• Nie Znaleziono Wyników

Some remarks on coefficients of double Fourier series

N/A
N/A
Protected

Academic year: 2021

Share "Some remarks on coefficients of double Fourier series"

Copied!
5
0
0

Pełen tekst

(1)

ANNALES SOCIETATIS MATHEMATICAE POLONAE Series I: COMMENTATIONES MATHEMATICAE X I (1967)

T. M

arkiew icz

(Poznań)

Some remarks on coefficients of double Fourier series

Let co{t) be a positive function defined in (0, 1>. An increasing sequence {%} of positive integers is called со-lacunary if it satisfies the two follow­

ing conditions (cf. [ 2 ])

oo

N

( 1 ) JF co(l/W'*) = 0{co(l{nN)}, ^ п ксоЩпк) = 0{nNco(l/nN)}

fc = JV f c = l

(JV = l , 2 , . . . ; w i > l ) . We say that the trigonometric series

OO

( 2 ) A amitn соs wii x соs % у

£,?■=l

is (co1; со2)-lacunarу if the sequences {m*} and {w?-} are oox- and a) 2 -lacunary, respectively.

Denote by Q(s,t) a positive function, defined and bounded for positive s , t < u /2 . Suppose that f ( x , y ) is 27i-periodic in each variable separately. Write

Г Г

The class of all functions / Lebesgue-integrable with pth power over the square Q = [ 0 , 2 тг; 0 , 2 тг], such that

{ / / l 4 ?*/(® , y)\pdxdy\',v < CQ(h, k)

for positive h , Тс <л:/ 2 , G being a constant depending only on /, will be denoted by Ap(Q). Analogously we define the class H ^ (Q ) of 27t- periodic functions f ( x , y) essentially bounded in the square Q and satis­

fying the inequality

esssupl у )I < CQ(h,lc).

Q

3 — P r a c e M a te m a ty c z n e X I (1967)

(2)

34 T. M a r k i e w i c z

Let co1, co2 be positive functions bounded in Q. If sup У)\ = ° (со 1 (Д)co 2 (fc)} as h , k - > 0

X,yeQ

we shall say that / belongs to the class Н(сог, a>2).

The symbols G,Ck (fc = 0 , 1 , 2 , . . . ) used below will denote some positive constants.

T

heorem

1. Let co1(t) and co2(t) be moduli of continuity satisfying (1).

Then the continuous function f ( x , y), with a double (wj, oo2)-lacunary Fourier series ( 2 ), belongs to the class H(co1, co2) if and only if its Fourier coeffi­

cients атьП. are of order 0 {co 1 (l/m i)co 2 (l/%)}.

Proof. S u f f i c i e n c y . It is easy to verify that, in virtue of (1),

Choose, for each h, k, 0 < h < 1 / т 1? 0 < Тс < 1 / щ , and integers M

= M(h)j N = N{k) such that l/m M+1 < h < 1 /mM, l / n N+1 < к < l / n N . We assumed that

F { x , y ) = A(f l 2kf { x —h , y - k ) has the Fourier series

OO

absolutely and uniformly convergent. Hence

T A$l,2kf(®-h,y-k) 4

г = 1

j=N+l

i= l j= 1

oo

N

OO

00

oooo

i=MĄ~ 1 /=JV+1

Therefore

v

Applying (1) and the well-known inequalities ([4], p. I l l ) wi(^) ^

2

a)i(ti)

t2 tx

т 2 (У ^ 2

^2 ^1

and when h < t 2,

(3)

we have

$1 < C1hkmM(o1 1

mM n^a>2 1

% N

c m . со^ 11п^)

1/тм l / nN

^ „ , 7 «hW co2{k)

< C2h k — — — ---— = C2co1(h) co2(k ) ;

h к

Л1 oo

S l <C„* y j n < 0 , k ? ę ^ J - ± - )

Jj“ ' mi '

i

J

n

+ 1 \ nil l / mM \ nN+ll Ci (o1(h)co2(k).

Similarly, we obtain

$ 3 < С&о)г{Ь) ы2{к), $ 4 < Ceco1(h)co2(k) , and the proof of sufficiency is completed.

N e c e s s i t y . Observing that

4 ат.уП. = i- JJ A % .'njn. f ^ x - - ^ — , y ~ ~ ^ ) ^ ^ s i n y y d x d y and

^тг/т^тг/и.,-/j ж --- , у ---

2 m,- 2 m 7 11 2 т * /

U a ( —

2 \ 2 % we have

hence

dxdy,

It is easily seen that in case of necessity the assumption of lacunarity can be omitted. Theorem 1 is also true for general (cox, co2)-lacunary Fourier series of the form (15) of [3]. In case of one variable the result was obtained by Rubinstein ([2], Th. 2 . 2 ).

