Professional Documents
Culture Documents
https://doi.org/10.1186/s13660-019-2048-0
1 Introduction
Let aij > 0, 1 ≤ i ≤ m, 1 ≤ j ≤ n, pj > 1, and nj=1 p–1
j = 1. The following Hölder inequality
is well known:
m p1
m
n
n pj
j
aij ≤ aij . (1)
i=1 j=1 j=1 i=1
b
n
n
b 1/pj
pj
fj (x) dx ≤ fj (x) dx .
a j=1 j=1 a
2 n
n
n
xi yi ≤ x2i y2i .
i=1 i=1 i=1
© The Author(s) 2019. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License
(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, pro-
vided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and
indicate if changes were made.
Yan and Gao Journal of Inequalities and Applications (2019) 2019:97 Page 2 of 12
Yang’s [13, 14] insights into inequalities have further led to several inferences. Qi’s [15,
16] several integral inequalities explain the integral form of the inequality and further
elaboration of the integral form of the Hölder inequality. Tian [17–20], Tian and Ha [21,
22], Tian, Ha, and Wang [23], Tian and Pedrycz [24], Kwon and Bae [25], Tian, Zhu and
Cheung [26], and Zhao and Cheung [27] give a series of meaningful improvements, gener-
alization, properties, and applications of Hölder’s inequality. The papers in the references
are of great guiding significance to the conception and promotion of this paper.
From the paper by Yang [14] we can get the following conclusions.
Let the function h : (–∞, +∞) → (0, +∞) be defined as
m n 1–t
1/pk
n pk t
h(t) = aij aik . (2)
k=1 i=1 j=1
m p1
m
n
n pj
j
h(0) = aij ≤ aij = h(1); (3)
i=1 j=1 j=1 i=1
n n m pj 1/pj
it is not difficult to see h(0) = m j=1 aij , h(1) = j=1 ( i=1 aij ) under the conditions
n i=1
pk > 1, k = 1, 2, . . . , n, and k=1 1/pk = 1. From (2) it follows that
⎧
⎨≥ 0, t ≥ 0,
h (t)
⎩≤ 0, t ≤ 0,
p p
and the equation holds if and only if t = 0 or aikk / nj=1 aij = ajkk / nj=1 aij for 1 ≤ i, j ≤ m,
k = 1, 2, . . . , n.
From (2) and (3), it is natural to consider the function h(t) more deeply. Similarly, we
can define the function g(t) related to the integral form of Hölder’s inequality:
n 1–t
1/pk
n b
p t
g(t) = fj (x) fk k (x) dx , t ∈ R, (4)
k=1 a j=1
b
n
n
b 1/pk
p
g(0) = fk (x) dx ≤ fk k (x) dx = g(1). (5)
a k=1 k=1 a
As can be seen from the paper of Yang [14], (2) and (4) are generalized to the case pj >
1, nk=1 1/pk = r. We will show that hr (t) and gr (t) are concave functions on R and that
mint∈R hr (t) = hr (0) and mint∈R gr (t) = gr (0); moreover, hr (t)t ≥ 0 and gr (t)t ≥ 0 for t ∈ R.
Therefore, we obtain refinements of (3) and (5):
where 0 ≤ hr (τ ) ≤ hr (1) and 0 ≤ gr (s) ≤ gr (1). We also obtain conditions at which the
equalities hold.
2 Main results
We begin this section with three lemmas, which will be used in the sequel.
From Yang [13] we get the following lemma.
⎧
t ⎨≥ 0, t ≥ 0,
(ln A – ln B) A – Bt
⎩≤ 0, t ≤ 0,
1
Lemma 2.2 ([14]) Let rpk > 1, nk=1 p1 = r, aij > 0, 1 ≤ i ≤ m, 1 ≤ j ≤ n, bi = ( nj=1 aij ) r ,
rp
k
dik = aik k / nj=1 aij , 1 ≤ i ≤ m, 1 ≤ k ≤ n. Then
m
n
1
bi ln dik = 0.
pk i=1
k=1
n
n rpk
1 1 a
ln dik = ln ikr
pk pk bi
k=1 k=1
n n
r
= r ln aik – ln bi
pk
k=1 k=1
n
aik
= r ln
bri
k=1
= 0.
