Next Article in Journal
Permutation Variation and Alternative Hyper-Sphere Decomposition
Previous Article in Journal
Single-Block Recursive Poisson–Dirichlet Fragmentations of Normalized Generalized Gamma Processes
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

New Sharp Double Inequality of Becker–Stark Type

Department of Mathematics, Zhejiang Gongshang University, Hangzhou 310018, China
Mathematics 2022, 10(4), 558; https://doi.org/10.3390/math10040558
Submission received: 26 January 2022 / Revised: 6 February 2022 / Accepted: 10 February 2022 / Published: 11 February 2022
(This article belongs to the Section Computational and Applied Mathematics)

Abstract

:
In this paper, we establish new sharp double inequality of Becker–Stark type by using a role of the monotonicity criterion for the quotient of power series and the estimation of the ratio of two adjacent even-indexed Bernoulli numbers. The inequality results are better than those in the existing literature.

1. Introduction

Becker and Stark [1] (or see Kuang [2] (5.1.102, p. 398)) obtained the following two-sided rational approximation for ( tan x ) / x :
Proposition 1.
Let 0 < x < π / 2 , then
8 π 2 4 x 2 < tan x x < π 2 π 2 4 x 2 .
Furthermore, 8 and π 2 are the best constants in (1).
In [3], Zhu and Hua obtained the following further result.
Proposition 2.
Let 0 < x < π / 2 , then
π 2 + 4 ( 8 π 2 ) x 2 / π 2 π 2 4 x 2 < tan x x < π 2 + π 2 12 x 2 / 3 π 2 4 x 2 .
Furthermore, α = 4 ( 8 π 2 ) / π 2 and β = π 2 12 / 3 are the best constants in (2).
Moreover, ref. [3] established a general refinement of the Becker–Stark inequalities as follows.
Proposition 3.
Let 0 < x < π / 2 , n , N 0 be natural numbers, B 2 n be the even-indexed Bernoulli numbers,
p n = 2 2 n + 2 ( 2 2 n + 2 1 ) π 2 ( 2 n + 2 ) ! | B 2 n + 2 | 2 2 n + 2 ( 2 2 n 1 ) ( 2 n ) ! | B 2 n | ,
and
α = 8 p 0 p 1 ( π / 2 ) 2 p N ( π / 2 ) 2 N ( π / 2 ) 2 N + 2 , β = p N + 1 .
Then
P 2 N ( x ) + α x 2 N + 2 π 2 4 x 2 < tan x x < P 2 N ( x ) + β x 2 N + 2 π 2 4 x 2
holds, where P 2 N ( x ) = a 0 + a 1 x 2 + + a N x 2 N . Furthermore, β and α are the best constants in (4).
Zhu [4] obtained a refinement of the Becker–Stark inequalities (1) in another way as follows.
Proposition 4.
Let 0 < x < π / 2 , then
8 π 2 4 x 2 + 2 π 2 π 2 9 6 π 4 ( π 2 4 x 2 ) < tan x x < 8 π 2 4 x 2 + 2 π 2 10 π 2 π 4 ( π 2 4 x 2 ) .
Furthermore, π 2 9 / 6 π 4 and 10 π 2 / π 4 are the best constants in (5).
Obviously, letting N = 0 in (4) gives (2). The double inequalities (2) and (5) can deduce the inequalities (1). Moreover, the left inequalities of (2) and (5) are not included each other while the upper estimate in (5) is less than the one in (2).
Numerous discussions on Becker–Stark inequality can be found in [5,6,7,8,9,10,11,12], as well as references therein. In 2015, Banjac, Markragić and Malešević [11] obtained the following results about the function ( tan x ) / x .
Proposition 5.
Let 0 < x < π / 2 , then
π 2 + π 2 3 4 x 2 + ( π 2 18 2 3 ) x 4 π 2 4 x 2 < tan x x < π 2 π 2 16 x 2 + 1 2 x 4 1 π 2 x 6 π 2 4 x 2
and
π 2 + π 2 12 3 x 2 + 384 4 π 4 3 π 4 x 4 π 2 4 x 2 < tan x x < π 2 + 72 8 π 2 π 2 x 2 + 16 π 2 160 π 4 x 4 π 2 4 x 2 .
Bagul and Chesneau looked closely at the Becker–Stark inequality from another perspective, and in [12] they got the following result.
Proposition 6.
For any x 0 , π / 2 , we have
1 + 128 π 4 x 2 5 π 2 12 x 2 π 2 4 x 2 2 < tan x x < 1 + 2 π 2 5 x 2 5 π 2 12 x 2 π 2 4 x 2 2
holds with the best constants 128 / π 4 and 2 π 2 / 5 .
Inspired by the above ideas, this paper considers the power series expansion of the following function
tan x x 2 1 π 2 4 x 2 2 x 2 λ x 2 = 2 3 π 4 λ + π 2 45 30 π 2 λ 240 17 π 2 λ 2 x 2 + 1 315 210 π 4 + λ 119 π 4 1680 π 2 + λ 2 62 π 4 952 π 2 + 3360 λ 3 x 4 + O x 6 .
Letting λ = 30 π 2 / 240 17 π 2 we can obtain the expression of the above function without x 2 , and draw the following inequality conclusion by using the property for the ratio of two adjacent even-indexed Bernoulli numbers and a role of the monotonicity criterion for the quotient of power series.
Theorem 1.
Let 0 < x < π / 2 . Then the double inequality
1 + 240 17 π 2 π 2 45 x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2 < tan x x < 1 + 240 17 π 2 1024 π 4 17 π 2 120 x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2
holds with the best constants 240 17 π 2 π 2 / 45 and 240 17 π 2 1024 / π 4 17 π 2 120 .

