• Aucun résultat trouvé

lf fd/ I f f @l oo} weighted Fourier algebras

N/A
N/A
Protected

Academic year: 2022

Partager "lf fd/ I f f @l oo} weighted Fourier algebras"

Copied!
6
0
0

Texte intégral

(1)

Sets of synthesis and sets of interpolation for

weighted Fourier algebras

U. ]~. TEwAI~I

Institut Mittag-Leffler, Djursholm, Sweden

w 1. Introduction. L e t A0(T ) denote the Banach algebra of continuous func.

*ions with absolutely convergent Fourier series. We define An(T ) = { f E C ( T ) : ~

lf(n)t

(1 + [hi) ~ < c~}, a > 0 .

n

We shall also be concerned with the Banach algebra of Lipschitz functions An(T), As(T ) and (A n N A)(T). We let A0(T ) = C(T) and ~I(T) = CI(T), [6;

pp. 42--43].

Let ~ e- C(T) be a regular Banach algebra such t h a t the maximal ideal space of /~ is T. For a closed subset E of T, we define

I ' ( E ) = { f e t ~ : f = O

on E } ,

_R(E) = R/IR(E)

is the restriction algebra of _R to E .

where .R'

f e _d(E),

B(E) = {f C C(E)

: sup

~eM(E)

~ 0

is the dual of R. _R(E)

[]f[I~

is defined b y

ilYlla =

I f f @l oo}

[I/~IIw <

is called the tilda algebra of _R(E). For

lf f d /

sup

u : ~ O

L e t I be a closed ideal in /~, then hull I is defined to be the set of common zeros of all functions in I. We say t h a t a closed subset E of T is of

synthesis

in if

IR(E)

is the only closed ideal in R whose hull is E a n d t h a t

ideal theorem

holds for E

in ~ if every closed ideal I in R whose hull is E is the intersection of all closed primary ideals containing I. We let

(2)

206 ARKIV F()~, 3 f . A T E M A T I K . VO]. 9 :No. 2

PM~(T) -~ (A~(T))'

and M~(T) = (A~(T))'

Let us remind ourselves t h a t when we talk of A~, the index ~ E [0, oo[ while in ease of 2~ a n d 2 ~ N A , a E [ 0 , 1 ] .

An element of P M ~ ( T ) will be called an ~-pseudomeasure and an element of M~(T) will be called an a-measure. B y PM~(E) [M~(E)] we shall denote the set of a-pseudomeasures [c~-measures] carried by E. E is called a set without true a-pseudomeasures if PM~(E) = M~(E) and E is called an Ha-set if A~(E) = A~(E).

I n Section 2, we shall discuss the Ditkin property for the algebras A~ and 2~ Cl A and prove the following:

THEOREM 1. Let E be a closed subset of T such that for an infinite sequence N ~O

{ j}}=l qf integers, the points 27~j/N (N E {Nj}) either belong to E or are at least at a distance 2 s I N from E. Then E is a set of synthesis for 2 ~ n A , 0 < o ~ < 1.

I n Section 3, we describe a totally disconnected perfect set of synthesis for A~(T), 0 < ~ < 1/2.

Finally in Section 4, we discuss H~-sets and sets without true ~-psendomeasures.

We prove t h a t only finite sets are Ha-sets. I n this process we shall also prove t h a t every function f E A~(E) can be extended to A~(T) without increasing its norm and t h a t the tilda algebra of A~(E) is A~(E).

I wish to express m y deep gratitude to Professor N. Th. Varopoulos for his generous advice a n d criticism and to the I n s t i t u t Mittag-Leffler for its hospitality.

w 2. Carl Herz [1] has proved t h a t if f E A~ is such t h a t f(to) = if(to) . . . fM(t0) = 0 t h e n there exists a sequence of functions f , E A~ such t h a t f , = 0

in a neighbourhood of t o and

I l f - - f f . i b = ~ 0 a s n - ~ o o .

We state the following consequence of his result.

TItEOEE~. Let E and F be closed subsets of T such that E is of synthesis in A~ and F has countable boundary. Then ideal theorem holds for E U F in A~.

I n particular, ideal theorem holds for a closed subset with countable boundary for the algebra A~.

I f 0 < c~ < 1, we observe t h a t usual ])itkin property holds for the algebra X~ Cl A. We shall now give the proof of Theorem 1.

For a positive integer n and f E 2~ Cl A, let us define f , C 2~ Cl A as follows:

f,(2:~j/n) = f(2z~j/n), 0 < j < n and linear in each interval [2~j/n,

2~(j + 1)In].

(3)

S E T S O F S Y : N T t I E S I S A I ~ D S E T S O:F I ~ T E I ! , P O L A T I O N : F O R V e ' E I G t t T E D : F O U I ~ I E I ~ A : L ( ~ E B R A S 207

L e t T , : ~ f3 A - + 2~ 13 A be t h e linear operator defined b y T,df) = fn. The r o u t i n e c o m p u t a t i o n shows t h a t I]T.][ < 3 for e v e r y n. Since piecewise linear functions are dense in X~ [3 A it follows t h a t i f iv C a n infinite sequence {Nj}

of integers, T N f - - > f in 2 ~ [ 3 A for all f E 2 ~ [ 3 A as i V - - > ~ .