Therefore
m m n
n
1 r 1
r
b ln dik = bi ln dik = 0.
pk i=1 i i=1
pk
k=1 k=1
n 1 n 1 rp
Lemma 2.3 ([14]) Let rpk > 1, k=1 pk = r, fk (x) > 0, F(x) = ( j=1 fj (x))
r , gk (x) = fk k /F r (x),
x ∈ [a, b], k = 1, 2, . . . , n. Then
n
p–1
k ln gk (x) = 0, x ∈ [a, b].
k=1
Yan and Gao Journal of Inequalities and Applications (2019) 2019:97 Page 4 of 12
n
n
1
rp
f k (x)
p–1
k ln gk (x) = ln k r
pk F (x)
k=1 k=1
n
1
= rpk ln fk (x) – r ln F(x)
pk
k=1
n
n
r
= ln fk (x) – ln F(x)
pk
k=1 k=1
n
= r ln fk (x) – r2 ln F(x)
k=1
n
k=1 fk (x)
= r ln
F r (x)
= 0.
Theorem 2.4 Let X = (X, μ) and (Y , ν) be positive measure spaces with μ(X) = 1 and 0 <
ν(Y ) < ∞. Let f (x, y) be a real-valued bounded measurable function on X × Y . Define the
function hr (t) : R → R+ by
hr (t) = exp ln G(y)etH(x,y) dν(y) dμ(x) , (6)
X Y
where G(y) = e X f (x,y) dμ(x) and H(x, y) = rf (x, y) – X f (x, y) dμ(x).
Then the discrete form of hr (t) can be further pushed out:
n m 1–t
p1
n rpk t k
hr (t) = Xij Xik , (7)
k=1 i=1 j=1
n
where rpk > 1, 1
k=1 pk = r, Xij > 0 (1 ≤ i ≤ m, 1 ≤ j ≤ n), and Xij are measurable.
n
1
μ= δj ,
p
j=1 j
m
ν= δi ,
i=1
Yan and Gao Journal of Inequalities and Applications (2019) 2019:97 Page 5 of 12
where δj and δi denote the unit mass concentrated at j and i, respectively. Then (X, 2X , μ)
and (Y , 2Y , ν) are positive measure spaces with μ(X) = 1 and 0 < ν(Y ) = m < ∞. Here 2{·}
means the set of all subsets of {·}. Let
pj
ef (j,i) = Xij .
Then
n
1 pj
exp f (x, y) dμ(x) = exp ln Xij
X p
j=1 j
n
= exp ln Xij
j=1
n
n
= exp ln Xij = Xij .
j=1 j=1
tH(x,y)
hr (t) = exp ln G(y)e dν(y) dμ(x)
X Y
= exp ln e X f (x,y) dμ(x) ertf (x,y)–t X f (x,y) dμ(x) dν(y) dμ(x)
X Y
f (x,y) dμ(x) 1–t f (x,y) rt
= exp ln eX e dν(y) dμ(x)
X Y
1–t
n
rp t
= exp ln Xij Xik k dν(y) dμ(x)
X Y j=1
n 1–t
m rp t
= exp ln Xij Xik k dμ(x)
X i=1 j=1
m n 1–t
n
1 rpk t
= exp ln Xij Xik
pk i=1 j=1
k=1
m n 1–t
p1
n rp t
k
= Xij Xik k .
k=1 i=1 j=1
n
Theorem 2.5 Let aij > 0 (1 ≤ i ≤ m, 1 ≤ j ≤ n), rpk > 1, k=1 1/pk = r. Define the positive
function
m n 1–t
1/pk
n rp t
hr (t) = aij aik k , t ∈ R. (8)
k=1 i=1 j=1
Yan and Gao Journal of Inequalities and Applications (2019) 2019:97 Page 6 of 12
Then the function hr (t) defined by (8) is concave, that is, hr (t) ≥ 0 for t ∈ R, and the equality
holds if and only if
rp rp
a k ajk k
nik = n , 1 ≤ i, j ≤ m, k = 1, 2, . . . , n. (9)
j=1 aij l=1 ajl
Proof It is easy to see from (2) that hr (t) can be expressed in the following equivalent form:
1 rp
Let bi = ( nj=1 aij ) r , dik = aik k /bri ,
m rp
p1
n a k t k
hr (t) = bri ikr .
i=1
bi
k=1
m r t
hr (t) 1
n
i=1 bi dik ln dik
Hr (t) = = m r t
hr (t) pk i=1 bi dik
k=1
m r t m r
n n
1 i=1 bi dik ln dik 1 i=1 bi ln dik
= m r t – m r
pk b d
i=1 i ik pk i=1 bi
k=1 k=1
⎧
n r r t
– djkt ) ⎨≥ 0, t ≥ 0,
1 1≤i<j≤m bi bj (ln dik – ln djk )(dik
= m r m r t
pk ( i=1 bi )( i=1 bi dik ) ⎩≤ 0, t ≤ 0.
k=1
n m m
n
1 ( r t r t 2 r t
ln dik )2
i=1 bi dik )( i=1 bi dik (ln dik ) ) – ( i=1 bi dik
Hr (t) = m r t 2 .
pk ( i=1 bi dik )
k=1
It is easy to see that Hr ≥ 0 and the equality holds at some t0 if and only if
Since hr (t) = hr (t)Hr (t), we have hr (t) = hr (t)[(Hr (t))2 + Hr (t)] ≥ 0, and hr (t) = 0 if and only
if Hr (t) = 0 and Hr (t) = 0.
