See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/307861334

Almost Everywhere Strong Summability of Fejer means of rectangular partial

sums of two-dimensional Walsh-Fourier Series

Article · September 2016

CITATIONS READS

0 85

1 author:

Ushangi Goginava
Ivane Javakhishvili Tbilisi State University
192 PUBLICATIONS   1,225 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

strong summability of Fourier series View project

generalized bounded variation View project

All content following this page was uploaded by Ushangi Goginava on 27 March 2018.

The user has requested enhancement of the downloaded file.

 ALMOST EVERYWHERE STRONG SUMMABILITY OF
FEJÉR MEANS OF RECTANGULAR PARTIAL SUMS OF
TWO-DIMENSIONAL WALSH-FOURIER SERIES
arXiv:1609.01640v1 [math.AP] 6 Sep 2016

USHANGI GOGINAVA

Abstract. It is proved a BMO-estimation for rectangular partial sums

of two-dimensional Walsh-Fourier series from which it is derived an al-
most everywhere exponential summability of rectangular partial sums
of double Walsh-Fourier series.

1. Introduction

We shall denote the set of all non-negative integers by N , the set of all
integers by Z and the set of dyadic rational numbers in the unit interval
I := [0, 1) by Q. In particular, each element of Q has the form 2pn for some
p, n ∈ N, 0 ≤ p ≤ 2n .
Let r0 (x) be the function defined by

1, if x ∈ [0, 1/2)
r0 (x) = , r0 (x + 1) = r0 (x) .
−1, if x ∈ [1/2, 1)
The Rademacher system is defined by
rn (x) = r0 (2n x) , n ≥ 1.
Let w0 , w1 , ... represent the Walsh functions, i.e. w0 (x) = 1 and if k =
2n1 + · · · + 2ns is a positive integer with n1 > n2 > · · · > ns then
wk (x) = rn1 (x) · · · rns (x) .
The Walsh-Dirichlet kernel is defined by
n−1
X
D0 (x) = 0, Dn (x) = wk (x) n ≥ 1.
k=0

Given x ∈ I, the expansion

0
2010 Mathematics Subject Classification 42C10 .
Key words and phrases: two-dimensional Walsh system, strong summability, a. e.
summability
The research was supported by Shota Rustaveli National Science Foundation grant
no.DI/9/5-100/13 (Function spaces, weighted inequalities for integral operators and prob-
lems of summability of Fourier series)
1
2 USHANGI GOGINAVA

∞
X
(1) x= xk 2−(k+1) ,
k=0

where each xk = 0 or 1, will be called a dyadic expansion of x. If x ∈ I\Q ,

then (1) is uniquely determined. For the dyadic expansion x ∈ Q we choose
the one for which lim xk = 0.
k→∞
The dyadic addition of x, y ∈ I in terms of the dyadic expansion of x and
y is defined by (see [16] or [34])
∞
X
x∔y = |xk − yk | 2−(k+1) .
k=0

Denote IN := [0, 2−N ),

IN (x) := x ∔ IN .
We consider the double system {wn (x) × wm (y) : n, m ∈ N} on the unit
square I2 = [0, 1) × [0, 1) .The notiation a . b in the whole paper stands for
a ≤ c · b, where c is an absolute constant.

The norm (or quasinorm) of the space Lp I2 is defined by
 1/p
Z
kf kp :=  |f |p  (0 < p < +∞) .
I2


If f ∈ L1 I2 , then
Z
fˆ (n, m) = f (x1 , x2 ) wn (x1 )wm (x2 )dx1 dx2
I2

is the (n, m)-th Fourier coefficient of f.

The rectangular partial sums of double Fourier series with respect to the
Walsh system are defined by
M
X −1 N
X −1
SM,N (x1 , x2 ; f ) = fˆ (m, n) wm (x1 )wn (x2 ).
m=0 n=0

Denote
n−1
X
Sn(1) (x1 , x2 ; f ) := fb(l, x2 ) wl (x1 ) ,
l=0
m−1
X
(2)
Sm (x1 , x2 ; f ) := fb(x1 , r) wr (x2 ) ,
r=0
where
Z Z
fb(l, x2 ) = f (x1 , x2 ) wl (x1 ) dx1 , fb(x1 , r) = f (x1 , x2 ) wr (x2 ) dx2 .
I I
 ALMOST EVERYWHERE SUMMABILITY 3
 

Recall the definition of BM O I2 space. 2

f ∈ L1 I . We say that f
 2Let
has bounded mean oscilation f ∈ BM O I if
 1/2
Z
 1 
kf kBM O := sup  |f − fQ |2  < ∞,
Q |Q|
Q

where Z
1
fQ := f
|Q|
Q
and the supremum is taken over all dyadic squares Q ⊂ I2 .
Let ξ := {ξn1 n2 : n1 , n2 = 0, 1, 2, ...} be an arbitrary sequence of numbers.
Taking  
n k k+1
δk := n , n ,
2 2
we define

2n1 −1 2n2 −1
X X
BM O [ξ] := sup ξ I n1 (t1 ) I n2 (t2 ) ,

0≤n1 ,n2 <∞
k k
1 2 δ
k1 δk
k1 =0 k2 =0
2
BM O
where IE is the characteristic function

of E ⊂ I2 .
We denote by L (log L)α I2 the class of measurable functions f , with
Z

α
|f | log+ |f | < ∞,
I2

where log+ u := I(1,∞) log u.

