Next Article in Journal
Adaptation of Residual-Error Series Algorithm to Handle Fractional System of Partial Differential Equations
Previous Article in Journal
SMIS: A Stepwise Multiple Integration Solver Using a CAS
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

New Bounds for the Modified Bessel Function of the First Kind and Toader-Qi Mean

Department of Mathematics, Zhejiang Gongshang University, Hangzhou 310018, China
Mathematics 2021, 9(22), 2867; https://doi.org/10.3390/math9222867
Submission received: 3 October 2021 / Revised: 27 October 2021 / Accepted: 9 November 2021 / Published: 11 November 2021
(This article belongs to the Section Mathematics and Computer Science)

Abstract

:
Let I p x be the modified Bessel function of the first kind of order p. The upper and lower bounds in the form of simple rational functions about cosh t and ( sinh t ) / t for the function I 0 x are obtained. The corresponding inequalities for the Toader-Qi mean do not match those in the existing literature.

1. Introduction

We know that the modified Bessel function of the first kind of order p is denoted as I p ( x ) (([1] p. 77), ([2] p. 374)) and has important applications in many fields of natural science. The function I p ( x ) denotes a class of solutions of the second-order differential equation ([2] p. 374)
x 2 y x + x y x x 2 + p 2 y x = 0 ,
which can be written as the infinite series
I p x = x 2 p n = 0 x / 2 2 n n ! Γ p + n + 1 , x R , p R \ { 1 , 2 , } ,
where Γ x is the gamma function ([2] p. 255). Moreover, the normalized modified Bessel function defined by I p : R [ 1 , + ) as I p ( x ) = 2 p Γ ( p + 1 ) x p I p ( x ) was considered in [3]. Then, we have
I p + 1 / 2 ( x ) = 2 p + 1 / 2 Γ ( p + 3 / 2 ) x ( p + 1 / 2 ) I p + 1 / 2 ( x ) .
Particularly, for p = 0 , p = 1 in the above, the function I p + 1 / 2 ( x ) reduces to some elementary functions, as follows ([1] p. 54):
I 1 / 2 ( x ) = π 2 x I 1 / 2 ( x ) = sinh x x , I 3 / 2 ( x ) = 3 x π 2 x I 3 / 2 ( x ) = 3 sinh x x 3 cosh x x 2 .
The properties and various inequalities of this special function I p ( x ) are widely discussed in References [3,4,5,6,7,8,9,10,11].
Throughout this article, we assume that a and b are two different positive real numbers. This paper focuses on
I 0 x = n = 0 x / 2 2 n n ! Γ n + 1 = n = 0 x 2 n 2 2 n n ! 2 ,
which is the modified Bessel function of the first kind of order 0, which is related to the Toader-Qi mean (see [12,13]) defined by
T Q a , b = 2 π 0 π / 2 a cos 2 θ b sin 2 θ d θ = a b I 0 ln a b .
Let
A A a , b = a + b 2 , G G a , b = a b , L L a , b = a b ln a ln b , I I a , b = e 1 b b a a 1 / b a
and
A ν A ν a , b = a ν + b ν 2 1 / ν if ν 0
be the arithmetic mean, geometric mean, logarithmic mean, exponential mean, and power mean of order ν , respectively. Then, an inequality chain (see [14,15,16])
G < L < A 1 / 3 < A + 2 G 3 < A 1 / 2 < 2 A + G 3 < A 2 / 3 < I < A ln 2 < A 1
holds. Alzer [17] also provided the following result:
A G < L I < L + I 2 < A + G 2 .
From now on, let us focus on the Toader-Qi mean of I 0 x . In 2017, Qi, Shi, Liu, and Yang [13] proved that the inequalities
L < T Q < A + G 2 < 2 A + G 3 < I
hold Yang and Chu ([18] Theorem 3.4) established the following results:
L 3 / 4 A 1 / 4 < T Q < L A 2 / 3 < 3 4 L + 1 4 A ,
