Abstract

The following inequality for 0<𝑝<1 and π‘Žπ‘›β‰₯0 originates from a study of Hardy, Littlewood, and PΓ³lya: βˆ‘βˆžπ‘›=1βˆ‘((1/𝑛)βˆžπ‘˜=π‘›π‘Žπ‘˜)𝑝β‰₯π‘π‘βˆ‘βˆžπ‘›=1π‘Žπ‘π‘›. Levin and Stečkin proved the previous inequality with the best constant 𝑐𝑝=(𝑝/(1βˆ’π‘))𝑝 for 0<𝑝≀1/3. In this paper, we extend the result of Levin and Stečkin to 0<𝑝≀0.346.

1. Introduction

Let 𝑝>1, and 𝑙𝑝 be the Banach space of all complex sequences 𝐚=(π‘Žπ‘›)𝑛β‰₯1. The celebrated Hardy's inequality [1, Theorem 326] asserts that for 𝑝>1 and any πšβˆˆπ‘™π‘βˆžξ“π‘›=1|||||1π‘›π‘›ξ“π‘˜=1π‘Žπ‘˜|||||π‘β‰€ξ‚΅π‘ξ‚Άπ‘βˆ’1π‘βˆžξ“π‘˜=1||π‘Žπ‘˜||𝑝.(1.1)

As an analogue of Hardy’s inequality, Theorem 345 of [1] asserts that the following inequality holds for 0<𝑝<1 and π‘Žπ‘›β‰₯0 with 𝑐𝑝=𝑝𝑝:βˆžξ“π‘›=11π‘›βˆžξ“π‘˜=π‘›π‘Žπ‘˜ξƒͺ𝑝β‰₯π‘π‘βˆžξ“π‘›=1π‘Žπ‘π‘›.(1.2) It is noted in [1] that the constant 𝑐𝑝=𝑝𝑝 may not be best possible, and a better constant was indeed obtained by Levin and Stečkin [2, Theorem 61]. Their result is more general as they proved, among other things, the following inequality [2, Theorem 62], valid for 0<π‘Ÿβ‰€π‘β‰€1/3 or 1/3<𝑝<1,π‘Ÿβ‰€(1βˆ’π‘)2/(1+𝑝) with π‘Žπ‘›β‰₯0:βˆžξ“π‘›=11π‘›π‘Ÿξƒ©βˆžξ“π‘˜=π‘›π‘Žπ‘˜ξƒͺ𝑝β‰₯𝑝1βˆ’π‘Ÿπ‘βˆžξ“π‘›=1π‘Žπ‘π‘›π‘›π‘Ÿβˆ’π‘.(1.3) We note here that the constant (𝑝/(1βˆ’π‘Ÿ))𝑝 is best possible, as shown in [2] by setting π‘Žπ‘›=π‘›βˆ’1βˆ’(1βˆ’π‘Ÿ)/π‘βˆ’πœ– and letting πœ–β†’0+. This implies inequality (1.2) for 0<𝑝≀1/3 with the best possible constant 𝑐𝑝=(𝑝/(1βˆ’π‘))𝑝. On the other hand, it is also easy to see that inequality (1.2) fails to hold with 𝑐𝑝=(𝑝/(1βˆ’π‘))𝑝 for 𝑝β‰₯1/2. The point is that in these cases 𝑝/(1βˆ’π‘)β‰₯1 so one can easily construct counterexamples.

A simpler proof of Levin and Stečkin's result (for 0<π‘Ÿ=𝑝≀1/3) is given in [3]. It is also pointed out there that, using a different approach, one may be able to extend their result to 𝑝 slightly larger than 1/3; an example is given for 𝑝=0.34. The calculation however is more involved, and therefore it is desirable to have a simpler approach. For this, we let π‘ž be the number defined by 1/𝑝+1/π‘ž=1 and note that by the duality principle (see [4, Lemma 2], but note that our situation is slightly different since we have 0<𝑝<1 with an reversed inequality), the case 0<π‘Ÿ<1,0<𝑝<1 of inequality (1.3) is equivalent to the following one for π‘Žπ‘›>0:βˆžξ“π‘›=1𝑛𝑛(π‘Ÿβˆ’π‘)/π‘ξ“π‘˜=1π‘Žπ‘˜π‘˜π‘Ÿ/𝑝ξƒͺπ‘žβ‰€ξ‚€π‘ξ‚1βˆ’π‘Ÿπ‘žβˆžξ“π‘›=1π‘Žπ‘žπ‘›.(1.4) The above inequality can be regarded as an analogue of a result of Knopp [5, 6], which asserts that Hardy's inequality (1.1) is still valid for 𝑝<0 if we assume π‘Žπ‘›>0. We may also regard inequality (1.4) as an inequality concerning the factorable matrix with entries 𝑛(π‘Ÿβˆ’π‘)/π‘π‘˜βˆ’π‘Ÿ/𝑝 for π‘˜β‰€π‘› and 0 otherwise. Here we recall that a matrix 𝐴=(π‘Žπ‘›π‘˜) is factorable if it is a lower triangular matrix with π‘Žπ‘›π‘˜=π‘Žπ‘›π‘π‘˜ for 1β‰€π‘˜β‰€π‘›. We note that the approach in [7] for the 𝑙𝑝 norms of weighted mean matrices can also be easily adopted to treat the 𝑙𝑝 norms of factorable matrices, and it is our goal in this paper to use this similar approach to extend the result of Levin and Stečkin. Our main result is the following.

Theorem 1. Inequality (1.2) holds with the best possible constant 𝑐𝑝=(𝑝/(1βˆ’π‘))𝑝 for any 1/3<𝑝<1/2 satisfying 2𝑝/(1βˆ’π‘)1βˆ’π‘π‘ξ‚Ά1/(1βˆ’π‘)βˆ’1βˆ’π‘π‘ξƒͺβˆ’ξ‚΅1+3βˆ’1/𝑝2ξ‚Ά1/(1βˆ’π‘)β‰₯0.(1.5) In particular, inequality (1.2) holds for 0<𝑝≀0.346.

It readily follows from Theorem 1 and our discussions above that we have the following dual version of Theorem 1.

Corollary 1. Inequality (1.4) holds with π‘Ÿ=𝑝 for any 1/3<𝑝<1/2 satisfying (1.5) and the constant is best possible. In particular, inequality (1.4) holds with π‘Ÿ=𝑝 for 0<𝑝≀0.346.

