounded˛-second variation.

Theorem 1. We haveRV.p;2;˛/.Œa; b/ ,!BV.2;˛/Œ.a; b/, with

V.2;˛/.f; Œa; b/.˛.b/ ˛.a//pp1VR.p;2;˛/.f; Œa; b/: (3.2) Proof. Let˘ be a block partition of type (3.1). Using the H¨older inequality we obtain

n

X

jD1

ˇˇf˛Œxj;4; xj;3 f˛Œxj;2; xj;1ˇ ˇ j˛.xj;4/ ˛.xj;1/jpp1

j˛.xj;4/ ˛.xj;1/jpp1

0

@

n

X

jD1

ˇˇf˛Œxj;4; xj;3 f˛Œxj;2; xj;1ˇ ˇ

p

j˛.xj;4/ ˛.xj;1/jp 1 1 A

1 p0

@

n

X

jD1

j˛.xj;4/ ˛.xj;1/j 1 A

p 1 p

VR.p;2;˛/.f; Œa; b/p1

.˛.b/ ˛.a//pp1:

Since the obtained inequality holds for all block partitions we obtain the desired

inequality (3.2).

Remark 2. We know (see [3]) that if f 2BV.2;˛/Œ.a; b/, then there exists the right and left˛-derivativef˛0C.x0/andf˛0 .x0/on eachx02.a; b/andf˛0C.a/and f˛0 .b/. The last result allow us to conclude that this is also true if f 2 RV.p;2;˛/.Œa; b/. In particular, there existsf˛0C.a/which we write asf˛0.a/.

Lemma 1. Letf WŒa; b !Rbe a function such thatf 2RV.p;2;˛/.Œa; b/and VR.p;2;˛/.f; Œa; b/D0. Then there exists; 2Rsuch thatf .x/D˛.x/C for allx2Œa; b.

Proof. SinceVR.p;2;˛/.f; Œa; b/D0we have that.p;2;˛/R .f; ˘ /D0for all block partitions ˘ of Œa; b. Let us consider the particular partition given by ˘0 D fa < xx < bg,

.p;2;˛/R .f; ˘0/D ˇ ˇ ˇ ˇ

f .b/ f .x/

˛.b/ ˛.x/

f .x/ f .a/

˛.x/ ˛.a/

ˇ ˇ ˇ ˇ

p 1

j˛.b/ ˛.a/jp 1 D0;

where

ˇ ˇ ˇ ˇ

f .b/ f .x/

˛.b/ ˛.x/

f .x/ f .a/

˛.x/ ˛.a/

ˇ ˇ ˇ ˇ

p

D0;

Direct calculations show that

f .x/Df .b/ f .a/

˛.b/ ˛.a/˛.x/Cf .a/˛.b/ f .b/˛.a/

˛.b/ ˛.a/

(5)

and now takingandin the following manner Df .b/ f .a/

˛.b/ ˛.a/; and Df .a/˛.b/ f .b/˛.a/

˛.b/ ˛.a/

we have the desired result.

Remark3. The setRV.p;2;˛/.Œa; b/can be equipped with a linear space structure considering the operator f 7!

f jRV.p;2;˛/.Œa; b/

defined in the space RV.p;2;˛/.Œa; b/given in the following way:

f jRV.p;2;˛/.Œa; b/

WD jf .a/j C jf˛0.a/j C

VR.p;2;˛/.f; Œa; b/p1

; (3.3) forf 2RV.p;2;˛/.Œa; b/.

Theorem 2. The operator f 7!

f jRV.p;2;˛/.Œa; b/

is a norm in the space RV.p;2;˛/.Œa; b/.

Proof. Let us takef such that

f jRV.p;2;˛/.Œa; b/

D0. From (3.3) this means that jf .a/j D 0, jf˛0.a/j D 0 and VR.p;2;˛/.f / D0. Since VR.p;2;˛/.f /D0 from Lemma1we deduce that

f .x/D˛.x/C; ; 2R:

Sincef˛0.x/Dforx2Œa; bwe conclude thatf˛0 is a constant function which is null atxDaand thusD0, from which we get thatf 0sincef .a/D0.

