You are on page 1of 10
Indian J. pure appl. Math., 28(4), 419-428, April 1994 SEQUENCE SPACES DEFINED BY ORLICZ FUNCTIONS S. D. PARASHAR’ AND B. CHOUDHARY? ‘Department of Mathematics, University of Dethi, Dethi 110 007 Department of Mathematics, Indian Institute of Technology, Hauz Khas New Delhi 110 016 (Received 4 June 1993; after revision 4 November 1993) In this paper we introduce some new sequence spaces using Orlicz function. We also examine some properties of these sequence spaces. § 1. Lindenstrauss and Tzafriri* used the idea of Orlicz function (see definition below) to construct the sequence space bre: | kel x p} xe forsome p> 0}. The space Jy with the norm p>o: 5 w( eta] becomes a Banach space which is called an Orlicz sequence space. Lindenstrauss and Tzafriri proved that every Orlicz sequence space ly contains a subspace isomor- phic to /, for some p= 1, thereby answering a general conjecture that every infinite dimensional Banach space contains a closed subspace isomorphic to cy or some Jy positively for a class of spaces (see Lindenstrauss* and Milman’ for discussion of this and related conjectures). The space ly is closely related to the space |, which is an Orlicz sequence space with M(x) = %:1sp [0,~), which is continuous, non decreasing and convex with M(0) = 0, M(x) > 0 for x > 0 and I] x || =ine 420 S.D. PARASHAR AND B. CHOUDHARY MQ) — @ as x -» ©, If convexity of Orlicz function M is replaced by MG + y) = M(x) +M(y) then this function is called modulus function, defined and discussed by Ruckle!? and Maddox’. Let p = (p,) be any sequence of positive real numbers. We define the following sequence spaces a Wo [re 5 («(-224)) <= rn p>] kel ° WM, p) = veoh § (aflaett)) eco eal for some p and / > 0} wap [se0:d 5 (u(lelyy ctnnse| for some p > 0 De Wtho)=|aee: sup 2 > (™(444)) <~trsaaep>o] kel ° When p, = 1, for all k, then Iy(p) becomes Jy. When M(x) = x then the family of sequences defined above become Kp), [C, 1, pl, [C. 1, plo and [C, 1, plo respectively. We denote W(M, p), Wo(M, p)and W.(M,p) as W(M), Wo(M) and Wz (M) when p, = 1, for each k. §2. In this section we shall establish some basic properties of the sequence spaces defined earlier. In order to discuss the properties of 4p) we assume that (p,) is bounded. Theorem 1 — Let H = sup Pp then i,(p) is a linear set over the set of complex * numbers C. PROOF : Let x y El,{p) and a,B EC. In order to prove the result we need to find some p, such that i fe(issmiyf= Since x, y Ely(p), therefore, there exists some positive p, and p2 such that 421 SEQUENCE SPACES 5 (a!) and 3 ((ni)f = Define 3 = max(2| a | 9,2 | B | 2). Since M is non decreasing and convex, BY (H( 18 “Ril “ib “Sb (af “ef Ca where C = max (1, 2-1). This proves that /,4p) is linear. Theorem 2 — 1,{p) is total paranormed space with ma OH ial) sine] von{ 3 [u( g(- x). By using Theorem 1 for a= = 1, we get g(x) = inf where H = max (1, sup p). PROOF : Clearly g(x) 8% + y) = atx) + Br). Since M(Q) = 0, we get inf {p'»"} = 0 for x = 0. Conversely, suppose (x) = 0, then 422 S.D. PARASHAR AND B. CHOUDHARY va om (3C35] | fe This implies that for a given € > 0, there exists some p, (0 < p,<€) such that inf Pe [3[e(syf |" Stop "= (Spey ye ‘Suppose x,,,* 0 for some m. Let ¢— 0, then it follows that Pm WH [Sls] ) which is a