Skip to main content

On a more accurate multidimensional Mulholland-type inequality

Abstract

In this paper, by using the way of weight coefficients and technique of real analysis, a more accurate multidimensional discrete Mulholland-type inequality with the best possible constant factor is given, which is an extension of the Mulholland inequality. The equivalent form, the operator expression with the norm as well as a few particular cases are also considered.

MSC:26D15, 47A07.

1 Introduction

Suppose that p>1, 1 p + 1 q =1, f(x),g(y)0, f L p ( R + ), g L q ( R + ), f p = { 0 f p ( x ) d x } 1 p >0, g q >0. We have the following Hardy-Hilbert integral inequality (cf. [1]):

0 0 f ( x ) g ( y ) x + y dxdy< π sin ( π / p ) f p g q ,
(1)

where the constant factor π sin ( π / p ) is the best possible. Assuming that a m , b n 0, a= { a m } m = 1 l p , b= { b n } n = 1 l q , a p = { m = 1 a m p } 1 p >0, b q >0, we have the following Hardy-Hilbert inequality with the same best constant π sin ( π / p ) (cf. [1]):

m = 1 n = 1 a m b n m + n < π sin ( π / p ) a p b q .
(2)

Inequalities (1) and (2) are important in analysis and its applications (cf. [16]). Also we have the following Mulholland inequality (cf. [1]):

m = 2 n = 2 a m b n ln m n < π sin ( π / p ) { m = 2 a m p m 1 p } 1 p { n = 2 b n q n 1 q } 1 q .
(3)

In 1998, by introducing an independent parameter λ(0,1], Yang [7] gave an extension of (1) for p=q=2. Yang [5] gave some extensions of (1) and (2) as follows: If λ 1 , λ 2 ,λR, λ 1 + λ 2 =λ, k λ (x,y) is a non-negative homogeneous function of degree −λ, with

k( λ 1 )= 0 k λ (t,1) t λ 1 1 dt R + ,

ϕ(x)= x p ( 1 λ 1 ) 1 , ψ(x)= x q ( 1 λ 2 ) 1 , f(x),g(y)0,

f L p , ϕ ( R + )= { f ; f p , ϕ : = { 0 ϕ ( x ) | f ( x ) | p d x } 1 p < } ,

g L q , ψ ( R + ), f p , ϕ , g q , ψ >0, then

0 0 k λ (x,y)f(x)g(y)dxdy<k( λ 1 ) f p , ϕ g q , ψ ,
(4)

where the constant factor k( λ 1 ) is the best possible. Moreover, if k λ (x,y) is finite and k λ (x,y) x λ 1 1 ( k λ (x,y) y λ 2 1 ) is decreasing with respect to x>0 (y>0), then for a m , b n 0,

a l p , ϕ = { a ; a p , ϕ : = { n = 1 ϕ ( n ) | a n | p } 1 p < } ,

b= { b n } n = 1 l q , ψ , a p , ϕ , b q , ψ >0, it follows that

m = 1 n = 1 k λ (m,n) a m b n <k( λ 1 ) a p , ϕ b q , ψ ,
(5)

where the constant factor k( λ 1 ) is still the best possible.

Clearly, for λ=1, k 1 (x,y)= 1 x + y , λ 1 = 1 q , λ 2 = 1 p , inequality (3) reduces to (1), while (5) reduces to (2). Some other results including the multidimensional Hilbert-type integral inequalities are provided by [824].

About half-discrete Hilbert-type inequalities with the non-homogeneous kernels, Hardy et al. provided a few results in Theorem 351 of [1]. But they did not prove that the constant factors are the best possible. However, Yang [25] gave a result with the kernel 1 ( 1 + n x ) λ (0<λ2) by introducing a variable and proved that the constant factor is the best possible. In 2011 Yang [26] gave a half-discrete Hardy-Hilbert inequality with the best possible constant factor. Zhong et al. [2733] investigated several half-discrete Hilbert-type inequalities with particular kernels. Using the way of weight functions and the techniques of discrete and integral Hilbert-type inequalities with some additional conditions on the kernel, a half-discrete Hilbert-type inequality with a general homogeneous kernel of degree λR and a best constant factor k( λ 1 ) is obtained as follows:

0 f(x) n = 1 k λ (x,n) a n dx<k( λ 1 ) f p , ϕ a q , ψ
(6)

(see Yang and Chen [34]). At the same time, a half-discrete Hilbert-type inequality with a general non-homogeneous kernel and the best constant factor is given by Yang [35].

In this paper, by using the way of weight coefficients and technique of real analysis, a more accurate multidimensional discrete Mulholland-type inequality with the best possible constant factor is given, which is an extension of (3). The equivalent form, the operator expression with the norm as well as a few particular cases are also considered.

