Skip to main content

Approximating trigonometric functions by using exponential inequalities

Abstract

In this paper, some exponential inequalities are derived from the inequalities containing trigonometric functions. Numerical examples show that one can achieve much tighter bounds than those of prevailing methods, which are presented by Cusa, Huygens, Chen and Sándor.

1 Introduction

Inequalities involving trigonometric and inverse trigonometric functions play an important role and have many applications in science and engineering [2, 8, 12, 17,18,19, 27]. The sinc function, defined as \(\frac{\sin (x)}{x}\), is often used in signal processing, optics, radio transmission, sound recording [12], has been studied in many references [1, 3,4,5, 7, 10, 13,14,15,16,17, 19, 21, 26, 28,29,30]. The study starts from the Jordan’s inequality [19], namely

$$ { \frac{2}{\pi } \leq \frac{\sin (x)}{x} \leq 1,\quad 0 \leq x \leq \frac{\pi }{2}. } $$
(1)

Later, the sinc function is bounded by using polynomials [7, 10, 17, 24], or by using exponential bounds [3, 4, 25].

Cusa–Huygens’s inequality is studied in [3, 4, 11, 20, 22, 23, 25], and gives

$$\begin{aligned}& \cos ^{\frac{1}{3}}(x) < \frac{\sin (x)}{x} < \frac{2+\cos (x)}{3},\quad 0 < x < \frac{\pi }{2}, \end{aligned}$$
(2)
$$\begin{aligned}& \biggl( \frac{2+\cos (x)}{3} \biggr)^{\theta } < \frac{\sin (x)}{x} < \biggl( \frac{2+ \cos (x)}{3} \biggr)^{\vartheta }, \end{aligned}$$
(3)

where the constants \(\theta = \frac{\ln (\pi /2)}{\ln (3/2)} \approx 1.113\) and \(\vartheta =1\) are the best possible.

Becker–Stark’s inequality, namely

$$ { \frac{8}{\pi ^{2}-4 x^{2}} < \frac{\tan (x)}{x} < \frac{\pi ^{2}}{\pi ^{2}-4 x^{2}}, \quad 0 < x < \frac{\pi }{2}, } $$
(4)

is studied in [3, 4, 10, 19, 29, 31]. In [32], Zhu provided improved bounds:

$$ { t_{1}(x) < \frac{\tan (x)}{x} < t_{2}(x), \quad 0< x< \frac{\pi }{2}, } $$
(5)

where \(t_{1}(x)=\frac{8}{\pi ^{2}-4 x^{2}}+\frac{2}{\pi ^{2}}-\frac{\pi ^{2}-9}{6 \pi ^{4}}(\pi ^{2}-4 x^{2})\) and \(t_{2}(x)=\frac{8}{\pi ^{2}-4 x^{2}}+\frac{2}{\pi ^{2}}-\frac{(10-\pi ^{2})}{\pi ^{4}}(\pi ^{2}-4 x^{2})\). Later, there in [6] further improved bounds were given, namely

$$ t_{3}(x) = \frac{\bar{t}_{3}(x)}{45 \pi ^{6} (\pi ^{2}-4 x^{2})} < \frac{ \tan (x)}{x} < \frac{\bar{t}_{4}(x)}{3 \pi ^{6} (\pi ^{2}-4 x^{2})}=t _{4}(x), $$
(6)

where \(\bar{t}_{3}(x)=45 \pi ^{8}+(-2 \pi ^{8}-3660 \pi ^{6}+36000 \pi ^{4}) x^{2}+(16 \pi ^{7}+21000 \pi ^{5}-208800 \pi ^{3}) x^{3}+(-48 \pi ^{6}-49440 \pi ^{4}+492480 \pi ^{2}) x^{4}+(64 \pi ^{5}+54240 \pi ^{3}-541440 \pi ) x^{5}+(-32 \pi ^{4}-23040 \pi ^{2}+230400) x^{6}\) and \(\bar{t}_{4}(x)=3 \pi ^{8}+(-12 \pi ^{6}+\pi ^{8}) x^{2}+(5280 \pi ^{3}-456 \pi ^{5}-8 \pi ^{7}) x^{3}+(-24768 \pi ^{2}+2272 \pi ^{4}+24 \pi ^{6}) x ^{4}+(40704 \pi -3808 \pi ^{3}-32 \pi ^{5}) x^{5}+(-23040+2176 \pi ^{2}+16 \pi ^{4}) x^{6}\).

Chen and Cheng established the following exponential bounds [3]:

$$ { \biggl(\frac{\pi ^{2}}{\pi ^{2}-4 x^{2}} \biggr)^{\frac{\pi ^{2}}{12}} \leq \frac{ \tan (x)}{x} \leq \frac{\pi ^{2}}{\pi ^{2}-4 x^{2}},\quad 0 < x < \frac{ \pi }{2}. } $$
(7)

Recently, Nishizawa established [25]

$$\begin{aligned}& \bigl(\cos (x) \bigr)^{\frac{\bar{\theta }_{1}(x)}{3}} \leq \frac{\sin (x)}{x} \leq \bigl(\cos (x) \bigr)^{\frac{\bar{\theta }_{2}(x)}{3}}, \quad 0 < x < \frac{ \pi }{2}, \end{aligned}$$
(8)
$$\begin{aligned}& \biggl(\frac{4 \pi }{\pi ^{2}-4 x^{2}} \biggr)^{(2x/\pi )^{2}} \leq \frac{1}{ \cos (x)} \leq \biggl(\frac{4 \pi }{\pi ^{2}-4 x^{2}} \biggr)^{\bar{\theta }_{3}(x)},\quad 0 < x < \frac{\pi }{2}, \end{aligned}$$
(9)

where

$$\begin{aligned}& \bar{\theta }_{1}(x) =\textstyle\begin{cases} 1, & 0< x \leq 1, \\ 2-x, & 1< x\leq 3/2, \\ 1/2, & 3/2< x< \pi /2, \end{cases}\displaystyle \qquad \bar{\theta }_{2}(x)=1-2x/ \pi , \quad \mbox{and} \\& \bar{\theta }_{3}(x) =\textstyle\begin{cases} x, & 0< x \leq 1, \\ 1, & 1< x< \pi /2. \end{cases}\displaystyle \end{aligned}$$

Motivated by Eqs. (5), (8) and (9), we provide some inequalities with much tighter bounds by using power exponential functions, which are described in Theorems 1.11.2.

Theorem 1.1

For every \(0 < x < \pi /2\), we have

$$ a(x)^{-\theta _{1}(x)} \leq \frac{1}{\cos (x)} \leq a(x)^{-\theta _{2}(x)}, $$
(10)

where \(a(x)= 1+\frac{\pi -4}{\pi } x -\frac{2 (\pi -2)}{\pi ^{2}} x ^{2}\), \(\theta _{1}(x)=-\frac{\pi }{2(\pi -4)} x- \frac{\pi ^{2}-4 \pi +8 }{4 (\pi -4)^{2}} x^{2} + \frac{\pi ^{4}+8 \pi ^{2}-128 \pi +256}{2 \pi ^{3} (\pi -4)^{2}} x^{3}\) and \(\theta _{2}(x)=- \frac{\pi }{2(\pi -4)} x +\frac{2 (\pi ^{2}+6 \pi -24)}{\pi ^{2} ( \pi -4)} x^{2}-\frac{ 2(\pi ^{2}+8 \pi -32)}{(\pi -4)\pi ^{3}}x^{3}\).

