• Nie Znaleziono Wyników

Non-convex integral functionals on Musielak-Orlicz spaces

N/A
N/A
Protected

Academic year: 2021

Share "Non-convex integral functionals on Musielak-Orlicz spaces"

Copied!
11
0
0

Pełen tekst

(1)

ANNALES SOCIETATIS MATHEMATICAE POLONAE Series E COMMENTATIONES MATHEMATICAE XXX (1990) ROCZNIKI POLSKIEGO TOWARZYSTWA MATEMATYCZNEGO

Séria I: PRACE MATEMATYCZNE XXX (1990)

Ry sz a r d Pl u c ie n n ik

(Poznan),

Sh e n Tia n

and

Yu w e n Wa n g

(Harbin) Non-convex integral functionals on Musielak-Orlicz spaces

Abstract. Non-convex integral functionals on Musielak-Orlicz spaces are discussed. A re­

presentation for the generalized gradient of an integral functional is given under some conditions.

A sufficient condition for the lower semicontinuity of a functional is given.

Introduction and preliminaries. It is well known that integral functionals on a Banach space are important non-linear functionals, which can be applied in optimization theory. In 1972 A. D. Ioffe and V. L. Levin (see [1]) discussed the subdifferential of a convex integral functional on L00. In 1979 A. Kozek extended these results and studied convex integral functionals on Orlicz spaces of vector-valued functions. In this paper we discuss integral functionals, without the convexity assumption, on Musielak-Orlicz spaces of vector-valued functions.

Let (T, I , p) be a measure space with a non-negative, complete, non-atomic and o--finite measure; (X , || • Ц*) denotes a separable real Banach space and (Y, || • ||y) its dual space. We denote by M(X) = M(T, I , X ) the linear space obtained from the set of all strongly /i-measurable functions x: T-> X by identifying the functions which are equal д-almost everywhere, and similarly for M(Y). Moreover, let <y, x> stand for the value of the functional ye Y at the point x eX . Obviously, for xeM (X ) and yeM(Y) the function <y(-), x(-)> is jU-measurable.

A function Ф: Tx [0, oo)->[0, oo) is called a Musielak-Orlicz function iff (i) for a.e. t e T the function Ф(г, •): [0, oo)->[0, oo) is convex, left-continuous on (0, oo), Ф(Г, 0) = 0, lim„_,.00Ф(^ и) = со and there exists a0 > 0 such that Ф(г, u0) < oo;

(ii) Ф ( , и ): T— > [0, oo) is /i-measurable for every we[0, oo).

The functional / ф: M (X)->[0, oo) defined by 1ф(х, X) = J Ф(г, \\x(t)\\x)dfi(t)

T

ls a convex pseudomodular.

The Musielak-Orlicz space is defined by

ЬФ(Т, X) = {

x g

M(X): / ф(гх, X) < oo for some r > 0).

~ Commentationes Math. 30.1

(2)

11

4 R. P l u c i e n n i k , S. T i a n a n d Y. W a n g

The set

d o m /ф =

{

x e

M(X ): 1

ф

(

х

,

X ) < oo}

is called the Musielak-0riiez class.

For every Musielak-Orlicz function Ф, we define the complementary function W:

T x [ 0 , o o ) -> [0 , oo)

by

Ф(£, v) = sup {uv — <P(t, и);

mg

[0, oo)}

for every teT,

i

;

g

[0, oo). The function Ф is a. Musielak-Orlicz function, too.

Moreover,

U V

Ф(£, u) = j (p(t, s)ds and • T(t, v) = j ij/{t, s)ds,

о

0

where q>(t, •) and iJ/(t, •) are mutually right inverse functions. Both q>(t, •) and IJ/(t,-) are increasing for a.e. teT.

The Luxemburg and Orlicz norms are introduced as follows:

||.х||(ф) = inf{r > 0: 1ф(х/г, X) ^ 1}, . \\х\[ф = sup \\ <y(f), x(t))dp(t)\.