2. Lemmas

In order to prove the main conclusion of this paper, the following three lemmas are needed.
Lemma 1
([13]). Let B 2 n be the even-indexed Bernoulli numbers, we have the following power series expansion
tan x = n = 1 2 2 n 1 2 n ! 2 2 n | B 2 n | x 2 n 1 ,
holds for all x ( π / 2 , π / 2 ) .
Lemma 2
([14,15,16,17]). Let B 2 n be the even-indexed Bernoulli numbers, we have
2 2 n 1 1 2 2 n + 1 1 ( 2 n + 2 ) ( 2 n + 1 ) π 2 < | B 2 n + 2 | | B 2 n | < 2 2 n 1 2 2 n + 2 1 ( 2 n + 2 ) ( 2 n + 1 ) π 2 .
Lemma 3
([18]). Let a n and b n ( n = 0 , 1 , 2 , · · · ) be real numbers, and let the power series A ( x ) = n = 0 a n x n and B ( x ) = n = 0 b n x n be convergent for x < R ( R + ) . If b n > 0 for n = 0 , 1 , 2 , , and if ε n = a n / b n is strictly increasing (or decreasing) for n = 0 , 1 , 2 , , then the function A ( x ) / B ( x ) is strictly increasing (or decreasing) on ( 0 , R ) ( R + ) .