It is straighforward to see that for2Rwe have that VR.p;2;˛/.f /D jjpVR.p;2;˛/.f /

and now by the definition of the operator (3.3) implies the homogeneity of the oper- ator under consideration.

Let us now prove the triangle inequality. We now introduce the following notation j˛.f /Df˛Œxj;4; xj;3 f˛Œxj;2; xj;1:

Letf; g2RV.p;2;˛/.Œa; b/and˘ be a block partition ofŒa; bas in (3.1), then .p;2;˛/R .f Cg; ˘ /D

n

X

jD1

jj˛.f Cg/jp j˛.xj;4/ ˛.xj;1/jp 1 D

n

X

jD1

jj˛.f /Cj˛.g/jp j˛.xj;4/ ˛.xj;1/jp 1

n

X

jD1

jj˛.f /Cj˛.g/jp 1

j˛.xj;4/ ˛.xj;1/jp 1jj˛.f /j

(6)

C

n

X

jD1

jj˛.f /Cj˛.g/jp 1

j˛.xj;4/ ˛.xj;1/jp 1jj˛.g/j

n

X

jD1

jj˛.f /Cj˛.g/jp 1 j˛.xj;4/ ˛.xj;1/j.pp1/2

jj˛.f /j j˛.xj;4/ ˛.xj;1/jpp1 C

n

X

jD1

jj˛.f /Cj˛.g/jp 1 j˛.xj;4/ ˛.xj;1/j.pp1/2

jj˛.g/j j˛.xj;4/ ˛.xj;1/jpp1

2 4

n

X

jD1

jj˛.f /Cj˛.g/jp j˛.xj;4/ ˛.xj;1/jp 1

3 5

p 1

p 2

4

n

X

jD1

jj˛.f /jp j˛.xj;4/ ˛.xj;1/jp 1

3 5

1 p

C 2 4

n

X

jD1

jj˛.f /Cj˛.g/jp j˛.xj;4/ ˛.xj;1/jp 1

3 5

p 1

p 2

4

n

X

jD1

jj˛.g/jp j˛.xj;4/ ˛.xj;1/jp 1

3 5

1 p

:

We therefore have

.p;2;˛/R .f Cg; ˘ /p1

.p;2;˛/R .f; ˘ /p1 C

.p;2;˛/R .g; ˘ /p1 and since this is true for all partitions, we have

VR.p;2;˛/.f Cg; Œa; b/VR.p;2;˛/.f; Œa; b/CVR.p;2;˛/.g; Œa; b/;

and now it easily follows the triangle inequality.

4. EMBEDDING WITHRV.p;2;˛/.Œa; b/

We will show that if p < q, then there exists an embedding between the spaces RV.p;2;˛/.Œa; b/andRV.q;2;˛/.Œa; b/. We will need this fact to show completeness ofRV.p;2;˛/.Œa; b/.

Theorem 3. If1 < q < p <1, thenRV.p;2;˛/.Œa; b/ ,!RV.q;2;˛/.Œa; b/, with f jRV.q;2;˛/.Œa; b/

max n

1; .˛.b/ ˛.a//1q p1o

f jRV.p;2;˛/.Œa; b/

; forf 2RV.p;2;˛/.Œa; b/.

Proof. Letf 2RV.p;2;˛/.Œa; b/and˘ be a block partition of Œa; bas in (3.1).