The proof of the theorem is completed.
Yan and Gao Journal of Inequalities and Applications (2019) 2019:97 Page 7 of 12
From our discussion we see that hr (t) ≥ 0 and the equality holds at some t0 ∈ R if and
only if (9) holds. In this case, hr (t) ≡ 0 and hr (t) = hr (0) = const. Since, by by the proof,
hr (t) = hr (0)Hr (0) = 0, which yields hr (t) = hr (0) = const, t ∈ R.
Corollary 2.6 If (9) holds, then hr (t) = const. Otherwise, hr (t) > 0, t ∈ R, and hr (t)t > 0 for
t = 0; in particular, for 0 = t1 < t2 < · · · < tN = 1, we have
m p1
m
n
n pj
j
aij ≤ aij .
i=1 j=1 j=1 i=1
n n p1
n
1
r r
1≤i<j≤m bi bj (ln dik – ln djk )(dik – djk ) pj
j
hr (1) = Hr (1)hr (1) = m r m r aij .
pk ( i=1 bi )( i=1 bi dik ) j=1 j=1
k=1
Theorem 2.8 Let fk (x) > 0, x ∈ [a, b], k = 1, 2, . . . , n, and fk ∈ Lpk [a, b]. Define the positive
function
n 1–t
p1
n b
rp t
k
gr (t) = fj (x) fk k (x) dx , t ∈ R. (10)
k=1 a j=1
The function gr (t) defined by (10) is concave, that is, gr (t) ≥ 0 for t ∈ R, and the equality
holds if and only if
rp
f k (x)
kn = ck = const. (11)
j=1 fj (x)
n 1 rp
Proof It is known from Lemma 2.3 that if F(x) = ( j=1 fj (x))
r and gk (x) = fk k (x)/F r (x),
then
n
b p1
k
gr (t) = F r (x)gkt (x) dx .
k=1 a
Yan and Gao Journal of Inequalities and Applications (2019) 2019:97 Page 8 of 12
n b
1
Gr (t) = ln F r (x)gkt (x) dx .
pk a
k=1
Gr (t)
b
n
1 F r (x)gkt (x) ln gk (x) dx
a
= b
pk r t
k=1 a F (x)gk (x) dx
b b
n
1 F r (x)gkt (x) ln gk (x) dx a F r (x)( nk=1 p1 ln gk (x)) dx
a k
= b – b
pk r t r (x) dx
k=1 a F (x)gk (x) dx a F
b b
n
1 F r (x)gkt (x) ln gk (x) dx F r (x) ln gk (x) dx
a a
= b – b
pk r t r
k=1 a F (x)gk (x) dx a F (x) dx
bb ⎧
n
1 F r (x)F r (y)(gkt (x) – gkt (y))(ln gk (x) – ln gk (y)) dx dy ⎨≥ 0, t ≥ 0,
a a
= b b
2pk r t r ⎩≤ 0, t ≤ 0.
k=1 a F (x)gk (x) dx a F (x) dx
rp
f k (x)
kn = ck = const.
j=1 fj (x)
bb
n
1 F r (x)F r (y)gkt (x)gkt (y)(ln gk (x) – ln gk (y))2 dx dy
Gr (t) = a a
b ≥ 0.
pk 2( a F r (x)gkt (x) dx)2
k=1
It is easy to see Gr (t) ≥ 0 and the equality holds at some t0 if and only if
rp
f k (x)
kn = ck = const.
j=1 fj (x)
Since gr (t) = gr (t)Gr (t), we have gr (t) = gr (t)[(Gr (t))2 + Gr (t)] ≥ 0, and gr (t) = 0 if and
only if Gr (t) = 0 and Gr (t) = 0.
Thus the proof of the theorem is completed.
