JAM Journal of Applied Mathematics 1687-0042 1110-757X Hindawi Publishing Corporation 273653 10.1155/2013/273653 273653 Research Article Optimal Lower Generalized Logarithmic Mean Bound for the Seiffert Mean Song Ying-Qing 1 Qian Wei-Mao 2 Jiang Yun-Liang 3 Chu Yu-Ming 1 Marcellán Francisco J. 1 School of Mathematics and Computation Sciences Hunan City University Yiyang Hunan 413000 China hncu.net 2 School of Distance Education Huzhou Broadcast and TV University Huzhou Zhejiang 313000 China 3 School of Information & Engineering Huzhou Teachers College Huzhou Zhejiang 313000 China hutc.zj.cn 2013 24 3 2013 2013 23 12 2012 26 02 2013 2013 Copyright © 2013 Ying-Qing Song et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We present the greatest value p such that the inequality P(a,b)>Lp(a,b) holds for all a,b>0 with ab, where P(a,b) and Lp(a,b) denote the Seiffert and pth generalized logarithmic means of a and b, respectively.

1. Introduction

For p, the pth generalized logarithmic mean Lp:(0,)2(0,) is defined by (1)Lp(a,b)={[bp+1-ap+1(p+1)(b-a)]1/p,ab,  p0,  -1,1e(bbaa)1/(b-a),ab,  p=0,b-alogb-loga,ab,  p=-1,a,a=b, and the Seiffert mean P:(0,)2(0,)  is defined by (2)P(a,b)={a-b4arctan(a/b)-π,ab,a,a=b.

It is well known that the generalized logarithmic mean Lp(a,b) is continuous and strictly increasing with respect to p for fixed a,b>0 with ab. The special cases of the generalized logarithmic mean are, for example, G(a,b)=ab=L-2(a,b) is the geometric mean, L(a,b)=(b-a)/(logb-loga)=L-1(a,b) is the logarithmic mean, I(a,b)=1/e(bb/aa)1/(b-a)=L0(a,b) is the identric mean, and A(a,b)=(a+b)/2=L1(a,b) is the arithmetic mean. The Seiffert mean P(a,b) can be rewritten as (see [2, equation (2.4)]) (3)P(a,b)={a-b2arcsin[(a-b)/(a+b)],ab,a,a=b.

Recently, the bivariate means have been the subject of intensive research. In particular, many remarkable inequalities and properties for the generalized logarithmic and the Seiffert means can be found in the literature .

In [1, 11], Seiffert proved that the inequalities (4)L-1(a,b)<P(a,b)<L0(a,b)=I(a,b),2πA(a,b)<P(a,b)<A(a,b),P(a,b)>G(a,b)A(a,b)L(a,b),P(a,b)>3A(a,b)G(a,b)A(a,b)+2G(a,b) hold for all a,b>0 with ab.

Sándor  presented the bounds for the Seiffert mean P(a,b) in terms of the arithmetic mean A(a,b) and geometric mean G(a,b) as follows: (5)12A(a,b)+12G(a,b)<P(a,b)<23A(a,b)+13G(a,b) for all a,b>0 with ab.

Hästö  proved that the double inequality Mα(a,b)<P(a,b)<Mβ(a,b) holds for all a,b>0 with ab if and only if αlog2/logπ and β2/3, where Mp(a,b)=((ap+bp)/2)1/p is the pth power mean of a and b.

In , the authors found the greatest value α and least value β such that the double inequality αC(a,b)+(1-α)G(a,b)<P(a,b)<βC(a,b)+(1-β)G(a,b) holds for all a,b>0 with ab, where C(a,b)=(a2+b2)/(a+b) is the contraharmonic mean of a and b.

Motivated by the first inequality in (4), Gao  gave the best possible constants λ and μ such that the double inequality λI(a,b)<P(a,b)<μI(a,b) holds for all a,b>0 with ab.

In , the author solved the following open problem proposed by Long and Chu : what is the smallest p (largest q) such that the inequality αA(a,b)+(1-α)G(a,b)<Lp(a,b)(>Lq(a,b)) holds for α(0,1/2)(α(1/2,1)) and all a,b>0 with ab?

Chu et al.  proved that the double inequality Lp(a,b)<T(a,b)<Lq(a,b) holds for all a,b>0 with ab if and only if p3 and qq0=3.152, where T(a,b)=(a-b)/[2arctan((a-b)/(a+b))] is the second Seiffert mean of a and b, and p0=3.152 is the unique solution of the equation (x+1)-1/x=2/π.

In , the authors answered the question: what are the greatest value p1=p1(q) and the least value p2=p2(q) such that the double inequality Lp1(a,b)<[L(aq,bq)]1/q<Lp2(a,b) holds for any q>0 with q1 and all a,b>0 with ab?

Motivated by the first inequality in (4), it is natural to ask what are the best possible generalized logarithmic mean bounds for the Seiffert mean P(a,b)? It is the aim of this paper to answer this question.

2. Preliminaries

In order to prove our main results, we need two lemmas, which we present in this section.

Lemma 1.

Let the function E:(-1,+)+ be defined with (6)(x+1)-1/x,x(-1,0)(0,+),1e,x=0. Then, E is a continuous and strictly increasing function.

Proof.

From (6), we clearly see that (7)limx0E(x)=1e=E(0),(8)limx-1+E(x)=0,limx+E(x)=1. If x(-1,0)(0,+), then simple computation yields (9)E(x)=(1+x)log(1+x)-xx2(1+x)E(x). If we define (10)F(x)=(1+x)log(1+x)-x, then (11)F(x)=log(1+x). Equation (11) implies that (12)F(x)<0,x(-1,0),F(x)>0,x(0,+). Equations (10) and (12) lead to (13)F(x)>limx0F(x)=0 for x(-1,0)(0,+).

From (9) and (10) together with (13), we clearly see that (14)E(x)>0 for x(-1,0)(0,+).

Therefore, the continuity of E follows from (6) and (7), and the strict monotonicity of E follows from (8), (14) and the continuity of E.

Remark 2.

From Lemma 1, we clearly see that for any fixed λ(0,1), there exists a unique x(-1,) such E(x)=λ. In particular, for λ=1/π, making use of Mathematica software, we get (15)E(-0.241)-1π=0.000166>0,E(-0.242)-1π=-0.000062<0.

Therefore, the unique solution of the equation E(x)=1/π belongs to the interval (-0.242,-0.241).

Lemma 3.

Let p(-0.242,-0.241) and let g:[1,+) be defined with g(t)=pt2p+3+2(p-1)t2p+2-(3p+2)t2p+1-p(2p+1)tp+3+(2p2+p+4)tp+2+(2p2+p+4)tp+1-p(2p+1)tp-(3p+2)t2+2(p-1)t+p. Then, there exists λ(1,+) such that g(t)<0 for t(1,λ) and g(t)>0 for t(λ,+).

Proof.

Let h(t)=t3-pg′′′(t). Further computations lead to (24)h(t)=2p(p+1)(2p+1)(2p+3)tp+3N+8p(p-1)(p+1)(2p+1)tp+2m-2p(2p-1)(2p+1)(3p+2)tp+1m-p(p+1)(p+2)(p+3)m×(2p+1)t3+p(p+1)(p+2)m×(2p2+p+4)t2+p(p-1)m×(p+1)(2p2+p+4)tm-p2(p-2)(p-1)(2p+1),(25)h(1)=0,(26)limt+h(t)=+,(27)h(t)=2p(p+1)(p+3)(2p+1)(2p+3)tp+2m+8p(p-1)(p+1)(p+2)(2p+1)tp+1m-2p(p+1)(2p-1)(2p+1)(3p+2)tpm-3p(p+1)(p+2)(p+3)(2p+1)t2m+2p(p+1)(p+2)(2p2+p+4)t+pm×(p-1)(p+1)(2p2+p+4),(28)h(1)=24p3(p+1)<0,(29)limt+h(t)=+,(30)h′′(t)=2p(p+1)(p+2)(p+3)(2p+1)(2p+3)tp+1n+8p(p-1)(p+1)2(p+2)(2p+1)tpn-2p2(p+1)(2p-1)(2p+1)(3p+2)tp-1n-6p(p+1)(p+2)(p+3)(2p+1)tm+2p(p+1)(p+2)(2p2+p+4),(31)h′′(1)=p3(72p2+156p+84)<0,(32)limt+h′′(t)=+,(33)h′′′(t)=2p(p+1)2(p+2)(p+3)(2p+1)m×(2p+3)tp+8p2(p-1)(p+1)2m×(p+2)(2p+1)tp-1-2p2(p-1)m×(p+1)(2p-1)(2p+1)m×(3p+2)tp-2-6p(p+1)m×(p+2)(p+3)(2p+1),(34)h′′′(1)=p2(112p4+312p3+352p2+192p+40)>0,(35)h(4)(t)=2p2(p+1)2(p+2)(p+3)m×(2p+1)(2p+3)tp-1m+8p2(p-1)2(p+1)2(p+2)(2p+1)tp-2m+2p2(1-p)(2-p)(1-2p)M×(p+1)(2p+1)(3p+2)tp-3>0.