Let us consider

.q;2;˛/R .f; ˘ /D

n

X

jD1

jj˛.f /jq j˛.xj;4/ ˛.xj;1/jq 1 D

n

X

jD1

jj˛.f /jq

j˛.xj;4/ ˛.xj;1/j.pp1/qj˛.xj;4/ ˛.xj;1/jppq:

(7)

Applying the H¨older inequality we obtain

.q;2;˛/R .f; ˘ / 2 4

n

X

jD1

jj˛.f /jp j˛.xj;4/ ˛.xj;1/jp 1

3 5

q p2

4

n

X

jD1

j˛.xj;4/ ˛.xj;1/j 3 5

p q p

Dh

.p;2;˛/R .f; ˘ /ipq

.˛.b/ ˛.a//ppq; from which

h

VR.q;2;˛/.f; Œa; b/i1q

.˛.b/ ˛.a//q1 p1h

VR.p;2;˛/.f; Œa; b/ip1 whence

f jRV.q;2;˛/.Œa; b/

maxn

1; .˛.b/ ˛.a//1q p1o

f jRV.p;2;˛/.Œa; b/

;

forf 2RV.p;2;˛/.Œa; b/.

Remark4. The proof of the above theorem remains valid ifqD1.

Corollary 1. Forp1we haveRV.p;2;˛/.Œa; b/ ,!BV.2;˛/Œ.a; b/;with f jBV.2;˛/Œ.a; b/

max n

1; .˛.b/ ˛.a//pp1o

f jRV.p;2;˛/.Œa; b/

:

The following corollary follows from the previous results and from results from [2].

Corollary 2. If1 < q < p <1, then

RV.p;2;˛/Œa; b ,!RV.q;2;˛/Œa; b ,!V.2;˛/Œa; b ,!˛-LipŒa; b ,!

,!RV.p;˛/Œa; b ,!RV.q;˛/Œa; b ,!˛-ACŒa; b ,!VŒa; b ,!BŒa; b:

5. RV.p;2;˛/.Œa; b/IS ABANACH SPACE

Theorem 4. The space RV.p;2;˛/.Œa; b/;

jRV.p;2;˛/.Œa; b/

is a Banach space.

Proof. Let.fk/k2Nbe a Cauchy sequence inRV.p;2;˛/.Œa; b/. Given" > 0there existsN"2Nsuch that

fq frjRV.p;2;˛/.Œa; b/

< "

(8)

ifq; r > N", from which we get the following system of inequalities 8

ˆˆ ˆˆ ˆˆ

<

ˆˆ ˆˆ ˆˆ :

jfq.a/ fr.a/j< ";

j.fq/0˛.a/ .fr/0˛.a/j< ";

VR.p;2;˛/.fq fp/ < "p:

From Corollary2we conclude that.fk/k2Nis a Cauchy sequence in˛-LipŒa; b;thus we have

fq fr j˛-LipŒa; b

< "K withKDmaxn

1; .˛.b/ ˛.a//pp1o , hence jfq.a/ fr.a/j Csup

x¤y

ˇ ˇ ˇ ˇ

.fq fr/.x/ .fq fr/.y/

˛.x/ ˛.y/

ˇ ˇ ˇ ˇ

< "K and so

jfq.x/ fr.x/j< K".1C˛.b/ ˛.a//; x2Œa; b:

This tell us that for eachx2Œa; b .fk.x//k2Nis a Cauchy sequence inR. SinceR is a complete space, we can definef WŒa; b !Rasx7!f .x/WDlimk!1fk.x/.

We are about to prove that:

(i) f 2RV.p;2;˛/.Œa; b/, and

(ii) .fk/k2Nconverges tof in theRV.p;2;˛/.Œa; b/-norm.

(i) Let˘ be a block partition ofŒa; bas in (3.1) Since.fk/k2Nis a Cauchy sequence inRV.p;2;˛/.Œa; b/the norm sequence