/ Ф ( у . Г К 1 T

For any

x g

L0(T,

A )

and y

g

L^T , Y) the function <y(

•), x ( • ) ) :

T-»(

— oo, oo)

is integrable and

|J <y(f), x(t))dp(t)I ^

||х||ф||у||т ,

|j <y(f), x{t))dp(t)I ^

||х!|(ф)!|у||г

T T

(see [3]).

Co n d i t i o n

A. A Musielak-Orlicz function Ф(£, и

)

is said to satisfy Condition A if there exist К ^ 1 and a positive ^-measurable function ô(t) on T such that

j Ф(£, 3(t))dp(t) < oo and Ф(£, 2и) ^ КФ(1, и)

T

for a.e. t e T and u ^ 0(t) (see [4]).

Let F be a subset of a Banach space U. We say that a function / : satisfies the Lipschitz condition if there exists a constant К > 0 such that l / M - / ( / ) l ^ K lly -y 'llu for еуегУ У.У'еК

We say that / satisfies the Lipschitz condition near x e U if there exists s > 0 such that / satisfies the Lipschitz condition on x + eBi , where B. is the unit ball of U.

We say that / satisfies the local Lipschitz condition if it satisfies the Lipschitz condition near every xe U .

De f in it io n 1.

Let

/ :

U -+R satisfy the Lipschitz condition near xe U . The expression

, 0/ ч г f ( y + tv)-f(y) TT

/

(x,

v) = limsup —--- ;--- , for v e u ,

V - * X

t l O

t

(3)

Non-convex integral functionals on Musielak-Orlicz spaces 115

is called the generalized directional derivative at x along v. The expression df{x) = /° (x , v) ^ <£, r>, for all veU},

where U* is the dual of U, is called the generalized gradient of / at x (cf. [5]).

De f in it io n

2. A function / : U -> R is said to be regular at x if the Gateaux derivative f'{x, v) of / at x along v exists and f ( x , v) = /° (x , v) for every v e U (cf. [5]).

An integral functional on. ЬФ(Т, X) is defined by f(x) = J f(x(t))dp(t)

T

for every х е Ь ф(Т,Х), where /(• ) : X ^ R (teT) is a class of functions on X and t-+ft(x) is measurable for every x e X .

Results. For simplicity, we introduce

Co n d i t io n

(M). The function

f - > / ((x )

is measurable for every x e X .

Co n d i t io n

(A). There exists a function K (-)eL r (T, R) such that

\ft(xl) - f t(x2)\ ^ \K(t)\ H x j-x J * for every t e T and for all x l5 x 2eX.

Co n d i t io n

(B). For every teT, the function / ( • ) satisfies the Lipschitz condition on X and there exists c > 0 such that

£ed/f(x) => U\\Y ^ c{\+(p(t, ||x||*)}

for all t

g

T and

x g X .

Le m m a 1.

Let tp(t,-) be the right derivative of

Ф /

,

• ),

for any fixed t e T and иЕЬф(Т, R).

(i) ||м||ф ^ 1 implies (p(-,|u(-)|)edom 1^ and ||<p(-, \u( • ^ 1.

(Ü) INL ^ 1 implies | г Ф (с\u(t)\)dp(t) ^ ||м||ф.

(iii) If Ф satisfies Condition A, then for any m > 0, there exists M > 0 such that ||и||ф ^ ш implies (/>(•, |n(-)|)e L^(T, R) and \\q)(-, \u{ • )|)||((p) ^ M.

Proof, (i) By the definition of Luxemburg norm, it is easy to show that 1Мф ^ 1 implies (p(-,\u(-)\)edomI4, and

/ (р[Ф(-,|п(-)|), R] < 1.

If not, we would have

1 < j T [t, (p(t,\u(t)\)]dp(t) ^

oo.

T

Without loss of generality, we may suppose that the integral is finite.