An alternative proof of Theorem 1 is given in Section 3, via an approach using the duality principle. In Section 4, we will study some inequalities which can be regarded as generalizations of (1.2). Motivations for considerations for such inequalities come both from their integral analogues as well as from their counterparts in the 𝑙𝑝 spaces. As an example, we consider the following inequality for 0<𝑝<1,0<𝛼<1/𝑝:βˆžξ“π‘›=1ξƒ©βˆžξ“π‘˜=π‘›π›Όπ‘˜π›Όβˆ’1π‘›π›Όπ‘Žπ‘˜ξƒͺ𝑝β‰₯𝛼𝑝1βˆ’π›Όπ‘π‘βˆžξ“π‘›=1π‘Žπ‘π‘›.(1.6) As in the case of (1.2), the above inequality does not hold for all 0<𝑝<1,0<𝛼<1/𝑝. In Section 4, we will however prove a result concerning the validity of (1.6) that can be regarded as an analogue to that of Levin and Stečkin's concerning the validity of (1.2).

Inequality (1.6) is motivated partially by integral analogues of (1.2), as we will explain in Section 4. It is also motivated by the following inequality for 𝑝>1,𝛼𝑝>1, π‘Žπ‘›β‰₯0:βˆžξ“π‘›=1ξƒ©π‘›ξ“π‘˜=1π›Όπ‘˜π›Όβˆ’1π‘›π›Όπ‘Žπ‘˜ξƒͺπ‘β‰€ξ‚΅π›Όπ‘ξ‚Άπ›Όπ‘βˆ’1π‘βˆžξ“π‘›=1π‘Žπ‘π‘›.(1.7) The above inequality is in turn motivated by the following inequality:βˆžξ“π‘›=1ξƒ©π‘›ξ“π‘˜=1π‘˜π›Όβˆ’1βˆ‘π‘›π‘–=1π‘–π›Όβˆ’1π‘Žπ‘˜ξƒͺπ‘β‰€ξ‚΅π›Όπ‘ξ‚Άπ›Όπ‘βˆ’1π‘βˆžξ“π‘›=1π‘Žπ‘π‘›.(1.8) Inequality (1.8) was first suggested by Bennett [8, pages 40-41]; see [9] and the references therein for recent progress on this. We point out here that it is easy to see that inequality (1.7) implies (1.8) when 𝛼>1; hence, it is interesting to know that, for which values of𝛼's, inequality (1.7) is valid. We first note that, on setting π‘Ž1=1 and π‘Žπ‘›=0 for 𝑛β‰₯2 in (1.7) that it is impossible for it to hold when 𝛼 is large for fixed 𝑝. On the other hand, when 𝛼=1, inequality (1.7) becomes Hardy's inequality, and hence one may expect it to hold for 𝛼 close to 1, and we will establish such a result in Section 5.

2. Proof of Theorem 1

First we need a lemma.

Lemma 1. The following inequality holds for 0≀𝑦≀1 and 1/2<𝑑<1: 𝑦1+2𝑑1+π‘‘βˆ’(1+𝑦)βˆ’π‘‘ξ‚΅1+(2π‘‘βˆ’1)𝑦2𝑑1+π‘‘βˆ’π‘¦π‘‘β‰₯0.(2.1)

Proof. We set π‘₯=𝑦/2𝑑 so that 0≀π‘₯≀1, and we recast the above inequality as 𝑓(π‘₯,𝑑)∢=(1+π‘₯)1+π‘‘βˆ’(1+2𝑑π‘₯)βˆ’π‘‘(1+(2π‘‘βˆ’1)π‘₯)1+π‘‘βˆ’2π‘₯β‰₯0.(2.2) Direct calculation shows that 𝑓(0,𝑑)=(πœ•π‘“/πœ•π‘₯)(0,𝑑)=0 and πœ•2π‘“πœ•π‘₯2(π‘₯,𝑑)=𝑑(1+𝑑)(1+π‘₯)π‘‘βˆ’1ξ€·1βˆ’(1+2𝑑π‘₯)βˆ’π‘‘βˆ’2(1+(2π‘‘βˆ’1)π‘₯)π‘‘βˆ’1(1+π‘₯)1βˆ’π‘‘ξ€ΈβˆΆ=𝑑(1+𝑑)(1+π‘₯)π‘‘βˆ’1𝑔(π‘₯,𝑑).(2.3) Note that πœ•π‘”πœ•π‘₯(π‘₯,𝑑)=(1+2𝑑π‘₯)βˆ’π‘‘βˆ’3(1+(2π‘‘βˆ’1)π‘₯)π‘‘βˆ’2(1+π‘₯)βˆ’π‘‘Γ—ξ€·2(4π‘‘βˆ’1)+4𝑑(4π‘‘βˆ’1)π‘₯+2𝑑(2π‘‘βˆ’1)(𝑑+2)π‘₯2ξ€Έβ‰₯0.(2.4) As 𝑔(0,𝑑)=0, it follows that 𝑔(π‘₯,𝑑)β‰₯0 for 0≀π‘₯≀1 which in turn implies the assertion of the lemma.

