alexa
Reach Us +44-1474-556909
Refined Estimates on Conjectures of Woods and Minkowski-I | OMICS International
ISSN: 2168-9679
Journal of Applied & Computational Mathematics
Make the best use of Scientific Research and information from our 700+ peer reviewed, Open Access Journals that operates with the help of 50,000+ Editorial Board Members and esteemed reviewers and 1000+ Scientific associations in Medical, Clinical, Pharmaceutical, Engineering, Technology and Management Fields.
Meet Inspiring Speakers and Experts at our 3000+ Global Conferenceseries Events with over 600+ Conferences, 1200+ Symposiums and 1200+ Workshops on Medical, Pharma, Engineering, Science, Technology and Business
All submissions of the EM system will be redirected to Online Manuscript Submission System. Authors are requested to submit articles directly to Online Manuscript Submission System of respective journal.

Refined Estimates on Conjectures of Woods and Minkowski-I

Kathuria L* and Raka M

Centre for Advanced Study in Mathematics, Panjab University, Chandigarh-160014, India

*Corresponding Author:
Kathuria L
Centre for Advanced Study in Mathematics
Panjab University, Chandigarh-160014
India
Tel: 08754216121
E-mail: [email protected]

Received Date: February 10, 2015; Accepted Date: March 23, 2015; Published Date: April 15, 2015

Citation: Kathuria L, Raka M (2015) Refined Estimates on Conjectures of Woods and Minkowski-I. J Appl Computat Math 4:209. doi: 10.4172/2168-9679.1000209

Copyright: © 2015 Kathuria L, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Visit for more related articles at Journal of Applied & Computational Mathematics

Keywords

Lattice; Covering; Non-homogeneous; Product of linear forms; Critical determinant; Korkine and Zolotare reduction; Hermite's constant; Centre density

Introduction

Let Equation ,Equationbe n real linear forms in n variables x1; : : : ; xn and having determinant Equation The following conjecture is attributed to H. Minkowski:

Conjecture I: For any given real numbers c1; : : : ; cn, there exists integers x1; : : : ; xn such that

Equation                      (1.1)

Equality is necessary if and only if after a suitable unimodular transformation the linear forms Li have the form 2cixlfor Equation

This result is known to be true for Equation For a detailed history and the related results,

Minkowski's Conjecture is equivalent to saying that [1]

Equation

where Equation is given by

Equation

Chebotarev proved the weaker inequality

Equation                    (1.2)

Since then several authors have tried to improve upon this estimate. The bounds have been obtained in the form

Equation             (1.3)

where Equation Clearly Equation by considering the linear forms Li=xi and Equation for Equation During 1949-1986, many authors such as Davenport, Woods, Bombieri, Gruber, Skubenko, Andrijasjan, Il'in and Malyshev obtained Vn for large n. obtained Equation for all n [2-4] improved Mordell's estimates for Equation Hans-Gill et al. [12,14] got improvements on the results of [5-8] for Equation Since recently Vn 9=29/2 has been established by the authors [9], we study Vn for Equation in a series of three papers.

In this paper we obtain improved estimates on Minkowski's Conjecture for n=10; 11 and 12. In next papers [10-12], we shall derive improved estimates on Minkowski's Conjecture for n=13; 14; 15 and for Equation respectively [13-16]. For sake of comparison, we give results by our improved Vn in Table 1.

  Estimates by Mordell Estimates by Il’in Estimates by Hans-Gill et al Our improved Estimates
n Vn Vn Vn Vn
10 2.899061 3.47989 24.3627506 27.60348
11 2.973102 3.52291 29.2801145 33.47272
12 3.040525 3.55024 32.2801213 39.59199
13 3.102356 3.57856 34.8475153 45.40041
14 3.159373 3.60209 37.8038391 51.26239
15 3.21218 3.61116 40.905198 57.00375
16 3.261252 3.61908 44.3414913 57.4702
17 3.306972 3.63924 47.2339309 57.67598
18 3.349652 3.66176 46.7645724 57.38876
19 3.389556 3.66734 47.2575897 60.09339
20 3.426907 3.67236 46.8640155 58.48592
21 3.461897 3.67692 46.0522028 56.42571
22 3.494699 3.68408 43.6612034 53.94142
23 3.525464 3.68633 37.8802374 50.98842
24 3.55433 3.68978 32.5852958 47.74632
25 3.581421 3.69295 27.8149432 42.39088
26 3.606852 3.69589 23.0801951 38.8657
27 3.630729 3.70012 17.3895105 31.93316
28 3.653149 3.70263 12.9938763 26.10663
29 3.674203 3.70497 9.5796191 19.96254
30 3.693976 3.70867 6.7664335 16.06884
31 3.712547 3.72558 4.745972 11.23872
32 3.729989     8.325879
33 3.746371     5.411488

Table 1: The weaker result.