contradiction. Therefore x,,, = 0 for each m. Finally, we prove scalar multiplication is continuous, Let 4 be any number. By definition, ey g(Ax)= inf } pr/#: eae) <1 Dl Then / oH som | onon 3[a 4) <1 n=1,33, where r= p/h. Since | A | smax(1, | A |%) therefore | A |?" < (max (1, | & |), Hence SEQUENCE SPACES 423 g(x) s (max (1, | & |)/4 inf me OH S[o(te] |] arfeenas- which converges to zero as g(x) converges to zero in 1(p). Now suppose 1, +0 and x is in I,(p). For arbitrary ¢ > 0, let N be a positive integer such that [one SCs for some p > 0. This implies that [S[egyf Jae let 0 < | A| <1, using convexity of M we get SPC 3 [el <(F Since M is continuous every where in [0, ©), then 3 [oS] ker vie is continuous at 0. So there is 1 > 8 > O such that | f(t) | K, then for n > K, va (S257 Je Pk WH [3 [w(ts21)] xt forn> K <1 Remark : It can be easily verified that when M(x) = x, then the paranorm defined in Jy(p) and paranorm defined in i(p) are same. ‘An Orlicz function M can always be represented (see Krasnoselskii and Rutitsky?, P-5) in the following integral form 404 §.D. PARASHAR AND B. CHOUDHARY Ma)= f goat 0 where g, known as the kernal of M, is right-differentiable for tz 0, q(0) = qd) > 0 for # > 0, q is non decreasing, and q{t) —* 0 as 1 ~. Theorem 3 — Let 1 0 with \,q() 21 Qed\ 2 : where q is the kernal associated with M, such that Lax | x0) (30 m{ Lee! 20), { (neon }e“(3)-(3) This implies that H dod re oe [ana |

1 sl. 4 ke (5 (u(t )) a kel e ‘Taking infimum of such p’s we get pO Nn inf } p42] DS (m(145*1)) slt 0 such that baa \Pk 3 (m()} == kel ® This implies that Lx| e Since M is non-decreasing, we get 3 (m(4t) =x (m( BP Hence x €h4q). § 3. In this section we study some properties of spaces W(M, p), Wo(M,p) and Wa (M, p) defined earlier. We first state two simple properties of these spaces. Theorem 5 — Let (p,) be bounded. Then W(M, p), Wo (M, p) and W. (M, p) are linear spaces. M ( )s 1 for sufficiently large values of i. 426 S.D. PARASHAR AND B. CHOUDHARY Theorem 6 — Let H = sup p, Then Wo(M,p) is a linear topological space Pr paranormed by g’ defined mou covomloe (eS (o(i3ty) | In order to discuss further result we need the following definition. Definition (Krasnoselskii and Rutitsky?, 25) — An Orlicz function M is said to satisfy 4,-condition for all values of u, if there exists, constant K > 0, such that M(2u) = KM(u) (uw 2 0). The A,-condition is equivalent to the satisfaction of inequality M(lu) = K-IM(u) for all values of u and for > 1. We write [C, 1] = Wu [C, to = Wo and [C, 1]. = Wa. Theorem 7 — Let M be an Orlicz function which satisfies A,-condition. Then W, © WM), Wo S WM) and WS We (M). PRoor : Let x€W,, then ia Sin & [ae—1 | +0 as nm, let & > 0 and choose 5 with 0<8 < 1 such that MQ) < © for Osts5. Write y= |x-L| and consider 3 Minds B+} where the first summation is over y,<6 and the second summation over y,> 8. Since M is continuous D «ne i and for y,>5, we use the fact that Yes y/5< 14+ y/5. Since M is non decreasing and convex, it follows that MO) sen yields z T WC WIM). Following similar arguments we can prove that WoC Wo(M) and WCW. @. Theorem 6 — (i) Let 0 < inf p»= ® € > 0 there exists a positive integer N such that 3 jee for all n =N. This implies that #3 (u(t) at 3 (m4). Therefore x € WM, p). Theorem 7 — Let 0

You might also like