We now describe a general approach towards establishing inequality (1.3) for 0<π‘Ÿ<1,0<𝑝<1. A modification from the approach in Section 3 of [3] shows that, in order for (1.3) to hold for any given 𝑝 with 𝑐𝑝,π‘Ÿ(=(𝑝/(1βˆ’π‘Ÿ))𝑝), it suffices to find a sequence 𝐰 of positive terms for each 0<π‘Ÿ<1 and 0<𝑝<1, such that for any integer 𝑛β‰₯1𝑛(π‘βˆ’π‘Ÿ)/(1βˆ’π‘)𝑀1+β‹―+π‘€π‘›ξ€Έβˆ’1/(1βˆ’π‘)β‰€π‘βˆ’1/(1βˆ’π‘)𝑝,π‘Ÿξƒ©π‘€π‘›βˆ’1/(1βˆ’π‘)π‘›π‘Ÿ/(1βˆ’π‘)βˆ’π‘€βˆ’1/(1βˆ’π‘)𝑛+1(𝑛+1)π‘Ÿ/(1βˆ’π‘)ξƒͺ.(2.5) We note here that if we study the equivalent inequality (1.4) instead, then we can also obtain the above inequality from inequality (2.2) of [3], on setting Λ𝑛=π‘›βˆ’(π‘Ÿβˆ’π‘)/𝑝,πœ†π‘›=π‘›βˆ’π‘Ÿ/𝑝 there. For the moment, we assume that 𝑐𝑝,π‘Ÿ is an arbitrary fixed positive number, and, on setting π‘π‘›π‘βˆ’1=𝑀𝑛/𝑀𝑛+1, we can recast the above inequality as ξƒ©π‘›ξ“π‘›π‘˜=1𝑖=π‘˜π‘π‘–π‘βˆ’1ξƒͺβˆ’1/(1βˆ’π‘)β‰€π‘βˆ’1/(1βˆ’π‘)𝑝,π‘Ÿπ‘›(π‘Ÿβˆ’π‘)/(1βˆ’π‘)ξ‚΅π‘π‘›π‘›π‘Ÿ/(1βˆ’π‘)βˆ’1(𝑛+1)π‘Ÿ/(1βˆ’π‘)ξ‚Ά.(2.6) The choice of 𝑏𝑛 in Section 3 of [3] suggests that, for optimal choices of the 𝑏𝑛's, we may have asymptotically π‘π‘›βˆΌ1+𝑐/𝑛 as 𝑛→+∞ for some positive constant 𝑐 (depending on 𝑝). This observation implies that 𝑛1/(1βˆ’π‘) times the right-hand side expression above should be asymptotically a constant. To take the advantage of possible contributions of higher-order terms, we now further recast the above inequality as1𝑛+π‘Žπ‘›ξ“π‘›π‘˜=1𝑖=π‘˜π‘π‘–π‘βˆ’1ξƒͺβˆ’1/(1βˆ’π‘)β‰€π‘βˆ’1/(1βˆ’π‘)𝑝,π‘Ÿπ‘›(π‘Ÿβˆ’π‘)/(1βˆ’π‘)(𝑛+π‘Ž)1/(1βˆ’π‘)ξ‚΅π‘π‘›π‘›π‘Ÿ/(1βˆ’π‘)βˆ’1(𝑛+1)π‘Ÿ/(1βˆ’π‘)ξ‚Ά,(2.7) where π‘Ž is a constant (may depend on 𝑝) to be chosen later. It will also be clear from our arguments below that the choice of π‘Ž will not affect the asymptotic behavior of 𝑏𝑛 to the first order of magnitude. We now choose 𝑏𝑛 so that𝑛(π‘Ÿβˆ’π‘)/(1βˆ’π‘)(𝑛+π‘Ž)1/(1βˆ’π‘)ξ‚΅π‘π‘›π‘›π‘Ÿ/(1βˆ’π‘)βˆ’1(𝑛+1)π‘Ÿ/(1βˆ’π‘)ξ‚Ά=π‘βˆ’π›Ό/(1βˆ’π‘)𝑝,π‘Ÿ,(2.8) where 𝛼 is a parameter to be chosen later. This implies that 𝑏𝑛=π‘βˆ’π›Ό/(1βˆ’π‘)𝑝,π‘Ÿπ‘›π‘/(1βˆ’π‘)(𝑛+π‘Ž)1/(1βˆ’π‘)+π‘›π‘Ÿ/(1βˆ’π‘)(𝑛+1)π‘Ÿ/(1βˆ’π‘).(2.9) For the so-chosen 𝑏𝑛's, inequality (2.7) becomesπ‘›ξ“π‘›π‘˜=1𝑖=π‘˜π‘π‘–π‘βˆ’1β‰₯(𝑛+π‘Ž)𝑐1+𝛼𝑝,π‘Ÿ.(2.10) We first assume that the above inequality holds for 𝑛=1. Then induction shows that it holds for all 𝑛 as long as 𝑏𝑛1βˆ’π‘β‰€π‘›+π‘Ž+π‘βˆ’(1+𝛼)𝑝,π‘Ÿβˆ’1.𝑛+π‘Ž(2.11) Taking into account the value of 𝑏𝑛, the above becomes (for 0≀𝑦≀1 with 𝑦=1/𝑛)ξ‚€ξ‚€1+π‘Ž+π‘βˆ’(1+𝛼)𝑝,π‘Ÿξ‚π‘¦ξ‚βˆ’11/(1βˆ’π‘)βˆ’(1+𝑦)βˆ’π‘Ÿ/(1βˆ’π‘)(1+π‘Žπ‘¦)1/(1βˆ’π‘)βˆ’π‘βˆ’π›Ό/(1βˆ’π‘)𝑝,π‘Ÿπ‘¦β‰₯0.(2.12) The first-order term of the Taylor expansion of the left-hand side expression above implies that it is necessary to have π‘βˆ’(1+𝛼)𝑝,π‘Ÿβˆ’(1βˆ’π‘)π‘βˆ’π›Ό/(1βˆ’π‘)𝑝,π‘Ÿ+π‘Ÿβˆ’1β‰₯0.(2.13) For fixed 𝑐𝑝,π‘Ÿ, the left-hand side expression above is maximized when 𝛼=1/π‘βˆ’1 with value π‘π‘βˆ’1/𝑝𝑝,π‘Ÿ+π‘Ÿβˆ’1. This suggests for us to take 𝑐𝑝,π‘Ÿ=(𝑝/(1βˆ’π‘Ÿ))𝑝. From now on we fix 𝑐𝑝,π‘Ÿ=(𝑝/(1βˆ’π‘Ÿ))𝑝 and note that in this case (2.12) becomesξ‚΅ξ‚΅1+π‘Ž+1βˆ’π‘Ÿπ‘ξ‚Άπ‘¦ξ‚Άβˆ’11/(1βˆ’π‘)βˆ’(1+𝑦)βˆ’π‘Ÿ/(1βˆ’π‘)(1+π‘Žπ‘¦)1/(1βˆ’π‘)βˆ’1βˆ’π‘Ÿπ‘π‘¦β‰₯0.(2.14)

We note that the choice of π‘Ž=0 in (2.14) with π‘Ÿ=𝑝 reduces to that considered in Section 3 of [3] (in which case the case 𝑛=1 of (2.10) is also included in (2.14)). Moreover, with π‘Ž=0 in the above inequality and following the treatment in Section 3 of [3], one is able to improve some cases of the previously mentioned result of Levin and Stečkin concerning the validity of (1.3). We will postpone the discussion of this to the next section and focus now on the proof of Theorem 1. Since the cases 0<𝑝≀1/3 of the assertion of the theorem are known, we may assume 1/3<𝑝<1/2 from now on. In this case we set π‘Ÿ=𝑝 in (2.14), and Taylor expansion shows that it is necessary to have π‘Žβ‰₯(3βˆ’1/𝑝)/2 in order for inequality (2.14) to hold. We now take π‘Ž=(3βˆ’1/𝑝)/2 and write 𝑑=𝑝/(1βˆ’π‘) to see that inequality (2.14) is reduced to (2.1) and Lemma 1 now implies that inequality (2.14) holds in this case. Inequality (1.2) with the best possible constant 𝑐𝑝=(𝑝/(1βˆ’π‘))𝑝 thus follows for any 1/3<𝑝<1/2 as long as the case 𝑛=1 of (2.10) is satisfied, which is just inequality (1.5), and this proves the first assertion of Theorem 1.

For the second assertion, we note that inequality (1.5) can be rewritten as2π‘‘π‘‘ξ€·π‘‘βˆ’π‘‘ξ€Έβˆ’1β‰₯(1+π‘Ž)1/(1βˆ’π‘),(2.15) where 𝑑 is defined as above. Note that 1/2<𝑑<1 for 1/3<𝑝<1/2 and both 2𝑑/𝑑 and π‘‘βˆ’π‘‘βˆ’1 are decreasing functions of 𝑑. It follows that the left-hand side expression of (2.15) is a decreasing function of 𝑝. Note also that for fixed π‘Ž, the right-hand side expression of (2.15) is an increasing function of 𝑝<1. As π‘Ž=(3βˆ’1/𝑝)/2 in our case, it follows that one just needs to check the above inequality for 𝑝=0.346 and the assertion of the theorem now follows easily.