We shall follow the Remak-Davenport approach. For the sake of convenience of the reader we give some basic results of this approach. Minkowski's Conjecture can be restated in the terminology of lattices as : Any lattice ^ of determinant d(^) in Rn is a covering lattice for the set

Equation

The weaker result (1.3) is equivalent to saying that any lattice ^ of determinant d(^) in Rn is a covering lattice for the set

Equation

Define the homogeneous minimum of ^ as

Equation

Proposition 1. Suppose that Minkowski Conjecture has been proved for dimensions 1, 2,…., n - 1: Then it holds for all lattices ^ in Rn for which MH(^)=0.

Proposition 2. If ^ is a lattice in Rn for Equation with with MH(^) ≠ 0 then there exists an ellipsoid having n linearly independent points of ^ on its boundary and no point of ^ other than O in its interior.

It is well known that using these results, Minkowski's Conjecture would follow from

Conjecture II. If ^ is a lattice in Rn of determinant 1 and there is a sphere |X| <R which contains no point of ^ other than O in its interior and has n linearly independent points of ^ on its boundary then ^ is a covering lattice for the closed sphere of radius Equation Equivalently, every closed sphere of radius Equation lying in Rn contains a point of ^.

They formulated a conjecture from which Conjecture-II follows immediately. To state Woods' conjecture, we need to introduce some terminology [17,18].

Let L be a lattice in Rn. By the reduction theory of quadratic forms introduced by a cartesian co-ordinate system may be chosen in Rn in such a way that L has a basis of the form [19-22],

Equation

where A1;A2; : : : ;An are all positive and further for each i=1; 2; : : : ; n any two points of the lattice in Rn-i+1 with basis

Equation

are at a distance atleast Ai apart. Such a basis of L is called a reduced basis [23].

Conjecture III (Woods): If A1A2…An=1 and Equation for each i then any closed sphere in Rn of radius Equation contains a point of L.

Woods [10] proved this conjecture for Equation Hans-Gill et al. [12] gave a unified proof of Woods' Conjecture for Equation Hans-Gill et al. [12,14] proved Woods' Conjecture for n=7 and n=8 and thus completed the proof of Minkowski's Conjecture for n=7 and 8 Woods [10,24] proved Conjecture and hence Minkowski's Conjecture for n=9. With the assumptions as in Conjecture III, a weaker result would be that

If Equationany closed sphere in Rn of radius Equation contains a point of L [25,26].

Hans-Gill et al. [12,14] obtained the estimates wn on Woods' Conjecture for Equation As w9=9 has been established by the authors [17] recently, we study wn for Equation in a series of three papers. In this paper we obtain improved estimates wn on Woods' Conjecture for n=10; 11 and 12. In next papers [18,19], we shall derive improved estimates wnon Woods' Conjecture for n=13; 14; 15 and for Equation respectively. Together with the following result of Hans-Gill et al. [12], we get improvements of wn for Equation

Proposition 3. Let L be a lattice in Rn with A1A2…An=1 and Equation for each i. Let Equation where ln and mn are real numbers. Then L is a covering lattice for the sphere Equation where Wn is defined inductively by

Equation

Here we prove

Theorem 1. Let n=10; 11; 12. If d(L)=A1 : : :An=1 and Equation for i=2;….; n, then any closed sphere in Rn of radius Equation contains a point of L, where Equation

The earlier best known values were w10=10:5605061, w11=11:9061976 and w12=13:4499927.

To deduce the results on the estimates of Minkowski's Conjecture we also need the following generalization of Proposition 1

Proposition 4. Suppose that we know

Equation

Let vn<min Vk1 Vk2…Vks , where the minimum is taken over all (k1; k2; ; ks) such that n=k1+k2+ : : :+ks, ki positive integers for all i and Equation . Then for all lattices in Rn with homogeneous minimum MH(<)=0, the estimate Vn holds for Minkowski's Conjecture.

Since by arithmetic-geometric inequality the sphere Equation is a subset of Equation Propositions 2 and 4 immediately imply

Theorem 2: The values of Vn for the estimates of Minkowski's Conjecture can be taken as Equation

For Equation these values are listed in Table 1. In Section 2 we state some preliminary results and in Sections 3-5 we prove Theorem 1 for n=10; 11 and 12.

Preliminary Results and Plan of the Proof

Let L be a lattice in Rn reduced in the sense of Korkine and Zolotare. Let (Sn) denotes the critical determinant of the unit sphere D Sn with centre O in Rn i.e.

Equation has no point other than O in the interior of Sn}

Let Equation be the Hermite's constant i.e. Equation is the smallest real number such that for any positive de nite quadratic form Q in n variables of determinant D, there exist integers u1; u2;…; un not all zero satisfying

Equation

It is well known that We write A2Bi=Bi.

