• Nem Talált Eredményt

Solution of the OQ. 2283

N/A
N/A
Protected

Academic year: 2022

Ossza meg "Solution of the OQ. 2283"

Copied!
7
0
0

Teljes szövegt

(1)

Solution of the OQ. 2283

R´obert Sz´asz and Aurel P´al Kup´an41

ABSTRACT.We determine the set ofα, for which the weighted H¨older mean Hα(a, b)is between theL(a, b)olya-Szeg˜o logarithmic and theI(a, b) exponential mean. Other results concerning this means are proved.

1. INTRODUCTION Let 0< a < b. The generalized H¨older mean is

Qα(a, b) = aα1 +b1α 2

!α

.

The P´olya &Szeg˝o logarithmic mean and the exponential mean are defined by L(a, b) = b−a

lnb−lna, I(a, b) = 1 e

bb aa

b1a .

In [4] the authors proved that, the inequality holds:

L(a, b)< I(a, b), a, b∈(0,∞).

Other results concerning these means were deduced in [1]. An exhaustive treatment of the topic can be found in [3].

The author of [2] proposed the following open question: determine the values α∈[2,∞) for which the inequalities hold

L(a, b)≤Qα(a, b)≤I(a, b) for all a, b∈(0,∞), a < b.

The aim of this paper is to determine the desired set of α and to deduce some other inequalities concerning these means.

41Received: 02.02.2009

2000Mathematics Subject Classification. 26D99

Key words and phrases. Inequalities, logarithmic means, exponential means.

(2)

2. PRELIMINARIES We will need in our work the following results.

Lemma 1. Letα∈(0,2] be a fixed number.If s∈(12,1) then the inequality 2α(sα(1−s) + (1−s)αs)<1

holds.

Proof. Let g1 : [12,1)→Rbe the function defined by the equality:

g1(s) = 2α(sα(1−s) + (1−s)αs).

By differentiation,

g1(s) = 2α[αs(1−s)(sα2−(1−s)α2) + (1−s)α−sα].

Since sα2−(1−s)α2 <0 and (1−s)α−sα<0 for s∈(12,1),shows g1(s)<0 for all s∈(12,1).

Thus the function g1 is strictly decreasing and the inequality g1(s)< g1(1

2) = 1, s∈(1 2,1) follows.

Lemma 2. Ifs∈(12,1) is a fixed number, then the function g2 : [2,3]→R defined by

g2(α) = 2α(sα(1−s) + (1−s)αs) is strictly increasing.

Proof. The expression ofg2 can be rewritten as follows:

g2(α) = 2s(1−s)[(2s)α1+ 2(1−s)α1

].

We get by differentiation:

g2(α) = 2s(1−s)[(2s)α1ln 2s+ 2(1−s)α1

ln 2(1−s) ].

Since 2s∈(1,2),2(1−s)∈(0,1) andα−1∈[1,2],follows that g2(α)≥2s(1−s)[2sln 2s+ 2(1−s) ln 2(1−s)

].

On the other hand for the derivative of the function g3: [12,1)→R, g3(s) = 2sln 2s+ 2(1−s) ln 2(1−s)

the following inequality holds:

g3(s) = ln s

1−s >0, for all s∈(1 2,1).

(3)

Thereforeg3 is strictly increasing andg3(s)> g3(12) = 0 for alls∈(12,1).

Thusg2(α)>0,for allα∈(2,3) and the assertion holds.

Lemma 3. [5]. II.2. Let n be a natural numbern≥2,and ak∈(0,∞), k= 1, n be real numbers, not all equal. The functionh: (0,∞)→R defined by the equality:

h(α) = 1 n

Xn k=1

a

1 α

k

!α

is strictly decreasing.

3. THE MAIN RESULT

Theorem 1. Letα∈(0,3] be a fixed number. For alla, b∈(0,∞), a < b the following inequality holds:

L(a, b)< Qα(a, b). (1)

Proof. Let x= ba,obviously x∈(1,∞).The inequality (1) is equivalent to x−1

lnx <

1 +xα1 2

α

, x∈(1,∞).

