Home Mathematics The monotonicity of ratios involving arc tangent function with applications
Article Open Access

The monotonicity of ratios involving arc tangent function with applications

  • Zhen-Hang Yang , King-Fung Tin EMAIL logo and Qin Gao
Published/Copyright: December 13, 2019

Abstract

In this paper, we investigate the monotonicity of the functions

x1x1a+23ax2+a2arctanx,x1x4π2+4π2x2+aarctanx

on (0, ∞) for a > 0, which not only gives relative errors of known bounds with quadratic for arctan x, but also yields some new accurate bounds. Moreover, the known bounds are extended and a more accurate estimate for arctan x is presented.

1 Introduction

In [1], Shafer proposed the elementary problem: Show that for all x > 0 the inequality

3x1+2x2+1<arctanx (1.1)

holds. This was proven in [2]. In [3], Qi, Zhang and Guo discussed the monotonicity of the function

xa+x2+1xarctanx

on (0, ∞), and sharpened and generalized Shafer’s inequality (1.1). Chen and Sun [4] further determined the best b, c such that the inequalities

bx1+ax2+1<arctanx<cx1+ax2+1 (1.2)

hold for a, x > 0. More refinements and sharpenings of inequality (1.1) can be seen in [5] and the recent paper [6].

A more general form of Shafer’s inequality (1.1) is that

Ba,b,cx=xc+bx2+a2<>arctanx

for all x > 0, where a, b > 0 and c + a > 0. For this, Shafer [7] established the following analytic inequalities:

8x3+80x2/3+25<arctanx (1.3)

for x > 0. Zhu [8] found a double inequality

8x3+80x2/3+25<arctanx<8x3+256x2/π2+25 (1.4)

holds for x > 0 with the best constants 80/3 and 256/π2. Alirezaei [9] provided other two sharp lower and upper bounds for arctan x, that is, the double inequality

π2x4+4π2x2+π242<arctanx<π2xπ26+4π2x2+36 (1.5)

holds for x > 0. Moreover, by observing the graph, he showed that the maximum relative errors of the lower and upper bounds are approximately smaller than 0.27% and 0.23%, respectively. Recently, Nishizawa [10] proved that

π2x4+4π2x2+π242<arctanx<π2x4+4π2x2+32 (1.6)

for x > 0, where (π2 – 4)2 and 32 are the best constants.

Other approximations for the arctangent function can be found in [11, 12].

To describe the coincidence of an odd function f(x) = arctan x with its approximation A(x) which is also odd, we use a similar suggestion as presented in [13] by Gasull and Utzet, which states that the function f is equal to A at 0 of order i ≥ 1 if f and A, and their derivatives up to order (i – 1) coincide at 0, that is,

limx0fxAxx2k1=0,k=1,3,...2i1. (1.7)

In a similar way, f and A are equal at infinity of order j ≥ 1 if

limxfxAxx1k=0 for k=1,2,...,j1. (1.8)

For i, j ≥ 0, that f and A are equal of order (i, j) if they are equal at 0 of order i and at infinity of order j.

Now, expanding in power series gives

arctanx=x13x3+15x5+Ox7, (1.9)
Ba,b,cx=1a+cxb2aa+c2x3+b23a+c8a3a+c3x5+Ox7 (1.10)

as x → 0, and

arctanx=π2arctan1x=π21x+13x3+O1x5, (1.11)
Ba,b,cx=1c/x+b+a2/x2=1bcb1xa22c22bb1x2+O1x3 (1.12)

as 1/x → 0.

  1. If arctan x and Ba,b,c(x) have a coincidence of order (2, 0), that is, the parameters a, b and c satisfy the relation (1.7) for k = 1, 2, then comparing respectively the coefficients of x and x3 in the Maclaurin expansions (1.9) and (1.10) we get

    a+c=1 and b2aa+c2=13,

    which indicates that

    b=2a3,c=1a. (1.13)

    Thus

    Ba,b,cx=x1a+2ax2/3+a2:=A2,0x;a. (1.14)

    Further, if arctan x and Ba,b,c(x) have a coincidence of order (3, 0), then comparing respectively the coefficients of x5 in the Maclaurin expansions (1.9) and (1.10) we have

    b23a+c8a3a+c3=15,

    which in combination with (1.13) yields a = 5/8. Therefore,

    A2,0x;58=x3/8+5x2/12+25/64:=A3,0x, (1.15)

    which is the lower bound given in (1.3). Likewise, if A(2,0) (x; a) satisfies (1.8) for k = 1, then

    1b=π2,

    which in combination with (1.13) yields a = 6/π2. That is,

    A2,0x;6π2=x16/π2+4x2/π2+36/π4:=A2,1x, (1.16)

    which is the upper bound given in (1.5).