2 A L π < T Q < A L ,
2 π A + 1 2 π L < T Q < λ 0 A + 1 λ 0 L ,
where λ 0 = 0.6766 Yang, Chu, and Song [19] and Yang, Tian, and Zhu ([20] Remark 4) obtained
e π L I < 2 4 / 3 π L A 2 / 3 < T Q < L A 2 / 3 < L I .
Let b > a > 0 and t = ln a / b . Then, those means mentioned above can be represented in terms of hyperbolic functions:
A G = cosh t , L G = sinh t t , I G = exp t tanh t 1 , T Q G = I 0 t , A p G = cosh 1 / p p t for p 0 .
Correspondingly, the inequalities (10)–(14) mentioned above are equivalent to
sinh t t < I 0 t < cosh t + 1 2 < 2 cosh t + 1 3 < exp t tanh t 1 ,
sinh t t 3 / 4 cosh t 1 / 4 < I 0 t < sinh t t cosh 3 / 2 2 t 3 < 3 4 sinh t t + 1 4 cosh t
sinh 2 t π t < I 0 t < sinh 2 t 2 t ,
2 π cosh t + 1 2 π sinh t t < I 0 t < λ 0 cosh t + 1 λ 0 sinh t t ,
and
e π sinh t t exp t tanh t 1 < 2 3 / 4 π sinh t t cosh 3 / 2 2 t 3 < I 0 t < sinh t t cosh 3 / 2 2 t 3 < sinh t t exp t tanh t 1
hold for t > 0 .
In this paper, we first consider finding the bound of T Q a , b in the following form:
x 1 A 2 + y 1 A L + z 1 G 2 A + w 1 G ,
which is equivalent to searching for a bound for T Q a , b / G = I 0 t in the form of
x 1 A 2 + y 1 A L + z 1 G 2 A + w 1 G G = x 1 A 2 + y 1 A L + z 1 G 2 A G + w 1 G 2 = x 1 A / G 2 + y 1 A / G L / G + z 1 A / G + w 1 = x 1 cosh t 2 + y 1 cosh t sinh t / t + z 1 cosh t + w 1 .
In this way, we can consider the power series expansion
cosh t + w 1 I 0 t x 1 cosh t 2 + y 1 cosh t sinh t / t + z 1 = x 1 y 1 z 1 + w 1 + 1 t 2 x 1 + 2 3 y 1 1 4 w 1 3 4 t 4 1 3 x 1 + 2 15 y 1 1 64 w 1 35 192 t 6 2 45 x 1 + 4 315 y 1 1 2304 w 1 77 3840 t 8 1 315 x 1 + 2 2835 y 1 1 147,456 w 1 143 114,688 t 10 2 14,175 x 1 + 4 155,925 y 1 1 14,745,600 w 1 46,189 928,972,800 + O t 12 .
When letting
x 1 y 1 z 1 + w 1 + 1 = 0 x 1 + 2 3 y 1 1 4 w 1 3 4 = 0 1 3 x 1 + 2 15 y 1 1 64 w 1 35 192 = 0 2 45 x 1 + 4 315 y 1 1 2304 w 1 77 3840 = 0
we can obtain that x 1 = 2263/11,456, y 1 = 20,475/22,912, z 1 = 1879/22,912 and w 1 = 31 / 179 and find that
cosh t + 31 179 I 0 t 2263 11,456 cosh t 2 + 20,475 22,912 sinh t cosh t t + 1879 22,912 = 209 21,995,520 t 8 16,217 15,242,895,360 t 10 + O t 12 ,
which motivates us to prove the inequality
I 0 t < 2263 11,456 cosh t 2 + 20,475 22,912 sinh t cosh t t + 1879 22,912 cosh t + 31 179 = 8284 t + 20,475 sinh 2 t + 4526 t cosh 2 t 256 t 179 cosh t + 31 , t > 0 .
Similarly, we can find the bound for T Q a , b in the form of
x 2 L 2 + y 2 L G + z 2 G 2 A + w 2 G
and obtain the exact bound by the approximation method used just now. The main conclusions of this paper are as follows.
Theorem 1.
The inequality
I 0 t < 2263 11,456 cosh t 2 + 20,475 22,912 sinh t cosh t t + 1879 22,912 cosh t + 31 179 = 8284 t + 20,475 sinh 2 t + 4526 t cosh 2 t 256 t 179 cosh t + 31
holds for t > 0 , or equivalently,
T Q a , b < 2263 11,456 A 2 + 20,475 22,912 A L + 1879 22,912 G 2 A + 31 179 G = 20,475 A L + 4526 A 2 + 1879 G 2 128 179 A + 31 G
holds for a , b > 0 with a b .
Theorem 2.
The inequality
I 0 t > 4659 568 sinh t t 2 + 7941 284 sinh t t 4877 568 cosh t + 1887 71 = 4659 cosh 2 t + 31,764 t sinh t 9754 t 2 4659 16 t 2 71 cosh t + 1887
holds for t > 0 , or equivalently,
T Q a , b > 4659 568 L 2 + 7941 284 L G 4877 568 G 2 A + 1887 71 G = 4659 L 2 + 15,882 G L 4877 G 2 8 71 A + 1887 G
holds for a , b > 0 with a b .