3. Proof of Theorem 1

Proof. 
Let
F ( x ) = tan x x 2 1 π 2 4 x 2 2 x 2 30 π 2 240 17 π 2 x 2 ,   0 < x < π 2 .
Then we can rewrite F ( x ) as
F ( x ) = tan x x 2 1 x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2 : = A ( x ) B ( x ) ,
where
A ( x ) = tan x x 2 1 , B ( x ) = x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2 .
By Lemma 1 we have
tan 2 x = sec 2 x 1 = tan x 1 = n = 1 2 2 n 2 2 n 1 2 n 1 2 n ! | B 2 n | x 2 n 2 1 = n = 2 2 2 n 2 2 n 1 2 n ! | B 2 n | x 2 n 2 ,
and
A ( x ) = n = 2 2 2 n 2 2 n 1 2 n 1 2 n ! | B 2 n | x 2 n 4 1 = n = 3 2 2 n 2 2 n 1 2 n 1 2 n ! | B 2 n | x 2 n 4 = : n = 3 a n x 2 n 4 ,
where
a n = 2 2 n 2 2 n 1 2 n 1 2 n ! | B 2 n | ,   n 3 .
At the same time, since
1 π 2 4 x 2 2 = n = 3 n 2 4 n 3 π 2 n 2 x 2 n 6 ,
we have
B ( x ) = x 2 30 π 2 240 17 π 2 x 2 n = 3 n 2 4 n 3 π 2 n 2 x 2 n 6 = 30 π 2 240 17 π 2 n = 3 n 2 4 n 3 π 2 n 2 x 2 n 4 n = 3 n 2 4 n 3 π 2 n 2 x 2 n 2 = 30 π 2 240 17 π 2 x 2 + 30 π 2 240 17 π 2 n = 4 n 2 4 n 3 π 2 n 2 x 2 n 4 n = 3 n 2 4 n 3 π 2 n 2 x 2 n 2 = 30 π 2 240 17 π 2 x 2 + 30 π 2 240 17 π 2 n = 4 n 2 4 n 3 π 2 n 2 x 2 n 4 n = 4 n 3 4 n 4 π 2 n 4 x 2 n 4 = 30 π 2 240 17 π 2 x 2 + n = 4 30 π 2 240 17 π 2 n 2 4 n 3 π 2 n 2 n 3 4 n 4 π 2 n 4 x 2 n 4 = : n = 3 b n x 2 n 4 ,
where
b 3 = 30 π 2 240 17 π 2 , b n = 30 π 2 240 17 π 2 n 2 4 n 3 π 2 n 2 n 3 4 n 4 π 2 n 4 ,   n 4 .
We find that
a 3 b 3 = a 4 b 4 = π 2 240 17 π 2 45 15.839 , a 5 b 5 = 31 π 6 1260 240 17 π 2 17 π 2 60 15.848 , a 6 b 6 = 691 π 8 340,200 240 17 π 2 17 π 2 80 15.855 ,
and can prove the fact that a n / b n n 6 is increasing, which is equivalent to: for all n 6 ,
a n b n < a n + 1 b n + 1 2 2 n 2 2 n 1 2 n 1 2 n ! B 2 n 30 π 2 240 17 π 2 n 2 4 n 3 π 2 n 2 n 3 4 n 4 π 2 n 4 < 2 2 n + 2 2 2 n + 2 1 2 n + 1 2 n + 2 ! B 2 n + 2 30 π 2 240 17 π 2 n 1 4 n 2 π 2 n n 2 4 n 3 π 2 n 2
B 2 n + 2 B 2 n > 2 2 n 2 2 n 1 2 n 1 2 n ! 2 2 n + 2 2 2 n + 2 1 2 n + 1 2 n + 2 ! 30 π 2 240 17 π 2 n 1 4 n 2 π 2 n n 2 4 n 3 π 2 n 2 30 π 2 240 17 π 2 n 2 4 n 3 π 2 n 2 n 3 4 n 4 π 2 n 4 .
From Lemma 2, the last inequality above holds as long as the following inequality is proved
2 2 n 1 1 2 2 n + 1 1 ( 2 n + 2 ) ( 2 n + 1 ) π 2 > 2 2 n 2 2 n 1 2 n 1 2 n ! 2 2 n + 2 2 2 n + 2 1 2 n + 1 2 n + 2 ! 30 π 2 240 17 π 2 n 1 4 n 2 π 2 n n 2 4 n 3 π 2 n 2 30 π 2 240 17 π 2 n 2 4 n 3 π 2 n 2 n 3 4 n 4 π 2 n 4 Δ : = 2 2 n 1 1 2 2 n + 1 1 π 2 > 2 2 n 1 2 n 1 4 2 2 n + 2 1 2 n + 1 30 π 2 240 17 π 2 n 1 4 n 2 π 2 n n 2 4 n 3 π 2 n 2 30 π 2 240 17 π 2 n 2 4 n 3 π 2 n 2 n 3 4 n 4 π 2 n 4 = : .
We compute to obtain
Δ = 1 2 h ( n ) π 2 4 × 2 2 n 1 2 × 2 2 n 1 2 n + 1 17 π 2 n 120 n 51 π 2 + 480 ,
where
h ( n ) = 20 168 17 π 2 2 2 n u ( n ) × 2 2 n + 1680 170 π 2 . u ( n ) = 3 2 n 2 17 π 2 120 + 5 n 168 17 π 2 221 π 2 + 2160 20 168 17 π 2 ,
Since 1680 170 π 2 > 0 , the fact h ( n ) > 0 holds for all n 6 when proving
2 2 n > u ( n ) ,   n 6 .
It is not difficult to prove (11) by mathematical induction. First, the above formula (11) is true for n = 6 due to
2 12 u ( 6 ) = 1 20 1,394,119 π 2 13,766,880 17 π 2 168 1725 > 0 .
Second, suppose (11) holds for m, that is
2 2 m > u ( m ) ,   m 6 .
Next, by (12) we have
2 2 m + 2 = 4 · 2 2 m > 4 · u ( m ) .
The inequality (11) is proved when proving
4 · u ( m ) > u ( m + 1 ) ,
that is
4 u ( m ) u ( m + 1 ) > 0 .
In fact,
4 u ( m ) u ( m + 1 ) = 3 20 102 π 2 720 m 6 2 + 901 π 2 5640 m 6 + 1122 π 2 2040 168 17 π 2 > 0
holds for all m 6 .
So a n / b n n 3 is increasing. From Lemma 3 we have that the function F ( x ) is increasing on 0 , π / 2 . Considering the reasons
lim x 0 + F ( x ) = π 2 240 17 π 2 45 15.839 , lim x π 2 F ( x ) = 1024 240 17 π 2 π 4 17 π 2 120 15.888 ,
the proof of Theorem 1 is completed. □