  2. If arctan x and Ba,b,c(x) have a coincidence of order (1, 1), that is, the parameters a, b and c satisfy the relations (1.7) for k = 1 and (1.8) for k = 1, then comparing respectively the coefficients of x in the Maclaurin expansions (1.9) and (1.10), and the constant items in the asymptotic expansions (1.11) and (1.12), we get

    a+c=1 and 1b=π2,

    which implies that

    Ba,b,cx=x1a+4x2/π2+a2:=A1,1x;a. (1.17)

    Analogously, we have

    A1,1x;6π2=x16/π2+4x2/π2+36/π4:=A2,1x,A1,1x;14π2=x4/π2+4x2/π2+14/π22:=A1,2x.

    Clearly, A(2,1)(x) and A(1,2)(x) are the upper and lower bounds given in (1.5).

  3. If arctan x and Ba,b,c(x) have a coincidence of order (0, 2), that is, the parameters a, b and c satisfy the relations (1.8) for k = 1, 2, then comparing the constant items and coefficients of x–1 in the asymptotic expansions (1.11) and (1.12), we get

    1b=π2 and cb=1.

    Thus

    Ba,b,cx=x4/π2+4x2/π2+a2:=A0,2x;a. (1.18)

    We also check that

    A0,2x;14π2=x4/π2+4/π2x2+14/π22:=A1,2x, (1.19)
    A0,2x;42π2=x4/π2+4x2/π2+32/π4:=A0,3x, (1.20)

    where A(1,2)(x) and A(0,3)(x) are clearly the lower and upper bounds given in (1.6).

From the inequalities (1.3), (1.5) and (1.6), we see clearly that there are two sharp lower bounds A(3,0)(x), A(1,2)(x) and two sharp upper bounds A(2,1)(x), A(0,3)(x) for arctan x, all of which have the form of Ba,b,c(x). Moreover, A(3,0)(x) and A(2,1)(x) are contained in the family of bounds A(2,0)(x; a), while A(1,2)(x) and A(0,3)(x) belong to the family of bounds A(0,2)(x; a). Inspired by these facts, the aim of this paper is to investigate the monotonicity of the ratios

R1x=arctanxA2,0x;a and R2x=arctanxA0,2x;a,

which gives new proofs of inequalities (1.3), (1.5) and (1.6). Moreover, as we all know, analytic inequality plays an important role in many different blanch of mathematics (See for example, [14, 15, 16, 17, 18]). By the obtained monotonicity of R1(x) and R2(x) we find some new inequalities, that is, new sharp bounds for arctan x. We show the maximum relative errors and maximum absolute errors estimating for arctan x by the four known sharp bounds mentioned above, and offer a new sharp bounds in the form of λ Ba,b,c(x) (λ > 0 with λ ≠ 1).

The main tool dealing with the monotonicity or R1 and R2 is two identities on the derivatives of ratio of two functions p and q, where p and q are twice differentiable on (a, b) (a < b) with q, q ≠ 0 on (a, b):

pq=qq2pqqp=qq2Hp,q (1.21)
Hp,q=pqq. (1.22)

Identities (1.21) and (1.22) (for short, IDR) were introduced in [19] by Yang. We remark that the auxiliary function Hp,q and its properties are very helpful to investigate those monotonicity of ratios of two functions, see for example, [20, 21, 22]. Similarly, the auxiliary function Hp,q together with the IDR (1.21) and (1.22) will be used effectively to prove Theorems 1 and 2.

2 Lemmas

The following three lemmas are used to prove Lemma 4.

Lemma 1

[23] For –∞ ≤ a < b ≤ ∞, let f and g be differentiable functions on (a, b) with f(a+) = g(a+) = 0 or f(b) = g(b) = 0. Assume that g(x) ≠ 0 for each x in (a, b). If f/g is increasing (decreasing) on (a, b) then so is f/g.

Lemma 2

[24] Let an and bn (n = 0, 1, 2, ...) be real numbers and let the power series A(t) = n=1 antn and B(t) = n=1 bntn be convergent for |t| < r. If bn > 0 for n = 0, 1, 2, ..., and an/bn is strictly increasing (or decreasing) for n = 0, 1, 2, ..., then the function tA(t)/B(t) is strictly increasing (or decreasing) on (0, r).

Lemma 3

[25] The following expansions

1sinx=1x+n=122n22n!B2nx2n1, (2.1)
cosxsin2x=1x2n=12n122n22n!B2nx2n2, (2.2)
cotx=1xn=122n2n!B2nx2n1, (2.3)
1sin2x=1x2+n=12n122n2n!B2nx2n2,|x|<π, (2.4)

hold for |x| < π, where Bn is the Bernoulli numbers.