fkjRV.p;2;˛/.Œa; b/

k2Nis bounded, that is, there existsM > 0such that

f jRV.p;2;˛/.Œa; b/

M fork2N. From this fact we have

2 4

n

X

jD1

ˇ ˇ ˇ ˇ

f .xj;4/ f .xj;3/

˛.xj;4/ ˛.xj;3/

f .xj;2/ f .xj;1/

˛.xj;2/ ˛.xj;1/ ˇ ˇ ˇ ˇ

p 1

j˛.xj;4/ ˛.xj;1/jp 1 3 5

1 p

D lim

k!1

2 4

n

X

jD1

ˇ ˇ ˇ ˇ

fk.xj;4/ fk.xj;3/

˛.xj;4/ ˛.xj;3/

fk.xj;2/ fk.xj;1/

˛.xj;2/ ˛.xj;1/ ˇ ˇ ˇ ˇ

p 1

j˛.xj;4/ ˛.xj;1/jp 1 3 5

1 p

lim

k!1

h

VR.p;2;˛/.fk; Œa; b/i1p lim

k!1

fkjRV.p;2;˛/.Œa; b/

M

for all partitions˘, then

VR.p;2;˛/.f /p1

M and thusf 2RV.p;2;˛/.Œa; b/. Using embedding we may observe thatf˛0.a/exists.

(9)

(ii) One more time let us consider˘ to be a block partition of Œa; bas in (3.1).

Letq; r > N", then

n

X

jD1

ˇ ˇ ˇ ˇ

.fq fr/.xj;4/ .fq fr/.xj;3/

˛.xj;4/ ˛.xj;3/

.fq fr/.xj;2/ .fq fr/.xj;1/

˛.xj;2/ ˛.xj;1/

ˇ ˇ ˇ ˇ

p

1

j˛.xj;4/ ˛.xj;1/jp 1 VR.p;2;˛/..fq fr/; Œa; b/ < "p:

Lettingr! 1the above expression becomes

n

X

jD1

ˇ ˇ ˇ ˇ

.fq f /.xj;4/ .fq f /.xj;3/

˛.xj;4/ ˛.xj;3/

.fq f /.xj;2/ .fq f /.xj;1/

˛.xj;2/ ˛.xj;1/

ˇ ˇ ˇ ˇ

p

1

j˛.xj;4/ ˛.xj;1/jp 1 < "p:

This holds for any partition ofŒa; b, thereforeVR.p;2;˛/.fq f / < "pwhenq > N". Leth2RCbe such thata < aCh < s < t b, then

ˇ ˇ ˇ ˇ

.fq f /.t / .fq f /.s/

˛.t / ˛.s/

.fq f /.aCh/ .fq f /.a/

˛.aCh/ ˛.a/

ˇ ˇ ˇ ˇ

p 1

j˛.t / ˛.s/jp 1 VR.p;2;˛/.fq f / < "p ifq > N". Lettingh!0we have

ˇ ˇ ˇ ˇ

.fq f /.t / .fq f /.s/

˛.t / ˛.s/ .fq f /0˛.a/

ˇ ˇ ˇ ˇ

p 1

j˛.t / ˛.s/jp 1 "p from which

ˇ ˇ ˇ ˇ

.fq f /.t / .fq f /.s/

˛.t / ˛.s/ .fq/0˛.a/C.f /0˛.a/

ˇ ˇ ˇ ˇ

p

"pj˛.t / ˛.s/jp 1: Sincefq; f 2RV.p;2;˛/.Œa; b/ ,!˛-LipŒa; b, we have

j.fq/0˛.a/ .f /0˛.a/j "j˛.b/ ˛.a/jpp1C ˇ ˇ ˇ ˇ

.fq f /.t / .fq f /.s/

˛.t / ˛.s/

ˇ ˇ ˇ ˇ "j˛.b/ ˛.a/jpp1C

fq f j˛-LipŒa; b

"j˛.b/ ˛.a/jpp1C"KD QK"

ifq > N", where

KQ Dmax n

K; .˛.b/ ˛.a//pp1o

(10)

since

fq f j˛-LipŒa; b

Dlimr!1

fq frj˛-LipŒa; b

< K". Finally, for q > N"we obtain

jfq.a/ f .a/j C j.fq/0˛.a/ .f /0˛.a/j Ch

VR.p;2;˛/.fq f; Œa; b/iq1

.KQC2/";

in other words

fq f jRV.p;2;˛/.Œa; b/

.KQC2/"

ifq > N"which means that.fk/k2Nconverges tof 2RV.p;2;˛/.Œa; b/in the norm

jRV.p;2;˛/.Œa; b/

.

6. RV.p;2;˛/.Œa; b/IS ABANACHALGEBRA

We are going to show thatRV.p;2;˛/.Œa; b/is closed under the multiplication of functions.