4. Remarks

Through the following analysis, we conclude that the left-hand side inequality of 9 does not match the one of 8 while the right-hand side inequality of 9 is better than the one of 8 .
Remark 1.
The left-hand side inequality of 9 does not match the one of 8 due to
1 + 240 17 π 2 π 2 45 x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2 > 1 + 128 π 4 x 2 5 π 2 12 x 2 π 2 4 x 2 2 240 17 π 2 π 2 45 x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2 > 128 π 4 x 2 5 π 2 12 x 2 π 2 4 x 2 2 240 17 π 2 π 2 45 30 π 2 240 17 π 2 x 2 > 128 π 4 5 π 2 12 x 2 240 π 6 17 π 8 69,120 x 2 < 30 π 8 28,800 π 2
x < 30 π 8 28,800 π 2 240 π 6 17 π 8 69,120 1.1549
Remark 2.
Since
1 + 240 17 π 2 1024 π 4 17 π 2 120 x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2 < 1 + 2 π 2 5 x 2 5 π 2 12 x 2 π 2 4 x 2 2 240 17 π 2 1024 π 4 17 π 2 120 30 π 2 240 17 π 2 x 2 < 2 π 2 5 π 2 12 x 2 5 4 x 2 10,880 π 2 360 π 6 + 51 π 8 153,600 < 85 π 10 600 π 8 76,800 π 2
x < 1 2 76,880 π 2 600 π 8 + 85 π 10 10,880 π 2 360 π 6 + 51 π 8 153,600 2.0294 ,
we can obtain that the right-hand side inequality of 9 is better than the one of 8 on 0 , π / 2 .

5. Conclusions

This paper established a new sharp double inequality of Becker–Stark type
1 + 240 17 π 2 π 2 45 x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2 < tan x x < 1 + 240 17 π 2 1024 π 4 17 π 2 120 x 2 30 π 2 240 17 π 2 x 2 π 2 4 x 2 2 ,
which holds for x ( 0 , π / 2 ) , where 240 17 π 2 π 2 / 45 and 240 17 π 2 1024 / π 4 17 π 2 120 are the best possible.

Funding

This research received no external funding.

