- Research
- Open access
- Published:
A two-point-Padé-approximant-based method for bounding some trigonometric functions
Journal of Inequalities and Applications volume 2018, Article number: 140 (2018)
Abstract
Inequalities are frequently used for solving practical engineering problem. There are two key issues of bounding inequalities; one is to find the bounds, and the other is to prove the bounds. This paper takes Wilker type inequalities as an example, presents a two-point-Padé-approximant-based method for finding the bounds, and it also provides a method to prove the bounds in a new way. It not only recovers the estimates in Mortici’s method, but it also provides new improvements of estimates obtained from prevailing methods. In principle, it can be applied for other inequalities.
1 Introduction
The Wilker inequality, which involves the trigonometric function
has been discussed in the recent past; see also [2, 3, 6–9, 11–15, 17–23] and the references therein, such as the following ones in [14, 18]:
where \(a(x)=\frac{8}{945} x^{2}\), \(b(x)=\frac{8}{945} x^{2}-\frac{16}{14\mbox{,}175} x^{4}\), \(c(x)=(\frac{160}{\pi^{5}}-\frac{16}{\pi^{3}})(\frac{\pi}{2}-x)\), \(d(x)=(\frac{160}{\pi^{5}}-\frac{16}{\pi^{3}})(\frac{\pi}{2}-x)+(\frac{960}{\pi^{6}}-\frac{96}{\pi^{4}})(\frac{\pi}{2}-x)^{2}\).
Recently, Nenezić, Malešević and Mortici provided inequalities within the extended interval \((0,\pi/2)\) [15], e.g., Eq. (7) extends both Eq. (4) and Eq. (5), while Eq. (6) extends the left side of Eq. (3). We have
where \(b_{1}(x)=\frac{8}{945} x^{2}-\frac{\alpha}{14\mbox{,}175} x^{4}\) with \(\alpha = \frac{480 \pi^{6} - 40\mbox{,}320 \pi^{4} + 3\mbox{,}628\mbox{,}800}{\pi^{8}} \approx 17.15041\).
In this paper, we consider
instead of \(f(x)\), which is bounded for \(x \in (0,\pi/2]\). Firstly, we present a two-point-Padé approximant-based method [1] to find the two bounding functions
such that
where \(l_{i}(x)\) and \(r_{i}(x)\) are unknown polynomials to be determined. Note that \(\cos(x)>0, \forall x \in (0,\pi/2)\), from Eq. (10), we obtain
Secondly, we also provide a new way for proving it.
2 The two-point-Padé approximant-based method and examples
Given an interval \([a,b] \subseteq [0,\pi/2]\). From Eq. (9), let
where \(p_{i}, q_{i} \geq 2\), \(\alpha_{i,j}\) and \(\beta_{i,j}\) are the unknowns to be determined, and \(i=1,2\); so there are \(n_{p}=p_{1}+p_{2}+2\) and \(n_{q}=q_{1}+q_{2}+2\) unknowns in \(L(x)\) and \(R(x)\) in Eq. (9), respectively. Let \(E_{1}(x)=F(x)-L(x)\) and \(E_{2}(x)=F(x)-R(x)\). For the sake of convenience, we introduce Theorem 3.5.1 in Page 67, Chap. 3.5 of [4] as follows.
Theorem 1
Let \(w_{0}, w_{1}, \ldots, w_{r}\) be \(r+1\) distinct points in \([a,b]\), and \(n_{0}, \ldots, n_{r}\) 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 \(g^{(i)}(w_{j})=f^{(i)}(w_{j})\), \(i=0,\ldots, n_{j}\), \(j=0,\ldots, r\). Then there exists \(\xi_{1}(t) \in [a,b]\) such that \(f(t)-g(t)=\frac{f^{(N+1)}(\xi_{1}(t))}{(N+1)!} \prod^{r}_{i=0} (t-w_{i})^{n_{i}+1}\).