Lemma 4 plays a key role in the proof of Theorems 1 and 2.

Lemma 4

Let

ϕ1t=sinttcostsin3t, (2.5)
ϕ2t=t22cos2tsin2t2tcostsin3t+tcostsint. (2.6)

Then φ2(t) > 0 for t ∈ (0, π /2), and φ1(t)/φ2(t) is strictly decreasing from (0, π/2) onto (4/π2, 15/32).

Proof

We write φ2(t) as

4ϕ2t=2t2+2tsin2tcos2t2sin22t:=ϕ42t.

Differentiating and expanding in power series by (2.3) and (2.4) yield

ϕ4ssin2s=3s3cosssins2ssin2ssin2s=3ssin2s3cosssins2s=n=26n22n2n!B2ns2n1>0, (2.7)

where s = 2t ∈ (0, π). Then φ4(2t) = φ4(s) > φ4(0) = 0, which implies φ2(t) > 0 for t ∈ (0, π/2).

Similarly, φ1(t) can be written as

ϕ1t=18scosssins2sin2s4cossssins+4,

where s = 2t. Then φ1(t)/φ2(t)

ϕ1tϕ2t=12scosssins2sin2s4cossssins+4s2+ssinscoss2sin2s=12124coss+ssins+s24s2+ssinscoss2sin2s:=1212ϕ3sϕ4s,

where

ϕ3s=4coss+ssins+s24.

Thus, to prove the desired monotonicity, it suffices to prove that φ3(s)/φ4(s) is strictly increasing on (0, π). Expanding in power series by (2.4), (2.1) and (2.2) leads to

ϕ3ssin2s=2s3sins+scosssin2s=2ssin2s3sins+scosssin2s=n=22n222n+2n+22n!B2ns2n1. (2.8)

Taking into account (2.7) and (2.8) we get

ϕ3sϕ4s=ϕ3s/sin2sϕ4s/sin2s=n=22n222n+2n+22n!B2ns2n1n=26n22n2n!B2ns2n1:=n=2uns2nn=2vns2n.

By Lemmas 1 and 2, it is enough to show that the sequence {un/vn}n≥2 is increasing. A direct verification yields

unvn=n222n+2n+23n22n,un+1vn+1unvn=4n+13n2+6n+46nn+1×22n.

Using the binomial theorem we arrive at

4n+13n2+6n+4>1+n+13+nn+12323n2+6n+4=32nn+1>0,

which gives the increasing property of the sequence {un/vn}n≥2.

An easy calculation gives

limt0ϕ1tϕ2t=1532 and limtπ/2ϕ1tϕ2t=4π2,

which completes the proof.□

Remark 1

Using the methods from [26, 27, 28, 29], one can directly prove the Lemma 4.

3 The monotonicity of R1 and inequalities

We now state and prove our first main result, which reveals the monotonicity pattern of R1 on (0, ∞).

Theorem 1

  1. If a ≥ 5/8, then the ratio

    R1x=1x1a+23ax2+a2arctanx

    is strictly increasing from (0, ∞) onto (1, πa/6 ). Therefore, the double inequality

    x1a+2ax2/3+a2<arctanx<aπ26x1a+2ax2/3+a2 (3.1)

    holds for x > 0.

  2. If 3π2+222 /(3π2) = a1 < a < 5/8, then there is an x0 > 0 such that R1 is strictly decreasing on (0, x0) and increasing on (x0, ∞). So the double inequality

    R1x0×x1a+2ax2/3+a2arctanx<max1,aπ2/6×x1a+2ax2/3+a2 (3.2)

    holds for x > 0. In particular, for a = 6/π2, we have

    c21x16/π2+4x2/π2+36/π4arctanx<x16/π2+4x2/π2+36/π4, (3.3)

    where c21 = R1(x0) = 0.9976914... is the best possible.

  3. If 0 < aa1 = 3π2+222 /(3π2) = 0.598..., then R1(x) is strictly decreasing from (0, ∞) onto ( πa/6 , 1). Then the double inequality

    aπ2/6×x1a+2ax2/3+a2arctanx<x1a+2ax2/3+a2 (3.4)

    holds for x > 0.

Proof

Making a change of variable x = tan t for t ∈ (0, π/2) yields

R1x=1x1a+23ax2+a2arctanx=1a+2a/3tan2t+a2tant/t:=ptqt,

where

pt=1a+23atan2t+a2,qt=tantt.

