You are on page 1of 7

Journal of Mathematical Analysis and Applications Volume 259, Issue 1 , 1 July 2001, Pages 219-225 doi:10.1006/jmaa.2000.

7422

THE CARLEMANS INEQUALITY FOR NEGATIVE POWER NUMBER


by Nguyen Thanh Long(#), Nguyen Vu Duy Linh (*) (#) Department of Mathematics and Computer Science, College of Natural Science ,Vietnam National University HoChiMinh City, 268 Ly Thuong Kiet Str.,Dist.10, HoChiMinh City, Vietnam. E-mail: longnt@netnam2.org.vn
(*)

Institute of Applied Mechanics in HoChiMinh City, 291 Dien Bien Phu Str.,Dist.03, HoChiMinh City, Vietnam. Address for correspondence : Nguyen Thanh Long. ABSTRACT By the method of indeterminate coefficients we prove the following inequality
1 (1 r ) r a n , n =1 r 1 r r r 1 2 an , n =1

r r r a1 + a 2 + ... + a n n n =1

if

1 r < 1, r 0, r < 1.

if

where a n 0 , n = 1,2,...

n =1

an < .

Keywords: Carlemans inequality. 1. INTRODUCTION

The following Carlemans inequality is well known (See [1], Chapt.9.12)


1 1 1p a1 + a 2 p + ... + a n p n n =1

p p an p 1 n =1

(1)

where a n 0 , n = 1,2,... ,

n =1

an <

and p > 1 .
Page 1 of 7

Letting p + , it follows from (1) that

(a1a2 ...an )
n =1

n =1

an .

(2)

In the practice, the inequality (2) is strict,i.e.,

(a1a2 ...an )
n =1

< e

n =1

an .

(3)

if a n 0 , n = 1,2,... , 0 < a n < .


n =1

The constant e is sharp in the sense that it can not be replaced by a smaller one. Recently, the inequality (3) is also improved by many authors, example:Yang Bicheng, L.Debnath[2] with

(a1a2 ...an )
n =1

< e

(1 2n + 2 ) a n ,
n =1

(4)

and in [3] Yan Ping ,Sun Guozheng with

(a1a2 ...an )
n =1

< e

(1 +
n =1
n =1

n+ 1

) 5

2a

(5)

where a n 0 , n = 1,2,... , 0 < a n < . We rewrite the inequality (1) with r = 1


r r r a1 + a 2 + ... + a n n n =1

p
1

as follows
r

(1 r )

an
n =1

(6)

where a n 0 , n = 1,2,... ,
2.MAIN RESULT

n =1

an <

and 0 < r < 1 .

In this paper, we shall prove the following theorem


Theorem 1. Let a n 0 , n = 1,2,... , and
n =1

a n < . Then we have

Page 2 of 7

r r r a1 + a 2 + ... + a n n n =1

1 (1 r ) r a n , n =1 r 1 r r r 1 2 an , n =1

if

1 r < 1, r 0, r < 1.

(7.a)

if

(7.b)

To prove Theorem 1, we first prove the following Lemmas.


Lemma 1.We have
t t a b t + b2 + ... + bn a 1 a ( 1 + 2 + ... + n ) 1 n b1 b2 bn n r , r < 1, r 0 . where a k 0 , bk > 0 , k = 1,2,..., n , t = 1 r Lemma 2.We have 1 1 1 1 < [(m ) (m + ) ] , 1+ 2 2 m for all m = 1,2,... , and > 0 .

r r r a1 + a 2 + ... + a n n

(8)

(9)

Lemma 3. For all n N , we have 1 (i) n (n 1) > (n ) 1 if 0 < < 1 , 2 1 (ii) n (n 1) < (n ) 1 if 1 < < 2 . 2 p Proof of Lemma 1. Let q = , Ak 0, Bk 0 , we have the Holders inequality p 1

(10) (11)

n Ak Bk Akp k =1 k =1
n

p n

q Bk k =1
q Bk k =1

if

p > 1,

(12)

and
n Ak Bk Akp k =1 k =1
n 1 p n 1 q

if

p <1 .

(13)

We obtain from (12),(13) that


n n n q Ak Bk Akp Bk k =1 k =1 k =1
p p q

, for all p > 1 or p < 0 .

(14)

a 1 r , Ak = ( k ) r , Bk = bk bk r Proof of Lemma 2. Using the following binomial series

Applying the inequality (14) with p =

we obtain (8).

Page 3 of 7

(1 + x) =

k =0

k C x k , for all , x R , 1 < x < 1 ,

(15)

where
k C =

( 1)...( k + 1) , k!

we have
1 1 k 1 ) = m C (m ) = m (1 . 2 2m 2m k =0 Simillarly k

(16)

1 k 1 (m + ) = m C . 2 2m k =0 k

(17)

It follows from (16),(17) that


1 1 k 1 (m ) (m + ) = m [(1) k 1]C 2 2 2m k =0 k

= 2m Proof of Lemma 3.