We state below some preliminary lemmas. Lemmas 1 and 2 are due to Woods [25], Lemma 3 is due to Korkine and Zolotare [21] and Lemma 4 is due to Pendavingh and Van Zwam [24]. In Lemma 5, the cases n=2 and 3 are classical results of Lagrange and Gauss; n=4 and 5 are due to Korkine and Zolotare [21] while n=6; 7 and 8 are due to Blichfeldt [3].

Lemma 1. If Equation then any closed sphere of radius Equation

in Rn contains a point of L.

Lemma 2. For a Fixed integer i with Equation denote by L1 the lattice in Ri with reduced basis

Equation

and denote by L2 the lattice in Rn-i with reduced basis

Equation

If any closed sphere in Ri of radius r1 contains a point of L1 and if any closed sphere in Rn-i of radius r2 contains a point of L2 then any closed sphere in Rn of radius Equation contains a point of L:

Lemma 3. For all relevant i,

Equation                          (2.1)

Lemma 4. For all relevant i,

Equation                          (2.2)

Throughout the paper we shall denote 0.46873 by ε .

Lemma 5. Equation for n=2; 3; 4; 5; 6; 7 and 8 respectively:

Lemma 6. For any integer s; Equation

Equation and Equation                  (2.4)

This is Lemma 4 of Hans-Gill et al. [12].

Lemma 7.

Equation              (2.5)

This is Lemma 6 of Hans-Gill et al. [14].

Remark 1. Let

Equation=the best centre density of packings of unit spheres in Rn;

Equation=the best centre density of lattice packings of unit spheres in Rn:

Then it is known that

Equation                             (2.6)

Equationand hence Equation is known for Equation Also Equation=4 has been proved by Cohn and Kumar [6]. For Equation using the bounds on Equation given by Cohn and Elkies [5] and inequality (2.6) we find that Equation, Equation,Equation,Equation

We assume that Theorem 1 is false and derive a contradiction. Let L be a lattice satisfying the hypothesis of the conjecture. Suppose that there exists a closed sphere of radius Equation in Rn that contains no point of L in Rn.

Since Bi=A2 i and d(L)=1; we have B1B2 : : :Bn=1:

We give some examples of inequalities that arise. Let L1 be a lattice in R4 with basis (A1; 0; 0; 0), (a2;1;A2; 0; 0); (a3;1; a3;2;A3; 0); (a4;1; a4;2; a4;3;A4); and Li for Equation be lattices in R1 with basis (Ai+3). Applying Lemma 2 repeatedly and using Lemma 1 we see that if Equation then any closed sphere of radius

Equation

contains a point of L: By the initial hypothesis this radius exceeds Equation Since Equation and Equation this results in the conditional inequality : if Equation then

Equation                        (2.7)

We call this inequality (4; 1;…; 1); since it corresponds to the ordered partition (4; 1;…; 1) of n for the purpose of applying Lemma 2. Similarly the conditional inequality (1;…; 1; 2; 1;…; 1) corresponding to the ordered partition (1;…; 1; 2; 1;…; 1) is : if Equation then

Equation                (2.8)

Since Equation(2.8) gives

Equation

One may remark here that the condition Equation is necessary only if we want to use inequality (2.8), but it is not necessary if we want to use the weaker inequality (2.9). This is so because if Equation using the partition (1; 1) in place of (2) for the relevant part, we get the upper bound Equationwhich is clearly less than 2Bi+1. We shall call inequalities of type (2.9) as weak inequalities and denote it by (1;…; 1; 2; 1;….; 1)w.

If Equation is an ordered partition of n, then the conditional inequality arising from it, by using Lemmas 1 and 2, is also denoted by Equation If the conditions in an inequality Equation are satisfied then we say that Equation holds. Sometimes, instead of Lemma 2, we are able to use induction. The use of this is indicated by putting (*) on the corresponding part of the partition. For example, if for n=10, B5 is larger than each of B6;B7;….;B10, and if Equation the inequality (4; 6*) gives

Equation            (2.10)

In particular the inequality ((n-1)*; 1) always holds. This can be written as

Equation                                          (2.11)

Also we have Equation because if B1<1, then Equation for each I contradicting B1B2:::Bn=1.

Using the upper bounds on and the inequality (2.5), we obtain numerical lower and upper bounds on Bn, which we denote by ln and mn respectively. We use the approach of Hans-Gill et al. [14], but our method of dealing with

Is somewhat different. In Sections 3-5 we give proof of Theorem 1 for n=10; 11 and 12 respectively. The proof of these cases is based on the truncation of the interval [ln;mn] from both the sides.

In this paper we need to maximize or minimize frequently functions of several variables. When we say that a given function of several variables in x; y; is an increasing/decreasing function of x; y;…., it means that the concerned property holds when function is considered as a function of one variable at a time, all other variables being fixed.