Denote by SnT (x, f ) the partial sums of the trigonometric Fourier series of
f and let
n
1 X T
σnT (x, f ) = Sk (x, f )
n+1
k=0
be the (C, 1) means. Fejér [1] proved that σnT (f ) converges to f uniformly
for any 2π-periodic continuous function. Lebesgue in [19] established almost
everywhere convergence of (C, 1) means if f ∈ L1 (T), T := [−π, π). The
strong summability problem, i.e. the convergence of the strong means
n
1 X  T p
(2) Sk (x, f ) − f (x) , x ∈ T, p > 0,
n+1
k=0
was first considered by Hardy and Littlewood in [17]. They showed that for
any f ∈ Lr (T) (1 < r < ∞) the strong means tend to 0 a.e., if n → ∞.
The Fourier series of f ∈ L1 (T) is said to be (H, p)-summable at x ∈ T , if
the values (2) converge to 0 as n → ∞. The (H, p)-summability problem in
L1 (T) has been investigated by Marcinkiewicz [24] for p = 2, and later by
Zygmund [44] for the general case 1 ≤ p < ∞. Oskolkov in [26] proved the
4 USHANGI GOGINAVA

following: Let f ∈ L1 (T) and let Φ be a continuous positive convex function

on [0, +∞) with Φ (0) = 0 and
ln Φ (t) = O (t/ ln ln t) (t → ∞) .
Then for almost all x
n
1 X  

(3) lim Φ SkT (x, f ) − f (x) = 0.
n→∞ n + 1
k=0

It was noted in [26] that Totik announced the conjecture that (3) holds
almost everywhere for any f ∈ L1 (T), provided
ln Φ (t) = O (t) (t → ∞) .
In [27] Rodin proved
Theorem R. Let f ∈ L1 (T). Then for any A > 0
n
1 X  


lim exp A SkT (x, f ) − f (x) − 1 = 0
n→∞ n + 1
k=0

for a. e. x ∈ T.
Karagulyan [18] proved that the following is true.
Theorem K. Suppose that a continuous increasing function Φ : [0, ∞) →
[0, ∞), Φ (0) = 0, satisfies the condition
log Φ (t)
lim sup = ∞.
t→+∞ t
Then there exists a function f ∈ L1 (T) for which the relation
n
1 X  

lim sup Φ SkT (x, f ) = ∞
n→∞ n + 1
k=0

holds everywhere on T.
For Walsh system Rodin [29] (see also Schipp [30]) proved that the fol-
lowing is true.
Theorem R2 (Rodin). If Φ(t) : [0, ∞) → [0, ∞), Φ(0) = 0, is an increasing
continuous function satisfying
log Φ(t)
(4) lim sup < ∞,
t→∞ t
then the partial sums of Walsh-Fourier series of any function f ∈ L1 (I)
satisfy the condition
n
1X
lim Φ (|Sk (x; f ) − f (x)|) = 0
n→∞ n
k=1
 ALMOST EVERYWHERE SUMMABILITY 5

almost everywhere on I.
In the paper [6] we established, that, as in trigonometric case [18], the
bound (4) is sharp for a.e. Φ-summability of Walsh-Fourier series. Moreover,
we prove
Theorem GGK. If an increasing function Φ(t) : [0, ∞) → [0, ∞) satisfies
the condition
log Φ(t)
lim sup = ∞,
t→∞ t
then there exists a function f ∈ L1 (I) such that
n
1X
lim sup Φ (|Sk (x; f )|) = ∞
n→∞ n
k=1

holds everywhere on I.
The two-dimensional Fejér summability of f ∈ L log+ L(I2 ) was proved
by Zygmund [44] for trigonometric Fourier series and by Móricz, Schipp and
Wade [25] (see also Weisz [40]) for Walsh-Fourier series. The two-dimensional
strong summability, i. e.
nX
1 −1 nX
2 −1
1
|Sk1 k2 (x1 , x2 ; f ) − f (x1 , x2 )|p → 0 a. e. as n → ∞
2n1 +n2
k1 =0 k2 =0

was shown by Gogoladze [13] for trigonometric Fourier series and for f ∈
L log+ L(T2 ). The same result for more-dimensional Walsh-Fourier series is
due to Rodin [28] (see also Weisz [41]). These results show that in the case
of two dimensional functions the (C; 1, 1) summability and (C; 1, 1) strong
summability we have the same maximal convergence spaces. That is, in both
cases we have L log+ L.
It is proved in ([12]) a BMO-estimation for quadratic partial sums of
two-dimensional trigonometric Fourier series from which it is derived an al-
most everywhere exponential summability of quadratic partial sums of dou-
ble Fourier series.
The results on strong summation and approximation of trigonometric
Fourier series have been extended for several other orthogonal systems. For
instance, concerning the Walsh system see Schipp [31, 32, 33], Fridli and
Schipp [2, 3], Leindler [20, 21, 22, 23], Totik [36, 37, 38], Rodin [28], Weisz
[41, 42], Gabisonia [4], Goginava, Gogoladze [11].
The problems of summability of multiple Fourier series have been inves-
tigated by Gogoladze [14, 15], Wang [39], Zhag [43], Glukhov [7], Goginava
[8, 9], Goginava, Gogoladze [10], Gat, Goginava, Karagulyan [5].

2. Main Results
In this paper we study a BMO-estimation for rectangular partial sums of
two-dimensional Walsh-Fourier series.
6 USHANGI GOGINAVA


Theorem 1. If f ∈ L (log L)2 I2 , then
 
Z
  1
 (x1 , x2 ) ∈ I2 : BMO [Sn n (x1 , x2 ; f )] > λ  . 1 + |f | (log |f |)2  .
1 2
λ
I2

The following theorem shows that the rectangular
sums of two-dimensional

Walsh-Fourier series of a function f ∈ L (log L)2 I2 are almost everywhere
exponentially summable to the function f . It will be obtained from the
previous theorem (see [12]) by using the John-Nirenberg theorem.


Theorem 2. Suppose that f ∈ L (log L)2 I2 . Then for any A > 0
2m
X1 −1 2m2 −1
X
1
lim (exp (A |Sn1 n2 (x1 , x2 ; f ) − f (x1 , x2 )|) − 1) = 0
m1 ,m2 →∞ 2m1 +m2
n1 =0 n2 =0

for a. e. (x1 , x2 ) ∈ I2 .