Theorem 1.2

For every \(0 < x < \pi /2\), we have

$$ \cos (x)^{\frac{1}{3}- \frac{2}{45} x^{2}+\frac{5}{124} x^{3}- \frac{41}{1000}x^{4}} \leq \frac{\sin (x)}{x} \leq \cos (x)^{ \frac{1}{3}-\frac{4}{3 \pi ^{2}} x^{2}}. $$
(11)

2 Proof of Theorem 1.1

Proof

Equation (10) is equivalent to

$$ {-\theta _{1}(x)} \cdot \ln \bigl(a(x) \bigr) \leq -\ln \bigl(\cos (x) \bigr) \leq {-\theta _{2}(x)} \cdot \ln \bigl(a(x) \bigr). $$
(12)

Step 1. Firstly, we prove that for every \(0 < x < \pi /2\),

$$ {-\theta _{1}(x)} \cdot \ln \bigl(a(x) \bigr) \leq -\ln \bigl(\cos (x) \bigr), $$

which is equivalent to

$$ D_{1}(x)= \ln \bigl(\cos (x) \bigr) {-\theta _{1}(x)} \cdot \ln \bigl(a(x) \bigr) \leq 0. $$
(13)

Let \(x_{1}=0.9\) and \(x_{2}=\frac{111 \cdot \pi }{256} \approx 1.362\).

Case 1.1. Proof of \(D_{1}(x) \leq 0\), \(\forall x \in (0,x_{1}]\).

Combining with Eq. (6), for every \(0< x \leq x_{1}\), we have that

$$\begin{aligned} D_{1}'(x) =&-\tan (x)-\frac{\theta _{1}(x) \cdot a'(x)}{a(x)} - \theta _{1}'(x) \cdot \ln \bigl(a(x) \bigr) \\ \leq& - t_{3}(x)\cdot x -\frac{\theta _{1}(x) \cdot a'(x)}{a(x)} - \theta _{1}'(x) \cdot \ln \bigl(a(x) \bigr) = D_{2}(x). \end{aligned}$$
(14)

It can be verified that \(\theta _{1}'(x) = \alpha _{2} x^{2} + \alpha _{1} x + \alpha _{0} = \alpha _{2} (x+\frac{\alpha _{1}}{2 \alpha _{2}})^{2} + \alpha _{3}\), where \(\alpha _{0} \approx 1.82\), \(\alpha _{1} \approx -3.59\), \(\alpha _{2} \approx 1.98>0\) and \(\alpha _{3} \approx 0.19>0\), so we have \(\theta _{1}'(x)>0\), \(\forall x \in (0, \pi /2)\). Let \(D_{3}(x)=\frac{D_{2}(x)}{\theta _{1}'(x)}\). It can be verified that

$$ D_{3}'(x)= \frac{(\sum^{9}_{i=0} \gamma _{1,i} B _{1,i}(x) ) \cdot x^{2}}{W_{1}(x) }, $$
(15)

where \(W_{1}(x)=((\pi x-2 x+\pi )\cdot (2 x+\pi )\cdot (-768 x^{2}-3 \pi ^{4} x^{2}-24 \pi ^{2} x^{2}+384 \pi x^{2}+\pi ^{5} x-4 \pi ^{4} x+8 \pi ^{3} x-4 \pi ^{4}+\pi ^{5})\cdot \pi )^{2} (\pi -2 x)\), \(B_{1,i}(x) = \frac{C_{9}^{i} \cdot x^{i} \cdot (x_{1}-x)^{9-i}}{(x_{1}-0)^{9}}\), and \(\gamma _{1,0} \approx -3.5 \cdot 10^{7}<0\), \(\gamma _{1,1} \approx -3.2 \cdot 10^{7}<0\), \(\gamma _{1,2} \approx -2.9 \cdot 10^{7}<0\), \(\gamma _{1,3} \approx -2.4 \cdot 10^{7}<0\), \(\gamma _{1,4} \approx -2.0 \cdot 10^{7}<0\), \(\gamma _{1,5} \approx -1.5 \cdot 10^{7}<0\), \(\gamma _{1,6} \approx -1.1 \cdot 10^{7}<0\), \(\gamma _{1,7} \approx -7.7 \cdot 10^{6}<0\), \(\gamma _{1,8} \approx -4.5 \cdot 10^{6}<0\), \(\gamma _{1,9} \approx -1.7 \cdot 10^{6}<0\). Note that \(W_{1}(x)>0\), \(\forall x \in [0,\pi /2)\) and \(B_{1,i}(x) \geq 0\), \(\forall x \in [0,x_{1}]\), and from Eq. (15) we have that \(D_{3}'(x)\leq 0\), \(\forall x \in [0,x_{1}]\). So

$$ D_{3}(x) \leq D_{3}(0)=0. $$
(16)

Combining Eq. (16) with \(\theta _{1}'(x)>0\), we have that \(D_{2}(x) \leq 0\), \(\forall x \in (0,x_{1}]\). Combining with Eq. (14) yields

$$ D_{1}'(x) \leq D_{2}(x) \leq 0,\quad \forall x \in (0,x_{1}]. $$
(17)

From Eq. (17), we have that

$$ D_{1}(x) \leq D_{1}(0) = 0,\quad \forall x \in (0,x_{1}]. $$
(18)

Case 1.2. Proof of \(D_{1}(x) \leq 0\), \(\forall x \in (x_{1},x_{2}]\).

For every \(x_{1}< x \leq x_{2}\), we have \(\theta _{1}'(x) \geq 0\), \(\ln (a(x))\leq 0\), \(\theta _{1}(x) \leq \theta _{1}(x_{2})\) and

$$ D_{1}(x) \leq \ln \bigl(\cos (x) \bigr) {-\theta _{1}(x_{2})} \cdot \ln \bigl(a(x) \bigr) = D _{4}(x). $$
(19)

Combining with Eq. (5) gives

$$\begin{aligned} D_{4}'(x) =& -\tan (x) - \theta _{1}(x_{2}) \frac{a'(x)}{a(x)} \\ \leq& - x \cdot t_{1}(x) - \theta _{1}(x_{2}) \frac{a'(x)}{a(x)} \\ =& \frac{(\sum^{6}_{i=0} \gamma _{2,i} \cdot x ^{i} )}{(\pi x-2 x+\pi ) (-2 x+\pi ) (\pi -4)^{2} \pi ^{4} (2 x+\pi )} = D_{5}(x), \end{aligned}$$
(20)

where \(\gamma _{2,6} \approx 1.95>0\), \(\gamma _{2,5} \approx 5.36>0\), \(\gamma _{2,4} \approx 56.7>0\), \(\gamma _{2,3} \approx 156.2>0\), \(\gamma _{2,2} \approx -265.4<0\), \(\gamma _{2,1} \approx -1050<0\), \(\gamma _{2,0} \approx 502.8>0\). So for every \(x_{1} < x \leq x_{2}\),

$$ \begin{aligned}\overset{6}{\underset{i=0}{\sum }} \gamma _{2,i} \cdot x^{i} &< \gamma _{2,6}\cdot x_{2}^{6} +\gamma _{2,5} \cdot x_{2}^{5} +\gamma _{2,4} \cdot x_{2}^{4} +\gamma _{2,3} \cdot x_{2}^{3} + \gamma _{2,0} + \gamma _{2,2} \cdot x_{1}^{2} + \gamma _{2,1} \cdot x_{1} \\ &\approx -29.68< 0. \end{aligned} $$

Combining with Eq. (20) yields

$$ D_{4}'(x) \leq D_{5}(x) \leq 0. $$
(21)

Combining with Eqs. (19)–(21), we have that

$$ D_{1}(x) \leq D_{4}(x) \leq D_{4}(x_{1}) \approx -0.0058 < 0, \quad \forall x \in (x_{1},x_{2}]. $$
(22)

Case 1.3. Proof of \(D_{1}(x) \leq 0\), \(\forall x \in (x_{2},\pi /2)\).

Combining with Eq. (5), for every \(x_{2}< x < \pi /2\), we have

$$\begin{aligned} D_{1}'(x) =&-\tan (x)-\frac{\theta _{1}(x) \cdot a'(x)}{a(x)} - \theta _{1}'(x) \cdot \ln \bigl(a(x) \bigr) \\ \geq& - t_{2}(x)\cdot x -\frac{\theta _{1}(x) \cdot a'(x)}{a(x)} - \theta _{1}'(x) \cdot \ln \bigl(a(x) \bigr) = D_{6}(x). \end{aligned}$$
(23)

Let \(D_{7}(x)=\frac{D_{6}(x)}{\theta _{1}'(x)}\). It can be verified that

$$ D_{7}'(x)= \frac{(\sum^{7}_{i=0} \gamma _{3,i} B _{2,i}(x) ) \cdot x^{2}}{W_{1}(x) }, $$
(24)

where \(B_{2,i}(x) = \frac{C_{7}^{i} \cdot (x-x_{2})^{i} \cdot (\pi /2-x)^{7-i}}{( \pi /2-x_{2})^{7}} \geq 0\), \(\forall x \in [x_{2},\pi /2)\), \(\gamma _{3,0} \approx 8.8 \cdot 10^{6}\), \(\gamma _{3,1} \approx 9.7 \cdot 10^{6}\), \(\gamma _{3,2} \approx 1.0 \cdot 10^{7}\), \(\gamma _{3,3} \approx 1.1 \cdot 10^{7}\), \(\gamma _{3,4} \approx 1.3 \cdot 10^{7}\), \(\gamma _{3,5} \approx 1.4 \cdot 10^{7}\), \(\gamma _{3,6} \approx 1.6 \cdot 10^{7}\) and \(\gamma _{3,7} \approx 1.8 \cdot 10^{7}\), such that \(\gamma _{3,i}>0\), \(i=0,1, \dots ,7\), and \(D_{7}'(x) \geq 0\), \(\forall x \in (x_{2},\pi /2)\). So we have

$$ D_{7}(x) \geq D_{7}(x_{2}) \approx 6.7>0, \quad \forall x \in (x_{2}, \pi /2). $$
(25)

Note that \(\theta _{1}'(x)>0\), \(\forall x \in (0,\pi /2)\), and, combining with Eqs. (23)–(25), we have that

$$ D_{1}'(x) \geq D_{6}(x) \geq 0, \quad \forall x \in (x_{2},\pi /2). $$
(26)

From Eq. (26), we obtain

$$ D_{1}(x) \leq D_{1}(\pi /2) = 0, \quad \forall x \in (x_{2},\pi /2). $$
(27)

Using Eqs. (18), (22) and (27), we have completed the proof of Eq. (13).

Step 2. Now we prove that for every \(0 < x < \pi /2\),

$$ -\ln \bigl(\cos (x) \bigr) \leq {-\theta _{2}(x)} \cdot \ln \bigl(a(x) \bigr), $$

which is equivalent to

$$ E_{1}(x)= \ln \bigl(\cos (x) \bigr) {-\theta _{2}(x)} \cdot \ln \bigl(a(x) \bigr) \geq 0. $$
(28)

Let \(x_{3}=\frac{98 \cdot \pi }{256} \approx 1.202\).

Case 2.1. Proof of \(E_{1}(x) \geq 0\), \(\forall x \in (0,1]\).

It can be verified that for every \(x \in (0,\pi /2)\),

$$ \textstyle\begin{cases} 0< a(x) \leq 1, \\ { \ln (a(x)) \leq (a(x)-1) - \frac{(a(x)-1)^{2}}{2} + \frac{(a(x)-1)^{3}}{3}. } \end{cases} $$
(29)

Combining Eq. (29) with Eq. (5), for every \(0< x \leq 1\), we have that

$$\begin{aligned} E_{1}'(x) =&-\tan (x)-\frac{\theta _{2}(x) \cdot a'(x)}{a(x)} - \theta _{2}'(x) \cdot \ln \bigl(a(x) \bigr) \\ \geq& - t_{2}(x)\cdot x -\frac{\theta _{2}(x) \cdot a'(x)}{a(x)} - \theta _{2}'(x) \cdot \ln \bigl(a(x) \bigr) \\ \geq& - t_{2}(x)\cdot x -\frac{\theta _{2}(x) \cdot a'(x)}{a(x)} - \theta _{2}'(x) \cdot \biggl( \bigl(a(x)-1 \bigr) - \frac{(a(x)-1)^{2}}{2} + \frac{(a(x)-1)^{3}}{3} \biggr) \\ =& \frac{(\sum^{8}_{i=0} \gamma _{4,i} {B}_{3,i}(x) ) \cdot x^{2}}{((\pi -2)x+\pi )(2 x+\pi ) }, \end{aligned}$$
(30)

where \(B_{3,i}(x)=C_{8}^{i} x^{i} (1-x)^{8-i}\), \(\gamma _{4,0} \approx 5.54\), \(\gamma _{4,1} \approx 5.28\), \(\gamma _{4,2} \approx 4.91\), \(\gamma _{4,3} \approx 4.43\), \(\gamma _{4,4} \approx 3.84\), \(\gamma _{4,5} \approx 3.16\), \(\gamma _{4,6} \approx 2.38\), \(\gamma _{4,7} \approx 1.52\) and \(\gamma _{4,8} \approx 0.632\). In Eq. (30), for every \(x \in (0,1)\), we have \(B_{3,i}(x) >0\), \(\gamma _{4,i}>0\) and \(((\pi -2)x+\pi )(2 x+\pi )>0\), which means that

$$ E_{1}'(x) \geq 0, \quad \forall x \in [0,1]. $$
(31)

From Eq. (31), we obtain

$$ E_{1}(x) \geq E_{1}(0) = 0, \quad \forall x \in [0,1]. $$
(32)

Case 2.2. Proof of \(E_{1}(x) \geq 0\), \(\forall x \in (1,x_{3}]\).

For every \(1< x \leq x_{3}\), note that \(\theta _{2}'(x) >0\) and \(\ln (a(x)) < 0\), hence

$$ E_{1}(x) \geq \ln \bigl(\cos (x) \bigr) - {\theta _{2}(1)} \cdot \ln \bigl(a(x) \bigr) = E _{2}(x). $$
(33)

Combining with Eq. (5), we have that

$$\begin{aligned} E_{2}'(x) =& -\tan (x) - \theta _{2}(1)\cdot \frac{a'(x)}{a(x)} \\ \leq& - x \cdot t_{1}(x) - \theta _{2}(1) \cdot \frac{a'(x)}{a(x)} \\ =& \frac{(\sum^{6}_{i=0} \gamma _{5,i} \cdot B _{4,i}(x) )}{(\pi x-2 x+\pi ) (-2 x+\pi ) (2 x+\pi )} = E_{3}(x), \end{aligned}$$
(34)

where \(B_{4,i}(x)= \frac{C_{6}^{i} (x-1)^{i} (x_{3}-x)^{6-i}}{(x_{3}-1)^{6}}\), \(\gamma _{5,0} \approx -3.98\), \(\gamma _{5,1} \approx -4.23\), \(\gamma _{5,2} \approx -4.45\), \(\gamma _{5,3} \approx -4.65\), \(\gamma _{5,4} \approx -4.81\), \(\gamma _{5,5} \approx -4.94\) and \(\gamma _{5,6} \approx -5.03\). In Eq. (34), for every \(1< x< x_{3}\), we have \(\gamma _{5,i}<0\), \(B_{4,i}(x) >0\) and \((\pi x-2 x+\pi ) (-2 x+\pi ) (2 x+\pi )>0\), which means that \(E_{3}(x) \leq 0\), \(\forall x \in (1,x_{3})\). Combining with Eq. (33), we get

$$ E_{2}'(x) \leq E_{3}(x) < 0,\quad \forall x \in (1,x_{3}). $$
(35)

Combining Eq. (35) with Eq. (33) yields

$$ E_{1}(x) \geq E_{2}(x) \geq E_{2}(x_{3}) \approx 0.0026>0,\quad \forall x \in (1,x_{3}). $$
(36)

Case 2.3. Proof of \(E_{1}(x) \geq 0\), \(\forall x \in (x_{3},\pi /2)\).