Differentiation yields

ptqt=2a3t2sinttcostsint3a2cos2t+2asin2t,ptqt=2a23tcost×3aϕ2t4ϕ1t2asin2t+3a2cos2t3tcostsint2, (3.5)

where φ1(t) and φ2(t) are defined by (2.5) and (2.6), respectively. As shown in Lemma 4, φ2(t) > 0 for t ∈ (0, π/2), φ1(t)/φ2(t) is strictly decreasing from (0, π/2) onto (4/π2, 15/32). Then the relation (3.5) can be written as

ptqt=23a2tcost×ϕ2t2asin2t+3a2cos2t3tcostsint2a43ϕ1tϕ2t.

Therefore, (p(t)/q(t)) > (<) 0 if and only if

asupt0,π/243ϕ1tϕ2t=58 or ainft0,π/243ϕ1tϕ2t=163π2,

while a ∈ (16/(3π2), 5/8), there is a t1 ∈ (0, π/2) such that (p(t)/q(t)) < 0 for t ∈ (0, t1) and (p(t)/q(t)) > 0 for t ∈ (t1, π/2).

On the other hand, we easily get

qt=tantt=2tsin2t2t2cos2t>0, (3.6)
Hp,qt=ptqtqtpt=2a3t2sinttcostsint3a2cos2t+2asin2t×tantt1a+23atan2t+a2
=a32sin3t3atcostsintcosttcostsint2asin2t+3a2cos2t1a0as t0,a+263πa1as tπ/2.

Now, we distinguish three cases to determine the monotonicity of p/q.

  1. a ≥ 5/8. By the relation (1.22), we have Hp,q = (p/q)q > 0, and so Hp,q(t) > Hp,q(0+) = 0 for t ∈ (0, π/2). The relation (1.21) in combination with q(t) > 0 and Hp,q(t) > 0 leads to (p(t)/q(t)) > 0 for t ∈ (0, π/2). Hence, for t ∈ (0, π/2),

    1=limt0ptqt<ptqt<limtπ/2ptqt=aπ26,

    which is equivalent to (3.1). This proves the first assertion of this theorem.

  2. 0 < a ≤ 16/(3π2). Likewise, we deduce that (p(t)/q(t)) < 0 for t ∈ (0, π/2). So the double inequality (3.4) holds for x > 0.

  3. 16/(3π2) < a < 5/8. As shown previously, by the relation (1.22) it is seen that Hp,q < 0 for t ∈ (0, t1) and Hp,q > 0 for t ∈ (t1, π/2). Since Hp,q(0+) = 0 and

    Hp,qπ2=a+263πa1>0if a1<a<58,0if 163 π 2<aa1,

    where a1 = 3π2+222 /(3π2) = 0.598..., we find that

  4. 16/(3π2) < aa1. We have Hp,q(t) < 0 for t ∈ (0, π/2), so is (p(t)/q(t)), which implies the double inequality (3.4) holds for x > 0.

    Combining Case 2 and Subcase 3.1 gives the third assertion of this theorem.

  5. a1 < a < 5/8. There is a t0 ∈ (t1, π/2) such that Hp,q(t) < 0 for t ∈ (0, t0) and Hp,q(t) > 0 for t ∈ (t0, π/2), and so is (p(t)/q(t)). That is, the ratio p/q is decreasing on (0, t0) and increasing on (t0, π/2). This leads to

    pt0qt0<ptqt<limt0ptqt=1 for t0,t0,pt0qt0<ptqt<limtπ/2ptqt=aπ26 for tt0,π2,

    that is,

    pt0qt0ptqt<max1,aπ26 for x0,,

    which is equivalent to the double inequality (4.2), where x0 = tan t0. In particular, for a = 6/π2, solving the equation Hp,q(t) = 0 gives t0 = 1.2900104..., then x0 = tan t0 = 3.467 341..., so c21 = R1(x0) = 0.9976914..., which proves the second assertion of this theorem.

    Thus we completes the proof.□

Taking a = 5/8, a1 = 3π2+222 /(3π2) in Theorem 1, we obtain two new sharp double inequalities.

Corollary 1

The following inequalities

8x3+80x2/3+25<arctanx<2π5/3x3+80x2/3+25, (3.7)
π2x4+4π2x2+23π2+222<arctanx<23π2+2+2x4+4π2x2+23π2+222 (3.8)

hold for all x > 0. All the bounds are sharp.

Remark 2

It is easy to verify that

A2,0x;aa=9xx2+3a6ax2+9a233a+6ax2+9a226ax2+9a2<0,aA2,0x;aa=9x2a1+a6ax2+9a23a233a+6ax2+9a226ax2+9a2>0,

that is, aA(2,0)(x; a) and a a A(2,0)(x; a) are decreasing and increasing on (0, ∞), respectively. Then taking a = 2/3, ∞ in inequalities (3.1) gives

3xx2+3<3x1+2x2+1<arctanx<πx1+2x2+1 (3.9)

for x > 0; taking a = 1/2, 7/12 in (3.4) yields

πx3+4x2+3<14πx5+56x2+49<arctanx<12x5+56x2+49<23x3+4x2+3

for x > 0.

As a direct consequence of Theorem 1 we immediately obtain the following

Proposition 1

The double inequality

β1x1a+2ax2/3+a2<arctanx<α1x1a+2ax2/3+a2 (3.10)

holds for x > 0 with the best constants

β1=1ifa58,R1x0ifa1<a<58,aπ26if0<aa1,andα1=aπ26ifa6π2,1if0<a6π2,

where a1 = 3π2+222 /(3π2) and R1(x0) is given in Theorem 1.

Using Proposition 1 with the decreasing property of aA(2,0)(x; a), we deduce the following corollary.

Corollary 2

Let a, b > 0. The double inequality

x1a+2ax2/3+a2<arctanx<x1b+2bx2/3+b2 (3.11)

holds for x > 0 if and only if a ≥ 5/8 = 0.625 and 0 < b ≤ 6/π2 = 0.607....

Remark 3

Clearly, the lower and upper bounds given in (3.11) for a = 5/8 and b = 6/π2, that are, A(3,0)(x) and A(2,1)(x), are the sharpest lower and upper bounds of arctan x of the form Ba,b,c(x).

Remark 4

The first inequality of (3.11) for a = 5/8 was first presented in [7], while the second one of (3.11) for b = 6/π2 appeared in [9]. It is easy to check that

8x3+256x2/π2+25π2xπ26+2π2x2+9>0

for all x > 0, the upper bound in (3.11) for b = 6/π2 is better than one in (1.4).

Taking into account Proposition 1 and the proof of Theorem 1 we obtain a new double inequality for arctan x.

Corollary 3

Let a, b > 0. The double inequality

a6πx1a+2ax2/3+a2<arctanx<b6πx1b+2bx2/3+b2 (3.12)

holds for x > 0 if and only if 0 < aa1 = 3π2+222 /(3π2) and b ≥ 6/π2.

Proof

We only prove that the left hand side inequality of (3.12) holds for x > 0 if and only if 0 < aa1. The sufficiency follows from Proposition 1 and the increasing property of a a A(2,0)(x; a) on (0, ∞). Suppose that the left hand side inequality of (3.12) holds for all x > 0. If a > a1, then a ≥ 5/8 or a1 < a < 5/8. If a ≥ 5/8, then by Theorem 1, the second inequality of (3.1) holds for all x > 0, which yields a contradiction with the assumption. If a1 < a < 5/8, then from Subcase 3.2 we see that

ptqt<limtπ/2ptqt=aπ26 for tt0,π2,

which is equivalent to

arctanx<a6πx1a+2ax2/3+a2 for xx0,,

where x0 = tan t0, which also yields a contradiction with the assumption, and the necessity follows. This completes the proof.□

Remark 5

Corollary 3 offers a new family of lower bounds in the form of λBa,b,c(x) > 0 with λ ≠ 1) for arctan x. And, a sharp lower bound is

a16πx1a1+2a1x2/3+a12=π2x4+4π2x2+23π2+222:=ξx. (3.13)

4 The monotonicity of R2 and inequalities

Our second main result is the following theorem, which exposes the monotonicity pattern of R2(x) on (0, ∞).

Theorem 2

  1. If aa0 = 2 3π2+222 /π4 = 0.363..., then the function

    R2x=1x4π2+4π2x2+aarctanx

    is strictly decreasing from (0, ∞) onto (1, 4/π2 + a ). Therefore, the double inequality

    x4/π2+4x2/π2+a<arctanx<4/π2+ax4/π2+4x2/π2+a (4.1)

    holds for x > 0.

  2. If 32/π4 < a < a0 = 2 3π2+222 /π4, then there is an x0 > 0 such that R2 is strictly increasing on (0, x0) and decreasing on (x0, ∞). So the inequalities

    min1,4/π2+a×x4/π2+4x2/π2+a<arctanxR2x0×x4/π2+4x2/π2+a (4.2)

    hold for x > 0, where x0 is the unique solution of the equation R2 (x) = 0 on (0, ∞). In particular, for a = (1 – 4/π2)2, we have

    x4/π2+4x2/π2+14/π22<arctanxc12x4/π2+4x2/π2+14/π22, (4.3)

    where c12 = R2(x0) = 1.0026766... is the best possible.