3. Auxiliary Results
Schipp in [30] introduce the following operator
  2 1/2
n −1 (l+1)2
Z
−n
2X n−1
X
   
Vn (x; f ) :=   2j−1 IIj (t) S2n (x ∔ t ∔ ej , f ) dt  ,
l=0 j=0
l2−n

Let
V (f ) := sup Vn (f ) .
n
The following theorem is proved by Schipp.
Theorem Sch1 ([30]). Let f ∈ L1 (I). Then
kf k1
µ {|V f | > λ} .
λ
and Z


kV f k . 1 + |f | log+ |f | f ∈ L log+ L (I) .
I

Theorem Sch2 ([30]). The following estimation holds

( 2n −1 )1/2
1 X 2
|Sm (x; f )| . Vn (x; |f |) .
2n m=0
Set
Vm1 m2 (x1 , x2 ; f )
 
m m (l1 +1)2
Z
−m1 (l +1)2−m2
2 Z
2 1 −1 2 2 −1 m1 −1 m2 −1
 X X  X
j1 −1
X
:=   2 IIj1 (t1 ) 2j2 −1 IIj2 (t2 )
l1 =0 l2 =0 j1 =0 j2 =0
l1 2−m1 l2 2−m2
 ALMOST EVERYWHERE SUMMABILITY 7

1/2
×S2m1 2m2 (x1 ∔ t1 ∔ ej1 , x2 ∔ t2 ∔ ej2 ; f ) dt1 dt2 )2 .
For a two-dimensional integrable function f we need to introduce the
following functions
(5) Vn(1) (x1 , x2 ; f )
  2 1/2
n −1 (l+1)2
Z
−n
2X n−1
X
  (1)  
: =  2j−1 IIj (t) S2n (x1 ∔ t ∔ ej , x2 ; f )2 dt  ,
l=0 j=0
l2−n

(6) Vn(2) (x1 , x2 ; f )

  2 1/2
n −1 (l+1)2
Z
−n
2X n−1
X
  (2)  
: =  2j−1 IIj (t) S2n (x1 , x2 ∔ t ∔ ej ; f ) dt  ,
l=0 j=0
l2−n
 
 
V (s) (x1 , x2 ; f ) := sup Vn(s) (x1 , x2 ; f ) , s = 1, 2.
n

Lemma 1. The following estimation holds

( 2m1 −1 2m2 −1
)1/2
1 X X 2
|f ∗ (Dn1 ⊗ Dn2 )| . Vm1 m2 (x1 , x2 ; |f |) .
2m1 +m2
n1 =0 n2 =0

Proof of Lemma 1. Let


−1, if j = 0, 1, ..., i − 1
εji := .
1, if j = i
In [30], Schipp proved that
n−1
X k
X
Dm (t) = IIk \Ik+1 (t) εkj 2j−1 wm (t + ej )
k=0 j=0
1
− wm (t) + (m + 1/2) IIn (t) , m < 2n .
2
Then we can write
(7) Hm1 m2 (x1 , x2 ; f )
( 2m1 −1 2m2 −1
)1/2
1 X X
:= |S2m1 2m2 (f ) ∗ (Dn1 ⊗ Dn2 )|2
2m1 +m2
2 n1 =0 n =0
 
 2m1 −1 2m2 −1
X X  Z m
X1 −1
1  S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; f )
≤  IIk1 \Ik1 +1 (t1 )
 2m1 +m2 
n1 =0 n2 =0 2 k1 =0
I
k1
X m
X2 −1

× εk1 j1 2j1 −1 wn1 (t1 ∔ ej1 ) IIk2 \Ik2 +1 (t2 )

j1 =0 k2 =0
8 USHANGI GOGINAVA

2 1/2
k2
X  

× εk2 j2 2j2 −1
wn2 (t2 ∔ ej2 ) dt1 dt2 
j2 =0  
 
 2m 1 −1 2m2 −1 Z m 1 −1
1 X X  X
+  S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; f ) IIk \Ik +1 (t1 )
 2m1 +m2  1 1
n1 =0 n2 =0  2 k1 =0
I
2 1/2
k1
X  
wn2 (t2 ) 
× εk1 j1 2j1 −1
wn1 (t1 ∔ ej1 ) dt1 dt2 
j1 =0
2  
 
 1 2m 1 −1 2m2 −1 Z m 1 −1
X X  X
+  S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; f ) IIk \Ik +1 (t1 )
 2m1 +m2  1 1
n1 =0 n2 =0 2  k1 =0
I
2 1/2
k1
X  

× εk1 j1 2j1 −1 wn1 (t1 ∔ ej1 ) (n2 + 1/2) IIm2 (t2 ) dt1 dt2 
j1 =0  
 
 2m1 −1 2m2 −1 Z m2 −1
1 X X  X
+  S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; f ) IIk \Ik +1 (t2 )
 2m1 +m2  2 2
n1 =0 n2 =0 2 k2 =0
I
2 1/2
k2
X  
w (t
n1 1 ) 
× εk2 j2 2j2 −1
wn2 (t2 ∔ ej2 ) dt1 dt2 
j2 =0
2  

 2m1 −1 2m2 −1 Z
1 X X
+ S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; f )
 2m1 +m2
n1 =0 n2 =0
I2
2 )1/2
wn1 (t1 ) wn2 (t2 ) 
× dt1 dt2 
2 2
 
 1 2m 1 −1 2m2 −1 Z
X X 
+  S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; f )
m
2 1 2 +m 
n1 =0 n2 =0  2
I
2 )1/2
wn1 (t1 ) 
× (n2 + 1/2) IIm2 (t2 ) dt1 dt2 
2
 
 1 2m 1 −1 2m2 −1 Z m 2 −1
X X  X
+  S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; f ) IIk \Ik +1 (t2 )
 2m1 +m2  2 2
n1 =0 n2 =0 2  k2 =0
I
2 1/2
k2  
X 
× εk2 j2 2j2 −1 wn2 (t2 ∔ ej2 ) (n1 + 1/2) IIm1 (t1 ) dt1 dt2 
j2 =0  
 ALMOST EVERYWHERE SUMMABILITY 9
 