Theorem 5. Letf; g2RV.p;2;˛/.Œa; b/. Thenfg2RV.p;2;˛/.Œa; b/.

Proof. Let˘ be a block partition ofŒa; bas in (3.1). Let us consider .p;2;˛/R .f g; ˘ /D

n

X

jD1

ˇˇ.fg/˛Œxj;4; xj;3 .fg/˛Œxj;2; xj;1ˇ ˇ

p

j˛.xj;4/ ˛.xj;1/jp 1 D

n

X

jD1

ˇˇ.fg/˛Œxj;4; xj;3 .fg/˛Œxj;2; xj;1ˇ ˇ

p 1

j˛.xj;4/ ˛.xj;1/jp 1 ˇ

ˇ.fg/˛Œxj;4; xj;3 .fg/˛Œxj;2; xj;1ˇ

ˇ: (6.1) Observe that the termˇ

ˇ.fg/˛Œxj;4; xj;3 .fg/˛Œxj;2; xj;1ˇ

ˇDWT can be written as T D jf .xj;4/g˛Œxj;4; xj;3Cg.xj;3/f˛Œxj;4; xj;3

f .xj;2/g˛Œxj;2; xj;1 g.xj;1/f˛Œxj;2; xj;1j and now adding and subtracting appropriate terms and grouping the terms we obtain

T D jf .xj;4/.g˛Œxj;4; xj;3 g˛Œxj;2; xj;1/Cg.xj;3/.f˛Œxj;4; xj;3 f˛Œxj;2; xj;1/

C.f .xj;4/ f .xj;2//g˛Œxj;2; xj;1C.g.xj;3/ g.xj;1//f˛Œxj;2; xj;1j: (6.2) From Corollary2we have

RV.p;2;˛/.Œa; b/ ,!˛-LipŒa; b and RV.p;2;˛/.Œa; b/ ,!BŒa; b:

This let us write

jf .xj;4/j kfk1; jg.xj;3/j kgk1 and

f ./ f ./

˛./ ˛./ Lip˛.f /; g./ g./

˛./ ˛./ Lip˛.g/;

(11)

for all; 2Œa; b. Using these estimates we obtain

jf .xj;4/ f .xj;2/jjg˛Œxj;2; xj;1j C jg.xj;3 g.xj;1/jjf˛Œxj;2; xj;1j Djf .xj;4/ f .xj;2/j

j˛.xj;4/ ˛.xj;2/jjg.xj;2/ g.xj;1/j

j˛.xj;2/ ˛.xj;1/jjj˛.xj;4/ ˛.xj;1/jj Cjf .xj;2/ f .xj;1/j

j˛.xj;2/ ˛.xj;1/j jg.xj;3/ g.xj;1/j

j˛.xj;3/ ˛.xj;1/jj˛.xj;4/ ˛.xj;1/j .Lip˛f /.Lip˛g/ ˛.xj;4 ˛.xj;2/C˛.xj;3/ ˛.xj;1//

replacing all this into (6.2) we have ˇˇ.fg/˛Œxj;4; xj;3 .fg/˛Œxj;2; xj;1ˇ

ˇ kfk1ˇ

ˇg˛Œxj;4; xj;3 g˛Œxj;2; xj;1ˇ

ˇC kgk1ˇ

ˇf˛Œxj;4; xj;3 f˛Œxj;2; xj;1ˇ ˇ C.Lip˛f /.Lip˛g/ ˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1//

:

Now replacing this last estimate into (6.1), separating summations and fixing expo- nents to apply the H¨older inequality we have