Combining with Eq. (5), for every \(x_{3}< x < \pi /2\), we have

$$\begin{aligned} E_{1}'(x) =&-\tan (x)-\frac{\theta _{2}(x) \cdot a'(x)}{a(x)} - \theta _{2}'(x) \cdot \ln \bigl(a(x) \bigr) \\ < & - t_{1}(x)\cdot x -\frac{\theta _{2}(x) \cdot a'(x)}{a(x)} - \theta _{2}'(x) \cdot \ln \bigl(a(x) \bigr) = E_{4}(x). \end{aligned}$$
(37)

Let \(E_{5}(x)=\frac{E_{4}(x)}{\theta _{2}'(x)}\). It can be verified that

$$ E_{5}'(x)= \frac{\sum^{8}_{i=0} \gamma _{6,i} B _{5,i}(x)}{W_{2}(x) }, $$
(38)

where \(W_{2}(x)=((\pi x-2 x+\pi ) (-2 x+\pi ) (2 x+\pi ) (-48 \pi x+192 x-6 \pi ^{2} x+\pi ^{3}))^{2} \geq 0\), \(B_{5,i}(x) = \frac{C_{8}^{i} \cdot (x-x_{3})^{i} \cdot (\pi /2-x)^{8-i}}{(\pi /2-x_{3})^{8}} \geq 0\), \(\forall x \in [x_{3},\pi /2)\), \(\gamma _{6,0} \approx -1.34 \cdot 10^{5}\), \(\gamma _{6,1} \approx -1.46 \cdot 10^{5}\), \(\gamma _{6,2} \approx -1.54 \cdot 10^{5}\), \(\gamma _{6,3} \approx -1.5 \cdot 10^{5}\), \(\gamma _{6,4} \approx -1.5 \cdot 10^{5}\), \(\gamma _{6,5} \approx -1.4 \cdot 10^{5}\), \(\gamma _{6,6} \approx -1.25 \cdot 10^{5}\), \(\gamma _{6,7} \approx -98875\) and \(\gamma _{6,8} \approx -64163\). In Eq. (38), for every \(x_{3}< x<\pi /2\), noting that \(W_{2}(x)>0\), \(\gamma _{6,i}<0\) and \(B_{5,i}(x) >0\), we have

$$ E_{5}'(x) \leq 0, \quad \forall x \in (x_{3},\pi /2). $$
(39)

From Eq. (39), we obtain

$$ E_{5}(x) \leq E_{5}(x_{3}) \approx -0.25< 0,\quad \forall x \in (x_{3}, \pi /2). $$
(40)

Combining Eq. (40) with Eq. (37), for every \(x \in (x_{3},\pi /2)\), and noting that \(\theta _{2}'(x)>0\), we get

$$ \textstyle\begin{cases} E_{4}(x) < 0, \quad \forall x \in (x_{3},\pi /2), \\ E_{1}(x) > E_{1}(\pi /2) =0,\quad \forall x \in (x_{3},\pi /2). \end{cases} $$
(41)

Using Eqs. (32), (36) and (41), we have completed the proof of Eq. (28).

Now combining Eq. (13) with Eq. (28), we have completed the proof of Eq. (10), and hence of Theorem 1.1. □

3 Proof of Theorem 1.2

3.1 Lemmas

We recall Theorem 3.5.1 in [9, Chap. 3.5, p. 67] as follows.

Theorem 3.1

Let \(w_{0}, w_{1}, \dots , w_{r}\) be \(r+1\) distinct points in \([a,b]\), and \(n_{0}, \dots , n_{r}\) be \(r+1\) integers ≥0. Let \(N=n_{0}+ \cdots + n_{r} + r\). Suppose that \(g(t)\) is a polynomial of degree N such that \(g^{(i)}(w_{j})=f^{(i)}(w_{j})\), \(i=0,\dots , n_{j}\), \(j=0,\dots , r \). Then there exists \(\xi _{1}(t) \in [a,b]\) such that \(f(t)-g(t)= \frac{f^{(N+1)}(\xi _{1}(t))}{(N+1)!} \overset{r}{\underset{i=0}{\prod }} (t-w_{i})^{n_{i}+1} \).

Lemma 3.2

For every \(0< x<\pi /2\), we have that

$$ \sin (x) \leq \frac{1}{120} x^{5}-\frac{1}{6} x^{3}+x = c(x). $$

Proof

Let \(H_{1}(x)=\sin (x) -( \frac{1}{120} x^{5}- \frac{1}{6} x^{3}+x)\). It can be verified that \(H_{1}(0)=H_{1}'(0)= \cdots = H^{(6)}(0)=0\) and \(H^{(7)}(0)=-1 \neq 0\). For every \(0< x<\pi /2\), using Theorem 3.1, there exists \(\psi (x) \in (0,\pi /2)\) such that

$$ H_{1}(x) = \frac{H^{(7)}(\psi (x))}{7!} (x-0)^{7} = \frac{-\cos ( \psi (x))}{7!} x^{7} \leq 0, $$

completing the proof. □

Lemma 3.3

For every \(0< x < \pi /2\), we have

$$ \ln ^{(7)} \bigl(\cos (x) \bigr) < 0, $$

where \(\ln ^{(7)}(\cos (x))\) denotes the seventh derivative.

Proof

For every \(0< x < \pi /2\), it can be verified that

$$\begin{aligned} \ln ^{(7)} \bigl(\cos (x) \bigr) =& \frac{-16 (45-30 \cdot \cos (x)^{2}+2 \cdot \cos (x)^{4})) \cdot \sin (x)}{(\cos (x))^{7}} \\ < &\frac{-16 ( 45 -30 ) \cdot \sin (x)}{(\cos (x))^{7}} < 0. \end{aligned}$$

This completes the proof. □

Lemma 3.4

For every \(0< x<\pi /2\), we have

$$ \varphi _{1}(x) = 1-x^{2}+\frac{x^{4}}{3}- \frac{2 x^{6}}{45} +\frac{x ^{8}}{315}-\frac{2 x^{10}}{14175}+\frac{2 x^{12}}{467775}- \frac{4 x ^{14}}{42567525} \leq \cos (x)^{2} = \kappa _{1}(x). $$

Proof

For every \(0< x<\pi /2\), it can be verified that \(\bar{\kappa }_{1}^{(i)}(0)=0\), \(i=0,1,\dots ,15\) and \(\bar{\kappa } _{1}^{(16)}(0)=32768>0\), where \(\bar{\kappa }_{1}(x)=\kappa _{1}(x)- \varphi _{1}(x)\). Employing Theorem 3.1, for every \(0< x<\pi /2\), there exists \(\xi _{2}(x) \in (0,\pi /2)\) such that

$$\begin{aligned} \kappa _{1}(x)-\varphi _{1}(x) =& \frac{\kappa _{1}^{(16)}(\xi _{2}(x))}{16!} (x-0)^{16} \\ =& \frac{32768 \cos (2\xi _{2}(x))}{16!} (x-0)^{16}, \quad \forall x \in (0,\pi /2). \end{aligned}$$
(42)

Note that \(\bar{\kappa }_{1}(\pi /2) = \frac{32768 \cos (2\xi _{2}( \pi /2))}{16!} (\pi /2)^{16} \approx 0.0000020>0\), and, on the other hand, \(\cos (2x)>0\), \(\forall x \in (0,\pi /4)\) and \(\cos (2x)<0\), \(\forall x \in (\pi /4,\pi /2)\), hence we have that \(\xi _{2}(\pi /2) \in (0,\pi /4)\) and then \(\xi _{2}(x) \in (\xi _{2}(0), \xi _{2}(\pi /2)) \in (0,\pi /4)\). Combining with Eq. (42), we get \(\kappa _{1}(x)-\varphi _{1}(x)>0\), \(\forall x \in (0,\pi /2)\), completing the proof. □