Corollary 2.9 If gk (x) = const, then gr (t) = const. Otherwise, gr (t) > 0, t ∈ R, and gr (t)t > 0
for t = 0; in particular, for 0 = t1 < t2 < · · · < tN = 1, we have
Next, we use the existing inequalities to derive the inequalities obtained in Kwon and
Bae [25].
n
Theorem 2.11 Let aij > 0, pk > 1, αkj ∈ R, 1 ≤ i ≤ m, 1 ≤ j, k ≤ n, 1
= 1, and
n k=1 pk
m n p1
m
n
n 1+pk αkj
k
aij ≤ aij . (12)
i=1 j=1 k=1 i=1 j=1
Moreover, for the integral form of this inequality, if fj (x) > 0 (j = 1, 2, . . . , n), x ∈ [a, b], –∞ <
a < b < +∞, and fj (x) ∈ C[a, b], then
b
n
n
b
n
p1
k
1+pk αkj
fj (x) dx ≤ fj (x) dx . (13)
a j=1 k=1 a j=1
Proof From (2) we can see that h(0) is the minimum point of h(t).
Thus h(0) ≤ h(t), that is,
m n 1–t
p1
m
n
n pk t k
aij ≤ aij aik .
i=1 j=1 k=1 i=1 j=1
Under the assumptions of (2), taking t = –pk αkj for j = k and t = αkk /(1 – 1
) for j = k with
n pk
k=1 αkj = 0, we have
m n p1
m
n
n 1+pk αkj
k
aij ≤ aij .
i=1 j=1 k=1 i=1 j=1
Similarly, the integral expression of Hölder’s inequality now becomes g(0) ≤ g(t), that is,
b
n
n
n
b
1–t p1
k
p t
fk (x) dx ≤ fj (x) fk k (x) dx , t ∈ R.
a k=1 k=1 a j=1
Yan and Gao Journal of Inequalities and Applications (2019) 2019:97 Page 10 of 12
Next, we make some changes to g(t). Taking t = –pk αkj for j = k and t = αkk /(1 – 1
) for
pk
j = k with nk=1 αkj = 0, we have
b
n
n
b
n
p1
k
1+pk αkj
fj (x) dx ≤ fj (x) dx .
a j=1 k=1 a j=1
m n p1
m
n
n 1+rpk αkj
k
aij ≤ aij . (14)
i=1 j=1 k=1 i=1 j=1
Moreover, for the integral form of this inequality, if fj (x) > 0 (j = 1, 2, . . . , n), x ∈ [a, b], –∞ <
a < b < +∞, and fj (x) ∈ C[a, b], then
b
n
n
b
n
p1
k
1+rpk αkj
fj (x) dx ≤ fj (x) dx . (15)
a j=1 k=1 a j=1
Proof From (8) we get that hr (0) is the minimum point of hr (t). Thus, hr (0) ≤ hr (t), that
is,
m n 1–t
p1
m
n
n rpk t k
aij ≤ aij aik .
i=1 j=1 k=1 i=1 j=1
Under the assumptions of (8), taking t = –rpk αkj for j = k and t = αkk /(1 – rp1 ) for j = k with
n k
k=1 αkj = 0, we have
m n p1
m
n
n 1+rpk αkj
k
aij ≤ aij .
i=1 j=1 k=1 i=1 j=1
Similarly, the integral expression of Hölder’s inequality now becomes gr (0) ≤ gr (t), that is,
b
n
n
n
b
1–t p1
k
rp t
fk (x) dx ≤ fj (x) fk k (x) dx , t ∈ R.
a k=1 k=1 a j=1
Similarly to discrete hr (t), we also make some changes to the continuous gr (t). Taking
t = –rpk αkj for j = k and t = αkk /(1 – p1 ) for j = k with nk=1 αkj = 0, we have
k
b
n
n
b
n
p1
k
1+rpk αkj
fj (x) dx ≤ fj (x) dx .
a j=1 k=1 a j=1
3 Conclusion
As is well known, the Hölder inequality and its derivative theorems play an important
role in mathematical analysis. In this paper, we have presented a wider range of discrete
function hr (t) and continuous function gr (t) based on the existing discrete function h(t)
and continuous function g(t). In addition, we give a proof and application of hr (t) and
gr (t), and, finally, by using the results obtained we get an extension of Hölder’s inequality
to the new inequality. In the future research, we will continue to explore other derived
inequalities of Hölder’s inequality.
Acknowledgements
The authors would like to express their sincere thanks to the anonymous referees for their great efforts to improve this
paper.
Funding
This work was supported by the Fundamental Research Funds for the Central Universities (No. 2015ZD29) and the Higher
School Science Research Funds of Hebei Province of China (No. Z2015137).
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
The authors read and approved the final manuscript.