L e t E a n d {Nj} be as in t h e s t a t e m e n t of Theorem 1. The condition on E implies t h a t if iV e{iVj} t h e n T N f = 0 on E for all f eI;~md(E). More- over, T N f vanishes on an open set which contains all b u t finitely m a n y points of E.

Given s > 0, choose iv E {iVj} so large t h a t l l f - <

N o w t h e D i t k i n p r o p e r t y for ~ [3 A implies t h e existence of a g E X~ 13 A such that g. T N f vanishes in a n e i g h b o u r h o o d of E a n d

IiT f - g " TNflb < .

Therefore I I f - g" TNfll < e where gTz~f = 0 in a n e i g h b o u r h o o d of E. This proves t h a t E is of synthesis for t h e algebra 2~ [3 A.

w 3. W e shall follow McGehee [4] in constructing a perfect t o t a l l y disconnected set of synthesis for t h e algebra A~, 0 < a < 1/2.

W e recall t h e following l e m m a due to McGehee [4] a b o u t f i n i t e l y s u p p o r t e d measures.

3.1. LE~MA. Let F = { x j : 1 < j <_k} be a finite set of distinct points of T.

Then given s > 0 there exists a number N N ( x 1 . . . x k ; s) such that for every

e M ( F )

m a x

I~(n)l

>_ (1 - s)II~llP~, for each m;

In--ml << N

where

II~tlPM = sup T/~(n) I

rt

W e construct E as t h e intersection of a decreasing sequence of closed sets E k, each E k being t h e u n i o n of s~ disjoint closed intervals each of length lk. L e t Ek = {x (1) , 9 9 9 x (sk)} be t h e set of the left end points o f the intervals c o n s t i t u t i n g E k. Given Ek, choose N k (by L e m m a 3.1) such t h a t for e v e r y m e a s u r e # E M(Ek),

sup L (n)l > 89 I~l -< Nk

Iqow choose t h e l e n g t h lk of t h e intervals in E k such t h a t t h e points of E~ are a t least 21k a p a r t a n d such t h a t

= o(1) (1)

W e shall n o w show t h a t E is a set of synthesis for t h e algebra A~, 0 G ~ < 1/2.

(4)

208 ~ Z K ] Z V F~JZ~ 1VIATE1VI2~TIK. VO1. 9 l'q'o. 2 3.2. L ~ A . To each S E PM~(E)

gk E M(Ek) such that

I S ( n ) - ~ ( n ) i

<C(~)lnI(skl~)~/'iiSli~

for all 1~ and n where C(~) is a constant depending only on ~.

I n particular,

we can associate a sequence of measures

(2)

lim ~k(n) ---- S(n) for all n . (3)

/~ --> co

Proof. We observe t h a t the formal integral of S is the L~-function S(n) e i"~ (where we have assumed t h a t S(0) 0) with the norm

~(x) ~ Z - - =

n r s n

[Jail2 _< C(c~)[ISl[p~ where C(~) is a positive constant depending only on a. The proof of the lemma is now exactly the same as t h a t of the lemma 4, p. 141--143 in [4].

3.3. TEEO~E~. E is a set of synthesis for the algebra A~, 0 _< ~ < 1/2.

Proof. Let S e PM~(E).

for all n

Therefore

B y (2) in L e m m a 3.2, we have IS(n) -

~(u) l _< c(~)Inl(sA)x/2ilSilpM~

and k. N o w b y our choice

sup I~k(n) l > 89 [I/~IIPM

19k(n) t

lf[~k[IPMc~ < 2 SUp < 211 "3 U C(O~)Nk(Sk~k) 112] liSlIPMo~

-- i~1<~, k (1 + In]) ~ --

~ o w since Nk(sklk)l/2~-0(1), it follows t h a t sup [l~tk][eM~ < ~ . This together k

with (3) of L e m m a 3.2 implies t h a t #k converges to S in the weak*-topology of PM~. This proves that E is of synthesis for the algebra A~.

Remark. B y changing the thinness condition (1), we can construct sets E which are of synthesis for the algebras A~, where ~ E [n, n -J- 1/2[, n is a positive integer.

w 4. We start b y stating the following proposition which relates the sets without true a-pseudomeasures and the Ha-sets.

4.1. :PRO:POSITIOn. Let E be a closed subset of T without true o~-pseudomeasures.

Then E is an Ha-set and the ideal theorem holds for E in A~.

(5)

S E T S O F SYI~TIIESIS A I ~ D S E T S O F I N T E R F O L A T I O I ~ FOI% W E I G H T E D ~ O U I % I E R A L G E B I % A S 209

W e n o w proceed to determine the H~-sets. W e need the following easy l e m m a . 4.2. LEM~-~. Let or > 0 and

l~ = sequences c = { c ~ } : s u p ~ ( 1 + lnl) ~ < oo , l~ = {sequences a = {a~}: ~ [a.l(1 § In[) ~ < oo},

n

c0,~ = {c e 12 : c~ = o ( I n l ~ ) } .