Lemma 3.5

For every \(0< x<\pi /2\), we have

$$ \ln ^{(6)} \biggl(\frac{x}{\sin (x)} \biggr)>0. $$

Proof

It can be verified that

$$ \ln ^{(6)} \biggl(\frac{x}{\sin (x)} \biggr) = \frac{8 \cdot \kappa _{2}(x)}{\sin (x)^{6} x^{6}}, $$
(43)

where \(\kappa _{2}(x)= (2 \cos (x)^{4}+11 \cos (x)^{2}+2)+ \cos (x)^{2} (45-45 \cos (x)^{2}+15 \cos (x)^{4}) -15\). Combining with Lemma 3.4, we have

$$ \kappa _{2}(x) \geq \bigl(2 \varphi _{1}(x)^{2}+11 \varphi _{1}(x)+2\bigr)+ \varphi _{1}(x) \bigl(45-45 \varphi _{1}(x)+15 \varphi _{1}(x)^{2}\bigr) -15 = \kappa _{3}(x), $$

where \(\kappa _{3}(x)=\frac{x^{10}}{5142140516927060521875} \bar{ \kappa }_{3}(x)\), and

$$\begin{aligned} \bar{\kappa }_{3}(x) =& -64 x^{30}+8736 x^{28}-685776 x^{26}+38749256 x^{24} -1619962344 x^{22} \\ &{}+53283946716 x ^{20}-1386097749036 x^{18} + 28235771493930 x^{16} \\ &{}-435529376632350 x ^{14} +4560089491932975 x^{12}-17910555851588400 x^{10} \\ &{}-349231679183512125 x^{8}+7231910390065486125 x^{6} \\ &{}-58866255075932679375 x^{4}+179566811702214811875 x^{2} \\ &{} +163242556092922556250 \\ \geq& \biggl(-64 \biggl(\frac{\pi }{2} \biggr)^{2} +8736 \biggr) x^{28}+ \biggl(-685776 \biggl(\frac{\pi }{2} \biggr)^{2} +38749256 \biggr) x^{24} \\ &{}+ \biggl(-1619962344 \biggl(\frac{ \pi }{2} \biggr)^{2}+53283946716 \biggr) x^{20}\\ &{}+ \biggl(-1386097749036 \biggl(\frac{\pi }{2} \biggr)^{2} +28235771493930 \biggr) x^{16}\\ &{} + \biggl(-435529376632350 \biggl( \frac{\pi }{2} \biggr)^{2} +4560089491932975 \biggr) x^{12}\\ &{}+ \biggl(-17910555851588400 \biggl(\frac{\pi }{2} \biggr)^{4} +7231910390065486125 \biggr) x^{6}\\ &{} + \biggl(-349231679183512125 \biggl(\frac{\pi }{2} \biggr)^{6} -58866255075932679375 \biggl(\frac{\pi }{2} \biggr)^{2}\\ &{}+179566811702214811875 \biggr) x^{2} +163242556092922556250 \\ \approx& 8578 x^{28} + 3.7 \times 10^{7} x^{24} + 4.9 \times 10^{10} x ^{20} + 2.4 \times 10^{13} x^{16} + 3.4 \times 10^{15} x^{12} \\ &{}+ 7.1 \times 10^{18} x^{6} + 2.9 \times 10^{19} x^{2}+1.6 \times 10^{20}>0,\quad \forall x \in (0,\pi /2). \end{aligned}$$

Combining with Eq. (43), for every \(0< x<\pi /2\), and noting that \(\sin (x)^{6} x^{6}>0\), we have \(\ln ^{(6)}(\frac{x}{\sin (x)})>0\), which completes the proof. □

Lemma 3.6

For every \(0< x<\pi /3\), we have

$$ \varphi _{2}(x) \geq \ln \bigl(\cos (x)\bigr), $$

where \(\varphi _{2}(x)= \frac{-x^{2}}{2}-\frac{x^{4}}{12}+\frac{(162 \sqrt{3} \pi +108 \pi ^{2}+\pi ^{4}-2916 \ln (2))\cdot x^{5}}{2 \pi ^{5}} -\frac{3 (324 \sqrt{3} \pi +162 \pi ^{2}+\pi ^{4}-4860 \ln (2))\cdot x^{6}}{4 \pi ^{6}}\).

Proof

Let \(\kappa _{4}(x) = \ln (\cos (x))-\varphi _{2}(x)\). It can be verified that

$$ \kappa _{4}^{(i)}(0)=0, \quad i=0,1,\dots ,4, \quad \text{and} \quad \kappa _{4}^{(j)}(\pi /3)=0,\quad j=0,1. $$

Using Theorem 3.1 and Lemma 3.3, for \(0< x<\pi /3\), there exists \(\xi _{3}(x) \in (0,\pi /3)\) such that

$$ \kappa _{4}(x)=\frac{ \kappa _{4}^{(7)}(\xi _{3}(x))}{7!}(x-\pi /3)^{2} \cdot x^{5} = \frac{\ln ^{(7)}(\cos (\xi _{3}(x)))}{7!}(x-\pi /3)^{2} \cdot x^{5} < 0, $$

which means that \(\ln (\cos (x))-\varphi _{2}(x) \leq 0\), and we complete the proof. □

Lemma 3.7

For every \(0< x<\pi /3\), we have

$$ \varphi _{3}(x)=\frac{x^{2}}{6}+\frac{x^{4}}{180}- \frac{(-14580 \ln (\frac{2 \sqrt{3}\cdot \pi }{9} )+270 \pi ^{2}+\pi ^{4})\cdot x ^{5}}{60 \pi ^{5}} \geq \ln \biggl(\frac{x}{\sin (x)} \biggr). $$

Proof

Let \(\kappa _{5}(x) = \ln (\frac{x}{\sin (x)})-\varphi _{3}(x)\). It can be verified that

$$ \kappa _{5}^{(i)}(0)=0, \quad i=0,1,\dots ,4, \quad \text{and} \quad \kappa _{5}(\pi /3)=0. $$

Now by Theorem 3.1 and Lemma 3.5, for \(0< x<\pi /3\), there exists \(\xi _{4}(x) \in (0,\pi /3)\) such that

$$ \kappa _{5}(x)=\frac{ \kappa _{5}^{(6)}(\xi _{4}(x))}{6!}(x-\pi /3) \cdot x^{5} = \frac{\ln ^{(6)}(\frac{\xi _{4}(x)}{\sin (\xi _{4}(x))})}{6!}(x-\pi /3)) \cdot x^{5} < 0, $$

which means that \(\ln (\frac{x}{\sin (x)})-\varphi _{3}(x) < 0\), \(\forall x \in (0,\pi /3)\), and we complete the proof. □

3.2 Proof of Theorem 1.2

Proof of Theorem 1.2

Step 1. Firstly, we prove that \(\frac{\sin (x)}{x} \leq \cos (x)^{ \theta _{3}(x)}\), \(\forall x \in (0,\pi /2)\), where \(\theta _{3}(x)= \frac{1}{3}-\frac{4}{3 \pi ^{2}} x^{2}\). This is equivalent to

$$ F_{1}(x) = \ln \bigl(\sin (x) \bigr)-\ln (x)-\theta _{3}(x) \cdot \ln \bigl(\cos (x) \bigr) \leq 0,\quad \forall x \in (0, \pi /2). $$
(44)

Combining with Lemma 3.2, we have that

$$ F_{1}(x) \leq \ln \bigl(c(x) \bigr)-\ln (x)-\theta _{3}(x) \cdot \ln \bigl(\cos (x) \bigr) = F_{2}(x), \quad \forall x \in (0,\pi /2). $$
(45)

For every \(0< x < \pi /2\), noting that \(\theta _{3}'(x)=-\frac{8 x}{3 \pi ^{2}} <0\) and \(\theta _{3}(x) >0\), and combining with Eq. (4), we have

$$\begin{aligned} F_{2}'(x) =& \frac{c'(x)}{c(x)} - \frac{1}{x}+\theta _{3}(x) \cdot \tan (x) -\theta _{3}'(x) \cdot \ln \bigl(\cos (x) \bigr) \\ \leq& \frac{c'(x)}{c(x)} - \frac{1}{x}+\theta _{3}(x) \cdot \tan (x) \\ < & \frac{c'(x)}{c(x)}- \frac{1}{x}+\theta _{3}(x) \cdot \frac{\pi ^{2} \cdot x}{\pi ^{2}-4 x^{2}} \\ =&\frac{(x^{2}-8) \cdot x^{3}}{3(x^{4}-20 x^{2}+120)} < 0, \quad \forall x \in (0,\pi /2). \end{aligned}$$
(46)

Combining Eq. (45) with Eq. (46), we obtain

$$ F_{1}(x) \leq F_{2}(x) < F_{2}(0) = 0, \quad \forall x \in (0,\pi /2). $$
(47)

This completes the proof of Eq. (44), and hence proves \(\frac{\sin (x)}{x} \leq \cos (x)^{\theta _{3}(x)}\).

Step 2. Secondly, we prove that \(\cos (x)^{\theta _{4}(x)} \leq \frac{ \sin (x)}{x}\), \(\forall x \in (0,\pi /2)\), where \(\theta _{4}(x)= \frac{1}{3}- \frac{2}{45} x^{2}+\frac{5}{124} x^{3}-\frac{41}{1000}x ^{4}\). This is equivalent to

$$ F_{3}(x) = \theta _{4}(x) \cdot \ln \bigl(\cos (x)\bigr) + \ln \biggl( \frac{x}{\sin (x)}\biggr)\leq 0, \quad \forall x \in (0, \pi /2). $$
(48)

Case 2.1. \(0< x<\pi /3\).

Noting that \(\theta _{4}(x) \geq 0\), and combining with Lemmas 3.6 and 3.7, we have that

$$ F_{3}(x) \leq \theta _{4}(x) \cdot \varphi _{2}(x) + \varphi _{3}(x) =\Biggl( \sum ^{5}_{i=0}\gamma _{7,i} B_{6,i}(x)\Biggr) \cdot x ^{5} { =G_{1}(x), } $$
(49)

where \(B_{6,i}=\frac{C_{5}^{i} \cdot x^{i}\cdot (\pi /3-x)^{5-i}}{( \pi /3)^{5}}\), and \(\gamma _{7,0} \approx -0.0068\), \(\gamma _{7,1} \approx -0.0067\), \(\gamma _{7,2} \approx -0.0071\), \(\gamma _{7,3} \approx -0.0072\), \(\gamma _{7,4} \approx -0.0070\), \(\gamma _{7,5} \approx -0.0041\). Noting that \(B_{6,i} \geq 0\) and \(\gamma _{7,i}<0\), \(i=0,1,\dots ,5\), and combining with Eq. (49), we obtain

$$ F_{3}(x) \leq G_{1}(x) < 0, \quad \forall x \in (0,\pi /3]. $$
(50)

Case 2.2. \(\pi /3< x<1.36\).

Let \(\varphi _{4}(x)\) be the sextic interpolation polynomial such that

$$ \begin{aligned}&\ln ^{(i)} \bigl(\cos (\pi /3) \bigr)=\varphi _{4}^{(i)}( \pi /3), \quad i=0,1,\dots ,4, \quad \text{and} \\ &\ln ^{(j)} \bigl( \cos (1.36) \bigr)=\varphi _{4}^{(j)}(1.36), \quad j=0,1, \end{aligned} $$

and \(\kappa _{6}(x) = \ln (\cos (x))-\varphi _{4}(x)\). We have that

$$ \kappa _{6}^{(i)}(\pi /3)=0,\quad i=0,1,\dots ,4, \quad \text{and} \quad \kappa _{6}^{(j)}(1.36)=0,\quad j=0,1. $$

Similar as in the proof of Lemma 3.6, by Theorem 3.1 and Lemma 3.3, for \(\pi /3< x<1.36\), there exists \(\xi _{5}(x) \in (\pi /3,1.36)\) such that

$$ \kappa _{6}(x)= \frac{\ln ^{(7)}(\cos (\xi _{5}(x)))}{7!}(x-1.36)^{2} \cdot (x-\pi /3)^{5} < 0, $$

which means that \(\ln (\cos (x))-\varphi _{4}(x) \leq 0\).

On the other hand, let \(\varphi _{5}(x)\) be the quintic interpolation polynomial such that

$$ \ln ^{(i)} \biggl(\frac{\pi /3}{\sin (\pi /3)} \biggr)=\varphi _{5}^{(i)}( \pi /3),\quad i=0,1, \dots ,4, \quad \text{and} \quad \ln \biggl( \frac{1.36}{\sin (1.36)} \biggr)=\varphi _{5}(1.36). $$

Similarly, let \(\kappa _{7}(x)=\ln (\frac{x}{\sin (x)})-\varphi _{5}(x)\), and then, for every \(\pi /3< x<1.36\), there exists \(\xi _{6}(x) \in ( \pi /3,1.36)\) such that

$$ \begin{aligned} &\kappa _{7}^{(i)}(\pi /3)=0,\quad i=0,1,\dots ,4, \quad \text{and} \quad \kappa _{7}(1.36)=0, \\ &\kappa _{7}(x) = \frac{\ln ^{(6)}( \frac{\xi _{6}(x) }{\sin (\xi _{6}(x) )})}{6!} (x-\pi /3)^{5} \cdot (x-1.36)< 0, \end{aligned} $$

which means that \(\ln (\frac{x}{\sin (x)}) \leq \varphi _{5}(x)\). Finally, for every \(\pi /3< x<1.36\), we have

$$ F_{3}(x) \leq \theta _{4}(x) \cdot \varphi _{4}(x) + \varphi _{5}(x) = \sum ^{10}_{i=0}\gamma _{8,i} B_{7,i}(x) =G_{2}(x), $$
(51)

where \(B_{7,i}(x) \approx \frac{C_{10}^{i} \cdot (x-\pi /3)^{i} \cdot (1.36-x)^{10-i}}{(1.36-\pi /3)^{10}}\), \(\gamma _{8,0} \approx -0.0052\), \(\gamma _{8,1} \approx -0.0054\), \(\gamma _{8,2} \approx -0.0055\), \(\gamma _{8,3} \approx -0.0056\), \(\gamma _{8,4} \approx -0.0055\), \(\gamma _{8,5} \approx -0.0052\), \(\gamma _{8,6} \approx -0.0048\), \(\gamma _{8,7} \approx -0.0041\), \(\gamma _{8,8} \approx -0.0033\), \(\gamma _{8,9} \approx -0.0026\) and \(\gamma _{8,10} \approx -0.0022\). Noting that \(B_{7,i}(x)\geq 0\) and \(\gamma _{8,i}<0\), and combining with Eq. (51), we have

$$ F_{3}(x) \leq G_{2}(x)< 0, \quad \forall x \in ( \pi /3,1.36]. $$
(52)

Case 2.3. \(1.36< x<1.54\).

Let \(\varphi _{6}(x)\) be the sextic interpolation polynomial such that

$$ \begin{aligned}&\ln ^{(i)} \bigl(\cos (1.36) \bigr)=\varphi _{6}^{(i)}( \pi /3), \quad i=0,1,\dots ,4, \quad \text{and} \\ & \ln ^{(j)} \bigl( \cos (1.54) \bigr)=\varphi _{6}^{(j)}(1.54), \quad j=0,1, \end{aligned} $$

and \(\kappa _{8}(x) = \ln (\cos (x))-\varphi _{6}(x)\). We have that

$$ \kappa _{8}^{(i)}(1.36)=0,\quad i=0,1,\dots ,4, \quad \text{and} \quad \kappa _{8}^{(j)}(1.54)=0, \quad j=0,1. $$

Similar as in the proof of Lemma 3.6, by Theorem 3.1 and Lemma 3.3, for \(1.36< x<1.54\), there exists \(\xi _{7}(x) \in (1.36,1.54)\) such that

$$ \kappa _{8}(x)= \frac{\ln ^{(7)}(\cos (\xi _{7}(x)))}{7!}(x-1.54)^{2} \cdot (x-1.36)^{5} < 0, $$

which means that \(\ln (\cos (x)) \leq \varphi _{6}(x)\).

On the other hand, let \(\varphi _{7}(x)\) be the quintic interpolation polynomial such that

$$ \ln ^{(i)} \biggl(\frac{1.36}{\sin (1.36)} \biggr)=\varphi _{7}^{(i)}(1.36), \quad i=0,1, \dots ,4, \quad \text{and} \quad \ln \biggl(\frac{1.54}{\sin (1.54)} \biggr)=\varphi _{7}(1.54). $$

Similarly, letting \(\kappa _{9}(x)=\ln (\frac{x}{\sin (x)})-\varphi _{7}(x)\), for every \(1.36< x<1.54\), there exists \(\xi _{8}(x) \in (1.36,1.54)\) such that

$$ \begin{aligned} &\kappa _{9}^{(i)}(1.36)=0,\quad i=0,1,\dots ,4, \quad \text{and} \quad \kappa _{9}(1.54)=0, \\ &\kappa _{9}(x) = \frac{\ln ^{(6)}( \frac{\xi _{8}(x) }{\sin (\xi _{8}(x) )})}{6!} (x-1.36)^{5} \cdot (x-1.54)< 0, \end{aligned} $$

which means that \(\ln (\frac{x}{\sin (x)}) \leq \varphi _{7}(x)\). Finally, for every \(1.36< x<1.54\), we have

$$ F_{3}(x) \leq \theta _{4}(x) \cdot \varphi _{6}(x) + \varphi _{7}(x) = \sum ^{10}_{i=0}\gamma _{9,i} \cdot B_{8,i}(x)=G _{3}(x), $$
(53)

where \(B_{8,i}(x)= \frac{C_{10}^{i} \cdot (x-1.36)^{i} \cdot (1.54-x)^{10-i}}{(1.54-1.36)^{10}}\), \(\gamma _{9,0} \approx -0.0022\), \(\gamma _{9,1} \approx -0.0019\), \(\gamma _{9,2} \approx -0.0018\), \(\gamma _{9,3} \approx -0.0020\), \(\gamma _{9,4} \approx -0.0025\), \(\gamma _{9,5} \approx -0.0029\), \(\gamma _{9,6} \approx -0.0024\), \(\gamma _{9,7} \approx -0.0021\), \(\gamma _{9,8} \approx -0.0068\), \(\gamma _{9,9} \approx -0.025\) and \(\gamma _{9,10} \approx -0.071\). Noting that \(B_{8,i}(x)> 0\), \(\forall x \in (1.36,1.54)\), and \(\gamma _{9,i}<0\), and combining with Eq. (53), we have that

$$ F_{3}(x) \leq G_{3}(x) < 0, \quad \forall x \in [1.36,1.54]. $$
(54)

Case 2.3. \(1.54< x<\pi /2\).

Let \(\varphi _{8}(x)\) be the quintic interpolation polynomial such that

$$ \ln ^{(i)} \biggl(\frac{1.54}{\sin (1.54)} \biggr)=\varphi _{8}^{(i)}(1.54), \quad i=0,1, \dots ,4, \quad \text{and} \quad \ln \biggl(\frac{\pi /2}{\sin (\pi /2)} \biggr)=\varphi _{8}(\pi /2). $$

Similarly, letting \(\kappa _{10}(x)=\ln (\frac{x}{\sin (x)})-\varphi _{8}(x)\), for every \(1.54< x<\pi /2\), there exists \(\xi _{9}(x) \in (1.54, \pi /2)\) such that

$$ \begin{aligned} &\kappa _{10}^{(i)}(1.54)=0, \quad i=0,1,\dots ,4, \quad \text{and} \quad \kappa _{10}(\pi /2)=0, \\ &\kappa _{10}(x) = \frac{\ln ^{(6)}( \frac{\xi _{9}(x) }{\sin (\xi _{9}(x) )})}{6!} (x-1.54)^{5} \cdot (x- \pi /2)< 0, \end{aligned} $$

which means that \(\ln (\frac{x}{\sin (x)}) \leq \varphi _{8}(x)\). Finally, for every \(1.54< x<\pi /2\), we have

$$ F_{3}(x) \leq \theta _{4}(x) \cdot \ln \bigl( \cos (1.54)\bigr) + \varphi _{8}(x) = \sum^{5}_{i=0} \gamma _{10,i} \cdot B_{9,i}(x)=G _{4}(x), $$
(55)

where \(B_{9,i}(x)=\frac{C_{5}^{i} \cdot (x-1.54)^{i} \cdot (\pi /2-x)^{5-i}}{( \pi /2-1.54)^{5}}\), \(\gamma _{10,0} = -0.071\), \(\gamma _{10,1} = -0.057\), \(\gamma _{10,2} = -0.043\), \(\gamma _{10,3} = -0.030\), \(\gamma _{10,4} = -0.01\) and \(\gamma _{10,5} = -0.0020\). Noting that \(B_{9,i}(x)> 0\), \(\forall x \in (1.54,\pi /2)\), and \(\gamma _{10,i}<0\), and combining with Eq. (55), we have

$$ F_{3}(x) \leq G_{4}(x)< 0, \quad \forall x \in [1.54,\pi /2]. $$
(56)

Using Eqs. (50), (52), (54) and (56), we have completed the proof of Eq. (48).

Combining Eqs. (44) and (48), we have completed the proof of Theorem 1.2. □

4 Comparisons and conclusion

Let \(f_{1}(x)=(\frac{4 \pi }{\pi ^{2}-4 x^{2}})^{(2x/\pi )^{2}}\), \(f_{2}(x)= a(x)^{-\theta _{1}(x)}\), \(f_{3}(x)= a(x)^{-\theta _{2}(x)}\) and \(f_{4}(x)=(\frac{4 \pi }{\pi ^{2}-4 x^{2}})^{\bar{\theta }_{3}(x)}\), and \(F_{i}(x)=\frac{1}{f_{i}(x)}\), \(i=1,2,3,4\). As shown in Fig. 1(a), \(F_{1}(x) \geq F_{2}(x) \geq F_{3}(x) \geq F_{4}(x)\), which means that Theorem 1.1 achieves much tighter bounds than those of Eq. (9).

Figure 1
figure 1

Plots of (a) \(F_{i}(x)\) and (b) \(p_{i}(x)\), \(i=1,2,3,4\)

Let \(p_{1}(x)=\bar{\theta }_{1}(x)/3\), \(p_{2}(x)=\frac{1}{3}- \frac{2}{45} x^{2}+\frac{5}{124} x^{3}-\frac{41}{1000}x^{4}\), \(p_{3}(x)=\frac{1}{3}-\frac{4}{3 \pi ^{2}} x^{2}\) and \(p_{4}(x)=\bar{ \theta }_{2}(x)/3\). As shown in Fig. 1(b), we have that \(p_{1}(x) \geq p_{2}(x) \geq p_{3}(x) \geq p_{4}(x)\), \(\forall x \in (0, \pi /2)\), combining with \(\cos (x) < 1\), \(\forall x \in (0,\pi /2)\), we have that

$$ \cos (x)^{p_{1}(x)} \leq \cos (x)^{p_{2}(x)} \leq \frac{\sin (x)}{x} \leq \cos (x)^{p_{3}(x)} \leq \cos (x)^{p_{4}(x)}, \quad \forall x \in (0, \pi /2). $$

which means that the bounds in Theorem 1.2 are tighter than in Eq. (8).