2. Lemmas

The precise power series representation of cosh t I 0 t is required to establish two theorems in this article. We thank Prof. Dr. H. W. Gould for compiling an important result on the finite sum of combinatorial numbers in his monograph [21], which we consider as the substance of Lemma 1.
Lemma 1
([21] (3.175)). Let n k be the number of different ways to choose k elements from a given set with n distinct elements, that is
n k = n ! k ! ( n k ) ! .
Then, for all x R ,
k = 0 n x 2 k x 2 k n k 2 2 k = 2 x 2 n .
Lemma 2.
We have
I 0 t cosh t = n = 0 4 n 1 ! ! 2 n ! 2 t 2 n = : n = 0 κ n t 2 n ,
where
κ n = 4 n 1 ! ! 2 n ! 2 .
Proof. 
Using the Cauchy product formula, we have
I 0 t cosh t = n = 0 1 2 2 n n ! 2 t 2 n n = 0 1 2 n ! t 2 n = n = 0 k = 0 n 1 2 2 k k ! 2 1 2 n 2 k ! t 2 n = : n = 0 κ n t 2 n ,
where
κ n = k = 0 n 1 2 2 k k ! 2 1 2 n 2 k ! .
Since
2 n 2 k = 2 n ! 2 k ! ( 2 n 2 k ) ! , 2 k ! = k ! 2 2 k k ,
we have
κ n = k = 0 n 1 2 2 k k ! 2 2 n 2 k ! = 1 2 n ! k = 0 n 2 k ! 2 2 k k ! 2 2 n 2 k = 1 2 n ! k = 0 n k ! 2 2 k k 2 2 k k ! 2 2 n 2 k = 1 2 n ! k = 0 n 1 2 2 k 2 n 2 k 2 k k .
Letting x = 2 n in (24) gives
k = 0 n 2 n 2 k 2 n 2 k n k 2 2 k = 4 n 2 n .
On the one hand, let n k = l . Then
k = 0 n 2 n 2 k 2 n 2 k n k 2 2 k = l = 0 n 2 n 2 n 2 l 2 l l 2 2 ( n l ) = 2 2 n l = 0 n 1 2 2 l 2 n 2 n 2 l 2 l l = 2 2 n k = 0 n 1 2 2 k 2 n 2 n 2 k 2 k k = 2 2 n k = 0 n 1 2 2 k 2 n 2 k 2 k k .
On the other hand, we have
4 n 2 n = ( 4 n ) ! ( 2 n ) ! 2 .
Therefore,
2 2 n k = 0 n 1 2 2 k 2 n 2 k 2 k k = ( 4 n ) ! ( 2 n ) ! 2 k = 0 n 1 2 2 k 2 n 2 k 2 k k = ( 4 n ) ! 2 2 n ( 2 n ) ! 2
and
κ n = 1 2 n ! ( 4 n ) ! 2 2 n ( 2 n ) ! 2 = 1 2 n ! ( 4 n ) ( 4 n 1 ) ( 4 n 2 ) ( 4 n 3 ) 4 · 3 · 2 · 1 2 2 n ( 2 n ) ! 2 = 1 2 n ! ( 4 n ) ( 4 n 2 ) 4 · 2 ( 4 n 1 ) ! ! 2 2 n ( 2 n ) ! 2 = 1 2 n ! 2 2 n ( 2 n ) ! ( 4 n 1 ) ! ! 2 2 n ( 2 n ) ! 2 = ( 4 n 1 ) ! ! ( 2 n ) ! 2 .
 □

3. Proofs of Theorems 1 and 2

3.1. Proof of Theorem 1

The inequality to be proved is
f 1 ( t ) = : 8284 + 20,475 1 t sinh 2 t + 4526 cosh 2 t > 256 179 cosh t + 31 I 0 t = : g 1 ( t ) , t > 0 .
Using the power series expansion of hyperbolic sine and hyperbolic cosine functions, we have
f 1 ( t ) = 8284 + 20,475 n = 0 2 2 n + 1 2 n + 1 ! t 2 n + 4526 n = 0 2 2 n 2 n ! t 2 n = 8284 + n = 0 20,475 · 2 2 n + 1 2 n + 1 ! + 4526 · 2 2 n 2 n ! t 2 n = 53,760 + 36,352 t 2 + 25,432 3 t 4 + 41,504 45 t 6 + n = 4 20,475 · 2 2 n + 1 2 n + 1 ! + 4526 · 2 2 n 2 n ! t 2 n .
Then, using Formulas (5) and (25), we obtain
g 1 ( t ) = 45,824 I 0 t cosh t + 7936 I 0 t = 45,824 n = 0 κ n t 2 n + 7936 n = 0 1 2 2 n n ! 2 t 2 n = n = 0 45,824 κ n + 7936 2 2 n n ! 2 t 2 n = 53,760 + 36,352 t 2 + 25,432 3 t 4 + 41,504 45 t 6 + n = 4 45,824 κ n + 7936 2 2 n n ! 2 t 2 n .
The inequality (27) is proved when the following inequality holds:
45,824 κ n + 7936 2 2 n n ! 2 < 20,475 · 2 2 n + 1 2 n + 1 ! + 4526 · 2 2 n 2 n ! , n 4
45,824 4 n 1 ! ! 2 n ! 2 + 7936 2 2 n n ! 2 < 20,475 · 2 2 n + 1 2 n + 1 ! + 4526 · 2 2 n 2 n ! , n 4
45,824 4 n 1 ! ! 2 n ! 2 < 20,475 · 2 2 n + 1 2 n + 1 ! + 4526 · 2 2 n 2 n ! 7936 2 2 n n ! 2 , n 4
Next, we prove (28) by mathematical induction. It is not difficult to verify that the above Formula (28) is true for n = 4 . Suppose (28) holds for m, that is
45,824 4 m 1 ! ! 2 m ! 2 < 20,475 · 2 2 m + 1 2 m + 1 ! + 4526 · 2 2 m 2 m ! 7936 2 2 m m ! 2 , m 4 .
By (29) we have
45,824 4 m + 3 ! ! 2 m + 2 ! 2 = 4 m + 3 4 m + 1 2 m + 2 2 2 m + 1 2 45,824 4 m 1 ! ! 2 m ! 2 < 4 m + 3 4 m + 1 2 m + 2 2 2 m + 1 2 20,475 · 2 2 m + 1 2 m + 1 ! + 4526 · 2 2 m 2 m ! 7936 2 2 m m ! 2 .
The inequality (28) is proved when proving
4 m + 3 4 m + 1 2 m + 2 2 2 m + 1 2 20,475 · 2 2 m + 1 2 m + 1 ! + 4526 · 2 2 m 2 m ! 7936 2 2 m m ! 2 < 20,475 · 2 2 m + 3 2 m + 3 ! + 4526 · 2 2 m + 2 2 m + 2 ! 7936 2 2 m + 2 m + 1 ! 2 ,
that is
4 m + 3 4 m + 1 2 m + 2 2 2 m + 1 2 20,475 · 2 2 m + 1 2 m + 1 ! + 4 m + 3 4 m + 1 2 m + 2 2 2 m + 1 2 4526 · 2 2 m 2 m ! 20,475 · 2 2 m + 3 2 m + 3 ! + 4526 · 2 2 m + 2 2 m + 2 ! < 4 m + 3 4 m + 1 2 m + 2 2 2 m + 1 2 7936 2 2 m m ! 2 7936 2 2 m + 2 m + 1 ! 2
or
2 × 2 2 m 70,909 m 68,754 m 2 + 36,208 m 3 + 6735 2 m + 3 ! m + 1 2 m + 1 2 < 3968 6 m + 6 m 2 + 1 m + 1 ! 2 2 2 m 2 m + 1 2 .
Since 36,208 m 3 − 68,754 m 2 − 70,909 m + 6735 > 0 for m 4 , the above inequality is obviously true.
The proof of the Theorem 1 has been completed.

3.2. Proof of Theorem 2

The conclusion to be proved is equivalent to
f 2 ( t ) = : 1136 t 2 cosh t + 30,192 t 2 I 0 t > 4659 cosh 2 t + 31,764 t sinh t 9754 t 2 4659 = : g 2 ( t ) , t > 0 .
Using Formulas (5) and (25), we obtain
f 2 ( t ) = 1136 t 2 I 0 t cosh t + 30,192 t 2 I 0 t = 1136 n = 0 κ n t 2 n + 2 + 30,192 n = 0 1 2 2 n n ! 2 t 2 n + 2 = n = 0 1136 κ n + 30,192 2 2 n n ! 2 t 2 n + 2 = 31,328 t 2 + 8400 t 4 + 4073 6 t 6 + 2153 60 t 8 + n = 4 1136 κ n + 30,192 2 2 n n ! 2 t 2 n + 2 .
Using the power series expansion of hyperbolic sine and hyperbolic cosine functions, we have
g 2 ( t ) = 4659 n = 0 2 2 n 2 n ! t 2 n + 31,764 n = 0 1 2 n + 1 ! t 2 n + 2 9754 t 2 4659 = 4659 n = 1 2 2 n + 2 2 n + 2 ! t 2 n + 2 + 31,764 n = 0 1 2 n + 1 ! t 2 n + 2 9754 t 2 4659 = 31,328 t 2 + 8400 t 4 + 4073 6 t 6 + 2153 60 t 8 + 4659 n = 4 2 2 n + 2 2 n + 2 ! t 2 n + 2 + 31,764 n = 4 1 2 n + 1 ! t 2 n + 2 = 31,328 t 2 + 8400 t 4 + 4073 6 t 6 + 2153 60 t 8 + n = 4 4659 · 2 2 n + 2 2 n + 2 ! + 31,764 2 n + 1 ! t 2 n + 2 .
The inequality (27) is proved when the following inequality
1136 κ n + 30,192 2 2 n n ! 2 > 4659 · 2 2 n + 2 2 n + 2 ! + 31,764 2 n + 1 !
holds for n 4 , that is,
1136 4 n 1 ! ! 2 n ! 2 + 30,192 2 2 n n ! 2 > 4659 · 2 2 n + 2 2 n + 2 ! + 31,764 2 n + 1 ! , n 4
or
1136 4 n 1 ! ! 2 n ! 2 > 4659 2 n + 2 ! 2 2 n + 2 + 31,764 2 n + 1 ! 30,192 2 2 n n ! 2 , n 4 .
It can be verified that the above formula holds when n = 4 . Suppose that Formula (31) holds for m, i.e.,
1136 4 m 1 ! ! 2 m ! 2 > 4659 2 m + 2 ! 2 2 m + 2 + 31,764 2 m + 1 ! 30,192 2 2 m m ! 2 , m 4 .
Next, we prove that Formula (31) is also true for m + 1 . By (32), we have
1136 4 m + 3 ! ! 2 m + 2 ! 2 = 4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 1136 4 m 1 ! ! 2 m ! 2 > 4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 4659 2 m + 2 ! 2 2 m + 2 + 31,764 2 m + 1 ! 30,192 2 2 m m ! 2 .
In this way, we complete the proof of (31) when proving
4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 4659 2 m + 2 ! 2 2 m + 2 + 31,764 2 m + 1 ! 30,192 2 2 m m ! 2 > 4659 2 m + 4 ! 2 2 m + 4 + 31,764 2 m + 3 ! 30,192 2 2 m + 2 m + 1 ! 2
4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 4659 2 m + 2 ! 2 2 m + 2 + 4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 31,764 2 m + 1 ! 4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 30,192 2 2 m m ! 2 > 4659 2 m + 4 ! 2 2 m + 4 + 31,764 2 m + 3 ! 30,192 2 2 m + 2 m + 1 ! 2
4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 4659 2 m + 2 ! 2 2 m + 2 + 4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 31,764 2 m + 1 ! 4659 2 m + 4 ! 2 2 m + 4 + 31,764 2 m + 3 ! > 4 m + 3 4 m + 1 2 m + 2 2 m + 1 2 30,192 2 2 m m ! 2 30,192 2 2 m + 2 m + 1 ! 2
6 74,544 m 3 + 170,830 m 2 + 107,157 m + 15,530 2 2 m + 127,056 m 5 + 719,984 m 4 + 1,503,496 m 3 + 1,413,498 m 2 + 577,046 m + 74,116 2 m + 4 ! m + 1 2 2 m + 1 2 > 15,096 6 m + 6 m 2 + 1 m + 1 ! 2 2 2 m 2 m + 1 2
74,544 m 3 + 170,830 m 2 + 107,157 m + 15,530 2 2 m + 127,056 m 5 + 719,984 m 4 + 1,503,496 m 3 + 1,413,498 m 2 + 577,046 m + 74,116 2 m + 4 ! > 2516 6 m + 6 m 2 + 1 m ! 2 2 2 m ,
that is,
m ! 2 2 2 m 2 m + 4 ! > 2516 6 m + 6 m 2 + 1 74,544 m 3 + 170,830 m 2 + 107,157 m + 15,530 2 2 m + 127,056 m 5 + 719,984 m 4 + 1,503,496 m 3 + 1,413,498 m 2 + 577,046 m + 74,116
