Volume 6, Number 1 (2015), 76 – 84
BOCHKAREV INEQUALITY FOR THE FOURIER TRANSFORM OF FUNCTIONS IN THE LORENTZ SPACES L2,r(R)
G.K. Mussabayeva, N.T. Tleukhanova Communicated by N.A. Bokayev Key words: Lorentz spaces, Fourier transform.
AMS Mathematics Subject Classification: 42B10.
Abstract. In this article we prove an analogue of Hardy -Littlewood- Stein inequality for the Fourier transform in Lorentz space L2,r(R).
1 Introduction
Let f(x) be a scalar-valued µ measurable function which is finite almost everywhere, we introduce the distribution function m(σ, f) defined by
m(σ, f) = µ({x:|f(x)|> σ}) and f∗ its non-increasing rearrangement of the function f.
f∗(t) = inf{σ :m(σ, f)6t}. (1.1) In [1], the Lorentz space Lp,q is defined that f ∈Lp,q,16p6∞,if and only if
kfkLp,q =
Z
R
t1pf∗(t)q dt t
1 q
<∞
when 16q <∞,
kfkLp,∞ = sup
t
t1pf∗(t) when q =∞.
Well-known classical theorem Hausdorff-Young [11], whereby iff(x)is a 1- periodic function and satisfying
1
Z
0
|f(x)|pdx <∞, 1< p 62, then following relation holds
+∞
X
n=−∞
|cn|p0
!p10
6
1
Z
0
|f(x)|pdx
1 p
, (1.2)
where
cn=
1
Z
0
f(x)e−2πinxdx, n ∈Z
are the Fourier coefficients of f(x) and p0 = p−1p .
An analogue of theorem Hausdorff-Young for the Fourier transform was proved by Titchmarsh [9],[10] and the inequality of Titchmarsh for Fourier integrals
∞
Z
−∞
fb(ξ)
p0
dξ
1 p0
6
∞
Z
−∞
|f(x)|pdx
1 p
, (1.3)
where
fb(ξ) =
∞
Z
−∞
f(x)e−2πixξdx
is Fourier transform of f(x).
The Hardy-Littlewood inequalities for the Lorentz spaces Lp,q(R) was proved by Stein [3]-[8]:
Let1< p <2, 16q 6∞and p0 = p−1p , then
∞
X
k=1
kp10c∗kq 1 k
!1q
6CkfkLp,q[0,1], (1.4)
where c∗k is a non-increasing rearrangement of the Fourier coefficients.
kfbkLp0,q(R) 6CkfkLp,q(R), (1.5) Unlike inequalities (1.2) and (1.3), relations (1.4) and (1.5) don’t cover the case ofp= 2.
S.V.Bochkarev in [2] for the Fourier transform on the torus has shown that the Hardy -Littlewood- Stein inequality is not the case.
Theorem A. Let {ϕn}∞n=1 be an orthonormal system of complex-valued functions on [0,1], such that
kϕnk∞6M, n= 1,2, ....
let f ∈L2,r[0,1], 2< r6∞. Then the inequality
sup
n∈N
1
|n|12(log(n+ 1))12−1r
n
X
m=1
a∗m 6CkfkL2,r[0,1] (1.6) holds, where an are the Fourier coefficients of the system {ϕn}∞n=1.
2 Main result
The aim of this paper is to obtain an analogue of Bochkarev theorem for the Fourier transform. The main result of this paper is the following.
Theorem 2.1. Let <N = {A =
N
S
i=1
Ai, where Ai are segments in R}. Then for any function f ∈L2,r(R) and 2< r <∞ we have the inequality
sup
N>8
sup
A⊂<N
1
|A|12 log2(1 +N)12−1r Z
A
fˆ(ξ)dξ
623kfkL2,r. (2.1)
To prove Theorem 2.1, we establish the following lemmas.