References

  1. Agarwal, R.P., Kim, Y.H., Sen, S.K.: A new refined Jordan’s inequality and its application. Math. Inequal. Appl. 12(2), 255–264 (2009)

    MathSciNet  MATH  Google Scholar 

  2. Alirezaei, G., Mathar, R.: Scrutinizing the average error probability for Nakagami fading channels. In: The IEEE International Symposium on Information Theory (ISIT’14), Honolulu, pp. 2884–2888 (2014)

    Chapter  Google Scholar 

  3. Chen, C.-P., Cheung, W.-S.: Sharp Cusa and Becker–Stark inequalities. J. Inequal. Appl. 2011, 136 (2011)

    Article  MathSciNet  Google Scholar 

  4. Chen, C.-P., Cheung, W.-S.: Sharpness of Wilker and Huygens type inequalities. J. Inequal. Appl. 2012, 72 (2012)

    Article  MathSciNet  Google Scholar 

  5. Chen, C.-P., Sandor, J.: Sharp inequalities for trigonometric and hyperbolic functions. J. Math. Inequal. 9(1), 203–217 (2015)

    Article  MathSciNet  Google Scholar 

  6. Chen, X.-D., Ma, J.Y., Jin, J.P., Wang, Y.G.: A two-point-Padé-approximant-based method for bounding some trigonometric functions. J. Inequal. Appl. 2018, 140 (2018) 1–15

    Article  Google Scholar 

  7. Chen, X.-D., Shi, J., Wang, Y., Xiang, P.: A new method for sharpening the bounds of several special functions. Results Math. 72(1–2), 695–702 (2017)

    Article  MathSciNet  Google Scholar 

  8. Cloud, M.J., Drachman, B.C., Lebedev, L.P.: Inequalities with Applications to Engineering. Springer, Berlin (2014)

    MATH  Google Scholar 

  9. Davis, P.: Interpolation and Approximation. Dover, New York (1975)

    MATH  Google Scholar 

  10. Debnath, L., Mortici, C., Zhu, L.: Refinements of Jordan–Stečkin and Becker–Stark inequalities. Results Math. 67(1–2), 207–215 (2015)

    Article  MathSciNet  Google Scholar 

  11. Lutovac, T., Malešević, B., Mortici, C.: The natural algorithmic approach of mixed trigonometric-polynomial problems. J. Inequal. Appl. 2017(116), 1 (2017)

    MathSciNet  MATH  Google Scholar 

  12. Lutovac, T., Malešević, B., Rašajski, M.: A new method for proving some inequalities related to several special functions. Results Math. 73(3), 100 (2018)

    Article  MathSciNet  Google Scholar 

  13. Makragic, M.: A method for proving some inequalities on mixed hyperbolic-trigonometric polynomial functions. J. Math. Inequal. 11(3), 817–829 (2017)

    Article  MathSciNet  Google Scholar 

  14. Malešević, B., Lutovac, T., Banjac, B.: A proof of an open problem of Yusuke Nishizawa for a power-exponential function. J. Math. Inequal. 12(2), 473–485 (2018)

    Article  MathSciNet  Google Scholar 

  15. Malešević, B., Lutovac, T., Rašajski, M., Mortici, C.: Extensions of the natural approach to refinements and generalizations of some trigonometric inequalities. Adv. Differ. Equ. 2018(90), 1 (2018)

    MathSciNet  MATH  Google Scholar 

  16. Malešević, B., Makragic, M.: A method for proving some inequalities on mixed trigonometric polynomial functions. J. Math. Inequal. 10(3), 849–876 (2016)

    Article  MathSciNet  Google Scholar 

  17. Malešević, B., Rašajski, M., Lutovac, T.: Refinements and generalizations of some inequalities of Shafer–Fink’s type for the inverse sine function. J. Inequal. Appl. 2017, 275 (2017)

    Article  MathSciNet  Google Scholar 

  18. Malešević, B., Rašajski, M., Lutovac, T.: Refined estimates and generalizations of inequalities related to the arctangent function and Shafer’s inequality. arXiv:1711.03786

  19. Mitrinović, D.S.: Analytic Inequalities. Springer, Berlin (1970)

    Book  Google Scholar 

  20. Mortici, C.: The natural approach of Wilker–Cusa–Huygens inequalities. Math. Inequal. Appl. 14(3), 535–541 (2011)

    MathSciNet  MATH  Google Scholar 

  21. Nenezić, M., Malešević, B., Mortici, C.: New approximations of some expressions involving trigonometric functions. Appl. Math. Comput. 283, 299–315 (2016)

    MathSciNet  Google Scholar 

  22. Neuman, E.: On Wilker and Huygens type inequalities. Math. Inequal. Appl. 15(2), 271–279 (2012)

    MathSciNet  MATH  Google Scholar 

  23. Neuman, E., Sandor, J.: On some inequalities involving trigonometric and hyperbolic functions with emphasis on the Cusa–Huygens, Wilker and Huygens inequalities. Math. Inequal. Appl. 13(4), 715–723 (2010)

    MathSciNet  MATH  Google Scholar 

  24. Nishizawa, Y.: Sharpening of Jordan’s type and Shafer–Fink’s type inequalities with exponential approximations. Appl. Math. Comput. 269, 146–154 (2015)

    MathSciNet  Google Scholar 

  25. Nishizawa, Y.: Sharp exponential approximate inequalities for trigonometric functions. Results Math. 71(3–4), 609–621 (2017)

    Article  MathSciNet  Google Scholar 

  26. Qi, F.: Extensions and sharpenings of Jordan’s and Kober’s inequality. J. Math. Technol. 12(4), 98–102 (1996)

    Google Scholar 

  27. Rahmatollahi, G., De Abreu, G.T.F.: Closed-form hop-count distributions in random networks with arbitrary routing. IEEE Trans. Commun. 60(2), 429–444 (2012)

    Article  Google Scholar 

  28. Rašajski, M., Lutovac, T., Malešević, B.: Sharpening and generalizations of Shafer–Fink and Wilker type inequalities: a new approach. J. Nonlinear Sci. Appl. 11(7), 885–893 (2018)

    Article  MathSciNet  Google Scholar 

  29. Sun, Z.-J., Zhu, L.: Simple proofs of the Cusa–Huygens-type and Becker–Stark-type inequalities. J. Math. Inequal. 7(4), 563–567 (2013)

    Article  MathSciNet  Google Scholar 

  30. Wu, S., Debnath, L.: A generalization of L’Hospital-type rules for monotonicity and its application. Appl. Math. Lett. 22, 284–290 (2009)

    Article  MathSciNet  Google Scholar 

  31. Zhu, L.: Sharp Becker–Stark-type inequalities for Bessel functions. J. Inequal. Appl. 2010, Article ID 838740 (2010)

    Article  MathSciNet  Google Scholar 

  32. Zhu, L.: A refinement of the Becker–Stark inequalities. Math. Notes 93(3–4), 421–425 (2013)

    Article  MathSciNet  Google Scholar 

Download references

Funding

This research work was partially supported by the National Science Foundation of China (61672009, 61761136010), Zhejiang Key Research and Development Project of China (LY19F020041, 2018C01030) and the Open Project Program of the National Laboratory of Pattern Recognition (201800006).

Author information

Authors and Affiliations

Authors

Contributions

All authors contributed equally to the manuscript and read and approved the final manuscript.

Corresponding author

Correspondence to Junyi Ma.

Ethics declarations

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Chen, XD., Ma, J. & Li, Y. Approximating trigonometric functions by using exponential inequalities. J Inequal Appl 2019, 53 (2019). https://doi.org/10.1186/s13660-019-1992-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13660-019-1992-z

Keywords