holds for m 4 .
Then, we use mathematical induction to prove (33). It is not difficult to verify that the above Formula (33) is true for m = 4 . Suppose that (33) holds for m = n , i.e.,
f ( n ) = : n ! 2 2 2 n 2 n + 4 ! > 2516 6 n + 6 n 2 + 1 74,544 n 3 + 170,830 n 2 + 107,157 n + 15,530 2 2 n + 127,056 n 5 + 719,984 n 4 + 1,503,496 n 3 + 1,413,498 n 2 + 577,046 n + 74,116 = : g ( n ) , n 4 .
By (34), we have
f ( n + 1 ) = n + 1 ! 2 2 2 n + 2 2 n + 6 ! = 4 n + 1 2 2 n + 6 2 n + 5 n ! 2 2 2 n 2 n + 4 ! > 4 n + 1 2 2 n + 6 2 n + 5 2516 6 n + 6 n 2 + 1 74,544 n 3 + 170,830 n 2 + 107,157 n + 15,530 2 2 n + 127,056 n 5 + 719,984 n 4 + 1,503,496 n 3 + 1,413,498 n 2 + 577,046 n + 74,116 ,
so the proof of (33) is complete when we can prove that
A B = : 2516 · 4 n + 1 2 6 n + 6 n 2 + 1 2 n + 6 2 n + 5 74,544 n 3 + 170,830 n 2 + 107,157 n + 15,530 2 2 n + 127,056 n 5 + 719,984 n 4 + 1,503,496 n 3 + 1,413,498 n 2 + 577,046 n + 74,116 > g ( n + 1 ) = 2516 6 n + 1 + 6 n + 1 2 + 1 74,544 n + 1 3 + 170,830 n + 1 2 + 107,157 n + 1 + 15,530 2 2 n + 2 + 127,056 n + 1 5 + 719,984 n + 1 4 + 1,503,496 n + 1 3 + 1,413,498 n + 1 2 + 577,046 n + 1 + 74,116 = : C D .
In fact,
( A D B C ) / 2516 = 9318 × 2 2 n 349 n + 4227 n 2 + 13,036 n 3 + 17,962 n 4 + 12,512 n 5 + 4296 n 6 + 576 n 7 18 10 588 n + 3 n + 2 n + 1 1238 n + 2702 n 2 + 2496 n 3 + 1008 n 4 + 144 n 5 + 177 > 0
holds for all n 4 . The above formula is obviously equivalent to
2 2 n > 5294 n + 1 n + 2 n + 3 1238 n + 2702 n 2 + 2496 n 3 + 1008 n 4 + 144 n 5 + 177 4659 349 n + 4227 n 2 + 13,036 n 3 + 17,962 n 4 + 12,512 n 5 + 4296 n 6 + 576 n 7 18 , n 4 .
We use mathematical induction for the third time in the process of proving Theorem 2. The Formula (36) is obviously true for n = 4 . We assume that (36) holds for n = m , that is,
2 2 m > 5294 m + 1 m + 2 m + 3 1238 m + 2702 m 2 + 2496 m 3 + 1008 m 4 + 144 m 5 + 177 4659 349 m + 4227 m 2 + 13,036 m 3 + 17,962 m 4 + 12,512 m 5 + 4296 m 6 + 576 m 7 18 .
It is proved below that (36) holds for n = m + 1 . By (37), we have
2 2 m + 2 = 4 · 2 2 m > 4 · 5294 m + 1 m + 2 m + 3 1238 m + 2702 m 2 + 2496 m 3 + 1008 m 4 + 144 m 5 + 177 4659 349 m + 4227 m 2 + 13,036 m 3 + 17,962 m 4 + 12,512 m 5 + 4296 m 6 + 576 m 7 18 ,
so the proof of (36) is complete when proving
4 · 5294 4659 m + 1 m + 2 m + 3 1238 m + 2702 m 2 + 2496 m 3 + 1008 m 4 + 144 m 5 + 177 349 m + 4227 m 2 + 13,036 m 3 + 17,962 m 4 + 12,512 m 5 + 4296 m 6 + 576 m 7 18 > 5294 4659 m + 2 m + 3 m + 4 1238 m + 1 + 2702 m + 1 2 + 2496 m + 1 3 + 1008 m + 1 4 + 144 m + 1 5 + 177 349 m + 1 + 4227 m + 1 2 + 13,036 m + 1 3 + 17,962 m + 1 4 + 12,512 m + 1 5 + 4296 m + 1 6 + 576 m + 1 7 18
a b = : 4 m + 1 1238 m + 2702 m 2 + 2496 m 3 + 1008 m 4 + 144 m 5 + 177 349 m + 4227 m 2 + 13,036 m 3 + 17,962 m 4 + 12,512 m 5 + 4296 m 6 + 576 m 7 18 > m + 4 1238 m + 1 + 2702 m + 1 2 + 2496 m + 1 3 + 1008 m + 1 4 + 144 m + 1 5 + 177 349 m + 1 + 4227 m + 1 2 + 13,036 m + 1 3 + 17,962 m + 1 4 + 12,512 m + 1 5 + 4296 m + 1 6 + 576 m + 1 7 18 = : c d .
We calculated that
a d b c = 248,832 m 13 + 5,505,408 m 12 + 55,199,232 m 11 + 332,101,728 m 10 + 1,337,449,104 m 9 + 3,805,389,696 m 8 + 7,862,263,692 m 7 + 11,923,195,508 m 6 + 13,236,235,612 m 5 + 10,578,592,104 m 4 + 5,876,921,975 m 3 + 2,125,982,239 m 2 + 440,485,650 m + 38,040,600 > 0 .