Inequalities (34) and (35) imply that (36)h′′′(t)>0 for t(1,).

From (31) and (32) together with (36), we clearly see that there exists t0(1,) such that h′′(t)<0 for t(1,t0) and h′′(t)>0 for t(t0,). Hence, h is strictly decreasing on (1,t0) and strictly increasing on (t0,).

It follows from (28) and (29) together with the monotonicity of h that there exists t1(1,) such that h is strictly decreasing on (1,t1) and strictly increasing on (t1,).

Equations (25)-(26) and the monotonicity of h lead to the conclusion that there exists t2(1,) such that h(t)<0 for t(1,t2) and h(t)>0 for t(t2,). Therefore, g is strictly decreasing on (1,t2) and strictly increasing on (t2,).

From (22) and (23) together with the monotonicity of g, we clearly see that there exists t3(1,) such that g is strictly decreasing on (1,t3) and strictly increasing on (t3,).

Equations (19)-(20) and the monotonicity of g imply that there exists t4(1,) such that g is strictly decreasing on (1,t4) and strictly increasing on (t4,).

Therefore, Lemma 3 follows from (16) and (17) together with the monotonicity of g.

3. Main Results Theorem 4.

Let E be as in Lemma 1 and let p be the unique solution of the equation E(x)=1/π. Then, for all a,b>0, ab, the inequality (37)Lp(a,b)<P(a,b) holds, and Lp(a,b) is the best possible lower generalized logarithmic mean bound for the Seiffert mean P(a,b).

Proof.

From Remark 2, it follows that (38)p(-0.242,-0.241).

We first prove that the inequality (37) holds. Without loss of generality, we assume that a>b. If t=a/b>1, then, from (1) and (2), we have (39)logLp(a,b)-logP(a,b)mm=1plogtp+1-1(p+1)(t-1)mmm+log(4arctant-π)-log(t-1), If (40)f(t)=1plogtp+1-1(p+1)(t-1)m+log(4arctant-π)-log(t-1), then simple computations lead to (41)limt1f(t)=limt+f(t)=0,(42)f(t)=(p+1)(tp-1)p(tp+1-1)(t-1)(4arctant-π)f1(t), where (43)f1(t)=-4arctant+π+2p(tp+1-1)(t-1)(p+1)t(t+1)(tp-1),limt1f1(t)=0,limt+f1(t)=+,(44)f1(t)=f2(t)(1+p)(t+1)2(tp-1)2t3/2, where (45)f2(t)=pt2p+3+2(p-1)t2p+2-(3p+2)t2p+1m-p(2p+1)tp+3+(2p2+p+4)tp+2m+(2p2+p+4)tp+1-p(2p+1)tpm-(3p+2)t2+2(p-1)t+p.

From (38) and (44)-(45) together with Lemma 3, we clearly see that there exists λ(1,+) such that f1 is strictly decreasing on (1,λ) and strictly increasing on (λ,). Then, (38) and (42)-(43) together with the monotonicity of f1 imply that there exists μ(1,+) such that f is strictly decreasing on (1,μ) and strictly increasing on (μ,+).

Therefore, Lp(a,b)<P(a,b) follows from (39)–(41) and the monotonicity of f.

Next, we prove that Lp(a,b) is the best possible lower generalized logarithmic mean bound for the Seiffert mean P(a,b).

For any 0<ε<-p, from (1) and (2), we get (46)limt+Lp+ε(t,1)P(t,1)=π(p+1+ε)-1/(p+ε). Lemma 1 and (46) lead to (47)limt+Lp+ε(t,1)P(t,1)>1.

Inequality (47) implies that, for 0<ε<-p, there exists T=T(ε)>1 such that Lp+ε(t,1)>P(t,1) for t(T,).

Theorem 5.

L 0 ( a , b ) is the best possible upper generalized logarithmic mean bound for the Seiffert mean P(a,b).

Proof.

For any 0<ε<1 and x>0, from (1) and (2), we have (48)[P(1+x,1)]ε-[L-ε(1+x,1)]εm=f(x)[(1+x)1-ε-1][4arctan1+x-π]ε, where f(x)=[(1+x)1-ε-1]xε-(1-ε)(4arctan1+x-π)εx.

When x0, then making use of the Taylor expansion, we get (49)f(x)=[(1-ε)x-12ε(1-ε)x2mm+16ε(1-ε)(1+ε)x3+o(x3)]xεm-(1-ε)[x-12x2+724x3+o(x3)]εx=124ε2(1-ε)x3+ε+o(x3+ε).

