http://jipam.vu.edu.au/
Volume 7, Issue 5, Article 178, 2006
SOME HARDY TYPE INTEGRAL INEQUALITIES
MEHMET ZEKI SARIKAYA AND HÜSEYIN YILDIRIM DEPARTMENT OFMATHEMATICS
FACULTY OFSCIENCE ANDARTS
KOCATEPEUNIVERSITY
AFYON-TURKEY sarikaya@aku.edu.tr
hyildir@aku.edu.tr
Received 20 March, 2006; accepted 20 November, 2006 Communicated by L. Pick
ABSTRACT. In this note, we obtain some new generalizations of the Hardy’s integral inequality by using a fairly elementary analysis. These inequalities generalize some known results and simplify the proofs of some existing results.
Key words and phrases: Hardy integral inequalities and Hölder’s inequality.
2000 Mathematics Subject Classification. 26D10 and 26D15.
1. INTRODUCTION
In [4], Hardy proved the following inequality. Ifp > 1, f ≥0and F(x) =
Z x
0
f(t)dt, then
(1.1)
Z ∞
0
F x
p
dx < qp Z ∞
0
fp(x)dx
unless f ≡ 0. The constant q = p(p−1)−1 is the best possible. This inequality plays an important role in analysis and its applications. It is obvious that, for parametersa andb such that0< a < b <∞,the following inequality is also valid
(1.2)
Z b
a
F x
p
dx < qp Z b
a
fp(x)dx,
where0 < R∞
0 fp(x)dx < ∞.The classical Hardy inequality asserts that if p > 1andf is a nonnegative measurable function on(a, b), then (1.2) is true unlessf ≡ 0a.e. in(a, b),where the constant here is best possible. This inequality remains true provided that0< a < b <∞.
ISSN (electronic): 1443-5756
c 2006 Victoria University. All rights reserved.
085-06
In particular, Hardy [3] in 1928 gave a generalized form of the inequality (1.1) when he showed that for anym6= 1, p > 1and any integrable functionf(x)≥0on(0,∞)for which
F(x) = ( Rx
0 f(t)dt for m >1, R∞
x f(t)dt for m <1, then
(1.3)
Z ∞
0
x−mFp(x)dx <
p
|m−1|
pZ ∞
0
x−m[xf(x)]pdx
unlessf ≡0,where the constant is also best possible.
Because of their fundamental importance in the discipline over the years much effort and time has been devoted to the improvement and generalization of Hardy’s inequalities (1.1), (1.2) and (1.3). These include, among others, the works in [1] – [9].
The objective of this paper is to obtain further generalizations of the classical Hardy integral inequality which will be useful in applications by using some elementary methods of analysis.
Throughout this paper, the left-hand sides of the inequalities exist if the right-hand sides exist.
2. MAINRESULTS
The following theorems are the main results of the present paper.
Theorem 2.1. Let p > 1, m > 1 be constants. Let f(x) be a nonnegative and integrable function on(0,∞)and letz(x)be differentiable in(0,∞)withz0(x) >0andz(0+) > 0.Let w(x)andr(x)be positive and absolutely continuous functions on(0,∞).Let
1− 1
m−1 z(x) z0(x)
w0(x)
w(x) + p m−1
z(x) z0(x)
r0(x) r(x) ≥ 1
λ >0.
Leta ∈(0,∞)be fixed and set
F(x) := 1 r(x)
Z x
a
r(t)z0(t)f(t)
z(t) dt, x∈(0,∞).
Then the inequality (2.1)
Z b
a
w(x) z0(x)
zm(x)Fp(x)dx≤
λp m−1
pZ b
a
w(x)z0(x)
zm(x)fp(x)dx
holds for allb ≥a.
Proof. Integrating the left-hand side of inequality (2.1) by parts gives
Z b
a
w(x)z0(x)
zm(x)Fp(x)dx =w(b)[z(b)]−m+1
−m+ 1 Fp(b) + 1 m−1
Z b
a
z−m+1w0Fpdx
− p m−1
Z b
a
z−m+1wr0(x)
r(x)Fpdx+ p m−1
Z b
a
w z0(x)
zm(x)Fp−1f dx.
