Internat. J. Math. & Math. Sci. VOL. 12 NO. 3 (1989) 579-582 579 FOURIER SERIES AND THE MAXIMAL OPERATOR ON THE WEIGHTED SPECIAL ATOM SPACES GERALDO SOARES DE SOUZA Department of Mathematics Auburn University Auburn University, AL 36849-5310 (Received November 24, 1987 and in revised form July 5, 1988) ABSTRACT. For an interval special atom looks like I . in b(t) [0,2] [(L(t) P (I I L with halves (R(t)], R, a weighted and where is a non- p ngative function satisfying some properties. We consider the weighted special atom space B(p) * formed by combinations of these weighted atoms. We showed that if f z B(p) linear then its Fourier series converges almost everywhere, using the Carleson-Hunt idea on their famous result about the almost everywhere convergence on KEY WORDS AND PHRASES. L-spaces. P Maximal operator, Lorentz Spaces and Fourier Series. 1980 AMS SUBJECT CLASSIFICATION CODE. I. Primary 42A20. INTRODUCTION. In [2] the following space was introduced. n n-0 n-0 weighted special atom, that is, a real valued function b(t)=--21’ which is either interval in III [0,21"], L or b(t)- is the left ha denotes the length of I, (|i|iPl IL or [XR(t)- XL(t)]’ I and R where is a non-negative, real valued function which is increasing, and is endowed with the norm llfllB(@)m Inf n-07, ,, ICnl all possible representations of is an E p(0) p and 0. B(p) where the infimum is taken over is a Banach space. For more details B(p) the reader is referred to [2]. a function on the almost everywhere convergence of theorem The Carleson-Hunt of f, series -, asserts that if f L then the Fourier p for L about this space in B(@) f. I is the right half. the characteristic function of XE [0,21.], b, defined on P denoted by < < S(f,x), P converges to f almost everywhere. (See [I] and [5].) Carleson-Hunt, we will prove that In this note, using basically the idea of B(@) then the operator condition on p if f with some additional f G.S. DE SOUZA 580 Tf(x) .ISn(f’x) sup n of Sn(f,x) where is the nth partial sum of the Fourier series L(). f, is bounded into some Banach space We would like to point out that other direct proofs of the almost everywhere B(p) convergence for functions in B(p) any function in For example, for some are also available. p However we prefer this one, satisfies Dini’s condition. because it is a consequence of the boundedness of the maximal operator and this boundedness could be useful in other contexts, for example in the interpolation of operators. One of the most important features of the space F functions I {z D on the disc I zl C F’(re 0 < i] is that under some F satisfying < dSdr l-r 0 where the dash means the derivative of For this z. with respect to [2]. characterization see 2. B(@) p, it can be identified with the space of analytic additional conditions on PRELIMINARIES. [0,-) p: [0,-) A function DEFINITION 2.1. is said to be in the class b, if it satisfies p(0) i) iv) lh0 v) h 0, @(s) ii) ds s 2 p(t) < p Cp(h) C dt C p() 0 < < f independent of b, We define the space (t)(t)dt measure and < (R)} where f* non-negative, decreasing function. Recall that if wf(x) t a, p. and h p(t) t a 2e log t {f:[0,2] L() by R is the decreasing rearrangement of If(x) > t}, y] f where seas means Lebesgue is a non-negative, decreasing function. In [6], G. G. Lorentz showed that and p(t) are L(#) Inf{y: seas {x; f*(t) defined by C with I. DEFINITION 2.2. I1 il, s 0 is decreasing, an absolute constant Example of functions in the class for (t) ---[- iii) is non-decreasing, lis sup n- k>n II is a norm if and only if L() Also he proved that is a is a Banach space. S (f x) is the nth partial sum of the Fourier series of f n then S (f,x) converges to f almost ISk(f,x) S(f,x) n >n everywhere if and only if 3. wf(x) 0 almost everywhere. MAIN RESULT The main result can be stated as follows. THEOREM. Let p be a function in the class operator defined by boundedly, that is, and S (f,x) n Tf(x) IITfll sp ISn(f,x) M[IflIB(p), where b,. maps M If B(p) B(p) f then the maximal L() into for (t) is a positive, absolute constant is the nth partial sum of the Fourier series of f p(t.__) t FOURIER SERIES AND THE MAXIMAL OPERATOR ON THE ATOM SPACES PROOF. [X fh(t (t) [-h,0) p(2h) g(t) estimate for sin[n+1/2]t t 2 sin (t). X > h 0 In fact let _< IS(g,x) n h- <2C ; Dn(t)dt _< A A where Sn(f,x) < Itl _< , < 0 dt Ch _< < x-h-- where x I Dn (t)dt. x-h satisfied by the Thus h 2C for x > 2h x x<-2h. ;; Dn(t)dt Recall that Therefore we have I ,t, x-h for -x )D n(x-t)dt is an absolute constant. _< Dn(t x-h and C where n S (g x) n IDn(t)l We now use the elementary inequality D f(x-a) which will follow from the is the Dirichlet kernel. Dirichlet kernels fa(x) where [O,h] [0,h] D (t) n a Consequently we just need to prove the result for ], (t) X fa T f First of all we notice that the operator maps B(p) into B(p) continuously. 581 is uniformly bounded in x; that is, and n (See [7], volume is an absolute constant. page 57. Consequently we have _< I) Tg(x) 2) Tg(x)_< 2C A x for ali Ixl for It We now evaluate > (t) for liTgll4, 2h. -----. p(t) [ITgl[,- ;02n (Tg)*(x) dx- ;02h (Tg)*(x) 2h @(x) 2n @(x) i A I dx + 2Ch dx. 0 12h Now using the conditions on llTgll _< AC The constant C p f(x) p(2h)2h llTfhll < E B(p) f(t) , l=nl < -. Cnbn(t) lITfll < Mn.0 Icnl we get If B(@) f , Cnbn(X) then where n-0 n-0 b n b (t) n where which implies complete. f(x) then n-0 n-0 almost everywhere. f dx Mp(2h) M and so if Let @ be in the class b,. Let PROOF. 12(Tg)*(x)0(X)2h x we get p(2h) + 2Ch (t) XLn (t)] and p(il n [(Rn IITfll _< MIIflIB(p)- The proof is to dx + may not be the same at every occurence. Consequently COROLLARY. . In fact, using I) and 2) we get [S (f x)} ICnl <- converges and the are weighted special atoms. Now observe that wf(x) lim sup n k>n ISk(f,x) S(f,x) satisfies wf(x) < 2Tf(x). Therefore >n 2n 0 w(x) p(t) dt other hand we see that _< [Tf(x)] *P(----!) 2 dt 0 wf and so wf L(). k wf_fk where fk(x) n-0’ Cnbn(X) and On the G.S. DE SOUZA 582 llf k fllB(p) 0 as k -. wf_fk(X) wf(x) Then consequently the theorem, that is, the boundedness of after definition 2.2 that We emphasize that for space B paper on p and B(p) L() Sn(f,x) t I/p, < L(p,l) is the space p in and T, implies that f(x) almost everywhere. p(t) The proof is complete. < the space [3], and so the results in this B(p) is the is a generalization of that result. We also point out that for f c B(p) fk)(X) 2T(f (t) so that the Fourier series of t, one can indeed show that there is a f, diverges almost everywhere. (See [4]). This answers a question asked of the author by Professor Guido Weiss concerning almost everywhere convergence of functions in B(p) for p(t) t. REFERENCES [I] CARLESON, L., On the convergence and growth of partial sums of Fourier series, Acta. Math. 116, (1966) 135-157. [2] BLOOM, S. and DE SOUZA, G. S., Atomic decomposition of the generalized Lipschitz spaces To appear, Illinois Journal of Mathematics. [3] DE SOUZA, G. S., On the convergence of Fourier series, Internat. J. [4] DE SOUZA, G. S. and SAMPSON, G., A function in Dirichlet space [5] HUNT, R. A., On the convergence of Fourier series, Proc. Conference Math. and Math. Sci. 7(4), (1984) 817-820. that its Fourier series diverges almost everywhere. [6] [7] B such Preprint. Southern Illinois University, Edwardville, IIi. 1967. LORENTZ, G. G., On the theory of spaces A, Pacific Journal of Math. I(1951), 411-429. ZYGMUND, A., Trigonometric Series, Cambridge University Press, London and New York, 1959. Division of Mathematics Foundations, Analysis and Topology Auburn University Auburn, AL 36849-5310