(4)

116 R. Pl uci e nni k, S. Tian and Y. Wang

Then

I V <p(t,\u(t) 1)

’ Iv [(p(-,\u{‘)\), Я] d/n(t)

< 1

/*.!>(•> MOI), R] 1 î 'f t , q>(t,\u(t)\)]dii(t) = 1.

Therefore, we get a contradiction as follows:

1 =

<

1

-MOI), к ] 1

Ivb>(' ,M -)I),R ] 1

^ [ N ’ MOI). R]

J (p(t,\u(t)\)] dn(t)

T

j[ÿ(Mu(t)|)+ï'[t><p(t,|«(t)l)]]^W

T

$\u(t)\(p(tt \u(t)\)dn(t) ^ ||м||ф,

which implies (i).

(ii) By (i) and using the estimate from the proof of (i), we have j Ф(г, \u(t)\)dfi{t) < 1

М

ф

( - ,М - ) М ]

j <P(t,\u(t)\)dfi{t)

^ J..| ^ . I [#>N01)+!?(*> </?(t,|M(t)l))] N ( 0 < \\и\\ф, which finishes the proof of (ii).

(iii) In view of (i), we can suppose that m > 1. By the monotonicity of (p(t,•) for a.e. teT, we have

2|u(r)|

\u(t)\(p(t,\u(t)\) ^ j (p{t,s)ds

2|u(f)|

^ J (p(t, s)ds = Ф(т 2|м(01) о

for a.e. teT. Hence

¥>[>, v(t,|u(t)|)] = |w(t)l l«(OI)—Ф(г,l«(t)l)

«S Ф(£,

2

|ы(г)|)+Ф(г,|и(

1

)|) « 24>(t, 2|»(t)|)

for a.e. teT. Since Ф satisfies Condition A, there exist К ^ 1 and <5(г) defined on T such that

j Ф(С S(t))dfi(t) < oo and 0(t, 2и) ^ КФ(

1

, u) for a.e. teT,

T

provided и ^ S(t). Hence, choosing an integer n0 > 1 such that 2 < 2m < 2"°,

(5)

Non-convex integral functionals on Musielak-Orlicz spaces 1 1 7

we have Ф(1, 2mu) ^ Ф(1, 2nou) ^ К"°Ф(1, u), whenever и ^ 5(1) for a.e. teT.

Let

Tx = {teT: \u(t)\ < mô(t)} and T2 = T \ ^ . Therefore

/^[<i!)(-,|M(-)|), Я] = J *^[r, <p(l,|w(l)|)]^u(l) ^ 2 |Ф(1, 2 \u(t)\)dp(t)

T T

= 2 j Ф[1, 2\u{t)\\dp{t) + 2 j Ф[1, 2m |u(l)|/m]d/i(f)

T i T 2

^ 2 j Ф(1, 2mô(t))dp.{t) + 2Kno J 0[t,\u{t)\jm\dp{t)

T i r 2

^ 2K"°{j Ф(1, 5(0)^(l) + l} = M < oo,

T

which completes the proof of (iii).

Le m m a

2. Let х( -) е Ь ф(Т,Х) and let f{-) satisfy the local Lipschitz condition and Condition (M). Define

f t(v) = f?(x(t), u)

for each fixed v e X and for any teT. Then the function t-*ft(v) is measurable.

P roof. Since /,(•) satisfies the local Lipschitz condition, we may express /,°(x(f), v) as the upper limit of

(1) ft(y(t) + lv)-ft{y(t))

A

with 2 j0 taking only rational values and y(t)-+x(t) taking values in a count­

able dense subset (xj,® t of X. But (1) defines a measurable function of t by Condition (M). Thus ff(x(t), v), being a countable upper limit of measurable functions of t, is measurable in t.

Le m m a 3. (i)

For х ( - ) е Ь ф(Т, X), we have

М ф = sup j |j/(f)l \\x(t)\\xdp(t) = III X III ф.