Proof of Theorem 1 for n=10

Here we have W10=10:3,B1<Equation10<2:2636302. Using (2.5), we have l10=0:4007<B10<1:9770808=m10.

The inequality (9*; 1) gives 9(B10) -1/9 + B10<10:3. But for 0:4398 B10 1:9378, this inequality is not true. Hence we must have either B10<0:4398 or B10>1:9378. We will deal with the two cases 0:4007< B10<0:4398 and 1:9378<B10<1:9770808 separately:

0:4007<B10<0:4398

Using the Lemmas 3 & 4 we have:

Equation

Claim(i) B2>1:7046

The inequality (2; 2; 2; 2; 2)w gives 2B2 +2B4 +2B6 +2B8 +2B10>10:3. Using (3.1), we find that this inequality is not true for B2 ≤ 1:7046. Hence we must have B2>1:7046.

Claim(ii) Equation

Suppose Equation then using (3.1) and that Equation we find that Equation and Equation So the inequality (1,4,4,1) holds, i.e.B1+4B2- Equation Applying AM-GM inequality we get Equation Now since Equation ,Equation and Equation we find that the left side is a decreasing function of B10 and B6. So replacing B10 by Equation and Equation we get Equation Now the left side is a decreasing function of B2, so replacing B2 by 1.8815 we find that Equationfor 1<B1<2:2636302, a contradiction. Hence we must have B2<1:8815.

Claim (iii) B3<1:5652

Suppose Equation From (3.1) we have B4B5B6<1:6524 and B8B9B10 <0:1702, so we find that Equationand Equation for B3>1:49.

Applying AM-GM to inequality (2,4,4) we get Equation Since Equation and Equation we find that left side is a decreasing function of B1 and B7. So we replace B1 by B2, B7 by Equation and get that Equation

But left side is a decreasing function of B3, so replacing B3 by 1.5652 we find that Equation for 1:7046<B2<1:8815, a contradiction. Hence we must have B3<1:5652.

Claim (iv) B1>1:9378

Suppose Equation Using (3.1) and that B3<1:5652, B2>1:7046, we find that B2 is larger than each of B3; B4;…;B10. So the inequality (1; 9,*) holds. This gives Equation which is not true for Equation So we must have B1>1:9378.

Claim (v) B3<1:5485

Suppose Equation We proceed as in Claim(iii) and replace B1 by 1.9378 and B7 by Equation to get that

Equation One easily checks that Equation for 1.5485≤ B3<1:5652 and 1:7046< B2<1:8815. Hence we have B3<1:5485.

Claim (vi) B1<2:0187

Suppose Equation Using (3.1) and Claims (ii), (v) we have B2B3B4<4:11. Therefore EquationAs Equation we see using (3.1) that B5 is larger than each of B6;B7,… ;B10. Hence the inequality (4; 6,*) holds. This gives Equation Left side is an increasing function of B2B3B4 and decreasing function of B1. So we can replace B2B3B4 by 4:11 and B1 by 2.0187 to find Equation a contradiction. Hence we have B1<2:0187.

Claim (vii) B4<1:337

Suppose Equation then using (3.1) we get Equation Applying AMGM to inequality (1,2,4,2,1) we have

Equation

Since B2>1:7046, Equation and Equation we find that left side is a decreasing function of B2, B8 and B10. So we can replace B2 by 1.7046; B8 by Equation and B10 by Equation to get

Equation

Now left side is a decreasing function of B4, replacing B4 by 1:337, we find that Equation for 1<B1<2:0187 and 1<B3<1:5485, a contradiction. Hence we have B4<1:337.

Claim (viii) B5<1:1492

Suppose Equation Using (3.1), we get B6B7B8<0:5445: Therefore Equation Also using Lemma 3 & 4, 2 Equation.So the inequality (4*; 4; 2) holds, i.e.4 Equation Now left side is a decreasing function of B5 and B9. So we replace B5 by 1.1492 and B9 by Equation and get that Equation Equationwhere x=B6B7B8. Using Lemma 3 & 4 we have x==B6B7B8 Equationand Equation It can be verified that Equation for Equation and Equation giving thereby a contradiction. Hence we must have B5<1:1492.

Claim (ix) B2<1:766.

Suppose Equation We have B3B4B5<2:3793. So Equation Also Equation Therefore B6 is larger than each of Equation Hence the inequality (1; 4; 5,*) holds. This gives Equation Left side is an increasing function of B3B4B5, a decreasing function of B2 and an increasing function of B1. One easily checks that this inequality is not true for B1<2:0187;

Equation and B3B4B5<2:3793: Hence we have B2<1:766.