Then (Co,~)' = 11.

The following proposition is of i n d e p e n d e n t interest. H o w e v e r , we shall use t h e m e t h o d of its proof.

4.3. PROPOSITION. Let E be a closed subset of T. Then for each f 6 A s ( E ) there exists F 6A~(T) such that f [ E - - - - f and [[F[[A~(T)= ItfNAAE ).

Proof. L e t N~(E) b e the annihilator of I~(T)(E) in PM~(T) and L ( E ) = Co, ~ fl N~(E). T h e n f induces a b o u n d e d linear functional f ' on L(E) with the n o r m ~ NfI[A~(E)" B y the H a h n - B a n a c h t h e o r e m there is an extension F ' of f ' to Co, ~ such t h a t ]IF'I[ ~ llflI~(E). [By L e m m a 4.2 this corresponds t o a function

2' C A~(T) such t h a t [[F[[Aa(T ) ~_~ lIf[iAa(E)" W e shall n o w show t h a t FIE = f. L e t t o b e an a r b i t r a r y point of E. Then O,o 6 L(E). Therefore

f(to) = f'(~,.) = F(~,0) = F(t0).

This completes t h e p r o o f of t h e proposition.

Remark. The idea of the a b o v e p r o o f goes b a c k to Y. K a t z n e l s o n a n d K.

D e L e e u w [3].

W e n o w p r o v e the m a i n proposition n e e d e d for our s t u d y of H~-sets. The proposition is interesting for its own sake.

4.4. PROI'OSITIOZ~. Let E be a closed subset of T. Then A s ( E ) = A~(E).

Proof. L e t f 6 A~(E). T h e n f induces a b o u n d e d linear functional f ' on M(E) as a subspace of N~(E). As in P r o p o s i t i o n 4.3 we can e x t e n d f ' to Co. ~ a n d t h u s get F 6 A,(T) such t h a t FIE ~ f, i.e. f 6 A s ( E ). This completes the proof.

W e n e e d the following easy facts: (i) L e t E be a closed subse~ o f T. Then for 0 < ~ < 1, 2~(E) = As(E ) .

(ii) I f E is an infinite c o m p a c t subset o f T, t h e n 2~(E) 4 : A s ( E ) for 0 < ~ 1 .

(6)

210 AI~KIV FOR MATE~IATIK. VO1. 9 N o . 2

T h e a b o v e facts t o g e t h e r w i t h P r o p o s i t i o n 4.4 i m p l y

4.5. PROPOSITION. I f E is an infinite compact subset of T then A~(E) =~ ~ ( E ) . O u r findings a b o u t H~-sets c a n be s u m m e d u p in t h e following:

4.6. T~EORE~. Let E be a compact subset of following are equivalent:

(i) E is a set without true ~-pseudomeasures.

(ii) E is an H~-set.

(iii) E is finite.

T and 0 < ~ < 1. T h e n the

References

1. HE,z, C. S., Two problems in the spectral synthesis of unbounded functions. Mimeographed notes (1964).

2. KArAtE, J. P., & SAI.E~, R., Ensembles Tarfaits et s~ries trigonometriques, Hermann, Paris, 1963.

3. I)ELEEuW, K., & KATZNELSO~, Y., On certain homomorphisms of quotients of group algebras.

Israel J. Math. 2 (1964), 120--126.

4. MCG~HEF~, 0. C., Certain isomorphisms between quotients of a group algebra. Pacific J.

Math. 21 (1967), 133--151.

5. VA~OrOULOS, N. T~., On a problem of A. Beurling. J. Functional Analysis 2 (1968), 24--30.

6. ZYG~U~D, A., Trigonometric series, 2nd ed., Cambridge University Press, 1959.

Received May, 1970 U. B. Tewari

Indian Institute of Technology Kanpur (U.P.)

India

Références

Documents relatifs

For this purpose we shall prove the following lemma, where an appropriate upper bound for the absolute value of a quasiregular mapping is found... Sttppose that

The purpose of this note is to show how the argument of Martio and Sarvas may be used to obtain a similar injectivity criterion for mappings of a uniform domain in

is an open hyperbolic triangular region qrith a vertex angle at least a at the vertex zn, and the hyperbolic distance from zn to the opposite side of the

have investigated the maximal distortion under a normalized ft-quasisymmetric func- tion f of the real line.. In particular, we investigate whether the functions

The best earlier result is the Kolmo- gorov-Seliverstov-Plessner theorem s.. We can now repeat the

Abstract. Completely invariant components of the Fatou sets of meromorphic maps are discussed. Positive answers are given to Baker's and Bergweiler's problems that

In the present paper, we find A-complementary spaces of Fourier coefficients to Orlicz spaces L~ and ~ of functions of two variables, and to the space C of

Among the classical results in t h e theory of absolutely convergent Fourier series are the following theorems [6, Vol.. Attempts to generalize these theorems have