Lemma 2.1. Let 43 < q < 2 < r < 4 and f ∈ Lq,r(R). Then for any measurable set A of finite measure of <N, the inequality
1
|A|1q Z
A
f(ξ)dξˆ
62 q
2−q
q10(r−2r )
kfkLq,r(R) (2.2)
holds.
Proof.
Z
A
f(ξ)dξˆ
6 Z
A
∞
Z
−∞
f(t)e−iξtdtdξ
6|A|kfkL1
and from the Plancherel’s theorem, we have
Z
A
fˆ(ξ)dξ
6|A|12kfkL2.
Let τ ∈(0,+∞). Define
f1(x) =
f(x), if |f(x)|6f∗(τ), 0, otherwise,
f0(x) =f(x)−f1(x).
Then, for any0< τ <∞, we have
Z
A
f(ξ)dξˆ
6 Z
A
fˆ0(ξ)dξ
+ Z
A
fˆ1(ξ)dξ
6|A|
τ
Z
0
f∗(s)ds+|A|12
∞
Z
τ
(f∗(s))2ds
1 2
.
Using Holder’s inequality, we obtain
Z
A
fˆ(ξ)dξ
6
|A|
τ
Z
0
t1qf∗(t) r dt
t
1
r
τ
Z
0
tr
0 q0dt
t
1 r0
+|A|12
∞
Z
τ
t1qf∗(t)r dt t
1
r
∞
Z
τ
t1−2qr−2r dt t
r−2 2r
6kfkLq,r |A|τ1−1q
1−1 q
1r−1
+|A|12τ12−1q
r(2−q) q(r−2)
1r−12! .
Now let τ = q(r−2)
r(2−q)
r−2r
q q−1
2(1−r)r
/|A| and using the 43 < q < 2 < r < 4, we can estimate
Z
A
f(ξ)dξˆ
62|A|1q q
2−q
q10(r−2r )
kfkLq,r.
The method of the proof in this paper is based on a study of the constants in inequalities of the form (2.2), namely their dependence on the relevant parameters.
Lemma 2.2. Let 2< p < 4, p0 = p−1p , 2< r <∞, f ∈Lp,r and A=
N
S
i=1
Ai, where Ai are segments in R. Then we have
1
|A|1pNp10−1p Z
A
f(ξ)dξˆ
6CkfkLp,r,
where
C = 4√ 2
r(p−2) p(r−2)
1p(2−rr )
. Proof. Let f ∈C0∞.
Z
A
fˆ(ξ)dξ
= Z
A
∞
Z
−∞
f(t)e−iξtdtdξ
6
N
X
i=1
Z
Ai
∞
Z
−∞
f(t)e−iξtdtdξ
6
N
X
i=1
∞
Z
−∞
|f(t)|
Z
Ai
e−iξtdξ
dt=
N
X
i=1
∞
Z
−∞
|f(t)|
bi
Z
ai
e−iξtdξ
dt=
N
X
i=1
∞
Z
−∞
|f(t)|
2 sin(bi−a2 i)t t
dt
6
N
X
i=1
∞
Z
−∞
|f(t)|min
(bi−ai),2 t
dt62
N
X
i=1
∞
Z
−∞
|f(t)|min
|Ai|,1 t
dt. (2.3)
From the Plancherel’s theorem we get Z
A
fˆ(ξ)dξ
6|A|12kfkL2. (2.4)
Consider the function f1(x) =
f(x), |f(x)|6f∗(τ), x∈R\
−τ2,τ2 0, otherwise,
f0(x) =f(x)−f1(x).