Final contradiction

As 2(B2+B4+B6+B8+B10)<2(1:766+1:337+0:9383+0:6597+0:4398)<10:3, the weak inequality (2; 2; 2; 2; 2)w gives a contradiction.

9378<B10<1:9770808

Here Equationand B2=(B1B3…B10)-1 EquationWhich implies Equationi.e. B2<1:75076.

Similarly

Equation

Equation

Equation

Equation

These respectively give B3<1:46138, B4<1:22883, B6<0:896058 and B8<0:721763. So we have Equation Also Equationand EquationApplying AMGM to inequality (4,2,2,1,1) we have 4B1 + 4B5 + 4B7 + B9 +B10 EquationWe find that left side is a decreasing function of B7 and B5, so can replace B7 by Equationand B5 by Equationthen it is a decreasing function of B1, so replacing B1 by B10 we have Equationwhich is not true for Equationand 1:9378<B10<1:9770808. Hence we get a contradiction.

Proof of Theorem 1 for n=11

Here we have w11=11.62, EquationUsing (2.5), we have l11=0:3673<B11<2:1016019=m11.

The inequality (10*; 1) gives 10:3 Equation But for Equationthis inequality is not true. So we must have either Equation or Equation

0:3673<B11<0:4409

Claim (i) B10<0:4692

Suppose Equationthen 2B10>B11, so (9*; 2) holds, i.e. 9 EquationAs left side is a decreasing function of B10, we can replace B10 by 0.4692 and find that it is not true for 0:3673<B11<0:4409.

Hence we must have B10<0:4692.

Using Lemmas 3 and 4 we have:

Equation

Equation

Equation                (4.1)

Claim (ii) B2>1:913

The inequality (2; 2; 2; 2; 2; 1) w gives 2B2+2B4+2B6+2B8+2B10+B11 > 11:62. Using (4.1) we find that this inequality is not true for B ≤1.913 so we must have B2>1:913.

Claim(iii) B3<1:761

Suppose Equationthen we have Equationand Equation

2. Applying AM-GM to the inequality (2,4,4,1) we get EquationOne easily finds that it is not true for Equationand Equation Hence we must have B3<1:761:

Claim (iv) B1<2:2436

Suppose EquationAs B2B3B4<2:13557×1:761×1:50151<5:6468, we have Equation Also Equation so B5 is larger than each of B6;B7…;B11. Hence the inequality (4; 7,*) holds. This gives Equation Left side is an increasing function of B2B3B4 and decreasing function of B1. One easily checks that the inequality is not true for B2B3B4<5:6468 and B1 ≥ 2:2436. Hence we have B1<2:2436.

Claim (v) B4<1.4465 and B2>1:9686

Suppose B4 ≥ 1.4465 We have B5B6B7<1:2569 and B9B10B11<0:1295. Therefore for B4>1:36, we have Equation and Equation So the inequality (1,2,4,4) holds. Applying AM-GM to inequality(1,2,4,4), we get Equation A simple calculation shows that this is not true for Equation,Equation, Equation and Equation Hence we have B4<1:4465.

Further if Equation then 2B2+2B4+2B6+2B8+2B10+B11<11:62. So the inequality (2; 2; 2; 2; 2; 1)w gives a contradiction.

Claim (vi) B4<1:4265 and B2>1:9888

Suppose Equation We proceed as in Claim (v) and get a contradiction with improved bounds on B2 and B4.

Claim (vii) B1<2:2056

Suppose Equation As B3B4B5<1:761 × 1:4265 × 1:3347<3:3529, we have Equation Also Equation so B6 is larger than each of B7;B8,…,B11. Hence the inequality (1; 4; 6*) holds, i.e. B1 + 4B2 - Equation

Claim (ix) B1<2:1669

Suppose Equation We proceed as in Claim(iv) and get a contradiction with improved bounds on B1, B2 and B4.

Claim (x) B4<1:403 and B2>2:012

Suppose Equation We proceed as in Claim(v) and get a contradiction with improved bounds on B2 and B4.

Final Contradiction:

As now B3B4B5<1:761×1:403 1:3347<3:2977, we have Equation for B2>2:012. Also Equation each of B7; B8; B11. Hence the inequality (1; 4; 6) holds. Proceeding as in Claim (viii) we find that this inequality is not true for B1<2:1669; B2>2:012 and B3B4B5<3:2977; giving thereby a contradiction.

2:018<B11<2:1016019

Here EquationTherefore using Lemmas 3 & 4 we have B10=(B1 B9B11)-1

Equation

Equation

Similarly

Equation which gives B4<1:37661.

Claim (i) B10<0:4402

The inequality (9*; 1; 1) gives Equation But this inequality is not true for Equation and 2:018<B11<2:1016019. Hence we must have B10<0:4402.