Equations (48) and (49) imply that for any 0<ε<1, there exists δ=δ(ε)>0 such that (50)P(1+x,1)>L-ε(1+x,1) for x(0,δ).

Therefore, Theorem 5 follows from inequalities (4) and (50).

Acknowledgments

This paper was supported by the Natural Science Foundation of China under Grant 61173123, the Natural Science Foundation of Zhejiang Province under Grants Z1110551 and LY12F02012, and the Natural Science Foundation of Huzhou City under Grant 2012YZ06.

Seiffert H. J. Problem 887 Nieuw Archief voor Wiskunde (4) 1993 11 2 176 Neuman E. Sándor J. On the Schwab-Borchardt mean Mathematica Pannonica 2003 14 2 253 266 MR2079318 ZBL1053.26015 Costin I. Toader G. Optimal evaluations of some Seiffert-type means by power means Applied Mathematics and Computation 2013 219 9 4745 4754 10.1016/j.amc.2012.10.091 MR3001522 Jiang W. D. Some sharp inequalities involving reciprocals of the Seiffert and other means Journal of Mathematical Inequalities 2012 6 4 593 599 Qian W. M. Long B. Y. Sharp bounds by the generalized logarithmic mean for the geometric weighted mean of the geometric and harmonic means Journal of Applied Mathematics 2012 2012 8 480689 10.1155/2012/480689 MR2923368 ZBL1244.26048 Shi H. N. Wu S. H. Refinement of an inequality for the generalized logarithmic mean Chinese Quarterly Journal of Mathematics 2008 23 4 594 599 MR2510382 ZBL1199.26118 Chen C. P. The monotonicity of the ratio between generalized logarithmic means Journal of Mathematical Analysis and Applications 2008 345 1 86 89 10.1016/j.jmaa.2008.03.071 MR2422636 ZBL1160.26012 Qi F. Chen S. X. Chen C. P. Monotonicity of ratio between the generalized logarithmic means Mathematical Inequalities & Applications 2007 10 3 559 564 10.7153/mia-10-52 MR2339549 ZBL1127.26021 Toader G. Seiffert type means Nieuw Archief voor Wiskunde (4) 1999 17 3 379 382 MR1743849 Pearce C. E. M. Pečarić J. Šimić V. On weighted generalized logarithmic means Houston Journal of Mathematics 1998 24 3 459 465 MR1686617 ZBL0966.26016 Seiffert H. J. Ungleichungen für einen bestimmten Mittelwert Nieuw Archief voor Wiskunde (4) 1995 13 2 195 198 MR1345569 ZBL0830.26008 Pearce C. E. M. Pečarić J. Some theorems of Jensen type for generalized logarithmic means Revue Roumaine de Mathématiques Pures et Appliquées 1995 40 9-10 789 795 MR1405116 ZBL0863.26021 Stolarsky K. B. The power and generalized logarithmic means The American Mathematical Monthly 1980 87 7 545 548 10.2307/2321420 MR600913 ZBL0455.26008 Sándor J. On certain inequalities for means III Archiv der Mathematik 2001 76 1 34 40 10.1007/s000130050539 MR1808740 ZBL0976.26015 Hästö P. A. Optimal inequalities between Seiffert's mean and power means Mathematical Inequalities & Applications 2004 7 1 47 53 10.7153/mia-07-06 MR2035348 ZBL1049.26006 Liu H. Meng X. J. The optimal convex combination bounds for Seiffert's mean Journal of Inequalities and Applications 2011 2011 9 686834 10.1155/2011/686834 MR2781425 ZBL1221.26037 Gao S. Q. Inequalities for the Seiffert's means in terms of the identric mean Journal of Mathematical Sciences: Advances and Applications 2011 10 1-2 23 31 MR2906381 Matejíčka L. Proof of one optimal inequality for generalized logarithmic, arithmetic, and geometric means Journal of Inequalities and Applications 2010 2010 5 902432 10.1155/2010/902432 MR2738681 ZBL1210.26033 Long B. Y. Chu Y. M. Optimal inequalities for generalized logarithmic, arithmetic, and geometric means Journal of Inequalities and Applications 2010 2010 10 806825 10.1155/2010/806825 MR2600201 ZBL1187.26015 Chu Y. M. Wang M. K. Wang G. D. The optimal generalized logarithmic mean bounds for Seiffert's mean Acta Mathematica Scientia B 2012 32 4 1619 1626 Jiang Y. L. Chu Y. M. Long B. Y. An optimal double inequality between logarithmic and generalized logarithmic means Journal of Applied Analysis. In press