4. Comparisons of New and Old Results

Through observations, in the literature, the more accurate upper and lower bounds of I 0 t are the second inequality in (11)
I 0 t < sinh t t cosh 3 / 2 2 t 3
and the left-hand side of the inequality in (11)
sinh t t 3 / 4 cosh t 1 / 4 < I 0 t .
Next, we compare them with the results of this paper.
(i) The numerical results show that inequality (20) is sharper than the one (38) on (0, 5.2904) while that inequality (38) is sharper than the one (20) on (5.2904, ∞). In other words, the two inequalities (21) and the second inequality in (11) cannot be compared.
(ii) The numerical results show that inequality (22) is sharper than the one (39) on (0, 4.916) while that inequality (39) is sharper than the one (22) on (4.916, ∞). In other words, the two inequalities (23) and the left-hand side of (11) cannot be compared.

5. Conclusions

Let I p x be the modified Bessel function of the first kind of order p. The double inequality
4659 cosh 2 t + 31,764 t sinh t 9754 t 2 4659 16 t 2 71 cosh t + 1887 < I 0 t < 8284 t + 20,475 sinh 2 t + 4526 t cosh 2 t 256 t 179 cosh t + 31
was proved to hold for t > 0 . The corresponding inequalities for a Toader-Qi mean are not in accordance with the conclusions in previous literature.

Funding

This research was funded by The Natural Science Foundation of China OF FUNDER grant number 61772025.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Acknowledgments

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

Conflicts of Interest

The author declares that he has no conflict of interest.