Now we have Equation< 0:58694, Equationand Equation

Claim (ii) B7<0:768

Suppose Equation Then Equation so (6*; 4; 1) holds. This gives Equation where x=B1B2 : : :B6. The function Equation has its maximum value at Equation Therefore Equationwhich is less than 11:62 for Equation 2:018<B11<2:1016019. This gives a contradiction.

Now Equation and Equation

Claim (iii) B2<1:795

Suppose Equation then Equation and Equation Applying AMGM to the inequality (1,4,4,1,1) p we get B1 + 4B2 + 4B6 + B10 + B11 - Equation We find that left side is a decreasing function of B6, so we first replace B6 by εB2 then it is a decreasing function of B2, so we replace B2 by 1.795 and get that

Equation Now Equationso Equation which can be verified to be at most 11.62 for Equation and 2:018<B1<2:393347, giving thereby a contradiction.

Claim (iv) B5<0:98392

Suppose Equation We have Equation and Equation Also Equation Applying AM-GM to the inequality (4; 4; 2; 1) we get Equation One can easily check that left side is a decreasing function of B9 and B1 so we can replace B9 by εB and B1 by B11 to get Equation Now the left side is a decreasing function of B5, so replacing B5 by 0.98392 we see that Equation and 2:018<B11 < 2:1016019, a contradiction.

Final Contradiction:

As in Claim(iv), we have Equation Also Equationeach of B6;B7,…,B10. Therefore the inequality (4; 6*; 1) holds, i.e. Equation Left side is an increasing function of B2B3B4 and B11 and decreasing function of B1. Using B5<0:98392, we haveEquation and Equation One easily checks that Equation for B2B3B4<1:795 × 1:47588 × 1:311894, B11<2:1016019 and Equation Hence we have a contradiction.

Proof of Theorem 1 for n=12

Here we have w12=13, Equation Using (2.5), we have l12 =0:3376<B12<2:2254706=m12 and using (2.3) we have Equation i.e Equation

The inequality (11*; 1) gives 11:62(B12)-1/11 +B12>13. But this is not true for Equation So we must have either B12<0:4165 or B12>2:17.

0:3376<B12<0:4165

Claim (i) B11<0:459

Suppose Equation then Equationand 2B11>B12, so (10*; 2) holds, i.e, Equation Left side is a decreasing function of B11, so we can replace B11 by .459 to find that Equation for 0:34425<B12<0:4165, a contradiction. Hence we have B11<0:459.

Claim (ii) B10<0:5432

Suppose Equation From Lemma 3, Equationand EquationB12.Therefore Equationand Equation so the inequality (9*; 3) holds, i.e. EquationOne easily checks that it is not true noting that left side is a decreasing function of B10. Hence we must have B10<0:5432.

Claim (iii) B9<0:6655

Suppose Equation then Equation So the inequality (8*; 4) holds. This gives Equation where x=B1B2 … B8. The function Equation has its maximum value at Equation so Equation for Equation where x=B1B2… B8. The function Equation has its maximum value at Equation so Equation for 0:6655 Equation This gives a contradiction.

Using Lemmas 3 & 4 we have:

Equation

Equation

Claim (iv) B2>1:828, B4>1:426, B6>1:019 and B8>0:715

Equation

Suppose Equation Then 2(B2+B4+B6+B8+B10+B12)<2(1:828+ 1:7384+1:1589+0:8148+0:5432+0:4165)<13, giving thereby a contradiction to the weak inequality (2; 2; 2; 2; 2; 2) w.

Similarly we obtain lower bounds on B4;B6 and B8 using (2; 2; 2; 2; 2; 2)w.

Claim(v) B2>2:0299

Suppose Equation Consider following two cases:

Case (i) B3>B4

We have B3>B4>1:426>each of B5,….,B12. So the inequality (2; 10*) holds, i.e. EquationThe left side is a decreasing function of B1, so replacing B1 by B2 we get 2B2 + 10:3 Equationwhich is not true for Equation

Case (ii) Equation

As B4>1:426>each of B5,….,B12, the inequality (3; 9*) holds, i.e. Equationwhere X=B2B3<min Equationsay. Now Equation is an increasing function of X for Equation and so Equation which can be seen to be less than 13. Hence we have B2>2:0299.

Claim (vi) B1>2:17 and B3<1:9517

Using (2.3) we have Equation Therefore B2>2:0299>each of B3,…,B12. So the inequality (1; 11*) holds, i.e. B1 + 11:62 Equation But this is not true for Equation So we must have B1>2:17: Again using (2.3) we have Equation

Claim (vii) B4<1:646

Suppose Equation From (5.1) and Claims (i)-(iii), we have Equation and Equation Applying AM-GM to the inequality (1,2,4,4,1) we get Equation We find that left side is a decreasing function of B2, B8 and B12. So we can replace B2 by 2:0299, B8 by "B4 and B12 by Equation. Then it turns a decreasing function of e 2B4, so can replace B4 by 1.646 to find that Equation,a contradiction. Hence we have B4<1:646.