I v ( ÿ , R ) ^ 1 T

(ii) For х( -) е Ь ф(Т,Х) and ÿ ( - ) e LV{T, R), we have J 1Я01 \\x{t)\\xdp{t) < III

t

III(«

p

) IWL>

T

where |||ÿ |||m = inf{r > 0: Iv [ÿ/r, R] ^ 1}.

Proof, (i). By the definition of || • ||ф, it is easy to see that

\\х\\ф= sup J <y(l), x{t)ydp(t) ^ sup S\\y{t)\\Y\\x{t)\\xdn(t)

I < r ( y , Y ) ^ l T 1 чг(у,у) ^ 1 T

= sup Ш 0 1 У * ( 0 М / Ф К sup j |ÿ(t)| \\x{t)\\x dp(t).

I v l II y(-) II Y, R) ^ 1 г I r i y . l t ) ^ 1 T

(6)

118 R. Pl uci e nni k, S. Ti an and Y. Wang

For every уеЬ^(Т, R) such that 1т{у, R) ^ 1 and for every x(-)eM(X) choose

z

(-)

e

M(Y) such that <z(f), x(t)> = ||x(t)||^ and \\z{t)\\Y = 1 for a.e. teT.

Define y(t) = z(t)\y(t)\ for teT. Then I^(y, У) = /^(ÿ, R) ^ 1 and j (y(t), x(t)}dp(t) = j \ÿ{t)\(z(t), x(t)}dg(t) = j |ÿ(t)| \\x(t)\\xdfi(t),

T T T

which implies that

||х||ф = sup j\y(t)\\\x{t)\\x dp(t).

I v ( y , R ) ^ 1 T

(ii) In virtue of part (i) of Lemma 3 and by the Holder inequality we have Л

у

(0И1*(0М

а

*(0 < lllÿlllc'P)HMH® = Illy 111(«р)||х||ф,

т

and the proof is complete.

Now, we will prove the main theorem.

Th e o r e m 1.

I f f ( - ) satisfies Condition (M) and Condition (A), then (i) / satisfies the Lipschitz condition on ЬФ(Т, X);

(ii) For every х е Ь ф(Т, X )

(2) df(x) c= J ôf,(x{t))dp(t);

T

(iii) I f for each teT, f ( - ) is regular at x(t), then f is regular at x, and

(3) df{x) = $dft(x{t))dg(t),

T

where

d/,(x(t)) = {yeL ^T , У): ff(x(t)) ^ <y(f), v(t)}, for all v e L 0(T, X)}.

P roof, (i) Let x l5 х 2е Ь ф{Т, X). Then, by Lemma 3 and Condition (A), we have

\f(x1) - f ( x 2)\ ^ J |/f(x1(t))-/f(x2(0)|^(r)

T

< f \K(t)\\\x1( t ) - x 2{t)\\x dfi{t) ^ IIIXHIm ||xj —х2||ф = Ц \х 1- х 2\\ф.

T

(ii) For every х е Ь ф(Т, X), Я > 0 and v e L 0(T, X), by Condition (A) and Lemma 3, we have

f(y{t) + b ( t ) ) - f t(y(t))

Я *sl*(t)INOIIx,

where у е Ь ф(Т, X), y->x in ЬФ(Т, X) and

\ \K(t)\ \\v(t)\\x dp(t) ^ HI K ||| m 1М1

ф

< °o.

(7)

N on-convex integral functionals on Musielak-Orlicz spaces 119

It follows from the Fatou lemma and (i) that

(5) -0 0 < /■>(*, ,) - iimsup + w

y-> X

y

^

ХЮ

< Iimsup

j

/ . H + ^ M ) - / , W )Jai(0

y(t)->x(t) J A

A | 0 ■

= 1 /° ( x (f)> v(t))dfi(t),

T

where in the fourth term y(t)->x(f) almost everywhere. According to the definition of the generalized gradient, in view of (5), we have

(6) J /,°(х(г), И 0 Ж 0 3= / V ») 3 v)

T

for any Çedf(x) and for all

v e

L0(T,X).