Sincem >1andF(b)≥0,we have Z b
a
w(x) z0(x)
zm(x)Fp(x)
1− 1
m−1 z(x) z0(x)
w0(x)
w(x) + p m−1
z(x) z0(x)
r0(x) r(x)
dx
=w(b)[z(b)]−m+1
−m+ 1 Fp(b) + p m−1
Z b
a
wz0(x)
zm(x)Fp−1f dx
≤ p m−1
Z b
a
wz0(x)
zm(x)Fp−1f dx.
Here, using the assumption onλ,we have Z b
a
w(x) z0(x)
zm(x)Fp(x)1 λdx
≤ Z b
a
w(x)z0(x)
zm(x)Fp(x)
1− 1
m−1 z(x) z0(x)
w0(x)
w(x) + p m−1
z(x) z0(x)
r0(x) r(x)
dx
≤ p m−1
Z b
a
wz0(x)
zm(x)Fp−1f dx.
By Hölder’s inequality, Z b
a
w(x) z0(x)
zm(x)Fp(x)dx≤ λp m−1
Z b
a
w z0(x) zm(x)Fpdx
1
q Z b
a
wz0(x) zm(x)fpdx
1 p
and thus on simplification, we have Z b
a
w(x) z0(x)
zm(x)Fp(x)dx≤
λp m−1
pZ b
a
w z0(x) zm(x)fpdx.
This proves the theorem.
Theorem 2.2. Letp, m, f, z, z0, wandrbe as in Theorem 2.1. Let
1− 1
m−1 z(x) z0(x)
w0(x)
w(x) − p m−1
z(x) z0(x)
r0(x) r(x) ≥ 1
λ >0.
Leta ∈(0,∞)be fixed and set
F(x) :=r(x) Z x
a
z0(t)f(t)
z(t)r(t)dt, x∈(0,∞).
Then the inequality (2.2)
Z b
a
w(x) z0(x)
zm(x)Fp(x)dx≤
λp m−1
pZ b
a
w(x)z0(x)
zm(x)fp(x)dx holds for allb ≥a.
Proof. This is similar to the proof of the Theorem 2.1.
Theorem 2.3. Letp, m, f, z, z0, wandrbe as in Theorem 2.1. Let
1− 1
m−1 z(x) z0(x)
w0(x)
w(x) + p m−1
z(x) z0(x)
r0(x) r(x) ≥ 1
λ >0.
Leta ∈(0,∞)be fixed and set
F(x) := 1 r(x)
Z x
x 2
r(t)z0(t)f(t)
z(t) dt, x∈(0,∞).
Then the inequality (2.3)
Z b
0
w(x)z0(x)
zm(x)Fp(x)dx≤
λp m−1
pZ b
0
w(x)[z(x)]p−m
[z0(x)]pq |g(x)|pdx holds where
g(x) = 1 r(x)
r(x)z0(x)f(x) z(x) − 1
2
r(x2)z0(x2)f(x2) z(x2)
.
Proof. Upon integrating by parts we have Z b
0
w(x)z0(x)
zm(x)Fp(x)dx =w(b)[z(b)]−m+1
−m+ 1 Fp(b) + 1 m−1
Z b
0
z−m+1w0Fpdx
− p m−1
Z b
0
z−m+1wr0(x)
r(x)Fpdx+ p m−1
Z b
0
z−m+1wFp−1|g(x)|dx, where
g(x) = 1 r(x)
r(x)z0(x)f(x) z(x) − 1
2
r(x2)z0(x2)f(x2) z(x2)
. Sincem >1andF(b)≥0we have
Z b
0
w(x) z0(x)
zm(x)Fp(x)
1− 1
m−1 z(x) z0(x)
w0(x)
w(x) + p m−1
z(x) z0(x)
r0(x) r(x)
dx
=w(b)[z(b)]−m+1
−m+ 1 Fp(b) + p m−1
Z b
0
wz0(x)
zm(x)Fp−1f dx
≤ p m−1
Z b
0
z−m+1wFp−1|g(x)|dx.