Then using the (2.3) and (2.4) we get
Z
A
f(ξ)dξˆ
6 Z
A
fˆ0(ξ)dξ
+ Z
A
fˆ1(ξ)dξ
6|A|12kf0kL2 + 2
N
X
i=1
∞
Z
−∞
|f1(x)|min
|Ai|, 1
|x|
dx
=B(I1+I2) We have
I1 =
∞
Z
−∞
|f0(x)|2dx
1 2
=
τ 2
Z
−τ
2
|f(x)|2dx+ Z
τ 26|x|<∞
|f0(x)|2dx
1 2
62
τ
Z
0
|f∗(t)|2dt
1 2
I2 = 2
N
X
i=1
∞
Z
−∞
|f1(x)|min
|Ai|, 1
|x|
= 2
N
X
i=1
Z
τ 26|x|<∞
|f1(x)|min
|Ai|, 1
|x|
62N
τ
Z
0
f1∗(t) 1 t+τdt=
= 2N
∞
Z
0
f∗(t+τ) dt
t+τ 62N
∞
Z
τ
f∗(t)dt t . Therefore,
Z
A
fˆ(ξ)dξ
62|A|12
τ
Z
0
(f∗(t))2dt
1 2
+ 2N
∞
Z
τ
f∗(t)dt t .
Applying H´older’s inequality, we obtain
Z
A
f(ξ)dξˆ
62|A|12
τ
Z
0
(f∗(t)t1p)rdt t
1
r
τ
Z
0
t1−2pr−2r dt t
r−2 2r
+2N
∞
Z
τ
f∗(t)tp1 r dt
t
1
r
∞
Z
τ
t−r
0 p dt
t
1 r0
6kfkLp,r 2|A|12
r(p−2) p(r−2)
2−r2r
τ12−1p + 2N pr−1r τ−1p
! .
Now let τ = N2
|A|(p−2p )2−rr p2(1−rr ) and using that 2 < p < 4 we have the following estimate:
Z
A
f(ξ)dξˆ
64√
2|A|1/p
p−2 p
1p(2r−1)
Np10−1pkfkLp,r.
Lemma 2.3. Let A =
N
S
i=1
Ai, where Ai are segments. Then for any 2 < p < 4 and 2< r <∞ the inequality
1
|A|12 Z
A
f(ξ)dξˆ
6CkfkL2,r (2.5)
holds, where
C= 8 p
p−2
1p(1−2r) p+ 2
p 14
N12−1p. Proof. Let τ >0,
f1(x) =
f(x), |f(x)|6f∗(τ) 0, otherwise, f0(x) =f(x)−f1(x).
Then, using Lemmas 2.1, 2.2 and 2< r <∞, we obtain
Z
A
fˆ(ξ)dξ
6 Z
A
fˆ0(ξ)dξ
+ Z
A
fˆ1(ξ)dξ
62kf0kLp0,r
p0 2−p0
1p(r−2r )
|A|p10
+4√
2kf1kLp,r
p−2 p
1p(2−rr )
|A|1pNp10−p1
62
τ
Z
0
tp10f∗(t)r dt t
1 r
|A|p10 p0
2−p0
1p(r−2r )
+4√ 2
∞
Z
0
t1pf∗(t+τ)rdt t
1 r
|A|1p
p−2 p
p1(2−rr )
Np10−1p. Further, we estimate each integral separately
τ
Z
0
tp10f∗(t)rdt t
1 r
=
τ
Z
0
t1−1pf∗(t)r dt t
1 r
6
τ
Z
0
t12f∗(t)rdt t
1 r
sup
06t6τ
t12−1p 6τ(12−1p)kfkL2,r.
∞
Z
0
tp1f∗(t+τ) r dt
t
1 r
6
τ
Z
0
t1p(t+τ)−12rdt t
sup
0<s<∞
s12f∗(s) r
+
∞
Z
τ
(t1pf∗(t+τ))rdt t
1 r
6
τ
Z
0
t1pτ−12rdt t sup
0<s<∞
r 2
s
Z
0
ξr2−1f∗(ξ)rdξ+
∞
Z
τ
t1pf∗(t)r dt t
1 r
6
τ(1p−12)rp
2kfkrL2,r +τ(1p−12)r
∞
Z
τ
t12f∗(t)r dt t
1 r
6τ(1p−12) p+ 2
2 1r
kfkL2,r.
