This paper derives the exact maximum value of a linear combination of five or seven cosines with positive angles summing to pi, expressed as a rational function of the coefficients, using algebraic methods.
Contribution
It provides a novel algebraic approach to determine the sharp bounds of cosine combinations with positive angles summing to pi.
Findings
01
Derived explicit sharp bounds as positive real fractions.
02
Extended the inequality analysis to five and seven cosine cases.
03
Used algebraic transformations to simplify the bounds expression.
Abstract
Given a positive linear combination of five (respectively seven) cosines, where the angles are positive and sum to pi, the aim of this article is to express the sharp bound of the combination as a Positive Real Fraction in the coefficients (hence cosine-free). The method uses algebraic and arithmetic manipulations with judicious transformations.
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsMathematics and Applications · History and Theory of Mathematics · graph theory and CDMA systems
Full text
The Pentagonal Inequality
Roy Barbara
1. Introduction
We recall the following result, conjectured by L. Fejes Tóth (see [References]) and proved by Lenhard H-C (see [References]).
Theorem 0**.**
(Tóth’s inequality) Let n≥3. Let x1,x2,...,xn and α1,α2,...,αn be positive real numbers with i=1∑nαi=π. Then, with xn+1=x1, we have the following inequality:
[TABLE]
Our interest lies in a sharp majoration of a sum as i=1∑naicosαi, where ai,αi>0 and i=1∑nαi=π. Results for the cases n=3,4 can be found in [References]. In this article, we focus our attention on the case n=5. In section 4, we highlight the sharpness of the result. In section 5, we quickly study the case n=7. The case n=6 remains open.
2. The Results
We present the Pentagonal Inequality in two forms: the strong form (theorem 1) and the normal form (theorem 2). Though theorem 1 is more precise, theorem 2 remains sometimes more practical.
Theorem 1**.**
Let a1,a2,...,a5 and α1,α2,...,α5 be positive real numbers with i=1∑5αi=π and such that a1≤a2≤a3≤a4≤a5. Then, we have the sharp inequality:
[TABLE]
[TABLE]
[TABLE]
Theorem 2**.**
Let a1,a2,...,a5 and α1,α2,...,α5 be positive real numbers with i=1∑5αi=π. Then, we have:
[TABLE]
[TABLE]
[TABLE]
We briefly indicate that the case of equality in either theorem 1 or 2 requires the condition:
[TABLE]
3. Proofs of Theorems 1 and 2
Let a1,a2,...,a5 be positive real numbers. A circular permutation of the ai can be represented by σ=(x1,x2,...,x5), with x1=a1, and where x2,x3,x4,x5 are a2,a3,a4,a5 in some order. There are 4!=24 such permutations.
In all what follows, we use the following notation:
Notation: Let σ=(x1,x2,...,x5) be a circular permutation of a1,a2,...,a5 (with x1=a1). We define ϕ(σ), also denoted by ϕ(x1,x2,...,x5) as follows:
[TABLE]
Lemma 1**.**
Let a1≤a2≤...≤a5 be positive real numbers. Then, there is a circular permutation of the ai, say σ0, such that ϕ(σ0) minimizes all the ϕ(σ). Namely, σ0 is the cycle σ0=(a1,a5,a2,a3,a4).
Equivalently, one might take σ1=(a1,a4,a3,a2,a5) since ϕ(σ1)=ϕ(σ0).
Proof.
After some amount of algebra, we find the following identities (and the result follows):
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
[TABLE]
∎
Lemma 2**.**
Let a1,a2,...,a5 and α1,α2,...,α5 be positive real numbers with i=1∑5αi=π. If σ=(b1,b2,...,b5) is any circular permutation of the ai (with b1=a1), then we have:
[TABLE]
Proof.
Set β1=α1,β2=α4,β3=α2,β4=α5, and β5=α3 (so i=1∑5βi=π).
We have x12=P−1(b12b22b32),x32=P−1(b22b32b42),x52=P−1(b32b42b52),x22=P−1(b42b52b12), and x42=P−1(b52b12b22). Summing, we obtain:
[TABLE]
On the other hand we have: x1x2cosβ1=b1cosα1,x2x3cosβ2=b4cosα4,x3x4cosβ3=b2cosα2,x4x5cosβ4=b5cosα5, and x5x1cosβ5=b3cosα3.
Hence, with x6=x1, we get:
[TABLE]
By theorem ‣ 1, we have i=1∑5xixi+1cosβi≤cos5π.i=1∑5xi2. From this, (2), (3), and cos5π=41+5, we obtain (1).
∎
Proofs of theorems 1 and 2: To get theorem 2, just apply lemma 2 to σ=(a1,a2,a3,a4,a5). To get theorem 1, apply lemma 2 to the cycle σ0=(a1,a5,a2,a3,a4), obtaining, in virtue of lemma 1, the sharpest form of the Pentagonal Inequality.
Note the following: If a1<a2<...<a5 (or more simply if a1≤a2≤...≤a5 and a2<a4), then, ϕ(a12,a52,a22,a32,a42)<ϕ(a12,a22,a32,a42,a52). Hence, in this case, while equality in theorem 1 might be reached, equality in theorem 2 never arises.
4. Sharpness of the Pentagonal Inequality
To see why the Pentagonal Inequality is sharp, we prove that Tóth’s inequality for n=5 can be derived from the Pentagonal Inequality (the normal form suffices): Indeed, let x1,x2,...,x5 and α1,α2,...,α5 be positive real numbers with i=1∑5αi=π. Given by hypothesis that theorem 2 holds, we prove (with x6=x1) that
[TABLE]
Set β1=α1,β2=α3,β3=α5,β4=α2, and β5=α4 (so i=1∑5βi=π). Set also a1=x1x2,a2=x3x4,a3=x5x1,a4=x2x3, and a5=x4x5.
2[2] Lenhard H-C. , Verallgemeinerung and Verscharfung der Erdos-Mordellschen Ungleichung fur Polygone , Arch. Math. Vol. XII, 1961, 311-314.
3[3] Faruk F. Abi-Khuzam , A trigonometric inequality and its geometric applications , Mathematical Inequalities & Applications, Vol. 3, Number 3, July 2000, 437-442.