By the assumption onλ,we have Z b
0
w(x) z0(x)
zm(x)Fp(x)1 λdx
≤ Z b
0
w(x)z0(x)
zm(x)Fp(x)
1− 1
m−1 z(x) z0(x)
w0(x)
w(x) + p m−1
z(x) z0(x)
r0(x) r(x)
dx
≤ p m−1
Z b
0
z(x)
z0(x)w z0(x)
zm(x)Fp−1|g(x)|dx.
By Hölder’s inequality Z b
0
w(x) z0(x)
zm(x)Fp(x)dx≤ λp m−1
Z b
0
w z0(x) zm(x)Fpdx
1q Z b
0
wzp−m(x)
[z0(x)]pq |g(x)|pdx
!1p
and thus on simplification, we have Z b
0
w(x) z0(x)
zm(x)Fp(x)dx≤
λp m−1
pZ b
0
wzp−m(x) [z0(x)]pq
|g(x)|pdx.
This proves the theorem.
Theorem 2.4. Letp, m, f, z, z0, wandrbe as in Theorem 2.1. Let
1− 1
m−1 z(x) z0(x)
w0(x)
w(x) + p m−1
z(x) z0(x)
r0(x) r(x) ≥ 1
λ >0.
Leta ∈(0,∞)be fixed and set
F(x) :=r(x) Z x
x 2
z0(t)f(t)
z(t)r(t)dt, x∈(0,∞).
Then the inequality (2.4)
Z b
0
w(x)z0(x)
zm(x)Fp(x)dx≤
λp m−1
pZ b
0
w(x)[z(x)]p−m
[z0(x)]pq |g(x)|pdx holds where
g(x) = r(x)
r(x)z0(x)f(x) z(x) −1
2
r(x2)z0(x2)f(x2) z(x2)
.
Proof. This is similar to the proof of the Theorem 2.3.
Theorem 2.5. Let p > q > 0, 1p + 1q = 1, and m < 1 be real numbers. Let z(x) be differentiable in (0,∞) with z0(x) > 0 and z(0+) > 0, let w(x) and r(x) be positive and absolutely continuous functions on(0,∞), nand letf : [0,∞)→[0,∞)be integrable so that
1 + 1 m−1
z(x) z0(x)
w0(x) w(x) − p
q 1 m−1
z(x) z0(x)
r0(x) r(x) ≥ 1
λ >0 a.e. for someλ >0.Letb ∈(0,∞)be fixed and set
F(x) := 1 r(x)
Z b
x
r(t)z0(t)f(t)
z(t) dt, x∈(0,∞).
Then the following inequality (2.5)
Z b
a
w(x) z0(x)
zm(x)Fpq (x)dx≤
λp m−1
pZ b
a
w(x)z0(x)
zm(x)fpq(x)dx holds for all0≤a≤b.
Proof. This is similar to the proof of the Theorem 2.1.
REFERENCES
[1] W.S. CHEUNG, Z. HANJSANDJ. PE ˇCARI ´C, Some Hardy-type inequalities, J. Math. Anal. Appl., 250(2) (2000), 621–634.
[2] G.H. HARDY, J.E. LITTLEWOOD AND G. PÓLYA, Inequalities, Cambridge University Press, (1952).
[3] G.H. HARDY, Notes on some points in the integral calculus, Messenger Math., 57 (1928), 12–16.
[4] G.H. HARDY, Notes on a theorem of Hilbert, Math. Z., 6 (1920), 314–317.
[5] B.G. PACHPATTE, On some variants of Hardy’s inequality, J. Math. Anal. Appl., 124 (1987), 495–
501.
[6] B.G. PACHPATTE, On some generalizations of Hardy’s integral inequality, J. Math. Anal. Appl., 234(1) (1999), 15–30.
[7] B.G. PACHPATTE, On a new class of Hardy type inequalities, Proc. R. Soc. Edin., 105A (1987), 265–274.
[8] B. YANG, Z. ZENGANDL. DEBNATH, On New Generalizations of Hardy’s Integral Inequality, J.
Math. Anal. Appl., 217(1) (1998), 321–327.
[9] J.E. PE ˇCARI ´CANDE.R. LOVE, Still more generalizations of Hardy’s inequality, J. Austral. Math.
Soc. Ser. A, 58 (1995), 1–11.