Let f t{v) = f t°(x{t),v) for v e X . By Lemma 2, the function t-*ft{v) is jU-measurable for every v e X . Define

f(v) = J

t "

for all

v e

L0(T, X). Then (5) can be written as follows:

for all

v e

L0(T,X). In virtue of Proposition 2.11 from [5], f(v) is a finite convex function and £ed/(0), where 5/(0) is the subdifferential of / at 0. By Theorem 3.1 and Corollary 3.11 in [2] there exist ÇeL^T, У) and a singular functional ,9 such that

<£, v} = j <£(f), v(t)}dfi(t) + &(v)

T

for all

v e

L0(T, X), where (e d /t(0) = ôft(x(t)) and #(i>) ^ 0 for all vEdomf.

Since f(v ) is finite for all

v e

L0{T, X), we have 5 = 0. Hence

<£> v) = f <((0> v(t)>dn(t)

T

for all

v e

L0(T, X), i.e.

f e $ dft(x(t))d f(t)

T

and so

df(x) c J dft(x{t))dfi{t).

T

(iii) For any

v e

L0(T, X), by the assumptions, we have

f t'{x{t), v(t)) = f t°(x(t), v(t)).

(8)

120 R. P l u c i e n n i k , S. T i a n a n d Y. W a n g

According to (5) and the Fatou lemma, we have

f°{x, v) ^ J f t°(x{t), v{t))dp{t) = J//(x(t), v{t))dp{t)

= jiim fl*M+Mo)-/.(*w^(t)

Y Я k О A

Я “ ^ 0 j* A

= liminf/(X + ;f ~ / ( x ) ^ / 4 x , 1,),

A->0 Л

i.e. f ( x , v) = f°(x,.v) for all v e L 0(T, X ).

Let £e j r 5/t(v:(t))d)u(t). Then there exist £(£)e df{x{t)) for a.e. te T such that

<£, t?> = j <C(0, v(t))dfi(t) < J / t°(x(f), ^ (0 )^ (0 = f°(x, v),

г г

so £ed/(x). From this and by part (ii) of our theorem, we have d/(x) = J 5/((x(t))d/r(t),

г

which finishes the proof of Theorem 1.

Th e o r e m

2. I f Ф(г, и) satisfies Condition Л and /Д-) satisfies Condition (M) and Condition (B), then

(i) / satisfies the Lipschitz condition on any hounded subset of L0(T, X );

(ii) For ebery x e L 0(T,X)

df(x) c= j df(x(t))dp(t);

T

(iii) I f for each teT, f ( - ) is regular at x(t), then f is regular at x and df(x) = J df(x(t))dp(t).

T

P roof, (i). Let x e L 0{T, X) and ||х||ф ^ m. For any u e L 0(T, X) such that

||w ||ф ^ m, by the mean-value theorem (cf. Th. 2.3.7 in [5]), we have (7) f(u{t))-f(x{t)) = <c(t), u(t)-x(t)),

where Ç(t)edft(x*[t)), x*(f) = 0(t)u(t) + (l — 0(t))x(t) and 0(t)e[0, 1] for all teT. By the monotonicity of '(p(t, • ) and Condition (B), there is a c > 0 such that

IIÉWIIr ^ c [ l+ ( ? ( t , ||х*(011лг)] = c[l+ç>(f,||O (t)M (O + (l-0 (O )^ (O |U )]

c[\ + (p(ti max(||u(t)||Xi ИОН*))]

< c[l +(p(t, \\u(t)\\x) + (p(t, ||x(t)||*)].

(9)

Non-convex integral functionals on Musielak-Orlicz spaces 121

Define

(8) y(t) = c[\ +(p(t, \\u(t)\\x) + (p(t, Цх(ОИл-)].