.p;2;˛/R .fg; ˘ / kfk1

n

X

jD1

ˇˇ.fg/˛Œxj;4; xj;3 .fg/˛Œxj;2; xj;1ˇ ˇ

p 1

j˛.xj;4/ ˛.xj;1/j.pp1/2

ˇˇg˛Œxj;4; xj;3 g˛Œxj;2; xj;1ˇ ˇ j˛.xj;4/ ˛.xj;1/jpp1 C kgk1

n

X

jD1

ˇˇ.fg/˛Œxj;4; xj;3 .fg/˛Œxj;2; xj;1ˇ ˇ

p 1

j˛.xj;4/ ˛.xj;1/j.pp1/2

ˇˇf˛Œxj;4; xj;3 f˛Œxj;2; xj;1ˇ ˇ j˛.xj;4/ ˛.xj;1/jpp1 C.Lip˛f /.Lip˛g/

n

X

jD1

ˇˇ.fg/˛Œxj;4; xj;3 .fg/˛Œxj;2; xj;1ˇ ˇ

p 1

j˛.xj;4/ ˛.xj;1/j.pp1/2 j˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/j

j˛.xj;4/ ˛.xj;1/jpp1

:

Applying in each summation the H¨older inequality we obtain .p;2;˛/R .fg; ˘ / kfk1

h

.p;2;˛/R .fg; ˘ /ipp1h

.p;2;˛/R .g; ˘ /ip1 C kgk1h

.p;2;˛/R .fg; ˘ /ipp1h

.p;2;˛/R .f; ˘ /ip1

(12)

C.Lip˛f /.Lip˛g/h

.p;2;˛/R .fg; ˘ /ipp1

2 4

n

X

jD1

j˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/jp j˛.xj;4/ ˛.xj;1/jp 1

3 5

1 p

:

Simplifying we have h

.p;2;˛/R .fg; ˘ / ip1

kfk1h

.p;2;˛/R .g; ˘ / i1p

C kgk1h

.p;2;˛/R .f; ˘ / ip1

C.Lip˛f /.Lip˛g/

2 4

n

X

jD1

j˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/jp j˛.xj;4/ ˛.xj;1/jp 1

3 5

1 p

(6.3) Simplifying the last bracket in (6.3), forj D1; : : : ; n, we can observe that

j˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/jp j˛.xj;4/ ˛.xj;1/jp 1

D ˇ ˇ ˇ ˇ

˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/

˛.xj;4/ ˛.xj;1/

ˇ ˇ ˇ ˇ

p 1

j˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/j D

ˇ ˇ ˇ ˇ

1C˛.xj;3/ ˛.xj;2/

˛.xj;4/ ˛.xj;1/ ˇ ˇ ˇ ˇ

p 1

j˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/j: (6.4) Sincexj;1< xj;2xj;3< xj;4we have

˛.xj;3/ ˛.xj;2/ < ˛.xj;4/ ˛.xj;1/I

˛.xj;4/ ˛.xj;2/ > 0I

˛.xj;3/ ˛.xj;1/ > 0:

Substituting this into (6.4)

j˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/jp j˛.xj;4/ ˛.xj;1/jp 1

2p 1 ˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/ and now summing we obtain

n

X

jD1

j˛.xj;4/ ˛.xj;2/C˛.xj;3/ ˛.xj;1/jp

j˛.xj;4/ ˛.xj;1/jp 1 2p.˛.b/ ˛.a//:

Substituting this inequality into (6.3) we have h

.p;2;˛/R .fg; ˘ /ip1

kfk1h

VR.p;2;˛/.g; Œa; b/ip1

C kgk1h

VR.p;2;˛/.f; Œa; b/ip1 C2.˛.b/ ˛.a//p1.Lip˛f /.Lip˛g/;

(13)

this last inequality holds for all partition˘ ofŒa; b, then h

VR.p;2;˛/.fg; Œa; b/ip1

kfk1h

VR.p;2;˛/.g; Œa; b/ip1 C kgk1h

VR.p;2;˛/.f; Œa; b/ip1

C2.˛.b/ ˛.a//p1.Lip˛f /.Lip˛g/ <1; (6.5)

from which it follows thatfg2RV.p;2;˛/.Œa; b/.

We now obtain the following corollary.

Corollary 3. Letf; g2RV.p;2;˛/.Œa; b/, then fgjRV.p;2;˛/.Œa; b/

P

f jRV.p;2;˛/.Œa; b/

gjRV.p;2;˛/.Œa; b/

C kfk1

gjRV.p;2;˛/.Œa; b/

C kgk1

f jRV.p;2;˛/.Œa; b/

: withP D2maxn

.˛.b/ ˛.a//p1; .˛.b/ ˛.a//2pp2o . Proof. Note that

Lip˛.f / kf j˛-LipŒa; bk

f jV.2;˛/Œa; b

max

n

1; .˛.b/ ˛.a//pp1o

f jRV.p;2;˛/.Œa; b/

and similarly tog. Then (6.5) can be written as

h

VR.p;2;˛/.fg; Œa; b/ip1

kfk1h

VR.p;2;˛/.g; Œa; b/ip1

C kgk1h

VR.p;2;˛/.f; Œa; b/ip1 C2 .˛.b/ ˛.a//p1 h

max n

1; .˛.b/ ˛.a//pp1oi2

f jRV.p;2;˛/.Œa; b/

gjRV.p;2;˛/.Œa; b/

:

REFERENCES

[1] J. Appell, J. Bana´s, and N. J. Merentes,Bounded variation and around. Berlin: de Gruyter, 2014.

doi:10.1515/9783110265118.

[2] R. E. Castillo and E. Trousselot, “On functions of.p; ˛/-bounded variation,”Real Anal. Exchange, vol. 34, no. 1, pp. 49–60, 2009.

[3] R. E. Castillo, H. Rafeiro, and E. Trousselot, “Space of functions with some generalization of bounded variation in the sense of de La Vall´ee Poussin.”J. Funct. Spaces, vol. 2015, p. 9, 2015, doi:10.1155/2015/605380.

[4] N. Merentes, “On functions of bounded.p; 2/-variation.”Collect. Math., vol. 43, no. 2, pp. 117–

123, 1992.

(14)

[5] T. Popoviciu, “Sur quelques propri´et´es des fonctions d’une variable r´eelle convexes d’ordre sup´erieur,”Bul. Soc. S¸tiint¸. Cluj, vol. 7, pp. 254–282, 1933.

[6] A. Vol’pert and S. Khudyaev,Analysis in classes of discontinuous functions and equations of math- ematical physics. Kluwer Academic Publishers, 1985.

Authors’ addresses

Ren´e Erl´ın Castillo

Universidad Nacional de Colombia, Colombia E-mail address:recastillo@unal.edu.co

Humberto Rafeiro

Pontificia Universidad Javeriana, Colombia E-mail address:silva-h@javeriana.edu.co

Eduard Trousselot

Universidad de Oriente, Venezuela

E-mail address:eddycharles2007@gmail.com

Hivatkozások

KAPCSOLÓDÓ DOKUMENTUMOK

Jordan (see [2]) introduced the notion of a function of bounded variation and established the relation between these functions and monotonic ones.. Later, the concept of

Key words: Riemann-Stieltjes integral, Functions of bounded variation, Lipschitzian func- tions, Integral inequalities, ˇ Cebyšev, Grüss, Ostrowski and Lupa¸s type inequali-

Key words and phrases: Riemann-Stieltjes integral, Functions of bounded variation, Lipschitzian functions, Integral inequal- ities, ˇ Cebyšev, Grüss, Ostrowski and Lupa¸s

Key words and phrases: Monotone sequences, Sequence of γ group bounded variation, Sine series.. 2000 Mathematics

We estimate the rate of the pointwise approximation by operators of Bleimann, Butzer and Hahn of locally bounded functions, and of functions having a locally bounded derivative..

It is observed that the analysis for our Bezier variant of new Bernstein Durrmeyer op- erators is different from the usual Bernstein Durrmeyer operators studied by Zeng and Chen

Waterman [3] have introduced the class ϕ ∧ BV (I) of functions of ϕ∧-bounded variation over I and have studied suffi- ciency conditions for the absolute convergence of Fourier series

Very recently, some authors studied some linear positive operators and obtained the rate of convergence for functions of bounded variation. For example, Bo- janic R. [3] estimated