We let s= xα1

1+xα1 and this leads to the next equivalent form of (1):

αln s

1−s−[(2s)α− 2(1−s)α

]>0, s∈(1

2,1). (2)

Consequently we have to study the function f1 : [1

2,1)→R, f1(s) =αln s

1−s−[(2s)α− 2(1−s)α

].

We mark out two cases.

We assume first α∈[0,2].Since f1(s) = α

s(1−s)[1−2α sα(1−s) +s(1−s)α ],

Lemma 1 implies f1(s)>0, s∈(12,1).Thus f1 is strictly increasing, and the desired inequality follows: f1(s)> f1(12) = 0, s∈[12,1).

The second case is α∈[2,3].

(4)

The equality f1(s) = s(1αs)[1−g1(s)], s∈(12,1) and Lemma 3 imply, that it is sufficient to prove

f1(s)>0, for all s∈(1 2,1)

in case if α= 3,and then the inequality follows for everyα ∈[2,3].

In case ifα= 3,we have f1(s) = 3

s(1−s)[1−23 s3(1−s) +s(1−s)3 ] and

g1(s) = 23 s3(1−s) +s(1−s)3 .

Since g1(s) = 8(1−2s)3<0, s∈(12,1),it follows that g1 is a decreasing mapping on (12,1).Thus g1(s)< g1(12) = 1 for alls∈(12,1).Hence

f1(s)>0, for all s∈(1

2,1) andα∈[2,3].

Consequentlyf1(s)> f1(12) for s∈(12,1),and (2) holds in this case too.

Remark 1. If α >3 then the inequality

L(a, b)≤Qα(a, b) (1)

does not hold for every 0< a < b.

Proof. We have to prove that (2) does not hold provided α >3.Letg1 be the function defined in the proof of Lemma 1. Sinceg1′′(12) =α(α−3)>0 the continuity of g′′1 implies the existence of a real numberε >0 so that g′′1(s)>0 for all s∈[12,12 +ε).Hence g1 is strictly increasing on [12,12 +ε).

Thusg1(s)> g1(12) = 0 for alls∈(12,12 +ε).Thus we get thatg1 is a strictly increasing mapping on [12,12 +ε) andg1(s)> g1(12) = 1, s∈(12,12 +ε).This leads to

f1(s)<0, s∈(1 2,1

2 +ε) and f1(s)< f1(1

2) = 0, s∈(1 2,1

2 +ε).

Hence (2) cannot be true for every s∈(12,1).

Theorem 2. Ifα∈[32,∞) then the following inequality holds

Qα(a, b)< I(a, b) for all a, b∈(0,∞), a < b. (3)

(5)

Proof. According to Lemma 3 we have to prove (3) only in case if α= 32. Using the notation x= ba the inequality

a23 +b23 2

32

< 1 e

bb aa

b1

a

, a, b∈(0,∞), a < b is equivalent to

x

x−1lnx−3 2ln

1 +x23 2

−1>0, x∈(1,∞). (4) Let

f2 : [1,∞)→R, f2(x) = x

x−1lnx− 3 2ln

1 +x23 2

−1.

By differentiation

f2(x) = 1 (x−1)2

−lnx+x−1−(x−1)2 1 x+x13

.

Let

u: [1∞)→R, u(x) =−lnx+x−1−(x−1)2 1 x+x13. Since

u(x) = (x−1)(x13 −1)3 3x23(x+x13)2

is positive for everyx∈(1,∞),it shows that u is strictly increasing on [1,∞).Therefore we haveu(x)> u(1) = 0, x∈(1,∞).Thus f2(x) is positive for every x∈(1,∞) and so f2 is strictly increasing on [1,∞) which means that the inequality

f2(x)> f2(1) = 0, x∈(1,∞) holds, and this is equivalent to (4).

Remark 2. If 0< α < 32 then the inequality

Qα(a, b)≤I(a, b) (5)

does not hold for all a, b∈(0,∞), a < b.