T heorem 2. Let 1 < p < 2 , l/p-j-l/q = 1. Suppose that f belongs to Л|^(12) and has the Fourier series

OO

«m.nCOS m x cosny m,n=o

(3)

(4)

36 T. M a r k i e w i c z

such that for some г, у > 0 , the numbers \am>n\m rn 11 are non-increasing in m, n ^ 1, separately. Then

am,n = 0 { ( - i - ) as m ^ l , n ^ l . l\ mn I \ 2m 2n / J

I n particular, if Q(s,t) = saf (a, /3 > 0) we have

(& rn. W . --

I ma+1/qn^+1,q 1

P ro o f. W rite F( x, y) = A ^ <tlkf ( x —h, y —k). The function F ( x , y ) has the Fo urie r series

4 ^ {0>m,n sin mh sin nk) sin mx sin n y . m,n=l

In view of Hausdorff-Young theorem ([5], p. 101)

OO X ^

j JV1 |4amnsinm^sin^|aJ < j— JJ ИйЬл/О» — h, y —k)\v dxdy\ .

B y the assumption,

{J/ y - m vdxdy)Vv < CQ(h, k),

Taking h = --- , к TC TC 2N л TC

0 <

в

<

2 , we get / 4 1

\q \

M

i \ l

N

and observing that sin0 > —

в

for

TC

w 1 1 \am>nmn\q < 11 1 4 amn sin mh sin nkf

' ' m = [ M / 2 ] + l « .= [iV /2 ] + l m = l w = i

< CqQq

\ 2M 7 2N Furthe r,

M N '

{ ~ w w f 1 1

i = [ M / 2 ] + l n = [ V / 2 ] + l

Hence

. „ = о { ( - к Г 0 Н , Н }

l\JfY/ \ 2 M ' 2NJ)

(5)

A similar theorem can be proved for general Fourier series (see [3], 7.2;

for one variable see [1], p. 428).

T

heorem

3. Let f e H (2)(Q) and let (3) be convergent to f(oc,y) for all pairs (%, y). Suppose that for some r , p > 0 , am>nm~rn~11 are non-nega­

tive and non-increasing in m , n ^ l , separately. Then

Proof. By the assumption,

d$l, 2 kf(oc—2h, 2 / —27b) = 0{Q(h, Щ for all {x,y).

It is easy to check that

oo

A{$, 2 kf{®— 2 h, у —Щ = 16 E am>n (sin mh sin nJc)2 cos mx cos ny.

m ,n = l

Therefore,

C,Q(h, ft) » l 4 * U / ( 0 - 2 7 ( , 0 —2ft)j

M N

> 16 I E am> n (sin mh sin n Jc)2.

m = [ M / 2 ] + l

n=[NI2] + l If h — tu /2М, Jc = n/2N we have

^

M N

E E a**.»“ 2»2 <

c

*

q

m=[MI2] + l n=[Nl2]+l

Beasoning now as in the proof of Theorem 2 we obtain the desired con­

clusion.

The last theorem is a slight generalization of a result given in [3], 7.4.

7C TC \

~ Ш ’ 2 N /

R eferences

[1] А. А. К он ю ш к ов , О классах Липшица, Известия Акад. Наук СССР, 21 (1957), рр. 423-448.

[2] А. И. Р у б и н ш т е й н , Об со-лакунарных рядах и о функциях классов Н ю, Мат. Сб. 65 (107), 2 (1964), рр. 239-271.

[3] R. T a b e r sk i, On double integrals and Fourier series, Ann. Polon. Math.

15 (1964), pp. 97-115.

[4] А. Ф. Тиман, Теория приближения функций действительного перемен­

ного, Москва 1960.

[5] A. Z y g m u n d , Trigonometric series, I, I I , Cambridge 1959.

INSTYTUT MATEMATYCZNY POLSKIEJ AKADEMII NAUK

Cytaty

Powiązane dokumenty

Key words and phrases: Fourier-Gegenbauer series, convergence, asymptotic estima- tion, Gegenbauer transform, strong derivative and of Gegenbauer integral, general- ized

Every weakly sequentially complete locally convex vector space, in particular every Hilbert space, satisfies condition (0) (this follows easily from the Orlicz-

They investigated the arithmetic properties of Q A and, in particular, they proved the following theorem:..

As consequences, we obtain classical theorems concerning absolute convergence of Fourier series, e.g.. Bernstein’s theorem, Zygmund’s theorem,

Clunie, On meromorphic schlicht functions,

Tripathir for his kind help and generous

Sharma for his valuable suggestions during the preparation of this

The Talagrand’s approach (see Theorem 4.3 in [7]) to Theorem 1.1 was based on the following idea: first prove (Proposition 3.4 and Theorem 4.3 in [7]) that for a given ϕ the