2k 1 C +1 2m k =0

2 k +1

1 > 2m C

1 = 1+ . 2m m

With n = 1 , then (10) and (11) evidently hold. 1 Let n 2 , put p = n . Applying (15) with x = 1 , = we obtain 2p 2 1 1 1 n = ( p + ) = p (1 + ) = p Ck ( ) k . 2p 2p 2 k =0 Similarly, 1 1 (n 1) = ( p ) = p Ck ( ) k . 2p 2 k =0 Hence, we obtain from (19) and (20) that
n (n 1) = p

(19)

(20)

k =0

[1 (1) k ]Ck (

1 k ) = 2 p 2p

k =0

C2k +1 ( 2 p ) 2k +1 .
k = 0,1,...

(21)

Note that C2 k +1 = and

( 1)( 2)...( 2k )
(2k + 1)!
if 1 < < 2 ,

>0

if 0 < < 1 ,

(22)

C2 k +1 < 0 Finally, we obtain:

k = 1,2,...

(23)

Page 4 of 7

- if 0 < < 1 , it follows from (21), (22) that 1 1 1 n (n 1) > 2 p C = p 1 = (n ) 1 . 2p 2 - if 1 < < 2 , we have also from (21), (23) that 1 1 = (n ) 1 . 2p 2 This completes the proof of Lemma 3 .
1 n (n 1) < 2 p C

(24)

(25)

Proof of Theorem 1. We prove the theorem by the method of indeterminate coefficients. Consider b1 , b2 ,... being the positive indeterminate coefficients. Put
t t t b1 + b2 + ... + bn n Applying the Lemma 1, we obtain

1 1 k n = n bk

where t =

r 1 r

(26)

r r r a1 + a 2 + ... + a n n n =1 N

k n ak = C n a n , for all N = 1,2,... , n =1 k =1 n =1

(27)

where
t . Cn = We shall prove that, for the chosen appropriate coefficients bn , n = 1,2,... , we have
n m m=n N

1 = bn

t t t 1 b1 + b2 + ... + bm m m=n m N

(28)

Cn C , where C is a constant only depending on r ( independing of n , N , a n , n = 1,2,... ). First, let > 0 such that r . (29) 1 t > 0 , t = 1 r Choosing
bm = m1 t (m 1)1 t

, m = 1,2,...

(30)

We have from (28) that


(n 1) Applying the Lemma 2, we obtain

Cn =

(n

1
1 t 1 t

m=n

m1+
1

(31)

m=n

m1+

<

m=n

[(m 2 ) (m + 2 ) ]
Page 5 of 7

1 1 1 1 [(n ) ( N + ) ] < (n ) . 2 2 2 It follows from (31),(32) that = Cn <


1
1 t

(32)
1

(n

(n 1)

1 t

1 (n ) 2

where = 1 t . We note that, if 0 < r < 1 , we choose , 0 < <

1 n (n 1) = (n 1 ) 1 2

(33)

1 r 1 1 = , then we have < 0 and r t t 0 < = 1 t < 1 . Using the Lemma 3 (i) , we deduce from (33) that 1 Cn < G ( , r ) . (34) 1

In the case of r < 0 , we choose , 0 < <

1 1 r 1 , then we have > 0 and = t r t 1 < = 1 t < 2 . Using again the Lemma 3 (ii) we also obtain the same inequality (34) from (33). 1 r Now, we shall choose 0 , 0 < 0 < such that r G ( , r ) G ( 0 , r ) for all (0, Consider the function
g ( ) = 1 G ( , r )
1

1 r ). r
1 t

(35)
1 r . r

= (1 t )

(36)

Then we have the derivative


1 (1 r ) . 1 r If 1 r < 1 , r 0 , then the derivative g ( ) changes from positive to negative at 0 = 1 r and the function g ( ) has a maximum in 0 = 1 r . Hence
1 1 g ( ) = (1 t ) t

(37)

g ( ) g ( 0 ) = g (1 r ) = (1 r ) If r < 1 , then it follows from (37) that 1 r g ( ) > 0 [0, ]. r

, [0,

1 r ]. r

(38)

(39)
1 r .We obtain r

Hence, the function g ( ) has a maximum in 0 =


1 r

1 r r 1 r 1 r g ( ) g ( 0 ) = g ( 2 [0, )= ]. r r r Finally, we obtain from (34), (36),(38),(40) that


Page 6 of 7

(40)

C n C (r ) =

inf
0< <

1 r r

G ( , r )

1 (1 r ) r , 1 r < 1 , r 0, 1 = = r 1 r sup g ( ) r , r < 1. 1 r r 1 2 0< <


r

Theorem 1 is proved completely.

REFERENCES

[1] G.H.Hardy, J.E.Littlewood, G.Polya, Inequalities, Cambridge Univ.Press, London, 1952. [2] Yang Bicheng, L.Debnath , Some inequalities involving the constant e and an application to Carlemans inequality, J.Math.Anal.Appl.223 (1998), 347-353. [3] Yan Ping ,Sun Guozheng , A strengthened Carlemans inequality, J.Math.Anal.Appl. 240 (1999), 290-293.

Page 7 of 7

You might also like