(6)

Proof. Inequality (5) is equivalent to:

xlnx−α(x−1) ln

1 +xα1 2

−x+ 1≥0, x∈(1,∞). (6) Suppose 0< α < 32.The derivatives of the function f3: [1,∞)→R defined by

f3(x) =xlnx−α(x−1) ln

1 +xα1 2

−x+ 1 are:

f3(x) = lnx−(x−1) xα11

1 +xα1 −αln

1 +x1α 2

,

f3′′(x) = 1−xα1

x(1 +xα1)+ (1−x)(α1 −1)x1α2−xα22 (1 +x1α)2 . Since

xlimց1

f3′′(x)

1−x = 3−2α 4α >0

there exists a positive real number ε >0,so that f3′′(x)<0, x∈(1,1 +ε).

Thusf3 is decreasing on (1,1 +ε) and f3(x)< f3(1) = 0, x∈(1,1 +ε).

Thereforef3 is also decreasing on (1,1 +ε) and it follows that

f3(x)< f3(1) = 0, x∈(1,1 +ε),and this inequality is in contradiction with (6).

Remark 3. If we denotex= ab,then the inequality

L(a, b)≥Qα(a, b). a, b∈(0,∞), a < b (7) is equivalent to

1

lnx− 1 +xα1 2(x−1)α1

!α

≥0, x∈(1,∞).

But

xlim→∞

1

lnx − 1 +xα1 2(x−1)α1

!α

=−1

(7)

for everyα∈(0,∞) fixed number. This means that inequality (7) cannot be true for any α∈(0,∞).

Conclusions

1. The inequalities

L(a, b)< Qα(a, b)< I(a, b), hold for all a, b∈(0,∞), a < b if and only if α∈[32,3].

2. Remark 1 shows that there is no α∈(0,∞) so that

Qα(a, b)≤L(a, b), for all a, b∈(0,∞), a < b.

Remark 4. A more general version of Theorem 1 and Theorem 2 can be found [6].

REFERENCES

[1] Anisiu Valeriu and Anisiu Mira Cristina, Refinement of Some Inequalities for Means, Revue D’Analyse et de Theorie de L’Approximation, Tome 35, No.1, 2006, pp.5-10

[2] Bencze Mih´aly, Open Question no 2283.Octogon Mathematical Magazine, Vol.14. No. 2. October 2006.

[3] Bullen P.S.,Handbook of Means and Their Inequalities,

Series:Mathematics and Its Applications, vol.560, 2nd ed., Kluwer Academic Publishers Group, Dordrecht, 2003

[4] Ivan, M. and Ra¸sa, I., Some Inequalities for Means, Seminar of Functional Equations, Approximation and Convexity, Cluj-Napoca, May 23-29,2000,pp.99-102

[5] P´olya, Gy.,Szeg˝o, G., Aufgaben und Lehrs¨atze aus der Analysis Springer Verlag. 1924

[6] Edward Neuman,A generalization of an inequality of JIA and CAU Journal of Inequalities in Pure and Applied Mathematics,Vol.5,Issue 1,

Hivatkozások

KAPCSOLÓDÓ DOKUMENTUMOK

Ezért a mintegy kóstolóba kapott anyag leget nyert, amin a hágai Állandó hivatal létesítése alapján felesleges volna a tudományos, nemzetközi s az egyéb

Write down the general solution of the

Mih´ aly Bencze OQ.. Determine all convex polygons having integer sides and integer diagonals.. Mih´ aly Bencze OQ. Determine all tetrahedrons having

OQ. Mih´aly Bencze OQ.. We have the following result, if AB = BA then the affirmation is true.. Mih´aly Bencze OQ.. Can be strongened

35 Regulation (EC) No 1906/2006 of the European Pariiament and of the Council laying down the rules for the participation of undertakings, research centres and universities in

We first establish some ˇ Cebyšev inequalities which generalize some results of Audréief [1], Beesack and Peˇcari´c [2], Dunkel [4], Fujimara [5, 6], Isayama [8], and Winckler

Referen- ces [l], [2] present only the four-dimensional solution by means of the electric scalar and vector potential for the case of known electric charge and current

1*.. A granular support IS prepared previously by some means and the solution of the active agent is sprayed on it. Subsequently the product is further ground. Active