  3. If 0 < a ≤ 32/π4 = 0.328..., then R2(x) is strictly increasing from (0, ∞) onto (4/π2 + a , 1). Therefore, the double inequality

    4/π2+ax4/π2+4x2/π2+a<arctanx<x4/π2+4x2/π2+a (4.4)

    holds for x > 0.

Proof

Making a change of variable x = tan t for t ∈ (0, π/2) yields

R2x=1x4π2+4π2x2+aarctanx=4/π2+4tant2/π2+atant/t:=ptqt,

where

pt=4π2+4π2tant2+a, qt=tantt.

Differentiation yields

ptqt=4πt2sinttsintcostπ2acos2t+4sin2t,
ptqt=4πtcost×aπ2ϕ2t8ϕ1t4sin2t+π2acos2t3tcostsint2=4πtcost×ϕ2t4sin2t+π2acos2t3tcostsint2a8π2ϕ1tϕ2t,

where φ1(t) and φ2(t) are defined by (2.5) and (2.6), respectively. As shown in Lemma 4, φ2(t) > 0 for t ∈ (0, π/2), φ1(t)/φ2(t) is strictly decreasing from (0, π/2) onto (4/π2, 15/32). Therefore, (p(t)/q(t)) > (<) 0 if and only if

asupt0,π/28π2ϕ1tϕ2t=154π2 or ainft0,π/28π2ϕ1tϕ2t=32π4,

while a ∈ (32/π4, 15/(4π2)), there is a t1 ∈ (0, π/2) such that (p(t)/q(t)) < 0 for t ∈ (0, t1) and (p(t)/q(t)) > 0 for t ∈ (t1, π/2).

On the other hand, we easily get

Hp,qt=ptqtqtpt=4πt2sinttsintcostπ2acos2t+4sin2t×tantt4π2+1cost4π2sin2t+acos2t=4sin3tπ2atcostsintcostππ2acos2t+4sin2ttcostsint4π2π2a+4a6π2aas t0,0as tπ/2,

Now, we distinguish three cases to determine the monotonicity of p/q.

  1. a ≥ 15/(4π2). By the relation (1.22), we have Hp,q = (p/q) q > 0, and so Hp,q(t) < Hp,q(π/2) = 0 for t ∈ (0, π/2). The relation (1.21) in combination with q(t) > 0 and Hp,q(t) < 0 leads to (p(t)/q(t)) < 0 for t ∈ (0, π/2). Therefore, for t ∈ (0, π/2),

    1=limtπ/2ptqt<ptqt<limt0ptqt=4π2+a,

    which is equivalent to (4.1).

  2. 0 < a ≤ 32/π4. Similarly, we deduce that (p(t)/q(t)) > 0 for t ∈ (0, π/2). So the inequalities (4.4) hold for x > 0, which proves the third assertion of this theorem.

  3. 32/π4 < a < 15/(4π2). As shown previously, by the relation (1.22) it is seen that Hp,q < 0 for t ∈ (0, t1) and Hp,q > 0 for t ∈ (t1, π/2). Since Hp,q(π/2) = 0 and

    Hp,q0=π2a+4a6π2a>0if 32π4<a<a0,0if a0a<154π2,

    where a0 = 2 3π2+222 /π4 = 0.363..., we find that

  4. a0a < 15/(4π2). We have Hp,q(t) < 0 for t ∈ (0, π/2), so is (p(t)/q(t)). So the double inequality (4.1) also holds for x > 0. This in combination with Case 1 proves the first assertion of theorem.

  5. 32/π4 < a < a0. There is a t0 ∈ (0, t1) such that Hp,q(t) > 0 for t ∈ (0, t0) and Hp,q(t) < 0 for t ∈ (t0, π/2), and so is (p(t)/q(t)). This yields

    4π2+a=limt0ptqt<ptqt<pt0qt0 for t0,t0,1=limtπ/2ptqt<ptqt<pt0qt0 for tt0,π2,

    that is,

    min1,4π2+a<ptqtpt0qt0 for x0,,

    which is equivalent to the double inequality (4.2), where x0 = tan t0. In particular, for a = (1 – 4/π2)2, solving the equation Hp,q(t) = 0 gives t0 = 0.9081516..., then x0 = tan t0 = 1.2814739..., so c12 = R2(x0) = 1.0026766......, which proves the second assertion of this theorem.

    The proof is finished.□

Taking a = a0 = 2 3π2+222 /π4, 32/π4 in Theorem 2, we obtain two new sharp double inequalities.