Therefore given the 2< r <∞, we have
Z
A
fˆ(ξ)dξ
62|A|1−1p p
p−2
1p(1−2r)
τ(12−1p)kfkL2,r
+4√ 2|A|1p
p−2 p
1p(2−rr ) p+ 2
2 12
Np10−1pτ(1p−12)kfkL2,r.
Taking τ(12−1p) = N
|A|
p10−1
p p+2
2
12 12
, we have
Z
A
f(ξ)dξˆ
68|A|12N12−1p p
p−2
1p(1−2r) p+ 2
2 14
kfkL2,r.
The proof of Theorem 2.1. Let A⊂ <N and |A|=N 68.
In the right part of the constant (2.5) we put p= 3. Then
8 p
p−2
1p(1−2r) p+ 2
2 14
N12−1p 620.
Therefore, from Lemma 2.3, we get 1
N12 log2(1 +N)12−1r Z
A
f(ξ)dξˆ
6 1 N12
Z
A
fˆ(ξ)dξ
620kfkL2,r.
Now let A⊂ <N so that N >8. Putp= log2 log2N
2N−2, i.e. p−22p = log1
2N, then p66.
Note also thatN12−1p =Nlog21N 62.
Using Lemma 2.3, we obtain 1
N12 Z
A
fˆ(ξ)dξ
623 log2(1 +N)
1 2− 1
log2N
(1−2r)kfkL2,r
623 log2(1 +N)12−1rkfkL2,r. Taking the top exact bound over all N of N we obtain
sup
N>8
sup
A⊂<N
1
N12 log2(1 +N)12−1r Z
A
fˆ(ξ)dξ
623kfkL2,r
Theorem is proved.
Acknowledgments
This work was supported by the grant of the Ministry of Education and Science of the Republic of Kazakhstan (project 3311/GF4)
References
[1] J. Bergh, J. L´ofstr´om,Interpolation Spaces, New York. 1976, 210p.
[2] S.V. Bochkarev,Hausdorff-Young-Riesz theorem in Lorentz spaces and multiplicative inequalities, Proc. Steclov Inst. Math. 219 (1997), 96-107.
[3] A.N. Kopezhanova, E.D. Nursultanov, L.-E. Persson,On inequalities for the Fourier transform of functions from Lorentz spaces, Mathematical notes. 90 (2011), no. 5, 785-788.
[4] E.D. Nursultanov,Net spaces and the Fourier transform, (Russian) Dokl. Akad. Nauk 361 (1998), no 5, 597–599.
[5] E.D Nursultanov, T.U. Aubakirov, Hardy-Littlewood theorem for Fourier series, Mathematical notes. 73 (2003), 340-347.
[6] E.D. Nursultanov,Interpolation properties of some anisotropic spaces and Hardy–Littlewood type inequalities,East J. Approx. 4 (1998), no. 2, 277-290.
[7] E.D. Nursultanov,Application of interpolational methods to the study of properties of functions of several variables, Mathematical Notes. 75 (2004), no.3, 341-351.
[8] M. Stein Elias,Interpolation of linear operators, Trans. Amer. Math. Soc. 83 (1956), 482-492.
[9] E.C. Titchmarsh,A contribution to the theory of Fourier transforms, Proc. London Math. Soc.
23 (1924), no. 2, 279-289.
[10] E.C. Titchmarsh,Introduction to the theory of Fourier integrals, Clarendon Press, Oxford 1937;
Russian transl. Gostekhizdat, Moscow-Leningrad 1948.
[11] A. Zygmund,Trigonometrical series, 2nd ed. Vols. I, II, Cambridge Univ. Press, New York 1959;
Russian transl. Mir, Moscow 1965.
Guliya Mussabayeva
L.N. Gumilyov Eurasian National University 2 Satpayev st,
010010 Astana, Kazakhstan E-mail: [email protected] Nazerke Tleukhanova
L.N. Gumilyov Eurasian National University 2 Satpayev st,
010010 Astana, Kazakhstan E-mail: [email protected]
Received: 03.02.2015