2 Some lemmas

Lemma 1 If ( 1 ) i h ( i ) (t)>0 (t>0; i=1,2), then for b>0, 0<α1,

( 1 ) i d i d x i h ( ( b + ln α x ) 1 α ) >0(x>1;i=1,2).
(7)

Proof We find

d d x h ( ( b + ln α x ) 1 α ) = 1 x h ( ( b + ln α x ) 1 α ) ( b + ln α x ) 1 α 1 ln α 1 x < 0 , d 2 d x 2 h ( ( b + ln α x ) 1 α ) = d d x [ 1 x h ( ( b + ln α x ) 1 α ) ( b + ln α x ) 1 α 1 ln α 1 x ] d 2 d x 2 h ( ( b + ln α x ) 1 α ) = 1 x 2 h ( ( b + ln α x ) 1 α ) ( b + ln α x ) 1 α 1 ln α 1 x d 2 d x 2 h ( ( b + ln α x ) 1 α ) = + 1 x 2 h ( ( b + ln α x ) 1 α ) ( b + ln α x ) 2 α 2 ln 2 α 2 x d 2 d x 2 h ( ( b + ln α x ) 1 α ) = + α ( 1 α 1 ) 1 x 2 h ( ( b + ln α x ) 1 α ) ( b + ln α x ) 1 α 2 ln 2 α 2 x d 2 d x 2 h ( ( b + ln α x ) 1 α ) = + ( α 1 ) 1 x 2 h ( ( b + ln α x ) 1 α ) ( b + ln α x ) 1 α 1 ln α 2 x d 2 d x 2 h ( ( b + ln α x ) 1 α ) = [ h ( ( b + ln α x ) 1 α ) ( b + ln α x ) ln x d 2 d x 2 h ( ( b + ln α x ) 1 α ) = + h ( ( b + ln α x ) 1 α ) ( b + ln α x ) 1 α ln α x d 2 d x 2 h ( ( b + ln α x ) 1 α ) = + b ( α 1 ) h ( ( b + ln α x ) 1 α ) ] 1 x 2 ( b + ln α x ) 1 α 2 ln α 2 x > 0 .

Then we have (7). □

If i 0 , j 0 N (N is the set of positive integers), α,β>0, we set

x α := ( k = 1 i 0 | x k | α ) 1 α ( x = ( x 1 , , x i 0 ) R i 0 ) ,
(8)
y β := ( k = 1 j 0 | y k | β ) 1 β ( y = ( y 1 , , y j 0 ) R j 0 ) .
(9)

Lemma 2 If sN, γ,M>0, Ψ(u) is a non-negative measurable function in (0,1], and

D M := { x R + s ; i = 1 s x i γ M γ } = { x ; i = 1 s ( x i M ) γ 1 } ,

then we have (cf. [36])

D M Ψ ( i = 1 s ( x i M ) γ ) d x 1 d x s = M s Γ s ( 1 γ ) γ s Γ ( s γ ) 0 1 Ψ ( u ) u s γ 1 d u .
(10)

Lemma 3 If sN, γ>0, ε>0, d=( d 1 ,, d s ) [ 1 2 , 1 ] s , then

A s ( ε ) : = m ln ( m + d ) γ s ε 1 i = 1 s ( m i + d i ) = Γ s ( 1 γ ) ε s ε / γ γ s 1 Γ ( s γ ) + O ( 1 ) ( ε 0 + ) .
(11)

Proof For M> s 1 / γ , we set

Ψ(u)= { 0 , 0 < u < s M γ , ( M u 1 / γ ) s ε , s M γ u 1 .

Then by the decreasing property and (10), it follows that

A s ( ε ) { x R + s ; x i e d i } ln ( x + d ) γ s ε d x i = 1 s ( x i + d i ) = u i = ln ( x i + d i ) { u R + s ; u i 1 } u γ s ε d u = lim M D M Ψ ( i = 1 s ( x i M ) γ ) d x 1 d x s = lim M M s Γ s ( 1 γ ) γ s Γ ( s γ ) s / M γ 1 ( M u 1 / γ ) s ε u s γ 1 d u = Γ s ( 1 γ ) ε s ε / γ γ s 1 Γ ( s γ ) .

In the following, by mathematical induction we prove that, for any sN,

A s (ε) O s (1)+ Γ s ( 1 γ ) ε s ε / γ γ s 1 Γ ( s γ ) ( ε 0 + ) .
(12)

For s=1, by the Hermite-Hadamard inequality (cf. [37]), it follows that

A 1 ( ε ) = m 1 = 1 2 ln 1 ε ( m 1 + d 1 ) m 1 + d 1 + m 1 = 3 ln 1 ε ( m 1 + d 1 ) m 1 + d 1 O 1 ( 1 ) + 5 2 ln 1 ε ( x + d 1 ) d x x + d 1 O 1 ( 1 ) + e d 1 ln 1 ε ( x + d 1 ) d x x + d 1 = u = ln ( x + d 1 ) O 1 ( 1 ) + 1 u 1 ε d u = O 1 ( 1 ) + 1 ε ,

and then (12) is valid. Assuming that (12) is valid for s1N, then for s, we set

A s ( ε ) = { m N s ; i 0 , m i 0 = 1 , 2 } ln ( m + d ) γ s ε 1 i = 1 s ( m i + d i ) + { m N s ; m i 3 } ln ( m + d ) γ s ε 1 i = 1 s ( m i + d i ) .

There exist constants a,b R + , such that

{ m N s ; i 0 , m i 0 = 1 , 2 } ln ( m + d ) γ s ε 1 i = 1 s ( m i + d i ) a + b { m N s 1 ; m i 1 } ln ( m + d ) γ ( s 1 ) ( 1 + ε ) 1 i = 1 s 1 ( m i + d i ) .