Acknowledgments

The author is thankful to reviewers for reviewers’ careful corrections to and valuable comments on the original version of this paper.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Becker, M.; Stark, E.L. On a hierarchy of quolynomial inequalities for tan x. Publ. Elektrotehničkog Fak. Ser. Mat. i Fiz. 1978, 602–633, 133–138. [Google Scholar]
  2. Kuang, J.-C. Applied Inequalities, 5th ed.; Shangdong Science and Technology Press: Jinan, China, 2021. [Google Scholar]
  3. Zhu, L.; Hua, J.-K. Sharpening the Becker-Stark inequalities. J. Inequal. Appl. 2010, 2010, 931275. [Google Scholar] [CrossRef] [Green Version]
  4. Zhu, L. A refinement of the Becker-Stark inequalities. Math. Notes 2013, 93, 421–425. [Google Scholar] [CrossRef]
  5. Chen, C.-P.; Cheung, W.-S. Sharp Cusa and Becker-Stark inequalities. J. Inequal. Appl. 2011, 2011, 136. [Google Scholar] [CrossRef] [Green Version]
  6. Wu, Y.T.; Bercu, G. New refinements of Becker-Stark and Cusa-Huygens inequalities via trigonometric polynomials method. Rev. Real Acad. Cienc. Exactas Físicas Nat. Ser. A Mat. 2021, 115, 87. [Google Scholar] [CrossRef]
  7. Nenezić, M.; Zhu, L. Some improvements of Jordan-Stečkin and Becker-Stark Inequalities. Appl. Anal. Discrete Math. 2018, 12, 244–256. [Google Scholar] [CrossRef] [Green Version]
  8. Zhu, L. Sharp Becker-Stark-Type inequalities for Bessel functions. J. Inequal. Appl. 2010, 2010, 838740. [Google Scholar] [CrossRef] [Green Version]
  9. Nishizawa, Y. Sharp Becker-Stark-type inequalities with power exponential functions. J. Inequal. Appl. 2015, 2015, 402. [Google Scholar] [CrossRef] [Green Version]
  10. Chouikha, A.R. New sharp inequalities related to classical trigonometric inequalities. J. Inequal. Spec. Func. 2020, 11, 27–35. [Google Scholar]
  11. Banjac, B.; Makragić, M.; Maleševixcx, B. Some notes on a method for proving inequalities by computer. Results Math. 2016, 69, 161–176. [Google Scholar] [CrossRef] [Green Version]
  12. Bagul, Y.J.; Chesneau, C. New sharp bounds for tangent function. Bull. Alla. Math. Soc. 2020, 34, 277–282. [Google Scholar]
  13. Jeffrey, A. Handbook of Mathematical Formulas and Integrals, 3rd ed.; Elsevier: San Diego, CA, USA, 2004. [Google Scholar]
  14. Qi, F. A double inequality for the ratio of two non-zero neighbouring Bernoulli numbers. J. Comput. Appl. Math. 2019, 351, 1–5. [Google Scholar] [CrossRef]
  15. Zhu, L. New bounds for the ratio of two adjacent even-indexed Bernoulli numbers. Rev. Real Acad. Cienc. Exactas Físicas y Nat. Ser. A Mat. 2020, 114, 83. [Google Scholar] [CrossRef]
  16. Zhu, L. Monotonicities of some functions involving multiple logarithm function and their applications. Rev. Real Acad. Cienc. Exactas FíSicas Nat. Ser. A Mat. 2020, 114, 139. [Google Scholar] [CrossRef]
  17. Yang, Z.H.; Tian, J.F. Sharp bounds for the ratio of two zeta functions. J. Comput. Appl. Math. 2020, 364, 112359. [Google Scholar] [CrossRef]
  18. Biernacki, M.; Krzyz, J. On the monotonicity of certain functionals in the theory of analytic functions. Can. Math. Bull. 1955, 2, 134–145. [Google Scholar]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Zhu, L. New Sharp Double Inequality of Becker–Stark Type. Mathematics 2022, 10, 558. https://doi.org/10.3390/math10040558

AMA Style

Zhu L. New Sharp Double Inequality of Becker–Stark Type. Mathematics. 2022; 10(4):558. https://doi.org/10.3390/math10040558

Chicago/Turabian Style

Zhu, Ling. 2022. "New Sharp Double Inequality of Becker–Stark Type" Mathematics 10, no. 4: 558. https://doi.org/10.3390/math10040558

APA Style

Zhu, L. (2022). New Sharp Double Inequality of Becker–Stark Type. Mathematics, 10(4), 558. https://doi.org/10.3390/math10040558

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop