• Nie Znaleziono Wyników

Cesàro sequence spaces of non-absolute type

N/A
N/A
Protected

Academic year: 2021

Share "Cesàro sequence spaces of non-absolute type"

Copied!
5
0
0

Pełen tekst

(1)

ANNALES SOCIETATIS MATHEMATICAE POLONAE Series I: COMMENTATIONES MATHEMATICAE X X (1978) ROCZNIKI POLSKIEGO TOWARZYSTWA MATEMATYCZNEGO

Séria I: PRACE MATEMATYCZNE X X (1978)

K g P e n g - N u n g and L e e P e n g - Y e e (Singapore)

1. Introduction. In [1] Jagers has determined the Kôthe duals of the Cesàro sequence spaces ces^, 1 < p < oo, where cesp consist of all real sequences x = {xk)keN such th a t

The case when p — 1 is trivial (see [4]). The case when p = oo has been considered by the present authors [3]. In this note, we shall define the Cesàro sequence spaces in a different way, th a t is, with norms \\x\\p which do"not satisfy the absolute property: = \\x\\p , where \x\ = (|%|)fc6iv>

and try to determine their associate spaces, i.e., the Kôthe duals.

2. Normed Kôthe sequence spaces. Let X be the set of all real se­

quences x = {xk)keN. A functional q from X into the non-negative extended real number system is called a sem i-norm if

(i) e(0) = 0, (ii) q(ox) = \a\q(x), (iii) Q{x + y ) ^ Q(x) + Q{y).

If instead of (i) q satisfies the condition th a t q{x) = 0 if and only if x = 0, then q is called a norm . We denote by X e the collection of all sequences x satisfying q{x) < oo. Obviously X e is a linear space, and we call X Q the norm ed K ôth e sequence space of non-absolute type with the semi-norm q. If X Q is complete with respect to the norm q, then X e is called a Banach sequence space of non-absolute type since we did not assume the absolute property.

From now on, let us assume th a t X e is a Banach sequence space of non-absolute type. Given a semi-norm q, we define a new semi-norm o'

as follows:

Cesàro sequence spaces of non-absolute type

oo

(2)

430 Ng P e n g -N u n g and Lee P e n g -Y e e

and we put q' (x) = o o i f the series JT xkyk does not converge for some y A=1

satisfying g(2/ ) < l . The semi-norm q' is called the associate semi-norm of q. The space X Q- consisting of all sequences x e X with q (x) < с о is called the associate space of X e. For any x = {xk)keN e X Q and у = (yk)keN e X Q we have

CO

\£хкУк\ < ?(*)$'(?)•

A = 1

A semi-norm q is said to be saturated if for every non-empty subset JE of N, there exists a non-empty subset F of E such th a t g(xF) < со,

where the sequence xF = (xk)keN is defined as xk = 1 if Te e F and xk = 0 if It $ F. I t is easy to see th a t q is saturated if and only if X Q contains all sequences of finite terms, i.e., sequences having only finitely many non-zero terms. The following theorem is a consequence of the B anaeh- Steinhaus theorem:

Th e o r e m 2 . 1 . Let q' be saturated, and y e X . Then y e X e. i f and

OQ

only i f £ x k yk < oo for all x e X e.

k = 1

3 . Cesàro sequence spaces. Let X p ( 1 < p < с о ) and X œ be respecti­

vely the spaces of all x e X with

and

for 1 < p < с о ,

n g N < OO.

Note th a t the above norms are saturated except for p = 1.

Th e o r e m 3.1. The space X p (1 < p < со) is a Banach sequence space of non-absolute type.

P ro o f. Let (x^)ieN be a Cauchy sequence in X p so th a t for given e > 0 we have \\x ^ ~ x^\\p < e for all i, j > X(e). We write х(г) = {х$)ы м . Then for fixed h, {a$){eN converges. If lim x $ = xK, then

i-*oo

for i ^ N(e) and all m.

Hence letting m->oo we have \\х^ — x\\p < e for all г > A(e). This proves th a t X p is complete for 1 < p < oo. The completeness of follows similarly.

(3)

Cesàrо sequence spaces 431 Theorem 3.2. cesp <= X p for 1 ^ p < oo and cesp Ф X p .

P ro o f. The inclusion is trivial. To see th a t cesp ф X p, we define a sequence (xk)keN by х г — —1 and xk = 2 ( —l ) fe, h = 2, 3, ... which is an element of X p bu t not of cesp for 1 < p < o o . Further, the sequence (xk)keat defined by xk = fc( —l ) fc is a member of Х ж b u t not of ces^. Finally the sequence {xk)keN defined by x t = —1 and xk = ( —l) fc (2fc —1) /7Ь(7c —1) for h = 2, 3, ... is an element of b u t not of cesl .

Theorem3.3. 1 < p < oo and a be defined on X p by o(x) = (o'n(a?))neiVr 1 n

where ffn(x) — — / Ф n k=i

Then a is an one-to-one bounded linear transfor­

mation from X p onto the sequence spaee lp with operator norm 1.

The proof is easy- The result will be used in an essential way in a proof in the next section.

4. The associate space of X p . Let Yq be the space of all sequences у e X such th a t

(1) \кУк\ ^ AT for all к e X , OO

(2) ?.q(y) = ( 2 \ к ( у к - у к+1)\а)119< OO for 1 < q < o o ,

&=i

and Aw(y) = su p {\h(yk - Ук+г)\', h e N} < o o .

We shall show th a t Y a is the associate space X'p of X p , where 1 (рф 4-1 lq = 1 and with Xq — ||*||p the associate norm of

Lemma 4.1. I f у — (yk)keN is an element in the associate space X'p of X p, then the sequence (hyk)keN is bounded. I n particular, when p = o o , hyk->0 as &->oo.

P ro o f. Let у = {yk)keN be an element of X'p . Then ^ x kyk < o o

for every x = (xk)keN in X p . This implies th a t xkyk->0 as &->oo. Since *=i x = (jfc( — l)*)*ejv is an element of J M, we have hyk->0 as ~k->oo. In general, write s. 1 VI

fc=l

and we have

(A) yjfcCA»* —(fc -l)e A_ j]->0 as Tc^oo

for all sequences (sk)keN in lp by Theorem 3.3. So (A) is also true for all sequences ( ( - l ) k \sk\)keN in lp, i.e.,

( — [ *|e*l + ( *— as &-^oo.

I t follows th a t Jcyksk-±0 as Tc-+oo. Now we claim th a t (hyk)keN is bounded.

For p = 1, it is easy. Suppose th a t the sequence is not bounded for 1 < p

< o o . Then there exists a subsequence (Цук.^бМ such th a t \Цук.\> j for j e X. Take r such th a t 0 < r < 1 and pr > 1, and define a sequence

(4)

432 N g P e n g -N u n g and Lee P e n g -Y e e

{ * ' k ) k e N in l P ЪУ

, (tyk)~r ^Ьеп к = ICj, j = 1, 2, 3, ..., SJç ----

0 otherwise.

Then we have Цук.*к. = (ЦУк-)1 r <l°es not tend to zero as j-> oо which leads to a contradiction. Therefore the sequence must be bounded.

Theorem 4.2. The associate space X p of X p (1 < p < oo) is the space Y q with the norm l q, where 1 /p + l/g = 1.

O O

P ro o f. Let у e X p . Then ] ? xkyk is convergent for all x e X p . hfow

k = 1

we apply Abel transformation and obtain

Л ^ к У к = £ +

fc=i A = 1

where an = — I xk . Since x e X p , it follows th a t an->0 as n-> oo for

*=i

1 < P < 00 and th a t (<rn)neN is bounded for p = o o . In view of Lemma 4.1, the last term in the above equality tends to zero as m~>oо and hence

£а>кУк = ]>]ЧУк-Ук+1)<*к-

/с=i *=i

Therefore applying Theorem 3.3 we see th a t the above series converges for all sequences a = (ak)keN belonging to lp . I t is known th a t (k (yk — 1) ) ^ belongs to lq, where 1 /p + l/q = 1 and th a t *

K(y) =

k = 1

O O

= sup {| £ xk ykI ; \\cc\\p < l} = ||y Hi.

*=i

Hence, together with Lemma 4.1, we have proved th a t X'p c Y q.

Conversely, let у — {yk)keN be an element in Y g. Since the norm Aq is saturated, then by Theorem 2.1 and Abel transformation we need only to prove th a t mykak->0 as &->oo for у e Y q and x e X p . The case when l < p < oo follows easily from the facts th a t {kyk)keN is bounded and <rfc->0 as Jc->oo. For p = oo, {ok)keN is bounded and lim yk = 0. I t

follows th a t k^°°

O O o o

Ы < ^ ьз- уш \ < ^ j \ y j - y j+1\.

j— k j = к

(5)

Cesàro sequence spaces 433 By letting ft->oo, we obtain th a t ~kyk-^- 0 as fc ^ o o . This completes the proof.

We remark th a t in defining X p , the lp norm involved may be replaced by Or liez or Luxemburg norm for an Orliez sequence space. Hence we have defined an Orliez space of non-absolute type. Its associate space can also be found. The details, together with its general theory, are being worked out and will appear in a separate paper.

R e f e r e n c e s

[1] A. A. J a g e r s , A note on Cesàro sequence spaces, Nienw Arch, voor Wiskunde (3) 22 (1974), p. 113-124.

[2] G. M. L e ib o w itz , A note on Cesàro sequence spaces, Tamkang J. Math. 2 (1971), p. 152-167.

[3] Ng P e n g -N u n g and Lee P e n g - Y ее, On the associate spaces of Cesàro sequence spaces, Nanta Math. 9 (1976).

[4] J. S. S h iu e , On the Cesàro sequence spaces, Tamkang J. Math. 2 (1970), p. 19-25.

[5] A. C. Z a a n en , Integration, Chapter 15, Amsterdam 1967.

NANYANG UNIVERSITY SINGAPORE

Cytaty

Powiązane dokumenty

Voor Wiskunde, III

Another example (different from Goodman’s example) has been constructed by W... Krzyzewski for suggesting the

Extending this idea we will introduce Hadamard matrices: such a matrix (of order q) gives sequences which can be generated by finite automata and which satisfy (2) where M 2 is

Compute terms of this sequence numbered from 3

Notice that for any family ∆ of functionals having a positive Weyl chamber we can define a set of simple roots in the way that (2.18) holds. For that we take the set ˜ ∆ of

The concepts of fuzzy sets and fuzzy set operations were first introduced by Zadeh [20] and subsequently several au- thors have discussed various aspects of the theory and

Abstract: In the present paper we introduced some seminormed difference sequence spaces combining lacunary sequences and Musielak- Orlicz function M = (M k ) over n-normed spaces

We would like to emphasize that the reasoning presented in this paper - i.e., the explicit formula for the Bellman function B - gives more information about the action of