We remark here that, in the proof of Theorem 1, instead of choosing 𝑏𝑛 to satisfy (2.8) (with π‘Ÿ=𝑝 and 𝑐𝑝,𝑝=(𝑝/(1βˆ’π‘))𝑝 there), we can choose 𝑏𝑛 for 𝑛β‰₯2 so that (𝑛+𝑐)1/(1βˆ’π‘)1𝑛𝑝/(1βˆ’π‘)βˆ’1(𝑛+1)𝑝/(1βˆ’π‘)𝑏𝑛ξƒͺ=(1βˆ’π‘)𝑝.(2.16) Moreover, note that we can also rewrite (2.7) for 𝑛β‰₯2 as (with π‘Ž replaced by 𝑐 and π‘Ÿ=𝑝, 𝑐𝑝,𝑝=(𝑝/(1βˆ’π‘))𝑝) 1𝑛+π‘π‘›βˆ’1ξ“π‘˜=1π‘›βˆ’1𝑖=π‘˜π‘π‘–π‘βˆ’1+1ξƒͺξƒͺβˆ’1/(1βˆ’π‘)≀1βˆ’π‘π‘ξ‚Άπ‘/(1βˆ’π‘)(𝑛+𝑐)1/(1βˆ’π‘)1𝑛𝑝/(1βˆ’π‘)βˆ’1(𝑛+1)𝑝/(1βˆ’π‘)𝑏𝑛ξƒͺ.(2.17) If we further choose 𝑏1 so that ξ‚΅1=1βˆ’π‘π‘ξ‚Άπ‘/(1βˆ’π‘)11βˆ’2𝑝/(1βˆ’π‘)𝑏1ξƒͺ,(2.18) then, repeating the same process as in the proof of Theorem 1, we find that the induction part (with 𝑐=(1/π‘βˆ’1)/2 here) leads back to inequality (2.14) (with π‘Ÿ=𝑝 and π‘Ž=(3βˆ’1/𝑝)/2 there) while the initial case (corresponding to 𝑛=2 here) is just (2.15), so this approach gives another proof of Theorem 1.

We end this section by pointing out the relation between the treatment in Sections 3 and 4 in [3] on inequality (1.2). We note that it is shown in Section 3 of [3] that, for any 𝑁β‰₯1 and any positive sequence 𝐰, we have𝑁𝑛=1π‘Žπ‘π‘›β‰€π‘ξ“π‘›=1π‘€π‘›βŽ›βŽœβŽœβŽπ‘ξ“π‘˜=π‘›ξƒ©π‘˜ξ“π‘–=1𝑀𝑖ξƒͺβˆ’1/(1βˆ’π‘)⎞⎟⎟⎠1βˆ’π‘ξƒ©π‘ξ“π‘˜=π‘›π‘Žπ‘˜ξƒͺ𝑝.(2.19) We now use 𝑀𝑛=βˆ‘π‘›π‘˜=1π‘€π‘˜βˆ’βˆ‘π‘›βˆ’1π‘˜=1π‘€π‘˜ and set (with πœˆπ‘=0) πœˆπ‘›=βˆ‘π‘π‘˜=𝑛+1ξ‚€βˆ‘π‘˜π‘–=1𝑀𝑖1/(π‘βˆ’1)ξ€·βˆ‘π‘›π‘–=1𝑀𝑖1/(π‘βˆ’1)(2.20) to see that inequality (2.19) leads to (with 𝜈0=0) 𝑁𝑛=1π‘Žπ‘π‘›β‰€π‘ξ“π‘›=1ξ‚€ξ€·1+πœˆπ‘›ξ€Έ1βˆ’π‘βˆ’πœˆ1βˆ’π‘π‘›βˆ’1ξ‚ξƒ©π‘ξ“π‘˜=π‘›π‘Žπ‘˜ξƒͺ𝑝.(2.21) The above inequality is essentially what is used in Section 4 of [3].

3. An Alternative Proof of Theorem 1

In this section we give an alternative proof of Theorem 1, using the following.

Lemma 2 (see, Lemma 2.4 [10]). Let {πœ†π‘–}βˆžπ‘–β‰₯1,{π‘Žπ‘–}βˆžπ‘–β‰₯1 be two sequences of positive real numbers, and let 𝑆𝑛=βˆ‘π‘›π‘–=1πœ†π‘–π‘Žπ‘–. Let 0≠𝑝<1 be fixed and let {πœ‡i}βˆžπ‘–β‰₯1,{πœ‚π‘–}βˆžπ‘–β‰₯1 be two positive sequences of real numbers such that πœ‡π‘–β‰€πœ‚π‘– for 0<𝑝<1 and πœ‡π‘–β‰₯πœ‚π‘– for 𝑝<0, then for 𝑛β‰₯2π‘›βˆ’1𝑖=2ξ‚€πœ‡π‘–βˆ’ξ€·πœ‡π‘žπ‘–+1βˆ’πœ‚π‘žπ‘–+1ξ€Έ1/π‘žξ‚π‘†π‘–1/𝑝+πœ‡π‘›π‘†π‘›1/π‘β‰€ξ€·πœ‡π‘ž2βˆ’πœ‚π‘ž2ξ€Έ1/π‘žπœ†11/π‘π‘Ž11/𝑝+𝑛𝑖=2πœ‚π‘–πœ†π‘–1/π‘π‘Žπ‘–1/𝑝.(3.1)

Following the treatment in Section 4 of [3], on first setting πœ‚π‘–=πœ†π‘–βˆ’1/𝑝, then a change of variables: πœ‡π‘–β†¦πœ‡π‘–πœ‚π‘– followed by setting πœ‡π‘žπ‘–βˆ’1=πœˆπ‘– and lastly a further change of variable: 𝑝↦1/𝑝, we can transform inequality (3.1) to the following inequality (with 𝜈1=0 here): π‘›βˆ’1𝑖=11+πœˆπ‘–ξ€Έ1βˆ’π‘πœ†π‘π‘–βˆ’πœˆ1βˆ’π‘π‘–+1πœ†π‘π‘–+1ξƒͺ𝑆𝑝𝑖+ξ€·1+πœˆπ‘›ξ€Έ1βˆ’π‘πœ†π‘π‘›π‘†π‘π‘›β‰€π‘›ξ“π‘–=1π‘Žπ‘π‘–.(3.2) Here the πœˆπ‘–'s are arbitrary nonnegative real numbers for 2≀𝑖≀𝑛. On setting πœˆπ‘›+1 to be any non-negative real number, we deduce immediately from the above inequality the following:𝑛𝑖=11+πœˆπ‘–ξ€Έ1βˆ’π‘πœ†π‘π‘–βˆ’πœˆ1βˆ’π‘π‘–+1πœ†π‘π‘–+1ξƒͺ𝑆𝑝𝑖≀𝑛𝑖=1π‘Žπ‘π‘–.(3.3)