Corollary 4

The following inequalities

π2x4+4π2x2+23π2+222<arctanx<23π2+2+2x4+4π2x2+23π2+222, (4.5)
22+1x2+π2x2+8<arctanx<π2/2x2+π2x2+8 (4.6)

for x > 0. All the bounds are sharp.

Remark 6

It is interesting that the double inequality (4.5) is the same as (3.8).

Remark 7

Obviously, aA(0,2)(x; a ) is decreasing on (0, ∞). And, a ↦ (4/π2 + a ) A(0,2)(x; a ) is increasing on (0, ∞) due to

4/π2+aA0,2x;aa=2xaπx2+4x2+aπ2aπ4x2+aπ24x2+aπ2+4/π2>0

for x, a > 0. Then taking a = 36/π4, 4/π2, 4/9 in (4.1) we obtain that for x > 0,

3π2/2x6+π9x2+π2<π2x/22+πx2+1<π2x/22+π2x2+9<arctanx<5x2+π2x2+9<π+2x2+πx2+1<π2+6x6+π9x2+π2.

Taking a = 32/π4, 3/π2, 0 in (4.4) we have that for x > 0,

2xπx+2<3π+4x4+π4x2+3<22+1x2+π2x2+8<arctanx<π2/2x2+π2x2+8<π2x4+π4x2+3<π2x2πx+4. (4.7)

As a direct consequence of Theorem 2, we have the following proposition.

Proposition 2

The double inequality

β2x4/π2+4x2/π2+a<arctanx<α2x4/π2+4x2/π2+a (4.8)

holds for x > 0 with the best constants

β2=1ifa14π22,4π2+aif0<a14π22,andα2=4π2+aifaa0,R2x0if32π4<a<a0,1if0<a32π4,

where a0 = 2 3π2+222 /π4 and R2(x0) is given in Theorem 2.

Proposition 2 with the decreasing property of aA(0,2)(x; a ) implies the following corollary.

Corollary 5

The double inequality

x4/π2+4x2/π2+a<arctanx<x4/π2+4x2/π2+b (4.9)

holds for x > 0 if and only if a ≥ (1 – 4/π2)2 = 0.353... and 0 < b ≤ 32/π4 = 0.328....

Remark 8

Obviously, the lower and upper bounds given in (4.9) for a = (1 – 4/π2)2 and b = 32/π4, that are, A(1,2)(x) and A(0,3)(x), are also the sharpest lower and upper bounds of arctan x of the form Ba,b,c(x).

Remark 9

Inequalities (4.9) for a = (1 – 4/π2)2 and b = 32/π4 were first proven in [10]. Clearly, we here give a new proof.

By Corollaries 2 and 5 with the increasing property of aA(1,1)(x; a) we easily deduce the relation between arctan x and A(1,1)(x; a).

Corollary 6

Let a, b > 0. The double inequality

x1a+4x2/π2+a2<arctanx<x1b+4x2/π2+b2 (4.10)

holds if and only if 0 < a ≤ 1 – 4/π2 and b ≥ 6/π2.

Proof

The necessity follows from

limxxarctanxx1a+4x2/π2+a2=π24a14π20,limx0+1x3arctanxx1b+4x2/π2+b2=13bb6π20.

The sufficiency follows from the first inequality of (4.9) for a = (1 – 4/π2)2 and the second inequality of (3.11) for b = 6/π2 with the facts that

A1,1x;aa=x4x2/π2+a2a4x2/π2+a21a+4x2/π2+a22>0

for x, a > 0.□

Remark 10

Corollary 6 shows that A(1,2)(x) and A(2,1)(x) are the sharpest lower and upper bounds of arctan x of the form Ba,b,c(x) in the sense that A(1,1)(x; a) < arctan x < A(1,1)(x; b).

Proposition 2 also contains another new double inequality.

Corollary 7

Let a, b > 0. The double inequality

4/π2+ax4/π2+4x2/π2+a<arctanx<4/π2+bx4/π2+4x2/π2+b (4.11)

holds for x > 0 if and only if 0 < a ≤ (1 – 4/π2)2 = 0.353... and ba0 = 2 3π2+222 /π4 = 0.363....

Remark 11

Corollary 7 provides a new family of upper bounds in the form of λBa,b,c(x) (λ > 0 with λ ≠ 1) for arctan x. And, a sharp upper bound is

4/π2+a0x4/π2+4x2/π2+a0=23π2+2+2x4+4π2x2+23π2+222:=ηx. (4.12)

5 Concluding remarks

For i, j ∈ {0, 1, 2, 3} such that i + j = 3, let A(i, j)(x) be given by (1.15), (1.16), (1.17) and (1.20), respectively.