Claim (viii) B1<2:4273

Suppose B1≥2:4273. Consider following two cases:

Case (i) B5>B6

Here B5>each of B6,…,B12 as B5 ≥ B1>1.137 >each of B7,…, B12. Also B2B3B4<2:2254706×1:9517×1:646<7:15. So EquationHence the inequality (4; 8*) holds. This gives Equation

Left side is an increasing function of B2B3B4 and decreasing function of B1. So we can replace B2B3B4 by 7.15 and B1 by 2.4273 to get a contradiction.

Case (ii) Equation

Using (5.1) we have Equation <1:1589 and so Equation Therefore Equation as B2>2:0299 and B3<1:9517. Also from Claim (iv), B6>1:019>each of B7,….,B12. Hence the inequality (1; 4; 7*) holds. This gives Equation 7(B1B2B3B4B5)-1/7>13: Left side is an increasing function of B3B4B5 and B1 and a decreasing function of B2. One can check that inequality is not true for B3B4B5<1:9517×1:5452 ×1:1589, B1<2:5217871 and for B2>2:0299: Hence we must have B1<2:4273:

Claim (ix) B5<1:396

Suppose Equation From (5.1), B6B7B8<0:925 and B10B11B12<0:104, so we have Equation and Equation Applying AMGM to the inequality (1,2,1,4,4) we get Equation+ Equation We find that left side is a decreasing function of B2 and B9. So we replace B2 by 2:0299 and B9 by ε B5. Now it becomes a decreasing function of B5 and an increasing function of B1 so replacing

B5 by 1.396 and B1 by 2.4273, we find that above inequality is not true for 1:522<B3<1:9517 and 1:426<B4<1:646, giving thereby a contradiction. Hence we must have B5<1:396.

Claim (x) B3>1:7855

Suppose Equation We have B4>1:426>each of B5;B6,…,B12, hence the inequality (1; 2; 9*) holds. It gives Equation It is easy to check that left side of above inequality is a decreasing function of B2 and an increasing function of B1 and B3. So replacing B1 by 2.4273, B3 by 1.7855 and B2 by 2.0299 we get -15<13; a contradiction. Hence we have B3>1:7855.

Claim (xi)Equation

Suppose Equation We have B3>1:7855>each of B4;B5,…,B12, hence the inequality (2; 10*) holds. It gives EquationThe left side is a decreasing function of B1 and an increasing function of B2, so replacing B1 by 2:17 and B2 by 2.0733 we get Equation a contradiction.

Claim (xii) B7<0:92 and B5<1:38

Suppose Equation Here we have B4B5B6<2:67 and B8B9B10<0:295, so Equation and EquationAlso Equation Applying AM-GM to the inequality (2,4,4,2) we get Equation We find that left side is a decreasing function of B1 and B11. So we can replace B1 by 2:17 and B11 by ε B7. Then left side becomes a decreasing function of B7 and an increasing function of B2, so can replace B7 by 0.92 and B2 by 2.2254706 to see that Equation for 1:7855<B3<1:9517 and 0:3376<B12<0:4156, a contradiction. Hence B7<0:92. Further Equation gives B5<1:38.

Claim (xiii) B6<1:097

Suppose Equation Here we have B3B4B5<4:44 and B7B8B9<0:5, so Equation and Equation Also Equation Applying AM-GM to the inequality (1,4,4,2,1) we get EquationWe find that left side is a decreasing function of B10, B12 and B11. So we can replace B10 by εB6 and B12 by 0.3376 and B11 by EquationB6. Then left side becomes a decreasing function of B6, so we can replace B6 by 1.097 to find that EquationHence we must have B6<1:097.

Claim (xiv) B5>B6 and Equation

First suppose B5 ≤ B6, then B4B5B6<1:646 × 1:0972<1:981 and Equation Also Equationeach of B8,…,B12. Hence the inequality (2; 4; 6*) holds, i.e. EquationNow the left side is a decreasing function of B1 and B3 as well; also it is an increasing function of B2 and B4B5B6. But one can check that this inequality is not true for B1>2:17, B3>1:7855, B2<2:2254706 and B4B5B6<1:981, giving thereby a contradiction. Further suppose Equation then as B5>B6>1:019>each of B7,…, B12, the inequality (4; 8*) holds. Now working as in Case (i) of Claim (viii) we get contradiction for B1>2:17 and B2B3B4<2:2254706 × 1:9517 × 1:646<7:14934.

Claim (xv) B3<1:9 and B1<2:4056

Suppose Equation,then for B4B5B6<1:646×1:38×1:097<2:492, Equation Also B7≥ε B3>0:89>each of B8,…,B12. Hence the inequality (2; 4; 6*) holds. Now working as in Claim (xiv) we get contradiction for B1>2:17, B2<2:2254706, B3>1:9 and B4B5B6<2:492. So B3<1:9. Further if B1≥ 2:4056, then Equation Equationcontradicting Claim (xiv).