Now we consider establishing inequality (1.3) for 0<π‘Ÿ<1,1/3<𝑝<1/2 in general, and, as has been pointed out in Section 1, we know this is equivalent to establishing inequality (1.4). Now, in order to establish inequality (1.4), it suffices to consider the cases of (1.4) with the infinite summations replaced by any finite summations, say from 1 to 𝑁β‰₯1 there. We now set 𝑛=𝑁,𝑝=π‘ž,πœ†π‘–=π‘–βˆ’π‘Ÿ/𝑝 in inequality (3.3) to recast it as (with 𝜈1=0, 𝑆𝑛=βˆ‘π‘›π‘˜=1π‘˜βˆ’π‘Ÿ/π‘π‘Žπ‘˜ here) 𝑁𝑛=11+πœˆπ‘›ξ€Έ1/(1βˆ’π‘)π‘›π‘Ÿ/(1βˆ’π‘)βˆ’πœˆ1/(1βˆ’π‘)𝑛+1(𝑛+1)π‘Ÿ/(1βˆ’π‘)ξƒͺπ‘†π‘žπ‘›β‰€π‘ξ“π‘›=1π‘Žπ‘žπ‘›.(3.4) Comparing the above inequality with (1.4), we see that inequality (1.4) holds as long as we can find non-negative πœˆπ‘›'s (with 𝜈1=0) such thatξ€·1+πœˆπ‘›ξ€Έ1/(1βˆ’π‘)π‘›π‘Ÿ/(1βˆ’π‘)βˆ’πœˆ1/(1βˆ’π‘)𝑛+1(𝑛+1)π‘Ÿ/(1βˆ’π‘)β‰₯𝑛(π‘βˆ’π‘Ÿ)/(1βˆ’π‘)𝑝1βˆ’π‘Ÿπ‘/(1βˆ’π‘).(3.5) Now, on setting for 𝑛β‰₯2, πœˆπ‘›=𝑛+π‘Žβˆ’1,(1βˆ’π‘Ÿ)/𝑝(3.6) and 𝑦=1/𝑛, we see easily that inequality (3.5) can be transformed into inequality (2.14). In the case of π‘Ÿ=𝑝, we further set π‘Ž=(3βˆ’1/𝑝)/2 to see that the validity of (2.14) established for this case in Section 2 ensures the validity of (3.5) for 𝑛β‰₯2. Moreover, with the above chosen 𝜈2 with π‘Ÿ=𝑝 and π‘Ž=(3βˆ’1/𝑝)/2, the 𝑛=1 case of (3.5) is easily seen to be equivalent to inequality (1.5), and hence this provides an alternative proof of Theorem 1.

4. A Generalization of Theorem 1

Let 0<𝑝<1,𝛼<1/𝑝, and let 𝑓(π‘₯) be a non-negative function. We note the following identity:ξ€œβˆž0ξ‚΅1π‘₯π›Όξ€œβˆžπ‘₯𝑓(𝑑)π‘‘π›Όβˆ’1𝑑𝑑𝑝𝑝𝑑π‘₯=ξ‚Άξ€œ1βˆ’π›Όπ‘βˆž0ξ‚΅1π‘₯π›Όξ€œβˆžπ‘₯𝑓(𝑑)π‘‘π›Όβˆ’1ξ‚Άπ‘‘π‘‘π‘βˆ’1𝑓(π‘₯)𝑑π‘₯.(4.1) In the above expression, we assume, 𝑓 is taken so that all the integrals converge. The case of 𝛼=1 is given in the proof of Theorem 337 of [1], and the general case is obtained by some changes of variables. As in the proof of Theorem 337 of [1], we then deduce the following inequality (with the same assumptions as above): ξ€œβˆž0ξ‚΅1π‘₯π›Όξ€œβˆžπ‘₯𝑓(𝑑)π‘‘π›Όβˆ’1𝑑𝑑𝑝𝑝𝑑π‘₯β‰₯ξ‚Ά1βˆ’π›Όπ‘π‘ξ€œβˆž0𝑓𝑝(π‘₯)𝑑π‘₯.(4.2) The above inequality can also be deduced from Theorem 347 of [1] (see also [11, equation (2.4)]). Following the way how Theorem 338 is deduced from Theorem 337 of [1], we deduce similarly from (4.1) the following inequality for 0<𝑝<1,0<𝛼<1/𝑝, and π‘Žπ‘›β‰₯0: βˆžξ“ξ…žπ‘›=11π‘›π›Όβˆžξ“π‘˜=𝑛((π‘˜+1)π›Όβˆ’π‘˜π›Ό)π‘Žπ‘˜ξƒͺ𝑝β‰₯𝛼𝑝1βˆ’π›Όπ‘βˆžξ“π‘›=11π‘›π›Όβˆžξ“π‘˜=𝑛((π‘˜+1)π›Όβˆ’π‘˜π›Ό)π‘Žπ‘˜ξƒͺπ‘βˆ’1π‘Žπ‘›.(4.3) The dash over the summation on the left-hand side expression above (and in what follows) means that the term corresponding to 𝑛=1 is to be multiplied by 1+1/(1βˆ’π›Όπ‘). It's easy to see here the constant is best possible (on taking π‘Žπ‘›=π‘›βˆ’1/π‘βˆ’πœ– and letting πœ–β†’0+). By HΓΆlder's inequality, the above inequality readily implies the following inequality: βˆžξ“ξ…žπ‘›=11π‘›π›Όβˆžξ“π‘˜=𝑛((π‘˜+1)π›Όβˆ’π‘˜π›Ό)π‘Žπ‘˜ξƒͺ𝑝β‰₯𝛼𝑝1βˆ’π›Όπ‘π‘βˆžξ“π‘›=1π‘Žπ‘π‘›.(4.4)

We are thus motivated to consider the above inequality with the dash sign removed, and this can be regarded as an analogue of inequality (1.2) with 𝑐𝑝=(𝑝/(1βˆ’π‘))𝑝, which corresponds to the case 𝛼=1 here. As in the case of (1.2), such an inequality does not hold for all 𝛼 and 𝑝 satisfying 0<𝑝<1 and 0<𝛼<1/𝑝. However, on setting π‘Žπ‘›=π‘›βˆ’1/π‘βˆ’πœ– and letting πœ–β†’0+, one sees easily that if such an inequality holds for certain 𝛼 and 𝑝, then the constant is best possible. More generally, we can consider the following inequality:βˆžξ“π‘›=11βˆ‘π‘›π‘–=1πΏπ›½π›Όβˆ’1(𝑖,π‘–βˆ’1)βˆžξ“π‘˜=π‘›πΏπ›½π›Όβˆ’1(π‘˜Β±1,π‘˜)π‘Žπ‘˜ξƒͺ𝑝β‰₯𝛼𝑝1βˆ’π›Όπ‘π‘βˆžξ“π‘›=1π‘Žπ‘π‘›,(4.5) where the function πΏπ‘Ÿ(π‘Ž,𝑏) for π‘Ž>0,𝑏>0,π‘Žβ‰ π‘, and π‘Ÿβ‰ 0,1 (the only case we will concern here) is defined as πΏπ‘Ÿπ‘Ÿβˆ’1(π‘Ž,𝑏)=(π‘Žπ‘Ÿβˆ’π‘π‘Ÿ)/(π‘Ÿ(π‘Žβˆ’π‘)). It is known [12, Lemma 2.1] that the function π‘Ÿβ†¦πΏπ‘Ÿ(π‘Ž,𝑏) is strictly increasing on ℝ. Here we restrict our attention to the plus sign in (4.5) for the case 𝛽>0,max(1,𝛽)≀𝛼 and to the minus sign in (4.5) for the case 0<𝛼<1 and 𝛽β‰₯𝛼. Our remark above implies that in either case (note that 𝐿𝛽(1,0) is meaningful) 𝑛𝑖=1πΏπ›½π›Όβˆ’1(𝑖,π‘–βˆ’1)≀𝑛𝑖=1πΏπ›Όπ›Όβˆ’1𝑛(𝑖,π‘–βˆ’1)=𝛼𝛼.(4.6) As we also have πΏπ›½π›Όβˆ’1(π‘˜Β±1,π‘˜)β‰₯π‘˜π›Όβˆ’1, we see that the validity of (4.5) follows from that of (1.6). We therefore focus on (1.6) from now on, and we proceed as in Section 3 of [3] to see that in order for inequality (1.6) to hold, it suffices to find a sequence 𝐰 of positive terms for each 0<𝑝<1, such that for any integer 𝑛β‰₯1ξƒ©π‘›ξ“π‘˜=1π‘€π‘˜ξƒͺ1/(π‘βˆ’1)≀𝛼𝑝1βˆ’π›Όπ‘π‘/(π‘βˆ’1)ξ€·π›Όπ‘›π›Όβˆ’1𝑝/(1βˆ’π‘)𝑀𝑛1/(π‘βˆ’1)𝑛𝛼𝑝/(1βˆ’π‘)βˆ’π‘€1/(π‘βˆ’1)𝑛+1(𝑛+1)𝛼𝑝/(1βˆ’π‘)ξƒͺ.(4.7) We now choose 𝐰 inductively by setting 𝑀1=1, and for 𝑛β‰₯1𝑀𝑛+1=𝑛+1/π‘βˆ’π›Όβˆ’1𝑛𝑀𝑛.(4.8) The above relation implies that π‘›ξ“π‘˜=1π‘€π‘˜=𝑛+1/π‘βˆ’π›Όβˆ’1𝑀1/π‘βˆ’π›Όπ‘›.(4.9) We now assume 0<𝑝<1/2 and note that, for the so-chosen 𝐰, inequality (4.7) follows (with π‘₯=1/𝑛) from 𝑓(π‘₯)β‰₯0 for 0≀π‘₯≀1, whereξ‚΅ξ‚΅1𝑓(π‘₯)=1+𝑝π‘₯ξ‚Άβˆ’π›Όβˆ’11/(1βˆ’π‘)βˆ’(1+π‘₯)βˆ’π›Όπ‘/(1βˆ’π‘)βˆ’1βˆ’π›Όπ‘π‘π‘₯.(4.10) As 𝑓(0)=π‘“ξ…ž(0)=0, it suffices to show π‘“ξ…žξ…ž(π‘₯)β‰₯0, which is equivalent to showing 𝑔(π‘₯)β‰₯0, where 𝑔(π‘₯)=(1/π‘βˆ’π›Όβˆ’1)2𝛼((π›Όβˆ’1)𝑝+1)(1βˆ’π‘)/(1βˆ’2𝑝)(1+π‘₯)(2+(π›Όβˆ’2)𝑝)/(1βˆ’2𝑝)βˆ’ξ‚΅ξ‚΅11+𝑝π‘₯ξ‚Ά.βˆ’π›Όβˆ’1(4.11) Nowπ‘”ξ…žξ‚΅(π‘₯)=(1/π‘βˆ’π›Όβˆ’1)2𝛼((π›Όβˆ’1)𝑝+1)(1βˆ’π‘)/(1βˆ’2𝑝)ξ‚΅2+(π›Όβˆ’2)𝑝1βˆ’2𝑝(1+π‘₯)(2+(π›Όβˆ’2)𝑝)/(1βˆ’2𝑝)βˆ’1βˆ’ξ‚΅1𝑝β‰₯ξ‚΅βˆ’π›Όβˆ’1(1/π‘βˆ’π›Όβˆ’1)2𝛼((π›Όβˆ’1)𝑝+1)(1βˆ’π‘)/(1βˆ’2𝑝)ξ‚΅2+(π›Όβˆ’2)π‘ξ‚Άβˆ’ξ‚΅11βˆ’2π‘π‘ξ‚Άβˆ’π›Όβˆ’1∢=β„Ž(𝛼,𝑝).(4.12) Suppose now 𝛼β‰₯1, then, when 1/𝑝β‰₯(𝛼+2)(𝛼+1)/2, we have 1/𝑝β‰₯𝛼(π›Όβˆ’1)𝑝+2𝛼+1 since 𝑝<1/2 so that both inequalities 1/π‘βˆ’π›Όβˆ’1β‰₯1 and 1/π‘βˆ’π›Όβˆ’1β‰₯𝛼((π›Όβˆ’1)𝑝+1) are satisfied. In this case we have ξ‚΅β„Ž(𝛼,𝑝)β‰₯(1/π‘βˆ’π›Όβˆ’1)2𝛼((π›Όβˆ’1)𝑝+1)(1βˆ’π‘)/(1βˆ’2𝑝)βˆ’ξ‚΅1𝑝β‰₯βˆ’π›Όβˆ’1(1/π‘βˆ’π›Όβˆ’1)2βˆ’ξ‚΅1𝛼((π›Όβˆ’1)𝑝+1)π‘ξ‚Άβˆ’π›Όβˆ’1β‰₯0.(4.13) It follows that 𝑔′(π‘₯)β‰₯0 and as 𝑔(0)β‰₯0, and we conclude that 𝑔(π‘₯)β‰₯0, and hence 𝑓(π‘₯)β‰₯0. Similar discussion leads to the same conclusion for 0<𝛼<1 when 𝑝≀1/(𝛼+2). We now summarize our discussions above in the following.

Theorem 2. Let 0<𝑝<1/2 and 0<𝛼<1/𝑝. Let β„Ž(𝛼,𝑝) be defined as in (4.12). Inequality (1.6) holds for 𝛼,𝑝 satisfying β„Ž(𝛼,𝑝)β‰₯0. In particular, when 𝛼β‰₯1, inequality (1.6) holds for 0<𝑝≀2/((𝛼+2)(𝛼+1)). When 0<𝛼≀1, inequality (1.6) holds for 0<𝑝≀1/(𝛼+2).

Corollary 2. Let 0<𝑝<1/2 and 0<𝛼<1/𝑝. Let β„Ž(𝛼,𝑝) be defined as in (4.12). When 𝛽>0,max(1,𝛽)≀𝛼, inequality (4.5) holds (where one takes the plus sign) for 𝛼,𝑝 satisfying β„Ž(𝛼,𝑝)β‰₯0. In particular, inequality (4.5) holds for 0<𝑝≀2/((𝛼+2)(𝛼+1)). When 0<𝛼<1,𝛽β‰₯𝛼, inequality (4.5) holds (where one takes the minus sign) for 𝛼,𝑝 satisfying β„Ž(𝛼,𝑝)β‰₯0. In particular, inequality (4.5) holds for 0<𝑝≀1/(𝛼+2).

We note here a special case of the above corollary: the case 0<𝛼<1 and 𝛽→+∞ leads to the following inequality, valid for 0<𝑝≀1/(𝛼+2): βˆžξ“π‘›=11βˆ‘π‘›π‘–=1π‘–βˆžπ›Όβˆ’1ξ“π‘˜=π‘›π‘˜π›Όβˆ’1π‘Žπ‘˜ξƒͺ𝑝β‰₯𝛼𝑝1βˆ’π›Όπ‘π‘βˆžξ“π‘›=1π‘Žπ‘π‘›.(4.14)

We further note here that if we set π‘Ÿ=𝛼𝑝 and π‘Ž=0 in inequality (2.14), then it is reduced to 𝑓(π‘₯)β‰₯0 for 𝑓(π‘₯) defined as in (4.10). Since the case 0<π‘Ÿ<𝑝≀1/3 is known, we need only to be concerned about the case 𝛼β‰₯1 here and we now have the following improvement of the result of Levin and Stečkin [2, Theorem 62].

Corollary 3. Let 0<𝑝<1/2 and 1≀𝛼<1/𝑝. Let β„Ž(𝛼,𝑝) be defined as in (4.12). Inequality (1.3) holds for π‘Ÿ=𝛼𝑝 for 𝛼,𝑝 satisfying β„Ž(𝛼,𝑝)β‰₯0. In particular, inequality (1.3) holds for π‘Ÿ=𝛼𝑝 for 𝛼,𝑝 satisfying 0<𝑝≀2/((𝛼+2)(𝛼+1)).

Just as Theorem 1 and Corollary 1 are dual versions to each other, our results above can also be stated in terms of their dual versions, and we will leave the formulation of the corresponding ones to the reader.

5. Some Results on 𝑙𝑝 Norms of Factorable Matrices

In this section we first state some results concerning the 𝑙𝑝 norms of factorable matrices. In order to compare our result to that of weighted mean matrices, we consider the following type of inequalities:βˆžξ“π‘›=1ξƒ©π‘›ξ“π‘˜=1πœ†π‘˜Ξ›π‘›π‘Žπ‘˜ξƒͺπ‘β‰€π‘ˆπ‘βˆžξ“π‘›=1π‘Žπ‘π‘›,(5.1) where 𝑝>1,π‘ˆπ‘ is a constant depending on 𝑝. Here we assume that the two positive sequences (πœ†π‘›) and (Λ𝑛) are independent (in particular, unlike in the weighted mean matrices case, we do not have Λ𝑛=βˆ‘π‘›π‘˜=1πœ†π‘˜ in general). We begin with the following result concerning the bound for π‘ˆπ‘.

Theorem 3. Let 1<𝑝<∞ be fixed in (5.1). Let π‘Ž be a constant such that Λ𝑛+π‘Žπœ†π‘›>0 for all 𝑛β‰₯1. Let 0<𝐿<𝑝 be a positive constant, and let 𝑏𝑛=ξ‚΅π‘βˆ’πΏπ‘πœ†ξ‚Άξ‚΅1+π‘Žπ‘›Ξ›π‘›ξ‚Άπ‘βˆ’1πœ†π‘›Ξ›π‘›+πœ†π‘›πœ†π‘›+1.(5.2) If, for any integer 𝑛β‰₯1, one has π‘›ξ“π‘˜=1πœ†π‘˜π‘›ξ‘π‘–=π‘˜π‘π‘–1/(π‘βˆ’1)β‰€π‘ξ€·Ξ›π‘βˆ’πΏπ‘›+π‘Žπœ†π‘›ξ€Έ,(5.3) then inequality (5.1) holds with π‘ˆπ‘β‰€(𝑝/(π‘βˆ’πΏ))𝑝.

We point out that the proof of the above theorem is analogue to that of Theorem 3.1 of [7], except that, instead of choosing 𝑏𝑛 to satisfy the equation (3.4) in [7], we choose 𝑏𝑛 so that ξ‚΅π‘π‘›πœ†π‘›βˆ’1πœ†π‘›+1Λ𝑝𝑛=ξ‚΅π‘βˆ’πΏπ‘ξ‚Άξ€·Ξ›π‘›+π‘Žπœ†π‘›ξ€Έπ‘βˆ’1.(5.4) We will leave the details to the reader, and we point out that, as in the case of weighted mean matrices in [7], we deduce from Theorem 3 the following.

Corollary 4. Let 1<𝑝<∞ be fixed in (5.1). Let π‘Ž be a constant such that Λ𝑛+π‘Žπœ†π‘›>0 for all 𝑛β‰₯1. Let 0<𝐿<𝑝 be a positive constant such that the following inequality is satisfied for all 𝑛β‰₯1 (with Ξ›0=πœ†0=0): ξ‚΅π‘βˆ’πΏπ‘πœ†ξ‚Άξ‚΅1+π‘Žπ‘›Ξ›π‘›ξ‚Άπ‘βˆ’1+Ξ›π‘›πœ†π‘›+1β‰€Ξ›π‘›πœ†π‘›ξ‚΅πœ†1+π‘Žπ‘›Ξ›π‘›ξ‚Άπ‘βˆ’1𝐿1βˆ’π‘ξ‚Άπœ†π‘›Ξ›π‘›+Ξ›π‘›βˆ’1Ξ›π‘›πœ†+π‘Žπ‘›βˆ’1Λ𝑛1βˆ’π‘.(5.5) Then inequality (5.1) holds with π‘ˆπ‘β‰€(𝑝/(π‘βˆ’πΏ))𝑝.

We now apply the previous corollary to the special case of (5.1) with πœ†π‘›=π›Όπ‘›π›Όβˆ’1,Λ𝑛=𝑛𝛼 for some 𝛼>1. On taking 𝐿=1/𝛼 and π‘Ž=0 in Corollary 4 and setting 𝑦=1/𝑛, we see that inequality (1.7) holds as long as we can show for 0≀𝑦≀11ξ‚΅ξ‚΅1βˆ’ξ‚Άπ‘π›Όπ›Όπ‘¦+(1βˆ’π‘¦)π›Όξ‚Άπ‘βˆ’11ξ‚΅ξ‚΅1βˆ’ξ‚Άπ‘π›Όπ›Όπ‘¦+(1+𝑦)1βˆ’π›Όξ‚Άβ‰€1.(5.6) We note first that, as (1βˆ’1/𝑝𝛼)𝛼𝑦+(1βˆ’π‘¦)𝛼≀(1βˆ’1/𝑝𝛼)𝛼𝑦+(1+𝑦)1βˆ’π›Ό, we need to have (1βˆ’1/𝑝𝛼)𝛼𝑦+(1βˆ’π‘¦)𝛼≀1 in order for the above inequality to hold. Taking 𝑦=1 shows that it is necessary to have 𝛼≀1+1/𝑝. In particular, we may assume 1<𝛼≀2 from now on, and it then follows from Taylor expansion that, in order for (5.6) to hold, it suffices to show thatξ‚΅11βˆ’π‘π‘¦+𝛼(π›Όβˆ’1)2𝑦2ξ‚Άπ‘βˆ’1ξ‚΅ξ‚΅11+1βˆ’π‘ξ‚Άπ‘¦+𝛼(π›Όβˆ’1)2𝑦2≀1.(5.7) We first assume 1<𝑝≀2, and in this case we use ξ‚΅11βˆ’π‘π‘¦+𝛼(π›Όβˆ’1)2𝑦2ξ‚Άπ‘βˆ’1ξ‚΅βˆ’1≀1+(π‘βˆ’1)𝑝𝑦+𝛼(π›Όβˆ’1)2𝑦2ξ‚Ά(5.8) to see that (5.7) follows from β„Ž1,𝛼,𝑝(𝑦)∢=𝛼(π›Όβˆ’1)𝑝2βˆ’ξ‚΅11βˆ’π‘ξ‚Ά2+𝛼(π›Όβˆ’1)(π‘βˆ’1)𝛼2𝑝(π‘βˆ’2)𝑦+2(π›Όβˆ’1)24(π‘βˆ’1)𝑦2≀0.(5.9) We now denote 𝛼1(𝑝)>1 as the unique number satisfying β„Ž1,𝛼1,𝑝(0)=0 and 𝛼2(𝑝)>1 the unique number satisfying β„Ž1,𝛼2,𝑝(1)=0 and let 𝛼0(𝑝)=min(𝛼1(𝑝),𝛼2(𝑝)). It is easy to see that both 𝛼1(𝑝) and 𝛼2(𝑝) are ≀1+1/𝑝 and that, for 1<𝛼≀𝛼0, we have β„Ž1,𝛼,𝑝(𝑦)≀0 for 0≀𝑦≀1.

Now suppose that 𝑝>2, then we recast (5.7) asξ‚΅11+1βˆ’π‘ξ‚Άπ‘¦+𝛼(π›Όβˆ’1)2𝑦2≀11βˆ’π‘π‘¦+𝛼(π›Όβˆ’1)2𝑦2ξ‚Ά1βˆ’π‘.(5.10) In order for the above inequality to hold for all 0≀𝑦≀1, we must have 𝛼(π›Όβˆ’1)𝑦2/2≀𝑦/𝑝. Therefore, we may from now on assume 𝛼(π›Όβˆ’1)≀2/𝑝. Applying Taylor expansion again, we see that (5.10) follows from the following inequality: ξ‚΅11+1βˆ’π‘ξ‚Άπ‘¦+𝛼(π›Όβˆ’1)2𝑦2ξ‚΅βˆ’1≀1+(1βˆ’π‘)𝑝𝑦+𝛼(π›Όβˆ’1)2𝑦2ξ‚Ά+𝑝(π‘βˆ’1)βˆ’(1/𝑝)𝑦+(𝛼(π›Όβˆ’1)/2)𝑦2ξ€Έ22.(5.11) We can recast the above inequality as β„Ž2,𝛼,𝑝(𝑦)∢=𝛼(π›Όβˆ’1)𝑝2βˆ’1βˆ’1/𝑝2+(π‘βˆ’1)𝛼(π›Όβˆ’1)𝑦2βˆ’π‘(π‘βˆ’1)𝛼2(π›Όβˆ’1)2𝑦28≀0.(5.12) We now denote 𝛼0(𝑝)>1 as the unique number satisfying 𝛼(π›Όβˆ’1)≀2/𝑝 and β„Ž2,𝛼0,𝑝(1)=0. It is easy to see that, for 1<𝛼≀𝛼0, we have β„Ž2,𝛼,𝑝(𝑦)≀0 for 0≀𝑦≀1. We now summarize our result in the following.

Theorem 4. Let 𝑝>1 be fixed, and let 𝛼0(𝑝) be defined as above, then inequality (1.7) holds for 1<𝛼≀𝛼0(𝑝).

As we have explained in Section 1, the study of (1.7) is motivated by (1.8). As (1.7) implies (1.8) and the constant (𝛼𝑝/(π›Όπ‘βˆ’1))𝑝 there is best possible (see [9]), we see that the constant (𝛼𝑝/(π›Όπ‘βˆ’1))𝑝 in (1.7) is also best possible. More generally, we note that inequality (4.7) in [9] proposes to determine the best possible constant π‘ˆπ‘(𝛼,𝛽) in the following inequality (πšβˆˆπ‘™π‘,𝑝>1,𝛽β‰₯𝛼β‰₯1):βˆžξ“π‘›=1|||||1βˆ‘π‘›π‘˜=1πΏπ›½π›Όβˆ’1(π‘˜,π‘˜βˆ’1)𝑛𝑖=1πΏπ›½π›Όβˆ’1(𝑖,π‘–βˆ’1)π‘Žπ‘–|||||π‘β‰€π‘ˆπ‘(𝛼,𝛽)βˆžξ“π‘›=1||π‘Žπ‘›||𝑝.(5.13) We easily deduce from Theorem 4 the following.

Corollary 5. Keep the notations in the statement of Theorem 4. For fixed 𝑝>1 and 1<𝛼≀𝛼0(𝑝), inequality (5.13) holds with π‘ˆπ‘(𝛼,𝛽)=(𝛼𝑝/(π›Όπ‘βˆ’1))𝑝 for any 𝛽β‰₯𝛼.

Acknowledgments

During this work, the author was supported by postdoctoral research fellowships at Nanyang Technological University (NTU). The author is also grateful to the referee for his/her helpful comments and suggestions.