We introduce the following constraints:
where \(N_{1} \geq n_{p}-k-1\) and \(N_{2} \geq n_{q}-l-1\). By selecting suitable k and \(N_{1}\), we can find \(n_{p}\) constraints for determining \(L(x)\); similarly, by selecting suitable l and \(N_{2}\), we can find \(n_{q}\) constraints for determining \(R(x)\). Combining Theorem 1 with Eq. (13), there exists \(\xi_{i}(x) \in [a,b], i=1,2\), such that
From Eq. (14), if \((-1)^{d} \cdot E_{1}^{(N_{1}+k+2)}(\xi_{1}(x)) \geq 0\), \(\forall x \in [a,b]\), we have \(E_{1}(x) \cdot (-1)^{N_{1}+1+d} \geq 0\), where \(d=0\) or \(d=1\); similarly, if \((-1)^{d} \cdot E_{2}^{(N_{2}+l+2)}(\xi_{2}(x)) \geq 0\), \(\forall x \in [a,b]\), we have \(E_{2}(x) \cdot (-1)^{N_{2}+1+d} \geq 0\). Based on the above observations, one may find the bounding functions in the above way.
We will show three examples which recover or refine previous Wilker type inequalities, including Eq. (2), Eq. (6) and Eq. (7), where \(c_{j}\) is a unknown coefficient to be determined by interpolation constraints.
Example 1
Let \(L_{1}(x)=2 \cos(x)+ c_{1} \sin(x)\) and \(R_{1}(x) = 2 \cos(x) + c_{2} \sin(x)\), \(E_{1,l}(x)=F(x)-L_{1}(x)\) and \(E_{1,r}(x)=F(x)-R_{1}(x)\), \(x \in [0,\pi/2]\). It can be verified that \(E_{1,i}^{(j)}(0)=0\), where \(j=0,1,2,3\), \(i=l,r\). By applying the constraints \(L_{1}(\pi/2)=F(\pi/2)\) and \(R_{1}^{(4)}(0)=F^{(4)}(0)\), we obtain \(c_{1}=\frac{16}{\pi^{4}}\) and \(c_{2}=\frac{8}{45}\), respectively, which recovers Eq. (2).
Example 2
Let \(L_{2}(x)=2 \cos(x)+ (c_{3}+c_{4} x + c_{5} x^{2}) x^{3} \sin(x)\) and \(R_{2}(x) = 2 \cos(x) + (c_{6}+c_{7} x^{2} +c_{8} x^{4}) x^{3} \sin(x)\), \(E_{2,l}(x)=F(x)-L_{2}(x)\) and \(E_{2,r}(x)=F(x)-R_{2}(x)\), \(x \in [0,\pi/2]\). It can be verified that \(E_{2,i}^{(j)}(0)=0\), where \(j=0,1,2,3\), \(i=l,r\). By applying the constraints \(L_{2}^{(j)}(0)=F^{(j)}(0)\), \(j=4,5,6\), we obtain \(c_{3} = \frac{8}{45}\), \(c_{4}=0\) and \(c_{5}=-\frac{8}{945}\), which recovers the left side of Eq. (6). By applying the constraints \(R_{2}^{(4)}(0)=F^{(4)}(0)\), \(R_{2}^{(5)}(0)=F^{(5)}(0)\) and \(R_{2}(\pi/2)=F(\pi/2)\), we obtain \(c_{6} = \frac{8}{45}\), \(c_{7}=-\frac{8}{945}\) and \(c_{8}=\frac{\alpha}{14\mbox{,}175}\), which recovers the right side of Eq. (6).
Example 3
Let \(L_{3}(x)=2 \cos(x)+ (c_{9}+c_{10} (\pi/2-x) ) x^{3} \sin(x)\), \(R_{3}(x) = 2 \cos(x) + (c_{11}+c_{12} (\pi/2-x) + c_{13}(x-\pi/2)^{2}) x^{3} \sin(x)\), \(E_{3,l}(x) = F(x)-L_{3}(x)\) and \(E_{3,r}(x) = F(x)-R_{3}(x)\), \(x \in [0,\pi/2]\). It can be verified that \(E_{3,i}^{(j)}(0)=0\), where \(j=0,1,2,3\), \(i=l,r\). By applying the constraints \(L_{3}(\pi/2)=F(\pi/2)\) and \(L_{3}'(\pi/2)=F'(\pi/2)\), we obtain \(c_{9} = \frac{16}{\pi^{4}}\) and \(c_{10}=\frac{160}{\pi^{5}}-\frac{16}{\pi^{3}}\), which recovers the left side of Eq. (7). By applying the constraints \(R_{3}^{(j)}(\pi/2)=F^{(j)}(\pi/2)\), \(j=0,1,2\), we obtain \(c_{11} = \frac{16}{\pi^{4}}\), \(c_{12}=\frac{160}{\pi^{5}}-\frac{16}{\pi^{3}}\) and \(c_{13}=\frac{960}{\pi^{6}}-\frac{96}{\pi^{4}}\), which recovers the right side of Eq. (7).
3 Results
This section finds other two bounding functions \(L(x)\) and \(R(x)\) to improve the bounds of Eq. (6) and Eq. (7). Combining Eq. (12) with Eq. (13), by setting \(p_{1}=q_{1}=4\), \(p_{2}=q_{2}=5\), \(k=8\), \(N_{1}=1\), \(l=7\) and \(N_{2}=2\), we obtain \(L(x)\) and \(R(x)\) in Eq. (10) as
where
In principle, more bounds can be found by setting different parameters in Eq. (12) and Eq. (13). The main result is as follows.
Theorem 2
We have \(L(x) \leq F(x) \leq R(x)\), \(\forall x \in [0,\pi/2]\).
Proof
(1) Firstly, we give the bounds of \(\sin(x)\), \(\cos(x)\) and \(\sin(2x)\). Let \(\Delta_{1,1}(x)=\sin(x)-P_{1}(x)\), \(\Delta_{1,2}(x)=\sin(x)-Q_{1}(x)\), \(\Delta_{2,1}(x)= \cos(x) - P_{2}(x)\), \(\Delta_{2,2}(x)= \cos(x) - Q_{2}(x)\), \(\Delta_{3,1}(x)= \sin(2x)/2-P_{3}(x)\), \(\Delta_{3,2}(x)= \sin(2x)/2-Q_{3}(x)\), where \(P_{1}(x)\), \(Q_{1}(x)\), \(P_{2}(x)\), \(Q_{2}(x)\), \(P_{3}(x)\) and \(Q_{3}(x)\) are polynomials of degree 12, 12, 13, 13, 15 and 15, respectively. By introducing the following constraints:
we can obtain \(P_{1}(x)=x-\frac{1}{6} x^{3}+\frac{1}{120} x^{5}-\frac{1}{5040} x^{7}+\frac{1}{362\mbox{,}880} x^{9} + \frac{\gamma_{1,1}}{{30\mbox{,}240 \pi^{11}}} x^{11} + \frac{\gamma_{1,2}}{{22\mbox{,}680 \pi^{12}}} x^{12}\), \(Q_{1}(x)=x-\frac{1}{6} x^{3}+\frac{1}{120} x^{5}-\frac{1}{5040} x^{7}+\frac{1}{362\mbox{,}880} x^{9} - \frac{\gamma_{1,3}}{60\mbox{,}480\pi^{10}} x^{10} +\frac{\gamma_{1,4}}{30\mbox{,}240 \pi^{11}} x^{11} - \frac{\gamma_{1,5}}{45\mbox{,}360 \pi^{12}} x^{12}\), \(P_{2}(x)=1-\frac{1}{2} x^{2}+\frac{1}{24} x^{4}-\frac{1}{720} x^{6}+\frac{1}{40\mbox{,}320} x^{8} - \frac{1}{3\mbox{,}628\mbox{,}800} x^{10} + \frac{\gamma_{2,1}}{604\mbox{,}800 \pi^{11}} x^{11} - \frac{\gamma_{2,2}}{302\mbox{,}400 \pi^{12}} x^{12}+ \frac{\gamma_{2,3}}{453\mbox{,}600 \pi^{13}} x^{13}\), \(Q_{2}(x)=1-\frac{1}{2} x^{2}+\frac{1}{24} x^{4}-\frac{1}{720} x^{6}+\frac{1}{40\mbox{,}320} x^{8} - \frac{1}{3\mbox{,}628\mbox{,}800} x^{10} + \frac{\gamma_{2,4}}{302\mbox{,}400 \pi^{12}} x^{12}- \frac{\gamma_{2,5}}{226\mbox{,}800\pi^{13}} x^{13}\), \(P_{3}(x)=x-\frac{2}{3} x^{3}+\frac{2}{15} x^{5}-\frac{4}{315} x^{7} +\frac{2}{2835} x^{9}-\frac{4}{155\mbox{,}925} x^{11}+ \frac{4}{6\mbox{,}081\mbox{,}075} x^{13}-\frac{\gamma_{3,1}}{6\mbox{,}081\mbox{,}075\pi^{13}} x^{14} + \frac{\gamma_{3,2}}{6\mbox{,}081\mbox{,}075 \pi^{14}} x^{15}\), \(Q_{3}(x)=x-\frac{2}{3} x^{3}+\frac{2}{15} x^{5}-\frac{4}{315} x^{7} +\frac{2}{2835} x^{9}-\frac{4}{155\mbox{,}925} x^{11}+ \frac{\gamma_{3,3}}{51\mbox{,}975 \pi^{12}} x^{13}-\frac{\gamma_{3,4}}{ 155\mbox{,}925\pi^{13}} x^{14} + \frac{\gamma_{3,5}}{155\mbox{,}925 \pi^{14}} x^{15}\), where \(\gamma_{1,1}= -743\mbox{,}178\mbox{,}240+340\mbox{,}623\mbox{,}360 \pi-11\mbox{,}612\mbox{,}160 \pi^{3}+112\mbox{,}896 \pi^{5}-480 \pi^{7}+\pi^{9}\), \(\gamma_{1,2}= -1\mbox{,}021\mbox{,}870\mbox{,}080+464\mbox{,}486\mbox{,}400 \pi-15\mbox{,}482\mbox{,}880 \pi^{3}+145\mbox{,}152 \pi^{5}-576 \pi^{7}+\pi^{9}\), \(\gamma_{1,3}= \pi^{9}-960 \pi^{7}+338\mbox{,}688 \pi^{5}-46\mbox{,}448\mbox{,}640 \pi^{3}+7\mbox{,}741\mbox{,}440 \pi^{2}+1\mbox{,}703\mbox{,}116\mbox{,}800 \pi-4\mbox{,}087\mbox{,}480\mbox{,}320\), \(\gamma_{1,4}= \pi^{9}-1440 \pi^{7}+564\mbox{,}480 \pi^{5}-81\mbox{,}285\mbox{,}120 \pi^{3}+15\mbox{,}482\mbox{,}880 \pi^{2}+3\mbox{,}065\mbox{,}610\mbox{,}240 \pi-7\mbox{,}431\mbox{,}782\mbox{,}400\), \(\gamma_{1,5}= \pi^{9}-1728 \pi^{7}+725\mbox{,}760 \pi^{5}-108\mbox{,}380\mbox{,}160 \pi^{3}+23\mbox{,}224\mbox{,}320 \pi^{2}+4\mbox{,}180\mbox{,}377\mbox{,}600 \pi-10\mbox{,}218\mbox{,}700\mbox{,}800\), \(\gamma_{2,1}= \pi^{10}-1200 \pi^{8}+564\mbox{,}480 \pi^{6}-116\mbox{,}121\mbox{,}600 \pi^{4}+8\mbox{,}515\mbox{,}584\mbox{,}000 \pi^{2}+7\mbox{,}431\mbox{,}782\mbox{,}400 \pi-96\mbox{,}613\mbox{,}171\mbox{,}200\), \(\gamma_{2,2}= \pi^{10}-1800 \pi^{8}+940\mbox{,}800 \pi^{6}-203\mbox{,}212\mbox{,}800 \pi^{4}+15\mbox{,}328\mbox{,}051\mbox{,}200 \pi^{2}+14\mbox{,}244\mbox{,}249\mbox{,}600 \pi-177\mbox{,}124\mbox{,}147\mbox{,}200\), \(\gamma_{2,3}= \pi^{10}-2160 \pi^{8}+ 1\mbox{,}209\mbox{,}600 \pi^{6}-270\mbox{,}950\mbox{,}400 \pi^{4}+20\mbox{,}901\mbox{,}888\mbox{,}000 \pi^{2}+20\mbox{,}437\mbox{,}401\mbox{,}600 \pi-245\mbox{,}248\mbox{,}819\mbox{,}200\), \(\gamma_{2,4}= \pi^{1}0-600 \pi^{8}+188\mbox{,}160 \pi^{6}-29\mbox{,}030\mbox{,}400 \pi^{4}+1\mbox{,}703\mbox{,}116\mbox{,}800 \pi^{2}+619\mbox{,}315\mbox{,}200 \pi-16\mbox{,}102\mbox{,}195\mbox{,}200\), \(\gamma_{2,5}= \pi^{1}0-720 \pi^{8}+241\mbox{,}920 \pi^{6}-38\mbox{,}707\mbox{,}200 \pi^{4}+2\mbox{,}322\mbox{,}432\mbox{,}000 \pi^{2}+928\mbox{,}972\mbox{,}800 \pi-22\mbox{,}295\mbox{,}347\mbox{,}200\), \(\gamma_{3,1}= 16(\pi^{12}-312 \pi^{10}+51\mbox{,}480 \pi^{8}-4\mbox{,}942\mbox{,}080 \pi^{6}+259\mbox{,}459\mbox{,}200 \pi^{4}-6\mbox{,}227\mbox{,}020\mbox{,}800 \pi^{2}+40\mbox{,}475\mbox{,}635\mbox{,}200)\), \(\gamma_{3,2}= 16 (\pi^{12}-468 \pi^{10}+85\mbox{,}800 \pi^{8}-8\mbox{,}648\mbox{,}640 \pi^{6}+467\mbox{,}026\mbox{,}560 \pi^{4}-11\mbox{,}416\mbox{,}204\mbox{,}800 \pi^{2}+74\mbox{,}724\mbox{,}249\mbox{,}600)\), \(\gamma_{3,3}= 32 (\pi^{10}-275 \pi^{8}+36\mbox{,}960 \pi^{6}-2\mbox{,}494\mbox{,}800 \pi^{4}+73\mbox{,}180\mbox{,}800 \pi^{2}-512\mbox{,}265\mbox{,}600)\), \(\gamma_{3,4}= 256(\pi^{10}-330 \pi^{8}+47\mbox{,}520 \pi^{6}-3\mbox{,}326\mbox{,}400 \pi^{4}+99\mbox{,}792\mbox{,}000 \pi^{2}-703\mbox{,}533\mbox{,}600)\), \(\gamma_{3,5}= 64(3 \pi^{10}-1100 \pi^{8}+166\mbox{,}320 \pi^{6}-11\mbox{,}975\mbox{,}040 \pi^{4}+365\mbox{,}904\mbox{,}000 \pi^{2}-2\mbox{,}594\mbox{,}592\mbox{,}000)\).
Combining Theorem 1 with Eq. (15), there exists \(\eta_{i}(x) \in [0,\pi/2]\), \(i=1,2,\ldots,6\), such that
So for \(\forall x \in [0, \pi/2]\), we have
i.e., \(Q_{1}(x) \geq \sin(x) \geq P_{1}(x)\), \(Q_{2}(x) \geq \cos(x) \geq P_{2}(x)\) and \(Q_{3}(x) \geq \frac{\sin(2x)}{2} \geq P_{3}(x)\).
(2) Secondly, we prove that \(\Delta_{4}(x)=(F(x)-L(x)) \cdot x^{2} \geq 0\), \(\forall x \in [0,\pi/2]\), which means that \(F(x) \geq L(x)\).
Note that \(l_{i}(x)\) and \(r_{i}(x)\) are polynomials of degree \(3+i\), \(i=1,2\), polynomials \(P_{1}(x)\), \(Q_{1}(x)\), \(P_{2}(x)\), \(Q_{2}(x)\), \(P_{3}(x)\) and \(Q_{3}(x)\) are of degree 12, 12, 13, 13, 15 and 15, respectively, by using Maple software, \(\forall x \in (0,\pi/2)\), we obtain
Combining Eq. (17) with Eq. (16), we have
where
and
Note that \(0< x^{i}<(\frac{\pi}{2})^{i}, i=2,3, \forall x \in (0,\pi/2)\), we have \(H_{1}(x) \geq (\rho_{1,0} + \rho_{1,2} \cdot (\frac{\pi}{2})^{2}+\rho_{1,3}\cdot (\frac{\pi}{2})^{3}) + \rho_{1,1} x + (\rho_{1,4}+\rho_{1,6}\cdot (\frac{\pi}{2})^{2}+\rho_{1,7}\cdot (\frac{\pi}{2})^{3}) x^{4} + \rho_{1,5} x^{5} + (\rho_{1,8}+\rho_{1,10} \cdot (\frac{\pi }{2})^{2}+\rho_{1,11} \cdot (\frac{\pi}{2})^{3}) x^{8} + \rho_{1,9} x^{9} + (\rho_{1,12}+\rho_{1,14}\cdot (\frac{\pi}{2})^{2}) x^{12} + \rho_{1,13} x^{13} \approx 9.6 \cdot 10^{8} x^{13}+4.3 \cdot 10^{9}*x^{1}2+1.5 \cdot 10^{13} x^{9}+5.0 \cdot 10^{13} x^{8}+4.2 \cdot 10^{16} x^{5}+1.2 \cdot 10^{17} x^{4}+1.5 \cdot 10^{19} x+3.8 \cdot 10^{19} > 0\), \(\forall x \in (0, \pi/2)\). It leads to \(\Delta_{4}(x) \geq 0\) and \(F(x) \geq L(x)\), \(\forall x \in [0, \pi/2]\).
(3) Finally, we prove that \(\Delta_{5}(x)=(F(x)-R(x)) \cdot x^{2} \leq 0\), \(\forall x \in [0,\pi/2]\), which means that \(F(x) \leq R(x)\). Combining Eq. (17) with Eq. (16), we have
where
and
Note that \(0< x^{i}<(\frac{\pi}{2})^{i}, i=2,3, \forall x \in (0,\pi/2)\), we have \(H_{2}(x) \leq (\rho_{2,0}+\rho_{2,2}\cdot (\frac{\pi }{2})^{2}+\rho_{2,3}\cdot (\frac{\pi}{2})^{3}) + \rho_{2,1} x + (\rho_{2,4}+\rho_{2,6}\cdot (\frac{\pi}{2})^{2}+\rho_{2,7}\cdot (\frac{\pi}{2})^{3}) x^{4} + \rho_{2,5} x^{5}+ (\rho_{2,8}+\rho_{2,10}\cdot (\frac{\pi}{2})^{2}+\rho_{2,11}\cdot (\frac{\pi }{2})^{3}) x^{8} + \rho_{2,9} x^{9} + (\rho_{2,12}+\rho_{2,14}\cdot (\frac{\pi }{2})^{2}) x^{12} + \rho_{2,13} x^{13} \approx -1.6 \cdot 10^{6} x^{13}-1.2 \cdot 10^{7} x^{12}-2.7 \cdot 10^{10} x^{9}-1.4 \cdot 10^{11} x^{8}-7.2 \cdot 10^{-13} x^{5}-3.7 \cdot 10^{14} x^{4}-2.5 \cdot 10^{16} x-1.1 \cdot 10^{17} < 0, \forall x \in (0,\pi/2)\). So we have \(\Delta_{5}(x) \leq 0\) and \(F(x) \leq R(x)\), \(\forall x \in [0, \pi/2]\).
From the above discussions, we have completed the proof. □
4 Discussions and conclusions
In principle, one can prove that \(L_{i}(x) \leq L(x) \leq F(x) \leq R(x) \leq R_{i}(x)\), \(\forall x \in [0,\pi/2]\) in a similar way, where \(L_{i}(x)\) and \(R_{i}(x)\), \(i=2,3\), are two bounding functions in Eq. (6) and Eq. (7), respectively. The maximum errors between \(F(x)\) and its different bounds are listed in Table 1. It shows that the bounds in this paper achieve a much better approximation than those of the bounds in Eq. (6) and Eq. (7).
The new method can be applied to refine the Becker–Stark inequality, which is studied in [5, 16, 24] and is known as
Zhu [24] refined it as
while it is refined in [16] as follows:
where \(\mu(x)=\frac{8}{\pi}(\frac{\pi}{2}-x) + (\frac{16}{\pi^{2}}-\frac{8}{3})(\frac{\pi}{2}-x)^{2}\).
By applying the method in Sect. 2 and using the form \(\frac{\sum^{6}_{i=0} \nu_{i} x^{i}}{\pi^{2}-4 x^{2}}\), one obtains the resulting bounds, \(\beta_{l}(x)= \frac{\kappa_{1}(x)}{45 \pi^{6} (\pi^{2}-4 x^{2})}\) and \(\beta_{r}(x)= \frac{\kappa_{2}(x)}{3 \pi^{6} (\pi^{2}-4 x^{2})}\), where \(\kappa_{1}(x)=45 \pi^{8}+(-2 \pi^{8} -3660 \pi^{6} +36\mbox{,}000 \pi^{4}) x^{2}+(16 \pi^{7} +21\mbox{,}000 \pi^{5} -208\mbox{,}800 \pi^{3}) x^{3} +(-48 \pi^{6} -49\mbox{,}440 \pi^{4} +492\mbox{,}480 \pi^{2}) x^{4}+(64 \pi^{5} +54\mbox{,}240 \pi^{3} -541\mbox{,}440 \pi) x^{5} +(-32 \pi^{4} -23\mbox{,}040 \pi^{2} +230\mbox{,}400) x^{6}\) and \(\kappa_{2}(x)=3 \pi^{8}+(-12 \pi^{6} +\pi^{8}) x^{2}+(5280 \pi^{3} -456 \pi^{5} -8 \pi^{7} x^{3})+ (-24\mbox{,}768 \pi^{2} +2272 \pi^{4} +24 \pi^{6}) x^{4}+(40\mbox{,}704 \pi -3808 \pi^{3} -32 \pi^{5}) x^{5}+(-23\mbox{,}040 +2176 \pi^{2} +16 \pi^{4}) x^{6}\), such that
By using the Maple software, \(\forall x \in (0,\frac{\pi}{2})\), it can be verified that \(\beta_{l}(x)-\alpha_{l}(x)=-\frac{(\pi-2x)^{3}}{90 \pi^{6}}\times(57\mbox{,}600 x^{3}-8 \pi^{4} x^{3}-5760 \pi^{2} x^{3}+4920 \pi^{3} x^{2}-48\mbox{,}960 \pi x^{2}+ 4 \pi^{5} x^{2}+6210 \pi^{2} x-630 \pi^{4} x-105 \pi^{5}+1035 \pi^{3}) \approx -\frac{(\pi-2x)^{3}}{90 \pi^{6}} (-28.1940986 x^{3}-37.4163 x^{2}-77.48403 x-40.57055)>0\), \(\beta_{r}(x) - \alpha_{r}(x)=\frac{1}{3 \pi^{6}} (\pi-2 x)^{2} x^{2} (-5760 x^{2}+544 \pi^{2} x^{2}+4 \pi^{4} x^{2}+4416 \pi x-408 \pi^{3} x-4 \pi^{5} x- 216 \pi^{2}+12 \pi^{4}+\pi^{6}) \approx \frac{1}{3 \pi^{6}} (\pi-2 x)^{2} x^{2} (-1.298840 x^{2}-1.36647 x-1.5362637)<0\), \(\beta_{l}(x)-\alpha_{2l}(x)=-\frac{(\pi-2x)^{3}}{45 \pi^{6}}(28\mbox{,}800 x^{3}-4 \pi^{4} x^{3}-2880 \pi^{2} x^{3}-24\mbox{,}480 \pi x^{2}+2 \pi^{5} x^{2}+ 2460 \pi^{3} x^{2}+3240 \pi^{2} x-330 \pi^{4} x-75 \pi^{5}+720 \pi^{3}) \approx -\frac{(\pi-2x)^{3}}{45 \pi^{6}}(-14.0970443 x^{3}-18.7081400 x^{2}-167.48179 x- 626.95716)>0\) and \(\beta_{l}(x)-\alpha_{2r}(x)=\frac{(\pi-2x)^{4}}{3 \pi^{6}}(\pi^{4} x^{2}-1440 x^{2}+136 \pi^{2} x^{2}-336 \pi x+34 \pi^{3} x-60 \pi^{2}+6 \pi^{4}) \approx \frac{(\pi-2x)^{4}}{3 \pi^{6}} (-0.324710 x^{2}-1.361725 x-7.7217177)<0\). So the bounds \(\beta_{l}(x)\) and \(\beta_{r}(x)\) achieve a better approximation than those results in both [24] and [16].