By Lemmas 1 and 2 and Holder’s inequality, it follows from (7) and (8) that

\f(u)-f(x)\ ^ $\ft(u(t))-ft(x(tj)\dfi{t)

T

= ^ j U(t)\\y\\u(t)-x{t)\\xdfi(t)

T T

< j 1Я01 \\u(t)-x{t)\\x dfi{t) ^ L\\u-x\\0,

T

where L = |||y ||| (<F) depends only on m.

(ii) For

x e

L 0(T, X ), у е Ь ф(Т, X), y->x and v e L 0{T, X), by the mean- value theorem, we have

(9) \ft(y(t) + AV(t))-ft(y(t))\ Ш , Н Ф

x = \< m , яо>1 for any X > 0, where Ç(t)edft(x*(t)) and

x*{t) = 0(t)(y(O + ^(O) + (l ~0(t))y(t) = X0(t)v{t) + y{t).

By the same proof as for (8) there exists a ÿ e L (f,(T, R) such that ||£(£)||y < y(t) and hence, by (9), we have

\ft(y(t) + Xv(t))-ft(y(tj)\ ^ |ÿ(OI \\v(t)\\x . Since

ЛЯ01 \\v(t)\\xdn(t) ^ Illy III (У)И1ф <

00

,

T

using the Fatou lemma, we obtain a formula analogous to (5). Therefore, repeating the argument in the proof of Th. 1 (ii), we have

df{x) c { df(x(t))d[i(t).

T

(iii) The proof is almost the same as the proof of Theorem 1 (iii).

Now, we will discuss the lower semicontinuity of the functional / To this end we will introduce some general definition concerning this fact.

De f in it io n 3.

A functional

/ :

U

R on a Banach space U is called lower semicontinuous if and only if its epigraph

Epi / = {(x, a)e U x R: / ( x ) ^ a | is closed in U xR.

Le m m a 4.

I f xn-+x in ЬФ(Т, X) and Ф satisfies Condition A, then

xn{t) — ->-0 almost everywhere in T.

(10)

1 2 2 R. P l u c i e n n i k , S. T i a n a n d Y. W a n g

P ro o f. Passing to a subsequence, we may assume that ||хи — х||ф < 2 "for all natural n. For each natural k, let

An,k = {teTk: ||xB(t)-x(0llx > V4»

where (Tk) is an increasing sequence of measurable subsets of T such that

00

[j Tk = T and ft{Tk) <oo for к = 1, 2, ...

k = 1

Set 00 00

^oo,к 0 LJ fe­

rn =ln=m Since \\хлп,к\\Ф < \\к(х„-х)\\ф < k/2n, we have

l lz w ll* ^ X \\Хлп,к\\ф+\\Х и Лп1к\\ф<к/2т~1 + \\х AnJ #

n — m n > j n > j

for each j ^ m ^ 1. Since Ф satisfies Condition Д, it follows that I I X u ^ fcU ° as j -кх),

n > j

so \\Хлх,к\\ф = 0 and hence р{Аж^ = 0. Putting к = 1 ,2 ,..., we obtain 00 00 00

r f U П U

{ t e T k :

iix „(0 -x (0 iix ^ i/fc }] = o,

fc = 1 m = 1 n = m

which shows that ||x„(f) — х(г)||х ->0 almost everywhere in T.

Th e o r e m

3. I f /Д-) satisfies Condition (M) and/Д-) is lower semicontinuous on X for almost every teT, then

(i) / is lower semicontinuous on L0(T, X)\

(ii) For every х е Ь ф(Т, X) the set function A^>f(%Ax) is countably additive on the o-algebra Z, provided /ДО) = 0 and f is proper, i.e. f { x ) > — oo for every х е Ь ф(Т,Х).

P roof, (i) Let (x„, a„) be a sequence in Epi/ convergent to (x, a) in L0( T , X ) x R , i.e. ||x„ — х||ф->0 and |a„ — a|-»0 as n->oo and /(x „ )^ a „ for every natural n. By Definition 3 it is sufficient to show that / (x) ^ a. By Lemma 4 we have ||x„(t) — x(t)||x ->0 a.e. in T. In view of the lower semicon­

tinuity of / ( • ) for a.e. teT, we have

f t(x{t)) ^ lim inf f(x(t)) a.e. in T, so by the Fatou lemma

/(* ) = I ^ J liminf/(x„(0)dju(0 ^ lim inf f f t(xn(t))dp(t) ^ a.

T T л ~ * oc f\ oo j

(11)

Non-convex integral functionals on Musielak-Orlicz spaces 123

(ii) Let х е Ь ф(Т, X) and A = |J* =1 An with pairwise disjoint Ane l . Then we have

ЯХл х) = Z / (Х

а

{

х

)+/(Х

впх

) (n> 1),

i = l

where Bn = By the lower semicontinuity of /, we have lim inf/(xBnx) ^ /(0) = 0.

Moreover, it is easy to prove that ||/в„х||ф|0 as n->cc. It follows that

П П

f (xAx) > limsup £ f(XAiX) + HminSf(xBnx) > limsup £ f ( x Aix )•

Л 00 j = l П-+СС HGO

On the other hand, since П

II z О и * - Х л * ) | | ф = \\Хвпх \\Ф1 0 , i = 1

we have

f(XAx) ^ liminf/ ( £ x^.x) = Hminf £ /Ou,*)-

n-* со j = l л oo г = 1

Thus

/ ( b * ) = Z /O u * )- i= 1

References

[1] A. D. I o f fe and V. L. L e v in , Subdifferentials of convex functions, Trans. Moscow Math. Soc.

26 (1972), 1-72.

[2] A. K o z e k , Convex integral functionals on Orlicz spaces, Comment. Math. 21 (1979), 109 135.

[3] R. P lu c ie n n ik , Conditions for compactness o f integral operator on Musielak-Orlicz space of vector-valued functions, Publ. Math. 35 (1988), 1-9.

[4] W u C o n g x in , W a n g T in g fu , C h en S h u t a o and W a n g Y u w en , Geometry of Orlicz Spaces, Harbin, 1986 (in Chinese).

[5] H. F. C la rk , Optim and Nonsmooth Analysis, Wiley-Interscience, New York 1983.

INSTYTUT MATEMATYKI, POLITECHNIKA POZNANSK.A

INSTITUTE OF MATHEMATICS, POZNAN TECHNICAL UNIVERSITY PIOTROWO ЗА, 60-965 POZNAN, POLAND

DEPARTMENT O F MATHEMATICS

HARBIN UNIVERSITY OF SCIENCE AND TECHNOLOGY 22 XUEFU ROAD, HARBIN, CHINA

Cytaty

Powiązane dokumenty

We shall generalize Theorem 4 [9] to Musielak-Orlicz spaces and we shall give some sufficient conditions for reflexivity of Musielak-Orlicz

In this paper the distinguished role is played by the methods of Banach function spaces the theory in which was created mainly by

Moreover, in this case the Mackey topology is semimetrizable and it is a unique semimetrizable locally convex topology between the weak topology o f X and the

In this paper we study the regularity of weak minimizers of integrals of type (1.1), under hypotheses (1.2) and (1.3).. Special emphasis will be placed on the integrands F (x, ξ)

In this paper we give the representation of linear functionals norm continuous on the space of finite elements Н0ф in the Hardy-Orlicz space Н*ф.. The linear functionals

Suppose an N -function i{/ does not satisfy the condition (d2).. Then t, is a linear functional norm continuous on

In the following we give a definition of the class of functionals that will be a basis for our considerations. We start with a brief summary of assumptions

It is proved in Theorem 1 that an additive functional separating these two types of functionals always exists.. Kaufman, of the well- known Mazur-Orlicz Theorem