Author details
1
North China Electric Power University, Baoding, P.R. China. 2 College of Science and Technology, North China Electric
Power University, Baoding, P.R. China.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
1. Qi, F.: An algebraic inequality. J. Inequal. Pure Appl. Math. 2(1), Article ID 13 (2001)
2. Qi, F., Chen, S.-X.: Complete monotonicity of the logarithmic mean. Math. Inequal. Appl. 10(4), 799–804 (2007)
3. Qi, F., Li, W.-H.: A unified proof of several inequalities and some new inequalities involving Neuman–Sándor mean.
Miskolc Math. Notes 15(2), 665–675 (2014)
4. Qi, F., Luo, Q.-M.: A simple proof of monotonicity for extended mean values. J. Math. Anal. Appl. 224(2), 356–359
(1998)
5. Tian, J.-F.: Note on common fixed point theorems in fuzzy metric spaces using the CLRg property. Fuzzy Sets Syst.
https://doi.org/10.1016/j.fss.2019.01.018
6. Yang, Zh.-H., Tian, J.-F.: Monotonicity rules for the ratio of two Laplace transforms with applications. J. Math. Anal.
Appl. 470(2), 821–845 (2019)
7. Yang, Zh.-H., Tian, J.-F.: A comparison theorem for two divided differences and applications to special functions.
J. Math. Anal. Appl. 464(1), 580–595 (2018)
8. Yang, Zh.-H., Tian, J.-F.: An accurate approximation formula for gamma function. J. Inequal. Appl. 2018, 56 (2018)
9. Yang, Zh.-H., Tian, J.-F.: Two asymptotic expansions for gamma function developed by Windschitl’s formula. Open
Math. 16, 1048–1060 (2018)
10. Yang, Zh.-H., Tian, J.-F.: Sharp inequalities for the generalized elliptic integrals of the first kind. Ramanujan J. 48(1),
91–116 (2019)
11. Yang, Zh.-H., Tian, J.-F.: A class of completely mixed monotonic functions involving the gamma function with
applications. Proc. Am. Math. Soc. 146(11), 4707–4721 (2018)
12. Zhang, S.-Q., Qi, F.: Some integral inequalities. J. Henan Norm. Univ. Nat. Sci. 27(1), 88–89 (1997)
13. Yang, X.-J.: A note on Hölder inequality. Appl. Math. Comput. 134, 319–322 (2003)
14. Yang, X.-J.: Hölder’s inequality. Appl. Math. Lett. 16, 897–903 (2003)
15. Qi, F.: Several integral inequalities. J. Inequal. Pure Appl. Math. 1(2), Article ID 19 (2000)
16. Qi, F., Luo, Q.-M.: Refinements and extensions of an inequality. Math. Inform. Q. 9(1), 23–25 (1999)
17. Tian, J.-F.: New property of a generalized Hölder’s inequality and its applications. Inf. Sci. 288, 45–54 (2014)
18. Tian, J.-F.: Reversed version of a generalized sharp Hölder’s inequality and its applications. Inf. Sci. 201, 61–69 (2012)
19. Tian, J.-F.: Inequalities and mathematical properties of uncertain variables. Fuzzy Optim. Decis. Mak. 10(4), 357–368
(2011)
20. Tian, J.-F.: Triple diamond-alpha integral and Hölder-type inequalities. J. Inequal. Appl. 2018, 111 (2018)
21. Tian, J.-F., Ha, M.-H.: Properties of generalized sharp Hölder’s inequalities. J. Math. Inequal. 11(2), 511–525 (2017)
22. Tian, J.-F., Ha, M.-H.: Extensions of Hölder’s inequality via pseudo-integral. Math. Probl. Eng. 2018, Article ID 4080619
(2018)
Yan and Gao Journal of Inequalities and Applications (2019) 2019:97 Page 12 of 12
23. Tian, J.-F., Ha, M.-H., Wang, Ch.: Improvements of generalized Hölder’s inequalities and their applications. J. Math.
Inequal. 12(2), 459–471 (2018)
24. Tian, J.-F., Pedrycz, W.: New refinements of generalized Hölder’s inequality and their applications. Math. Inequal. Appl.
19(3), 805–822 (2016)
25. Kwon, E.-G., Bae, E.-K.: On a continuous form of Hölder inequality. J. Math. Anal. Appl. 343, 585–592 (2008)
26. Tian, J.-F., Zhu, Y.-R., Cheung, W.-S.: N-tuple Diamond-Alpha integral and inequalities on time scales. Rev. R. Acad.
Cienc. Exactas Fís. Nat., Ser. A Mat. https://doi.org/10.1007/s13398-018-0609-6
27. Zhao, C.-J., Cheung, W.-S.: Hölder’s reverse inequality and its application. Publ. Inst. Math. 99(113), 211–216 (2016)