References
Baker, G.A. Jr., Graves-Morris, P.: Padé Approximants. Cambridge University Press, New York (1996)
Banjac, B., Makragić, M., Malešević, B.: Some notes on a method for proving inequalities by computer. Results Math. 69(1), 161–176 (2016)
Chen, C.P., Paris, R.B.: Series representations of the remainders in the expansions for certain trigonometric functions and some related inequalities. Math. Inequal. Appl. 20(4), 1003–1016 (2017)
Davis, P.J.: Interpolation and Approximation. Dover Publications, New York (1975)
Debnath, L., Mortici, C., Zhu, L.: Refinements of Jordan–Steckin and Becker–Stark inequalities. Results Math. 67(1–2), 207–215 (2015)
Jiang, W.D., Luo, Q.M., Qi, F.: Refinements and sharpening of some Huygens and Wilker type inequalities. Math. Inequal. Appl. 6(1), 19–22 (2014)
Lutovac, T., Malešsević, B., Mortici, C.: The natural algorithmic approach of mixed trigonometric-polynomial problems. J. Inequal. Appl. 2017, 116 (2017)
Malešević, B., Banjac, B., Jovović, I.: A proof of two conjectures of Chao–Ping Chen for inverse trigonometricfunctions. J. Math. Inequal. 11(1), 151–162 (2017)
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)
Malešević, B., Lutovac, T., Rašajski, M., et al.: Extensions of the natural approach to refinements and generalizations of some trigonometric inequalities. Adv. Differ. Equ. 2018(1), 90 (2018)
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 (2018)
Malešević, B., Makragic, M.: A method for proving some inequalities on mixed trigonometric polynomial functions. J. Math. Inequal. 10, 849–876 (2015)
Mortici, C.: The natural approach of Wilker–Cusa–Huygens inequalities. Math. Inequal. Appl. 14, 535–541 (2011)
Mortici, C.: A subtly analysis of Wilker inequation. Appl. Math. Comput. 231, 516–520 (2014)
Nenezić, M., Malesević, B., Mortici, C.: New approximations of some expressions involving trigonometric functions. Appl. Math. Comput. 283, 299–315 (2016)
Nenezić, M., Zhu, L.: Some improvements of Jordan–Steckin and Becker–Stark inequalities. Appl. Anal. Discrete Math. 12, 244–256 (2018)
Neuman, E.: Wilker and Huygens-type inequalities for the generalized trigonometric and for the generalized hyperbolic functions. Appl. Math. Comput. 230(3), 211–217 (2014)
Sumner, J.S., Jagers, A.A., Vowe, M., Anglesio, J.: Inequalities involving trigonometric functions. Am. Math. Mon. 98(3), 264–267 (1991)
Wilker, J.B.: Problem E-3306. Am. Math. Mon. 96, 55 (1989)
Wu, S.H., Li, S.G., Bencze, M.: Sharpened versions of Mitrinovic–Adamovic, Lazarevic and Wilker’s inequalities for trigonometric and hyperbolic functions. J. Nonlinear Sci. Appl. 9(5), 2688–2696 (2016)
Wu, S.H., Srivastava, H.M.: A further refinement of Wilker’s inequality. Integral Transforms Spec. Funct. 19(10), 757–765 (2008)
Wu, S.H., Yu, H.P., Deng, Y.P., et al.: Several improvements of Mitrinovic–Adamovic and Lazarevic’s inequalities with applications to the sharpening of Wilker-type inequalities. J. Nonlinear Sci. Appl. 9(4), 1755–1765 (2016)
Yang, Z.H., Chu, Y.M., Zhang, X.H.: Sharp Cusa type inequalities with two parameters and their applications. Appl. Math. Comput. 268, 1177–1198 (2015)
Zhu, L.: A refinement of the Becker–Stark inequalities. Math. Notes 93(3–4), 421–425 (2013)
Acknowledgements
The authors would like to thank the editor and the anonymous referees for their valuable suggestions and comments which helped us to improve this paper greatly.
Funding
This research work was partially supported by the National Science Foundation of China (61672009, 61761136010), Zhejiang Key Research and Development Project of China (2018C01030) and the Open Project Program of the National Laboratory of Pattern Recognition (201800006).
Author information
Authors and Affiliations
Contributions
All authors contributed equally to the manuscript and read and approved the final manuscript.
Corresponding author
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.
About this article
Cite this article
Chen, XD., Ma, J., Jin, J. et al. A two-point-Padé-approximant-based method for bounding some trigonometric functions. J Inequal Appl 2018, 140 (2018). https://doi.org/10.1186/s13660-018-1726-7
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13660-018-1726-7