 2m1 −1 2m2 −1 Z
1 X X 
+  S2m1 2m2 (x1 + t1 , x2 + t2 ; f )
 2m1 +m2 
n1 =0 n2 =0 2
I
2 )1/2
wn2 (t2 ) 
× (n1 + 1/2) IIm1 (t1 ) dt1 dt2 
2
 
 1 2m 1 −1 2m2 −1 Z
X X 
+  S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; f )
 2m1 +m2 
n1 =0 n2 =0  2
I
2 o1/2
× (n1 + 1/2) IIm1 (t1 ) (n2 + 1/2) IIm2 (t2 ) dt1 dt2 
9
X
:= Ri .
i=1
There is a suitable vector
n o
βn(1)
1 n2
(x ,
1 2x ) : 0 ≤ n 1 < 2m1
, 0 ≤ n 2 < 2m2

such that
2m
X1 −1 2m2 −1 
X 2
 (1) 
βn1 n2 (x1 , x2 ) = 1
n1 =0 n2 =0
and
(8) 2(m1 +m2 )/2 R1
 
Z mX
1 −1 m
X1 −1

= 2j1 −1  εk1 j1 IIk1 \Ik1 +1 (t1 ∔ ej1 )

j1 =0 k1 =j1
I2
 
m
X2 −1 m
X2 −1

× 2j1 −1  εk2 j2 IIk \Ik2 +1 (t1 ∔ ej2 )

2
j2 =0 k2 =j2

×S2m1 2m2 (x1 ∔ t1 ∔ ej1 , x2 ∔ t2 ∔ ej2 ; f )

2m
X1 −1 2m2 −1
X
× βn(1)
1 n2
(x1 , x2 ) wn1 (t1 ) wn2 (t2 ) dt1 dt2
n1 =0 n2 =0
Z mX
1 −1 m
X2 −1

≤ 2j1 −1 IIj1 (t1 ) 2j2 −1 IIj2 (t2 )

j1 =0 j2 =0
I2
×S2m1 2m2 (x1 + t1 + ej1 , x2 + t2 + ej2 ; |f |)
 m1 
2 X −1 2m
X2 −1 
 
× βn(1)
1 n2
(x ,
1 2x ) w (t
n1 1 ) wn2 2  dt1 dt2 .
(t )
 
n1 =0 n2 =0
10 USHANGI GOGINAVA

Analogously, we can prove that

Z mX
1 −1
(m1 +m2 )/2
(9) 2 R2 . 2j1 −1 IIj1 (t1 )
j1 =0
I2

×S2m1 2m2 (x1 ∔ t1 ∔ ej1 , x2 ∔ t2 ∔ e0 ; |f |)

 m1 
2 X −1 2m
X2 −1 
 (2) 
× βn1 n2 (x1 , x2 ) wn1 (t1 ) wn2 (e0 ) wn2 (t2 ) dt1 dt2 ,
 
n1 =0 n2 =0

(10) 2(m1 +m2 )/2 R4

Z mX
2 −1

. 2j2 −1 IIj2 (t2 ) S2m1 2m2 (x1 + t1 + e0 , x2 + t2 + ej2 ; |f |)

j2 =0
I2
2m1 −1 2m2 −1 
 X X 
 
× βn(4)
1 n2
(x 1 , x 2 ) wn 1 (t 1 ) wn 1 (e0 ) wn 2 (t 2 ) dt1 dt2 .
 
n1 =0 n2 =0

(11) 2(m1 +m2 )/2 R5

Z
. S2m1 2m2 (x1 ∔ t1 ∔ e0 , x2 ∔ t2 ∔ e0 ; |f |)
I2
2m1 −1 2m2 −1 
 X X 
 (5) 
× βn1 n2 (x1 , x2 ) wn1 (t1 ) wn1 (e0 ) wn2 (t2 ) wn2 (e0 ) dt1 dt2 ,
 
n1 =0 n2 =0
where
2m1 −1  2
X  (s) 
βn1 (x1 , x2 ) = 1 (x1 , x2 ) ∈ I2 , s = 2, 4, 5.
n1 =0
Now, we estimate R3 . There is a suitable vector
n o
βn(3)
1
(x 1 , x 2 ) : 0 ≤ n 1 < 2m1

such that
2m1 −1  2
X  (3) 
βn1 (x1 , x2 ) = 1, (x1 , x2 ) ∈ I2
n1 =0
and
(12) 2(m1 +m2 )/2 R3
Z m
X1 −1
3m2 /2
≤ c2 2j1 −1 IIj1 (t1 ) S2m1 2m2 (x1 ∔ t1 ∔ ej1 , x2 ∔ t2 ; |f |)
I×Im2 j1 =0
 m1 
2 X −1 
 
× βn(3) (x ,
1 2x ) w (t )
n1 1  dt1 dt2 .
 1

n1 =0
 ALMOST EVERYWHERE SUMMABILITY 11

Analogously, we can prove that

Z
(m1 +m2 )/2 (3/2)m2
(13) 2 R6 . 2 S2m1 2m2 (x1 + t1 + e0 , x2 + t2 ; |f |)
I×Im2
 m1 
2 X −1 
 
× βn(6) (x ,
1 2x ) wn1 (e0 ) wn1 1  dt1 dt2 ,
(t )
 1

n1 =0

Z m
X2 −1
(m1 +m2 )/2 (3/2)m1
(14) 2 R7 . 2 2j2 −1 IIj2 (t2 )
Im1 ×I j2 =0

×S2m1 2m2 (x1 + t1 , x2 + t2 + ej2 ; |f |)

 m1 
2 X −1 
 
× βn(7) (x ,
1 2x ) wn2 2  dt1 dt2 ,
(t )
 1

n1 =0
Z
(15) 2(m1 +m2 )/2 R8 . 2(3/2)m1 S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ∔ e0 ; |f |)
Im1 ×I
2m2 −1 
 X 
 (8) 
× βn2 (x1 , x2 ) wn2 (e0 ) wn2 (t2 ) dt1 dt2 ,
 
n2 =0
Z
(m1 +m2 )/2 (3/2)(m1 +m2 )
(16) 2 R9 . 2 S2m1 2m2 (x1 ∔ t1 , x2 ∔ t2 ; |f |) ,
Im1 ×Im2

where
2m1 −1  2
X  (s) 
βn1 (x1 , x2 ) = 1 (x1 , x2 ) ∈ I2 , s = 6, 7
n1 =0
and
2m2 −1  2
X  (8) 
βn2 (x1 , x2 ) = 1 (x1 , x2 ) ∈ I2 .
n2 =0
Set
2m
X1 −1 2m2 −1
X
(1)
Pm 1 m2
(x1 , x2 ) := βn(1)
1 n2
(x1 , x2 ) wn1 (t1 ) wn2 (t2 ) .
n1 =0 n2 =0

Then from (8) we have

2(m1 +m2 )/2 R1
1 −1 2m2 −1    (l1 +1)2
Z
−m1 (l +1)2−m2
2 Z
2m
X X m
X1 −1
 (1) l1 l2 
= P , 2j1 −1 IIj1 (t1 )
 m1 m2 2m1 2m2 
l1 =0 l2 =0 j1 =0
l1 2−m1 l2 2−m2
12 USHANGI GOGINAVA

m
X2 −1

× 2j2 −1 IIj2 (t2 ) S2m1 2m2 (x1 ∔ t1 ∔ ej1 , x2 ∔ t2 ∔ ej2 ; |f |) dt1 dt2
j2 =0
 
2m
X1 −1 2m2 −1 
X  2 1/2
 (1) l1 l2 
≤ Pm m , m2  
 1 2 2m 1 2
l1 =0 l2 =0
 
(l1 +1)2−m1
Z (l2 +1)2
Z
−m2
2m
X1 −1 2m2 −1
X m
X1 −1 m
X2 −1
  j1 −1
×  2 IIj1 (t1 ) 2j2 −1 IIj2 (t2 )
l1 =0 l2 =0 j1 =0 j2 =0
l1 2−m1 l2 2−m2
1/2
×S2m1 2m2 (x1 ∔ t1 ∔ ej1 , x2 ∔ t2 ∔ ej2 ; |f |) dt1 dt2 )2 .
Since
2m1 −1 2m2 −1   
X X  (1)
Pm m l1 l2 2
 1 2 ,
2m1 2m2 
l1 =0 l2 =0
(l1 +1)2
Z
−m1 (l +1)2−m2
2 Z
2m1 −1
X 2m2 −1
X  2
m1 +m2  (1) 
=2 Pm1 m2 (t1 , t2 ) dt1 dt2
l1 =0 l2 =0
l 2−m1 l 2−m2
Z  1 2
2
 (1) 
2m1 +m2 Pm 1 m2
(t 1 2  dt1 dt2 ≤ 2
, t ) m1 +m2

I2
we have
(17) R1 . Vm1 m2 (x1 , x2 ; |f |)
Analogously, from (9)-(16) we can prove that (s = 2, ..., 9)
(18) Rs . Vm1 m2 (x1 , x2 ; |f |) .
Combining (7), (17) and (18) we conclude the proof of Theorem 1. 
Lemma 2. The following estimation holds
 
Vm1 m2 (x1 , x2 ; |f |) . V (1) x1 , x2 ; V (2) (|f |) .

Proof of Lemma 2. There is a suitable vector

{al1 l2 (x1 , x2 ) : 0 ≤ l1 < 2m1 , 0 ≤ l2 < 2m2 }
such that
2m
X1 −1 2m2 −1
X
|al1 l2 (x1 , x2 )|2 = 1
l1 =0 l2 =0
and
Vm1 m2 (x1 , x2 ; |f |)
(l1 +1)2
Z
−m1
2m
X1 −1 m
X1 −1

= 2j1 −1 IIj1 (t1 )

l1 =0 j1 =0
l1 2−m1
 ALMOST EVERYWHERE SUMMABILITY 13

(l2 +1)2
Z
−m2
2m
X2 −1 m
X2 −1

× al1 l2 (x1 , x2 )  2j2 −1 IIj2 (t2 )
l2 =0 j2 =0
l2 2−m2
×S2m1 2m2 (x1 + t1 + ej1 , x2 + t2 + ej2 ; |f |) dt2 ) dt1
(l1 +1)2−m1 m 1/2 m
2m
X1 −1 Z m
X1 −1 2X
2 −1 2X2 −1
j1 −1  2
≤ 2 IIj1 (t1 ) |al1 l2 (x1 , x2 )|
l1 =0 j1 =0 l2 =0 l2 =0
l1 2−m1

(l2 +1)2
Z
−m2
m
X2 −1

 2j2 −1 IIj2 (t2 )
j2 =0
l2 2−m2
2 1/2
(2)
×S2m2 (x1 ∔ t1 ∔ ej1 , x2 ∔ t2 ∔ ej2 ; |f |) dt2 dt1
 1/2 (l1 +1)2−m1
2m
X1 −1 2m
X2 −1 Z m
X1 −1
2
≤  |al1 l2 (x1 , x2 )| 2j1 −1 IIj1 (t1 )
l1 =0 l2 =0 j1 =0
l1 2−m1

×V (2) (x1 + t1 + ej1 , x2 ; |f |) dt1

 1/2  
(l1 +1)2−m1
2m1 −1 2m2 −1
X X 2m1 −1
X Z m
X1 −1
2  
≤  |al1 l2 (x1 , x2 )|   2j1 −1 IIj1 (t1 )
l1 =0 l2 =0 l1 =0 j1 =0
l1 2−m1
2 1/2  
×V (2) (x1 ∔ t1 ∔ ej1 , x2 ; |f |) dt1 . V (1) x1 , x2 ; V (2) .
Lemma 2 is proved. 

4. Proof of Main Results

Proof of Theorem 1. Set
2n
X1 −1 2n2 −1
X
fn1 ,n2 (x1 , x2 , t1 , t2 ) := Sk1 ,k2 (x1 , x2 ; f ) Iδn1 (t1 ) Iδn2 (t2 ) ,
k1 k2
k1 =0 k2 =0



J1 := j1 2−m , (j1 + 1) 2−m , J2 := j2 2−m , (j2 + 1) 2−m .
Then we can write (n1 ≤ n2 )
(19) kfn1 ,n2 (x1 , x2 , ·, ·)kBM O

Z
 1
= sup sup |fn1 ,n2 (x1 , x2 , t1 , t2 )
m 0≤j1 ,j2 <2m |J1 × J2 |
J1 ×J2
2 1/2
Z 
1 
− fn1,n2 (x1 , x2 , u1 , u2 ) du1 du2  dt1 dt2 
|J1 × J2 | 
J1 ×J2
14 USHANGI GOGINAVA
!
≤ sup sup + sup sup + sup sup
m≤n1 0≤j1 ,j2 <2m n1 <m≤n2 0≤j1 ,j2 <2m m>n2 0≤j1 ,j2 <2m

Z
 1
|fn1 ,n2 (x1 , x2 , t1 , t2 )
|J1 × J2 |
J1 ×J2
2 1/2
Z 
1 
− fn1,n2 (x1 , x2 , u1 , u2 ) du1 du2  dt1 dt2 
|J1 × J2 | 
J1 ×J2
:= P1 (n1 , n2 ) + P2 (n1 , n2 ) + P3 (n1 , n2 ) .
h 
j1 j1 +1
et n1 ≤ n2 < m. Since fn1 ,n2 (x1 , x2 , t1 , t2 ) is constant on 2m , 2m ×
h 
j2 j2 +1
2m , 2m for fixed (x1 , x2 ) ∈ I2 we conclude that

(20) P3 (n1 , n2 ) = 0.
Let m ≤ n1 . Then for P1 we can write
P1 (n1 , n2 )
 
Z (j1 +1)2 n −m
X1 −1 (j2 +1)2
n −m
X2 −1

= sup sup 22m 

m≤n1 0≤j1 ,j2 <2m  k1 =j1 2n1 −m k2 =j2 2n2 −m
J1 ×J2

Sk1 ,k2 (x1 , x2 ; f ) Iδn1 (t1 ) Iδn2 (t2 ) − 22m

k1 k2

Z (j1 +1)2n1 −m −1 (j2 +1)2n2 −m −1

X X
× Sk1 ,k2 (x1 , x2 ; f )
J1 ×J2 k1 =j1 2n1 −m k2 =j2 2n2 −m

2 1/2

×Iδn1 (u1 ) Iδn2 (u2 ) du1 du2  dt1 dt2
k1 k2

(j1 +1)2n1 −m −1 (j2 +1)2n2 −m −1
X X
= sup sup 2m−n1 2m−n2
m≤n1 0≤j1 ,j2 <2m
k1 =j1 2n1 −m k2 =j2 2n2 −m

Sk ,k (x1 , x2 ; f ) − 2m−n1 2m−n2
1 2
2 1/2
(j1 +1)2n1 −m −1 (j2 +1)2n2 −m −1
X X 

× Sk1 ,k2 (x1 , x2 ; f ) 
k1 =j1 2n1 −m k2 =j2 2n2 −m 

2mX 1 −1 2m2 −1
X
= sup sup 2−m1 2−m2
m1 ≤n1 ,m2 ≤n2 0≤j1 <2m1 ,0≤j2 <2m2
l1 =0 l2 =0

Sl +j 2m1 ,l +j 2m2 (x1 , x2 ; f ) − 2−m1 2−m2
1 1 2 2
 ALMOST EVERYWHERE SUMMABILITY 15

2 1/2
2m
X1 −1 2m2 −1
X 

× Sq1+j1 2m1 ,q2 +j2 2m2 (x1 , x2 ; f )  .
q1 =0 q2 =0 
Since
Sl1 +j1 2m1 ,l2 +j2 2m2 (x1 , x2 ; f )
= Sj1 2m1 ,j2 2m2 (x1 , x2 ; f ) + Sl1 ,j22m2 (x1 , x2 ; f wj1 2m1 ) wj1 2m1 (x1 )
+Sj12m1 ,l2 (x1 , x2 ; f wj2 2m2 ) wj2 2m2 (x2 )
+Sl1 ,l2 (x1 , x2 ; f wj1 2m1 ⊗ wj2 2m2 ) wj1 2m1 (x1 ) wj2 2m2 (x2 )
we can write

2m
X1 −1 2m2 −1
X
2 −m1 −m2
(21) P1 (n1 , n2 ) ≤ sup sup
m1 ≤n1 ,m2 ≤n2 0≤j1 <2m1 ,0≤j2 <2m2
l1 =0 l2 =0

Sl1 ,l2 (x1 , x2 ; f wj12m1 ⊗ wj2 2m2 ) − 2−m1 −m2

2 1/2
2m
X 1 −1 2m2 −1
X 

−× Sq1 ,q2 (x1 , x2 ; f wj1 2m1 ⊗ wj2 2m2 ) 
q1 =0 q2 =0 

2m
X1 −1

2 −m
+ sup sup 1
|Sl1 ,j2 2m2 (x1 , x2 ; f wj1 2m1 )
m1 ≤n1 ,m2 ≤n2 0≤j1 <2m1 ,0≤j2 <2m2
l1 =0
2 1/2
2m1 −1 
X 
− 2−m1 Sq1,j2 2m2 (x1 , x2 ; f wj1 2m1 ) 
q1 =0 

2m
X2 −1

+ sup sup 2−m2 |Sj1 2m1 ,l2 (x1 , x2 ; f wj2 2m2 )

m1 ≤n1 ,m2 ≤n2 0≤j1 <2m1 ,0≤j2 <2m2
l2 =0
2 1/2
2m
X2 −1 

− 2−m1 Sj12 1 ,q2 (x1 , x2 ; f wj2 2 2 ) 
m m

q2 =0 
:= P11 (n1 , n2 ) + P12 (n1 , n2 ) + P13 (n1 , n2 ) .
From Lemmas 1 and 2 we obtain
P11 (n1 , n2 ) . sup sup Vm1 ,m2 (x1 , x2 ; |f wj12m1 ⊗ wj2 2m2 |)
m1 ≤n1 ,m2 ≤n2 0≤j1 <2m1 ,0≤j2 <2m2
 
. V (1) x1 , x2 ; V (2) (|f |) .
Consequently, by Theorem Sch 1 we can write
 
 
(22)  2
P11 (n1 , n2 ) > λ 
 (x1 , x2 ) ∈ I : nsup
,n 1 2
16 USHANGI GOGINAVA
 
Z Z
1 
. V (2) (x1 , x2 ; |f |) dx1  dx2
λ
I I
 
Z Z
1 
. |f (x1 , x2 )| log+ |f (x1 , x2 )| dx2 + 1 dx1
λ
I I
 
Z
1
. |f | log+ |f | + 1 .
λ
I2
Since
(1)
Sl1 ,j2 2m (x1 , x2 ; f wj12m ) = Sl1 (x1 , x2 ; Sj2 2m (f ) wj1 2m )
from Lemma 3 we have
P12 (n1 , n2 )
 1/2
2m 1 −1  2
X  (1) 
. sup sup 2−m1 Sl1 (x1 , x2 ; Sj2 2m2 (f ) wj1 2m1 ) 
m1 ≤n1 ,m2 ≤n2 0≤j1 <2m1 ,0≤j2 <2m2
l1 =0
  
 (2) 
≤ sup sup V (1) x1 , x2 ; Sj2 2m2 (f ) wj1 2m1 
m
m1 ≤n1 ,m2 ≤n2 0≤j1 <2 ,0≤j2 <2
1 m 2
 
(2)
≤ V (1) x1 , x2 ; S∗ (f ) ,
where  
(2)  
S∗ (f ) := sup Sn(2) (f ) .
n
+

2

2
If f ∈ L log L I2 . Then f (x1 , ·) ∈ L log+ L (I) for a. e. x1 ∈ I,
(2)
and from the well-known theorem (see [35]) ∈ L1 (I) for a. e.
S∗ (x1 , ·, f )
x1 ∈ I. Moreover
Z Z
(2)
2
S∗ (x1 , x2 ; f ) dx2 . |f (x1 , x2 )| log+ |f (x1 , x2 )| dx2 + 1
I I
for a. e. x1 ∈ I.
Hence,
 
 
(23)  (x1 , x2 ) ∈ I2 : sup P12 (n1 , n2 ) > λ 
 n1 ,n2

n   o
 (2) 
.  (x1 , x2 ) ∈ I2 : V (1) x1 , x2 ; S∗ (f ) > λ 
 
Z Z
1  (2)
. S∗ (x1 , x2 ; f ) dx1  dx2
λ
I I
 
Z Z
1 
2
. |f (x1 , x2 )| log+ |f (x1 , x2 )| dx2 + 1 dx1
λ
I I
 ALMOST EVERYWHERE SUMMABILITY 17
 
Z
1
2
. |f | log+ |f | + 1 .
λ
I2
Analogously, we can prove that
 
 
(24)  (x1 , x2 ) ∈ I2 : sup P13 (n1 , n2 ) > λ 
 n1 ,n2

 
Z
1
2
. |f | log+ |f | + 1 .
λ
I2
Combining (21)- (24) we get
 
 
(25)  2
P1 (n1 , n2 ) > λ 
 (x1 , x2 ) ∈ I : nsup
1 ,n2
 
Z
1
2
.  |f | log+ |f | + 1 .
λ
I2
Analogously, we can prove that
 
 
(26)  2
P2 (n1 , n2 ) > λ 
 (x1 , x2 ) ∈ I : nsup
1 ,n2
 
Z
1
2
.  |f | log+ |f | + 1 .
λ
I2
Combining (19), (20), (25) and (26) we complete the proof of Theorem
1. 

References
[1] Fejér L., Untersuchungen über Fouriersche Reihen, Math. Annalen, 58 (1904), 501–
569.
[2] Fridli S., Schipp F., Strong summability and Sidon type inequalities, Acta Sci. Math.
(Szeged) 60 (1995), no. 1-2, 277–289.
[3] Fridli S., Schipp F., Strong approximation via Sidon type inequalities, J. Approx.
Theory 94 (1998), 263–284.
[4] Gabisonia O. D., On strong summability points for Fourier series, Mat. Zametki. 5,
14 (1973), 615–626.
[5] Gát G., Goginava U., Karagulyan G., Almost everywhere strong summability of
Marcinkiewicz means of double Walsh-Fourier series. Anal. Math. 40 (2014), no. 4,
243–266.
[6] Gát G., Goginava U., Karagulyan G., On everywhere divergence of the strong Φ-
means of Walsh-Fourier series. J. Math. Anal. Appl. 421 (2015), no. 1, 206–214.
[7] Glukhov V. A., Summation of multiple Fourier series in multiplicative systems (Rus-
sian), Mat. Zametki 39 (1986), no. 5, 665–673.
[8] Goginava U., Almost everywhere convergence of (C, α)-means of cubical partial sums
of d-dimensional Walsh-Fourier series. J. Approx. Theory 141 (2006), no. 1, 8–28.
[9] Goginava U., The weak type inequality for the Walsh system. Studia Math. 185
(2008), no. 1, 35–48.
18 USHANGI GOGINAVA

[10] Goginava U., Gogoladze L., Strong approximation by Marcinkiewicz means of two-
dimensional Walsh-Fourier series, Constr. Approx. 35 (2012), no. 1, 1–19.
[11] Goginava U., Gogoladze L., Convergence in measure of strong logarithmic means of
double Fourier series. Izv. Nats. Akad. Nauk Armenii Mat. 49 (2014), no. 3, 39–49;
translation in J. Contemp. Math. Anal. 49 (2014), no. 3, 109–116
[12] Goginava U., Gogoladze L., Karagulyan G., BMO-estimation and almost everywhere
exponential summability of quadratic partial sums of double Fourier series. Constr.
Approx. 40 (2014), no. 1, 105–120.
[13] Gogoladze L., The (H, k)-summability of multiple trigonometric Fourier series. (Rus-
sian) Izv. Akad. Nauk SSSR Ser. Mat. 41 (1977), no. 4, 937–958.
[14] Gogoladze L., On the exponential uniform strong summability of multiple trigono-
metric Fourier series, Georgian Math. J. 16 (2009), 517–532.
[15] Gogoladze L. D., Strong means of Marcinkiewicz type (Russian), Soobshch. Akad.
Nauk Gruzin. SSR 102 (1981), no. 2, 293–295.
[16] Golubov B. I., Efimov A.V., Skvortsov V.A., Series and transformations of Walsh,
Moscow, 1987 (Russian); English translation, Kluwer Academic, Dordrecht, 1991.
[17] Hardy G. H., Littlewood J. E., Sur la series de Fourier d’une fonction a carre
sommable, Comptes Rendus (Paris) 156 (1913), 1307–1309.
[18] Karagulyan G. A., Everywhere divergent Φ-means of Fourier series (Russian), Mat.
Zametki 80 (2006), no. 1, 50–59; translation in Math. Notes 80 (2006), no. 1-2, 47–56.
[19] Lebesgue H., Recherches sur la sommabilite forte des series de Fourier, Math. Annalen
61 (1905), 251–280.
[20] Leindler L., Über die Approximation im starken Sinne, Acta Math. Acad. Hungar,
16 (1965), 255–262.
[21] Leindler L., On the strong approximation of Fourier series, Acta Sci. Math. (Szeged)
38 (1976), 317–324.
[22] Leindler L., Strong approximation and classes of functions, Mitteilungen Math. Sem-
inar Giessen, 132 (1978), 29–38.
[23] Leindler L., Strong approximation by Fourier series, Akadémiai Kiadó, Budapest,
1985.
[24] Marcinkiewicz J., Sur la sommabilité forte de séries de Fourier (French), J. London
Math. Soc. 14, (1939).162–168.
[25] Móricz, F., Schipp, F., Wade, W. R., Cesàro summability of double Walsh-Fourier
series. Trans. Amer. Math. Soc. 329 (1992), no. 1, 131–140.
[26] Oskolkov K. I., Strong summability of Fourier series. (Russian) Studies in the theory
of functions of several real variables and the approximation of functions, Trudy Mat.
Inst. Steklov. 172 (1985), 280–290, 355.
[27] Rodin, V. A., The space BMO and strong means of Fourier series, Anal. Math. 16
(1990), no. 4, 291–302.
[28] Rodin V. A., BMO-strong means of Fourier series, Funct. anal. Appl. 23 (1989),
73–74, (Russian)
[29] Rodin,The space BMO and strong means of Fourier-Walsh series. (Russian) Mat.
Sb. 182 (1991), no. 10, 1463–1478; translation in Math. USSR-Sb. 74 (1993), no. 1,
203–218
[30] Schipp F., On the strong summability of Walsh series, Publ. Math. Debrecen 52
(1998), no. 3-4, 611–633.
[31] Schipp F., Über die starke Summation von Walsh-Fourier Reihen, Acta Sci. Math.
(Szeged), 30 (1969), 77–87.
[32] Schipp F., On strong approximation of Walsh-Fourier series, MTA III. Oszt. Kozl.
19(1969), 101–111 (Hungarian).
[33] Schipp F., Ky N. X., On strong summability of polynomial expansions, Anal. Math.
12 (1986), 115–128.
 ALMOST EVERYWHERE SUMMABILITY 19

[34] Schipp F., Wade W., Simon P., Pál P., Walsh Series, an Introduction to Dyadic
Harmonic Analysis, Adam Hilger, Bristol, New York, 1990.
[35] Tateoka J., On almost everywhere convergence of Walsh-Fourier series. Proc. Japan
Acad. 44 1968 647–650.
[36] Totik V., On the strong approximation of Fourier series, Acta Math. Sci. Hungar 35
(1980), 151–172.
[37] Totik V., On the generalization of Fejér’s summation theorem, Functions, Series,
Operators; Coll. Math. Soc. J. Bolyai (Budapest) Hungary, 35, North Holland,
Amsterdam-Oxford-New-Yourk, 1980, 1195–1199.
[38] Totik V., Notes on Fourier series: Strong approximation, J. Approx. Theory, 43
(1985), 105–111.
[39] Wang, Kun Yang. Some estimates for the strong approximation of continuous periodic
functions of the two variables by their sums of Marcinkiewicz type (Chinese), Beijing
Shifan Daxue Xuebao 1981, no. 1, 7–22.
[40] Strong summability of more-dimensional Ciesielski-Fourier series. East J. Approx. 10
(2004), no. 3, 333–354.
[41] Weisz F., Strong Marcinkiewicz summability of multi-dimensional Fourier series, Ann.
Univ. Sci. Budapest. Sect. Comput. 29 (2008), 297–317.
[42] Weisz F., Lebesgue points of double Fourier series and strong summability. J. Math.
Anal. Appl. 432 (2015), no. 1, 441–462.
[43] Zhang Y., He X., On the uniform strong approximation of Marcinkiewicz type for
multivariable continuous functions, Anal. Theory Appl. 21 (2005), 377–384.
[44] Zygmund A., Trigonometric series. Cambridge University Press, Cambridge, 1959.

U. Goginava, Department of Mathematics, Faculty of Exact and Natural

Sciences, Ivane Javakhishvili Tbilisi State University, Chavchavadze str. 1,
Tbilisi 0128, Georgia
E-mail address: zazagoginava@gmail.com

View publication stats