References

  1. Watson, G.N. A Treatise on the Theory of Bessel Functions; Cambridge University Press: Cambridge, UK, 1962. [Google Scholar]
  2. Abramowitz, M.; Stegun, I. Handbook of Mathematical Functions; U.S. National Bureau of Standards: Washington, DC, USA, 1964.
  3. Baricz, Á. Some inequalities involving generalized Bessel functions. Math. Ineq. Appl. 2007, 10, 827–842. [Google Scholar] [CrossRef]
  4. Baricz, Á. Generalized Bessel Functions of the First Kind. Ph.D. Thesis, Babes-Bolyai University, Cluj-Napoca, Romania, 2008. [Google Scholar]
  5. Baricz, Á. Jordan-type inequalities for generalized Bessel functions. J. Inequal. Pure Appl. Math. 2008, 9, 39. [Google Scholar]
  6. Baricz, Á.; Wu, S.H. Sharp Jordan-type inequalities for Bessel functions. Publ. Math. Debr. 2009, 74, 107–126. [Google Scholar]
  7. Baricz, Á.; Neuman, E. Inequalities involving generalized Bessel functions. J. Inequal. Pure Appl. Math. 2005, 6, 126. [Google Scholar] [CrossRef]
  8. Baricz, Á. Functional inequalities involving Bessel and modified Bessel functions of the first kind. Expo. Math. 2008, 26, 279–293. [Google Scholar] [CrossRef] [Green Version]
  9. Baricz, Á. Landen-type inequality for Bessel functions. Comput. Methods Funct. Theory 2005, 5, 373–379. [Google Scholar] [CrossRef]
  10. Baricz, Á. Functional inequalities involving special functions. J. Math. Anal. Appl. 2006, 319, 450–459. [Google Scholar] [CrossRef] [Green Version]
  11. Baricz, Á. Functional inequalities involving special functions II. J. Math. Anal. Appl. 2007, 327, 1202–1213. [Google Scholar] [CrossRef] [Green Version]
  12. Toader, G. Some mean values related to the arithmetic-geometric mean. J. Math. Anal. Appl. 1998, 218, 358–368. [Google Scholar] [CrossRef] [Green Version]
  13. Qi, F.; Shi, X.-T.; Liu, F.-F.; Yang, Z.-H. A double inequality for an integral mean in terms of the exponential and logarithmic means. Period Math. Hung. 2017, 75, 180–189. [Google Scholar] [CrossRef]
  14. Lin, T.-P. The power mean and the logarithmic mean. Am. Math. Mon. 1974, 81, 879–883. [Google Scholar] [CrossRef]
  15. Stolarsky, K.B. Generalizations of the Logarithmic Mean. Math. Mag. 1975, 48, 87–92. [Google Scholar] [CrossRef]
  16. Páles, Z. Inequalities for differences of powers. J. Math. Anal. Appl. 1988, 131, 271–281. [Google Scholar] [CrossRef] [Green Version]
  17. Alzer, H. Ungleichungen für Mittelwerte. Arch. Math. 1986, 47, 422–426. [Google Scholar] [CrossRef]
  18. Yang, Z.-H.; Chu, Y.-M. On approximating the modified Bessel function of the first kind and Toader-Qi mean. J. Inequal. Appl. 2016, 2016, 40. [Google Scholar] [CrossRef] [Green Version]
  19. Yang, Z.-H.; Chu, Y.-M.; Song, Y.-Q. Sharp bounds for Toader-Qi mean in terms of logarithmic and identric means. Math. Inequal. Appl. 2016, 19, 721–730. [Google Scholar] [CrossRef] [Green Version]
  20. Yang, Z.-H.; Tian, J.-F.; Zhu, Y.-R. New sharp bounds for the modified Bessel function of the first kind and Toader-Qi mean. Mathematics 2020, 8, 901. [Google Scholar] [CrossRef]
  21. Gould, H.W. Combinatorial Identities; Morgantown Printing and Binding: Morgantown, WV, USA, 1972. [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 Bounds for the Modified Bessel Function of the First Kind and Toader-Qi Mean. Mathematics 2021, 9, 2867. https://doi.org/10.3390/math9222867

AMA Style

Zhu L. New Bounds for the Modified Bessel Function of the First Kind and Toader-Qi Mean. Mathematics. 2021; 9(22):2867. https://doi.org/10.3390/math9222867

Chicago/Turabian Style

Zhu, Ling. 2021. "New Bounds for the Modified Bessel Function of the First Kind and Toader-Qi Mean" Mathematics 9, no. 22: 2867. https://doi.org/10.3390/math9222867

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