By the assumption of mathematical induction for s1, we find

{ m N s 1 ; m i 1 } ln ( m + d ) γ ( s 1 ) ( 1 + ε ) 1 i = 1 s 1 ( m i + d i ) O s 1 ( 1 ) + Γ s 1 ( 1 γ ) ( 1 + ε ) ( s 1 ) ( 1 + ε ) / γ γ s 2 Γ ( s 1 γ ) ,

and then

{ m N s ; i 0 , m i 0 = 1 , 2 } ln ( m + d ) γ s ε 1 i = 1 s ( m i + d i ) O s (1).

By Lemma 1 and the Hermite-Hadamard inequality (cf. [37]), we obtain

{ m N s ; m i 3 } ln ( m + d ) γ s ε 1 i = 1 s ( m i + d i ) { x R + s ; x i 5 2 } ln ( x + d ) γ s ε 1 i = 1 s ( x i + d i ) d x { x R + s ; x i e d i } ln ( x + d ) γ s ε 1 i = 1 s ( x i + d i ) d x = u i = ln ( x i + d i ) { u R + s ; u i 1 } u γ s ε d u = Γ s ( 1 γ ) ε s ε / γ γ s 1 Γ ( s γ ) .

Hence we prove that (12) is valid for sN. Therefore, we have (11). □

Lemma 4 If C is the set of complex numbers and C =C{}, z k C{z|Rez0,Imz=0} (k=1,2,,n) are different points, the function f(z) is analytic in C except for z i (i=1,2,,n), and z= is a zero point of f(z) whose order is not less than 1, then for αR, we have

0 f(x) x α 1 dx= 2 π i 1 e 2 π α i k = 1 n Res [ f ( z ) z α 1 , z k ] ,
(13)

where 0<Imlnz=argz<2π. In particular, if z k (k=1,,n) are all poles of order 1, setting φ k (z)=(z z k )f(z) ( φ k ( z k )0), then

0 f(x) x α 1 dx= π sin π α k = 1 n ( z k ) α 1 φ k ( z k ).
(14)

Proof By [[38], p.118], we have (13). We find

1 e 2 π α i = 1 cos 2 π α i sin 2 π α = 2 i sin π α ( cos π α + i sin π α ) = 2 i e i π α sin π α .

In particular, since f(z) z α 1 = 1 z z k ( φ k (z) z α 1 ), it is obvious that

Res [ f ( z ) z α 1 , a k ] = z k α 1 φ k ( z k )= e i π α ( z k ) α 1 φ k ( z k ).

Then by (13), we obtain (14). □

Example 1 For sN, we set

k λ (x,y)= k = 1 s 1 ( x λ / s + c k y λ / s ) (0< c 1 << c s ,0<λs).

For 0< λ 1 i 0 , 0< λ 2 j 0 , λ 1 + λ 2 =λ, by (14), we find

k s ( λ 1 ) : = 0 k = 1 s 1 t λ / s + c k t λ 1 1 d t = u = t λ / s s λ 0 k = 1 s 1 u + c k u s λ 1 λ 1 d u = π s λ sin ( π s λ 1 λ ) k = 1 s c k s λ 1 λ 1 j = 1 ( j k ) s 1 c j c k R + .
(15)

In particular, for s=1, we obtain

k 1 ( λ 1 )= 1 λ 0 u ( λ 1 / λ ) 1 u + c 1 du= π λ sin ( π λ 1 λ ) c 1 λ 1 λ 1 .

Definition 1 For sN, 0<α,β1, 0< c 1 << c s , 0<λs, 0< λ 1 i 0 , 0< λ 2 j 0 , λ 1 + λ 2 =λ, τ=( τ 1 ,, τ i 0 ) [ 1 2 , 1 ] i 0 , σ=( σ 1 ,, σ j 0 ) [ 1 2 , 1 ] j 0 , ln(m+τ)=(ln( m 1 + τ 1 ),,ln( m i 0 + τ i 0 )) R + i 0 , ln(n+σ)=(ln( n 1 + σ 1 ),,ln( n j 0 + σ j 0 )) R + j 0 , we define two weight coefficients w λ ( λ 2 ,n) and W λ ( λ 1 ,m) as follows:

w λ ( λ 2 , n ) : = m ln ( n + σ ) β λ 2 ln ( m + τ ) α λ 1 i 0 k = 1 s [ ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] i = 1 i 0 ( m i + τ i ) , W λ ( λ 1 , m ) : = n ln ( m + τ ) α λ 1 ln ( n + σ ) β λ 2 j 0 k = 1 s [ ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] j = 1 j 0 ( n j + σ j ) ,
(16)

where m = m i 0 = 1 m 1 = 1 and n = n j 0 = 1 n 1 = 1 .

Lemma 5 Let the assumptions as in Definition  1 be fulfilled. Then:

  1. (i)

    we have

    w λ ( λ 2 ,n)< K 2 ( n N j 0 ) ,
    (17)
W λ ( λ 1 ,m)< K 1 ( m N i 0 ) ,
(18)

where

K 1 := Γ j 0 ( 1 β ) β j 0 1 Γ ( j 0 β ) k s ( λ 1 ), K 2 := Γ i 0 ( 1 α ) α i 0 1 Γ ( i 0 α ) k s ( λ 1 ),
(19)

and k s ( λ 1 ) is indicated by (15);

  1. (ii)

    for p>1, 0<ε<pmin{ λ 1 ,1 λ 2 }, setting λ ˜ 1 = λ 1 ε p , λ ˜ 2 = λ 2 + ε p , we have

    0< K ˜ 2 ( 1 θ ˜ λ ( n ) ) < w λ ( λ ˜ 2 ,n),
    (20)

where

θ ˜ λ ( n ) : = 1 k s ( λ ˜ 1 ) 0 i 0 λ / ( α s ) / ln ( n + σ ) β λ / s v s λ 1 λ 1 k = 1 s ( v + c k ) d v θ ˜ λ ( n ) = O ( 1 ln ( n + σ ) β λ ˜ 1 ) ,
(21)
K ˜ 2 = Γ i 0 ( 1 α ) k s ( λ ˜ 1 ) α i 0 1 Γ ( i 0 α ) R + .
(22)

Proof By Lemma 1, the Hermite-Hadamard inequality (cf. [37]), (10), and (15), it follows that

w λ ( λ 2 , n ) < ( 1 2 , ) i 0 ln ( n + σ ) β λ 2 ln ( x + τ ) α λ 1 i 0 d x k = 1 s [ ln ( x + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] i = 1 i 0 ( x i + τ i ) = u i = ln ( x i + τ i ) { u R + i 0 ; u i > ln ( 1 2 + τ i ) } ln ( n + σ ) β λ 2 u α λ 1 i 0 k = 1 s [ u α λ / s + c k ln ( n + σ ) β λ / s ] d u R + i 0 n σ β λ 2 u α λ 1 i 0 k = 1 s [ u α λ / s + c k ln ( n + σ ) β λ / s ] d u = lim M D M ln ( n + σ ) β λ 2 M λ 1 i 0 [ i = 1 j 0 ( u i M ) α ] ( λ 1 i 0 ) / α k = 1 s { M λ s [ i = 1 i 0 ( u i M ) α ] λ α s + c k ln ( n + σ ) β λ s } d u = lim M M i 0 Γ i 0 ( 1 α ) α i 0 Γ ( i 0 α ) 0 1 ln ( n + σ ) β λ 2 M λ 1 i 0 t ( λ 1 i 0 ) / α t i 0 α 1 k = 1 s ( M λ s t λ α s + c k ln ( n + σ ) β λ s ) d t = lim M M λ 1 Γ i 0 ( 1 α ) α i 0 Γ ( i 0 α ) 0 1 ln ( n + σ ) β λ 2 t λ 1 α 1 k = 1 s ( M λ s t λ α s + c k ln ( n + σ ) β λ s ) d t = t = ln ( n + σ ) β α M α v α s / λ s Γ i 0 ( 1 α ) λ α i 0 1 Γ ( i 0 α ) 0 v s λ 1 λ 1 k = 1 s ( v + c k ) d v = Γ i 0 ( 1 α ) α i 0 1 Γ ( i 0 α ) k s ( λ 1 ) = K 2 .

Hence, we have (17). In the same way, we have (18).

By the decreasing property and (10), similarly to the proof of (11), we find

w λ ( λ ˜ 2 , n ) > { x R + i 0 ; x i e τ i } ln ( n + σ ) β λ ˜ 2 ln ( x + τ ) α λ ˜ 1 i 0 d x k = 1 s [ ln ( x + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] i = 1 i 0 ( x i + τ i ) = u i = ln ( x i + τ i ) ln ( n + σ ) β λ ˜ 2 { u R + i 0 ; u i 1 } u α λ ˜ 1 i 0 d u k = 1 s [ u α λ / s + c k ln ( n + σ ) β λ / s ] = ln ( n + σ ) β λ ˜ 2 lim M D M { i = 1 i 0 ( u i M ) α } λ ˜ 1 i 0 α M λ ˜ 1 i 0 d u 1 d u i 0 k = 1 s [ { i = 1 i 0 ( u i M ) α } λ α s M λ s + c k ln ( n + σ ) β λ / s ] = M i 0 Γ i 0 ( 1 α ) α i 0 Γ ( i 0 α ) ln ( n + σ ) β λ ˜ 2 lim M i 0 M α 1 t λ ˜ 1 i 0 α M λ ˜ 1 i 0 t i 0 α 1 k = 1 s [ t λ α s M λ s + c k ln ( n + σ ) β λ / s ] d t = s Γ i 0 ( 1 α ) λ α i 0 1 Γ ( i 0 α ) i 0 λ / ( α s ) ln ( n + σ ) β λ / s v s λ ˜ 1 λ 1 d v k = 1 s ( v + c k ) = K ˜ 2 ( 1 θ ˜ λ ( n ) ) > 0 , 0 < θ ˜ λ ( n ) = s λ k s ( λ ˜ 1 ) 0 i 0 λ / ( α s ) / ln ( n + σ ) β λ / s v s λ 1 λ 1 k = 1 s ( v + c k ) d v 0 s λ k s ( λ ˜ 1 ) k = 1 s c k 0 i 0 λ / ( α s ) / ln ( n + σ ) β λ / s v s λ ˜ 1 λ 1 d v 0 = 1 λ ˜ 1 k s ( λ ˜ 1 ) k = 1 s c k i 0 λ ˜ 1 / α ln ( n + σ ) β λ ˜ 1 .

Hence, we have (20) and (21). □

3 Main results and operator expressions

Setting Φ(m):= i = 1 i 0 ( m i + τ i ) p 1 ln ( m + τ ) α p ( i 0 λ 1 ) i 0 (m N i 0 ) and Ψ(n):= j = 1 j 0 ( n j + σ j ) q 1 ln ( n + σ ) β q ( j 0 λ 2 ) j 0 (n N j 0 ), wherefrom

[ Ψ ( n ) ] 1 p = j = 1 j 0 ( n j + σ j ) 1 ln ( n + σ ) β p λ 2 j 0 ,

we have the following.

Theorem 1 If sN, 0<α,β1, 0< c 1 << c s , 0<λs, 0< λ 1 i 0 , 0< λ 2 j 0 , λ 1 + λ 2 =λ, τ [ 1 2 , 1 ] i 0 , σ [ 1 2 , 1 ] j 0 , then for p>1, 1 p + 1 q =1, a m , b n 0, 0< a p , Φ , b q , Ψ <, we have

I : = n m a m b n k = 1 s [ ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] < K 1 1 p K 2 1 q a p , Φ b q , Ψ ,
(23)

where the constant factor

K 1 1 p K 2 1 q = [ Γ j 0 ( 1 β ) β j 0 1 Γ ( j 0 β ) ] 1 p [ Γ i 0 ( 1 α ) β i 0 1 Γ ( i 0 α ) ] 1 q k s ( λ 1 )
(24)

is the best possible ( k s ( λ 1 ) is indicated by (15)).

Proof By the Hölder inequality (cf. [37]), we have

I = n m 1 k = 1 s [ ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] × [ ln ( m + τ ) α ( i 0 λ 1 ) / q ln ( n + σ ) β ( j 0 λ 2 ) / p i = 1 i 0 ( m i + τ i ) 1 / q j = 1 j 0 ( n j + σ j ) 1 / p a m ] × [ ln ( n + σ ) β ( j 0 λ 2 ) / p ln ( m + τ ) α ( i 0 λ 1 ) / q j = 1 j 0 ( n j + σ j ) 1 / p i = 1 i 0 ( m i + τ i ) 1 / q b n ] { m W λ ( λ 1 , m ) i = 1 i 0 ( m i + τ i ) p 1 ln ( m + τ ) α p ( i 0 λ 1 ) i 0 a m p } 1 p × { n w λ ( λ 2 , n ) j = 1 j 0 ( n j + σ j ) q 1 ln ( n + σ ) β q ( j 0 λ 2 ) j 0 b n q } 1 q .

Then by (17) and (18), we have (23).

For 0<ε<pmin{ λ 1 ,1 λ 2 }, λ ˜ 1 = λ 1 ε p , λ ˜ 2 = λ 2 + ε p , we set

a ˜ m = ln ( m + τ ) α i 0 + λ 1 ε p 1 i = 1 i 0 ( m i + τ i ) , b ˜ n = ln ( n + σ ) β j 0 + λ 2 ε q 1 j = 1 j 0 ( n j + σ j ) ( m N i 0 , n N j 0 ) .

Then by (11) and (20)-(22), we obtain

a ˜ p , Φ b ˜ q , Ψ = { m i = 1 i 0 ( m i + τ i ) p 1 ln ( m + τ ) α p ( i 0 λ 1 ) i 0 a ˜ m p } 1 p a ˜ p , Φ b ˜ q , Ψ = × { n j = 1 j 0 ( n j + σ j ) q 1 ln ( n + σ ) β q ( j 0 λ 2 ) j 0 b ˜ n q } 1 q a ˜ p , Φ b ˜ q , Ψ = { m ln ( m + τ ) α i 0 ε 1 i = 1 i 0 ( m i + τ i ) } 1 p a ˜ p , Φ b ˜ q , Ψ = × { n ln ( n + σ ) β j 0 ε 1 j = 1 j 0 ( n j + σ j ) } 1 q a ˜ p , Φ b ˜ q , Ψ = 1 ε [ Γ i 0 ( 1 α ) i 0 ε / α α i 0 1 Γ ( i 0 α ) + ε O ( 1 ) ] 1 p [ Γ j 0 ( 1 β ) j 0 ε / β β j 0 1 Γ ( j 0 β ) + ε O ˜ ( 1 ) ] 1 q ,
(25)
I ˜ : = n [ m a ˜ m k = 1 s ( ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ) ] b ˜ n I ˜ = n w λ ( λ ˜ 2 , n ) ln ( n + σ ) β j 0 ε 1 j = 1 j 0 ( n j + σ j ) I ˜ > K ˜ 2 n ( 1 O ( 1 ln ( n + σ ) β λ ˜ 1 ) ) ln ( n + σ ) β j 0 ε 1 j = 1 j 0 ( n j + σ j ) I ˜ = K ˜ 2 [ Γ j 0 ( 1 β ) ε j 0 ε / β β j 0 1 Γ ( j 0 β ) + O ˜ ( 1 ) O ( 1 ) ] .
(26)

If there exists a constant K K 1 1 p K 2 1 q , such that (23) is valid when replacing K 1 1 p K 2 1 q by K, then we have

( K 2 + o ( 1 ) ) [ Γ j 0 ( 1 β ) j 0 ε / β β j 0 1 Γ ( j 0 β ) + ε O ˜ ( 1 ) ε O ( 1 ) ] < ε I ˜ < ε K a ˜ p , φ b ˜ q , ψ = K [ Γ i 0 ( 1 α ) i 0 ε / α α i 0 1 Γ ( i 0 α ) + ε O ( 1 ) ] 1 p [ Γ j 0 ( 1 β ) j 0 ε / β β j 0 1 Γ ( j 0 β ) + ε O ˜ ( 1 ) ] 1 q .

For ε 0 + , we find

Γ j 0 ( 1 β ) Γ i 0 ( 1 α ) k s ( λ 1 ) β j 0 1 Γ ( j 0 β ) α i 0 1 Γ ( i 0 α ) K [ Γ i 0 ( 1 α ) α i 0 1 Γ ( i 0 α ) ] 1 p [ Γ j 0 ( 1 β ) β j 0 1 Γ ( j 0 β ) ] 1 q ,

and then K 1 1 p K 2 1 q K. Hence, K= K 1 1 p K 2 1 q is the best possible constant factor of (23). □

Theorem 2 With the assumptions of Theorem  1, for 0< a p , Φ <, we have the following inequality with the best constant factor K 1 1 p K 2 1 q :

J : = { n [ Ψ ( n ) ] 1 p ( m a m k = 1 s [ ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] ) p } 1 p < K 1 1 p K 2 1 q a p , Φ ,
(27)

which is equivalent to (23).

Proof We set b n as follows:

b n := [ Ψ ( n ) ] 1 p ( m a m k = 1 s ( ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ) ) p 1 .

Then it follows that J p = b q , Ψ q . If J=0, then (27) is trivially valid, since 0< a p , Φ <; if J=, then it is a contradiction since the right hand side of (27) is finite. Suppose that 0<J<. Then by (23), we find

b q , Ψ q = J p =I< K 1 1 p K 2 1 q a p , Φ b q , Ψ ,

namely, b q , Ψ q 1 =J< K 1 1 p K 2 1 q a p , Φ , and then (27) follows.

On the other hand, assuming that (27) is valid, by the Hölder inequality, we have

I = n ( Ψ ( n ) ) 1 q [ m a m k = 1 s ( ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ) ] × [ ( Ψ ( n ) ) 1 q b n ] J b q , Ψ .
(28)

Then by (27), we have (23). Hence (27) and (23) are equivalent.

By the equivalency, the constant factor K 1 1 p K 2 1 q in (27) is the best possible. Otherwise, we would reach a contradiction by (28) that the constant factor K 1 1 p K 2 1 q in (23) is not the best possible. □

For p>1, we define two real weight normal discrete spaces l p , φ and l q , ψ as follows:

l p , φ : = { a = { a m } ; a p , Φ = { m Φ ( m ) a m p } 1 p < } , l q , ψ : = { b = { b n } ; b q , Ψ = { n Ψ ( n ) b n q } 1 q < } .

With the assumptions of Theorem 2, in view of J< K 1 1 p K 2 1 q a p , Φ , we have the following definition.

Definition 2 Define a multidimensional Hilbert-type operator T: l p , Φ l p , Ψ 1 p as follows: For a l p , Φ , there exists an unique representation Ta l p , Ψ 1 p , satisfying for n N j 0 ,

(Ta)(n):= m a m k = 1 s [ ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] .
(29)

For b l q , Ψ , we define the following formal inner product of Ta and b as follows:

(Ta,b):= n m a m b n k = 1 s [ ln ( m + τ ) α λ / s + c k ln ( n + σ ) β λ / s ] .
(30)

Then by Theorem 1 and Theorem 2, for 0< a p , φ , b q , ψ <, we have the following equivalent inequalities:

(Ta,b)< K 1 1 p K 2 1 q a p , Φ b q , Ψ ,
(31)
T a p , Ψ 1 p < K 1 1 p K 2 1 q a p , Φ .
(32)

It follows that T is bounded since

T:= sup a ( θ ) l p , Φ T a p , Ψ 1 p a p , Φ K 1 1 p K 2 1 q .
(33)

Since the constant factor K 1 1 p K 2 1 q in (32) is the best possible, we have:

Corollary 1 With the assumptions of Theorem  2, T is defined by Definition  2, it follows that

T= K 1 1 p K 2 1 q = [ Γ j 0 ( 1 β ) β j 0 1 Γ ( j 0 β ) ] 1 p [ Γ i 0 ( 1 α ) α i 0 1 Γ ( i 0 α ) ] 1 q k s ( λ 1 ).
(34)

Remark 1 (i) Setting Φ 1 (m):= i = 1 i 0 ( m i + 1 ) p 1 ln ( m + 1 ) α p ( i 0 λ 1 ) i 0 (m N i 0 ) and Ψ 1 (n):= j = 1 j 0 ( n j + 1 ) q 1 ln ( n + 1 ) β q ( j 0 λ 2 ) j 0 (n N j 0 ), then putting τ=σ=1 in (23) and (27), we have the following equivalent inequalities with the best constant factor K 1 1 p K 2 1 q :

n m a m b n k = 1 s [ ln ( m + 1 ) α λ / s + c k ln ( n + 1 ) β λ / s ] < K 1 1 p K 2 1 q a p , Φ 1 b q , Ψ 1 ,
(35)
{ n [ Ψ 1 ( n ) ] 1 p ( m a m k = 1 s [ ln ( m + 1 ) α λ / s + c k ln ( n + 1 ) β λ / s ] ) p } 1 p < K 1 1 p K 2 1 q a p , Φ 1 .
(36)

Hence, (23) and (27) are more accurate inequalities than (35) and (36).

  1. (ii)

    Putting i 0 = j 0 =1, λ=s, ϕ 1 (m):= ( m + 1 ) p 1 ln p ( 1 λ 1 ) 1 (m+1) (mN) and ψ 1 (n):= ( n + 1 ) q 1 ln q ( 1 λ 2 ) 1 (n+1) (nN), in (32), we have the following new inequality:

    m = 1 n = 1 a m b n k = 1 s ln ( m + 1 ) ( n + 1 ) c k < π sin ( π λ 1 ) k = 1 s j = 1 ( j k ) s c k λ 1 1 c j c k a p , ϕ 1 b q , ψ 1 .
    (37)

In particular, for s= c k =1, λ 1 = 1 q , λ 2 = 1 p in (37), we can deduce (4). Hence, (23) is an extension of (4).

References

  1. Hardy GH, Littlewood JE, Pólya G: Inequalities. Cambridge University Press, Cambridge; 1934.

    MATH  Google Scholar 

  2. Mitrinović DS, Pečarić JE, Fink AM: Inequalities Involving Functions and Their Integrals and Derivatives. Kluwer Academic, Boston; 1991.

    Book  MATH  Google Scholar 

  3. Yang BC: Hilbert-Type Integral Inequalities. Bentham Science Publishers, Sharjah; 2009.

    Google Scholar 

  4. Yang BC: Discrete Hilbert-Type Inequalities. Bentham Science Publishers, Sharjah; 2011.

    Google Scholar 

  5. Yang BC: The Norm of Operator and Hilbert-Type Inequalities. Science Press, Beijing; 2009.

    Google Scholar 

  6. Yang BC: Two Types of Multiple Half-Discrete Hilbert-Type Inequalities. Lambert Academic Publishing, Saarbrücken; 2012.

    Google Scholar 

  7. Yang BC: On Hilbert’s integral inequality. J. Math. Anal. Appl. 1998, 220: 778–785. 10.1006/jmaa.1997.5877

    Article  MathSciNet  MATH  Google Scholar 

  8. Yang BC, Brnetić I, Krnić M, Pečarić JE: Generalization of Hilbert and Hardy-Hilbert integral inequalities. Math. Inequal. Appl. 2005,8(2):259–272.

    MathSciNet  MATH  Google Scholar 

  9. Krnić M, Pečarić JE: Hilbert’s inequalities and their reverses. Publ. Math. (Debr.) 2005,67(3–4):315–331.

    MATH  Google Scholar 

  10. Yang BC, Rassias TM: On the way of weight coefficient and research for Hilbert-type inequalities. Math. Inequal. Appl. 2003,6(4):625–658.

    MathSciNet  MATH  Google Scholar 

  11. Yang BC, Rassias TM: On a Hilbert-type integral inequality in the subinterval and its operator expression. Banach J. Math. Anal. 2010,4(2):100–110. 10.15352/bjma/1297117244

    Article  MathSciNet  MATH  Google Scholar 

  12. Azar L: On some extensions of Hardy-Hilbert’s inequality and applications. J. Inequal. Appl. 2009., 2009: Article ID 546829

    Google Scholar 

  13. Arpad B, Choonghong O: Best constant for certain multilinear integral operator. J. Inequal. Appl. 2006., 2006: Article ID 28582

    Google Scholar 

  14. Kuang JC, Debnath L: On Hilbert’s type inequalities on the weighted Orlicz spaces. Pac. J. Appl. Math. 2007,1(1):95–103.

    MathSciNet  MATH  Google Scholar 

  15. Zhong WY: The Hilbert-type integral inequality with a homogeneous kernel of − λ -degree. J. Inequal. Appl. 2008., 2008: Article ID 917392

    Google Scholar 

  16. Hong Y: On Hardy-Hilbert integral inequalities with some parameters. J. Inequal. Pure Appl. Math. 2005.,6(4): Article ID 92

  17. Zhong WY, Yang BC: On a multiple Hilbert-type integral inequality with the symmetric kernel. J. Inequal. Appl. 2007., 2007: Article ID 27962 10.1155/2007/27962

    Google Scholar 

  18. Yang BC, Krnić M: On the norm of a multi-dimensional Hilbert-type operator. Sarajevo J. Math. 2011,7(20):223–243.

    MathSciNet  MATH  Google Scholar 

  19. Krnić M, Pečarić JE, Vuković P: On some higher-dimensional Hilbert’s and Hardy-Hilbert’s type integral inequalities with parameters. Math. Inequal. Appl. 2008, 11: 701–716.

    MathSciNet  MATH  Google Scholar 

  20. Krnić M, Vuković P: On a multidimensional version of the Hilbert-type inequality. Anal. Math. 2012, 38: 291–303. 10.1007/s10476-012-0402-2

    Article  MathSciNet  MATH  Google Scholar 

  21. Adiyasuren V, Batbold T: Some new inequalities similar to Hilbert-type integral inequality with a homogeneous kernel. J. Math. Inequal. 2012,6(2):183–193.

    Article  MathSciNet  MATH  Google Scholar 

  22. Adiyasuren V, Batbold T, Krnić M: On several new Hilbert-type inequalities involving means operators. Acta Math. Sin. Engl. Ser. 2013,29(8):1493–1514. 10.1007/s10114-013-2545-x

    Article  MathSciNet  MATH  Google Scholar 

  23. Rassias MT, Yang BC: A multidimensional half-discrete Hilbert-type inequality and the Riemann zeta function. Appl. Math. Comput. 2013, 225: 263–277.

    Article  MathSciNet  Google Scholar 

  24. Li YJ, He B: On inequalities of Hilbert’s type. Bull. Aust. Math. Soc. 2007,76(1):1–13. 10.1017/S0004972700039423

    Article  MATH  Google Scholar 

  25. Yang BC: A mixed Hilbert-type inequality with a best constant factor. Int. J. Pure Appl. Math. 2005,20(3):319–328.

    MathSciNet  MATH  Google Scholar 

  26. Yang BC: A half-discrete Hilbert-type inequality. J. Guangdong Univ. Educ. 2011,31(3):1–7.

    Google Scholar 

  27. Zhong WY: A mixed Hilbert-type inequality and its equivalent forms. J. Guangdong Univ. Educ. 2011,31(5):18–22.

    MATH  Google Scholar 

  28. Zhong WY: A half discrete Hilbert-type inequality and its equivalent forms. J. Guangdong Univ. Educ. 2012,32(5):8–12.

    MATH  Google Scholar 

  29. Zhong JH, Yang BC: On an extension of a more accurate Hilbert-type inequality. J. Zhejiang Univ. Sci. Ed. 2008,35(2):121–124.

    MathSciNet  MATH  Google Scholar 

  30. Zhong JH: Two classes of half-discrete reverse Hilbert-type inequalities with a non-homogeneous kernel. J. Guangdong Univ. Educ. 2012,32(5):11–20.

    Google Scholar 

  31. Zhong WY, Yang BC: A best extension of Hilbert inequality involving several parameters. J. Jinan Univ., Nat. Sci. 2007,28(1):20–23.

    Google Scholar 

  32. Zhong WY, Yang BC: A reverse Hilbert’s type integral inequality with some parameters and the equivalent forms. Pure Appl. Math. 2008,24(2):401–407.

    MathSciNet  Google Scholar 

  33. Rassias MT, Yang BC: On half-discrete Hilbert’s inequality. Appl. Math. Comput. 2013, 220: 75–93.

    Article  MathSciNet  MATH  Google Scholar 

  34. Yang BC, Chen Q: A half-discrete Hilbert-type inequality with a homogeneous kernel and an extension. J. Inequal. Appl. 2011., 2011: Article ID 124 10.1186/1029-242X-2011-124

    Google Scholar 

  35. Yang BC: A half-discrete Hilbert-type inequality with a non-homogeneous kernel and two variables. Mediterr. J. Math. 2012,10(2):677–692. 10.1007/s00009-012-0213-5

    Article  MathSciNet  MATH  Google Scholar 

  36. Yang BC: Hilbert-type integral operators: norms and inequalities. In Nonlinear Analysis: Stability, Approximation, and Inequalities. Edited by: Pardalos PM, Georgiev PG, Srivastava HM. Springer, New York; 2012:771–859. Chapter 42

    Chapter  Google Scholar 

  37. Kuang JC: Applied Inequalities. Shangdong Science Technic Press, Jinan; 2004.

    Google Scholar 

  38. Pan YL, Wang HT, Wang FT: On Complex Functions. Science Press, Beijing; 2006.

    Google Scholar 

Download references

Acknowledgements

This work is supported by the National Natural Science Foundation of China (No. 61370186), and 2013 Knowledge Construction Special Foundation Item of Guangdong Institution of Higher Learning College and University (No. 2013KJCX0140).

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Bicheng Yang.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

QC participated in the design of the study and performed the numerical analysis. BY carried out the mathematical studies, participated in the sequence alignment and drafted the manuscript. All authors read and approved the final manuscript.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as 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.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Chen, Q., Yang, B. On a more accurate multidimensional Mulholland-type inequality. J Inequal Appl 2014, 322 (2014). https://doi.org/10.1186/1029-242X-2014-322

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/1029-242X-2014-322

Keywords