Claim (xvi) B4<1:58 and B1<2:373

Suppose Equation then for B5B6B<1:38 × 1:097 × 0:92<1:393, Equation Also B8 Equation B4>0:74>each of B9,…,B12. Hence the inequality (1; 2; 4; 5*) holds, i.e. -19=B1 + 4B2 -Equation Left side is a decreasing function of B2 and B4.

So we replace B2 by 2.0733 and B4 by 1.58. Then it becomes an increasing function of B1, B3 and B5B6B7. So we replace B1 by 2.4056, B3 by 1.9 and B5B6B7 by 1.393 to find that -19<13, a contradiction. Further if B1 ≥ 2:373, then Equation contradicting Claim (xiv).

Final Contradiction:

We have B3B4B<1:9×1:58×1:38<4:15. Therefore Equation >2 Also B6>1:019>each of B7,…,B12. Hence the inequality (1; 4; 7*) holds. Now we get contradiction working as in Case (ii) of Claim (viii).

5.2 2:17<B12<2:2254706

Here Equation 2.17Using Lemma 3 and 4, we have

B11=(B1B2…B10B12)-1<Equation

Claim (i) Either B11<0:4307 or B11>1:818

Suppose 0:4307 ≤B11≤1.818 The inequality (10*; 1; 1) gives 10:3 Equationwhich is not true for 0:4307 ≤B11≤1.818 and 2:17<B12<2:2254706. So we must have either B11<0:4307 or B11>1:818

Claim (ii) B11<0:4307

Suppose B11 ≥ 0.4307 then using Claim(i) we have B11>1:818. Now we have using Lemmas 3 & 4,

Equation This gives B2<1:777.

EquationThis gives B3<1:487

EquationThis gives B4<1:213.

EquationThis gives B6<0:826.

Equation This gives B7<0:697.

EquationThis gives B8<0:559.

EquationThis gives B9<0:478.

Equation

Therefore we have Equationand Equation each of B6,…,B10. So the inequality (4; 6*; 1; 1) holds, i.e.4B1 Equation (B1B2B3B4B11B12)-1/6 +B11+B12>13 Now the left side is an increasing function of B2B3B4, B11 and of B12 as well. Also it is a decreasing function of B1. So we replace B2B3B4 by 1:777 × 1:487 × 1:213, B11 by 1.8223, B12 by 2.2254706 and B1 by 2.17 to arrive at a contradiction. Hence we must have B11<0:4307.

Claim (iii) B10<0:445

Suppose Equation then 2B10>B11. So the inequality (9*; 2; 1) holds, i.e. Equation Equation B10 and B12>2:2254706, the left side is an increasing function of B12 and a decreasing function of B10, so replacing B12 by 2.2254706 and B10 by 0.445 we find that Equation,for 3 4(0:445)<B11<0:4307, a contradiction. Hence we must have B10<0:445.

Using Lemmas 3 and 4 we have:

Equation

Equation

Equation

Claim (iv) B3<1:62

Suppose Equation From (5.2), we have B4B5B6<1:712 and B8B9B10<0:178, so Equation and Equation Applying AM-GM to the inequality (2,4,4,1,1) we get Equation We find that left side is a decreasing function of B1, B7 and B11. So we can replace B1 by B12, B7 by ε B3 and B11 by e 2B3. Then it becomes a decreasing function of B3, so replacing B3 by 1.62 we find that φ 21<13; for 1:6275<B2<2:0255 and 2:17<B12<2:2254706, a contradiction. Hence we must have B3<1:62.

Claim (v) B12>2:196

Suppose Equation From (5.2), we have B2B3B4<4:674 and Equation Also Equation each of B6,…, B11. Therefore the inequality (4; 7*; 1) holds, i.e. EquationLeft side is an increasing function of B2B3B4 and of B12 as well. Also it is a decreasing function of B1. So we can replace B2B3B4 by 4.674, B12 by 2.196 and B1 by 2.17 to get Φ 22<13, a contradiction. Hence we must have B12>2:196.

Final Contradiction

Now we have B1 ≥B12>2:196. We proceed as in Claim(v) and use (4; 7*; 1). Here we replace B2B3B4 by 4.674, B12 by 2.2254706 and B1 by 2.196 to get Φ22<13, a contradiction.

References

Select your language of interest to view the total content in your interested language
Post your comment

Share This Article

Article Usage

  • Total views: 12242
  • [From(publication date):
    April-2015 - Sep 21, 2019]
  • Breakdown by view type
  • HTML page views : 8450
  • PDF downloads : 3792
Top