Remark 12

From Corollaries 2, 5 and 6 we see that A(3,0)(x) and A(1, 2)(x) are the sharpest lower bounds of arctan x of the form Ba,b,c(x), while A(2,1)(x) and A(0,3)(x) are the sharpest upper bounds of arctan x of the form Ba,b,c(x). Then we have

maxA3,0x,A1,2x<arctanx<minA2,1x,A0,3x. (5.1)

Numeric computations show that the two sharpest lower bounds A(3,0)(x) and A(1,2)(x) are not comparable for all x > 0, so are the two sharpest upper bounds A(2,1)(x) and A(0,3)(x).

Remark 13

From the inequalities (3.3), (3.7), (4.3) and (4.6) we obtain that for all x > 0,

0.0023139...=11c211A2,1xarctanx<0,0<1A3,0xarctanx<11215π=0.013752...,0<1A1,2xarctanx11c12=0.0026694...,0.022030...=1π242+11A0,3xarctanx<0.

These indicate that the maximum relative errors of the four sharpest bounds of arctan x equal 0.0023139..., 0.013 752..., 0.0026694... and 0.022030....

Remark 14

It has been shown in [9] that D(1,2)(x) = arctan xA(1,2)(x) is strictly increasing on (0, x12) and decreasing on (x12, ∞), while D(2,1)(x) = arctan xA(2,1)(x) is strictly decreasing on (0, x21) and increasing on (x21, ∞), where

x12=π24210π2π28π48π216=1.67133...,x21=5π24812π2π10π2=4.36812....

These actually reveal the maximum absolute errors estimating for arctan x by A(1,2)(x) and A(2,1)(x). More precisely, we have

0<arctanxA1,2xD1,2x12=0.0025995...,0.0030381...=D2,1x21arctanxA2,1x<0

for all x > 0, which means that the maximum absolute errors of bounds A(1,2)(x) and A(2,1)(x) equal 0.0025995... and 0.0030381..., respectively.

To show the maximum absolute errors of bounds A(3,0)(x) and A(0,3)(x), we need to prove Theorem 3.

Theorem 3

(i) The difference D(3,0)(x) = arctan xA(3,0)(x) is strictly increasing on (0, ∞), and therefore, the double inequality

0<arctanxA3,0x<π22515=0.021602...

holds for all x > 0. (ii) The difference D(0,3)(x) = arctan xA(0,3)(x) is strictly decreasing on (0, x03) and increasing on (x03, ∞), where

x03=8π48π216π12π2=0.66178....

Consequently, the double inequality

0.0080482...=D0,3x03arctanxA0,3x<0

holds for all x > 0.

Proof

  1. Differentiation yields

    D3,0x=1x2+1880x2/3+25+3+640x2/380x2/3+2580x2/3+25+32=8x2/3+1080x2/3+25104x2+5x2+180x2/3+2580x2/3+25+32>0,

    where the inequality holds due to

    83x2+10803x2+252104x2+52=512027x6>0.

    This leads to

    0=D3,00<D3,0x<D3,0=π22515=0.021602....
  2. Analogously, we have

    D0,3x=1x2+1π22π2x2+8+4+2π4x2π2x2+82π2x2+8+42=412π2π2x2+84π28x2+1π2x2+82π2x2+8+42.

    Then

    sgnD0,3x=sgn12π2π2x2+84π28=sgn12π22π2x2+816π282=sgnx2x032.

    It is therefore deduced that D0,3 (x) < 0 for x ∈ (0, x03) and D0,3 (x) > 0 for x ∈ (x03, ∞), so we arrive at

    0.0080482...=D0,3x03D0,3x<maxD0,30,D0,3=0,

    which completes the proof.□

Remark 15

Taking into account Remarks 13, 14 and Theorem 3, we have the following table:

Sharp bounds Type Maximum absolute errors Maximum relative errors
A(3,0)(x) lower 0.021602... 0.01352...
A(2,1)(x) upper 0.0030381... 0.0023139...
A(1,2)(x) lower 0.0025995 .. 0.0026694...
A(0,3)(x) upper 0.0080482... 0.022030...

From the table, we see that A(1,2)(x) and A(2,1)(x) are respectively better than A(3,0)(x) and A(0,3)(x) in the sense that both of the maximum absolute errors and maximum relative errors are minimum on (0, ∞).

Remark 16

As mentioned in Remarks 5 and 11, inequalities (3.12) and (4.11) offer a new family of bounds in the form of λBa,b,c(x) (λ > 0 with λ ≠ 1) for arctan x. As sharp bounds, by (3.8) or (4.5), we have ξ(x) < arctan x < η (x) for all x > 0. Similar to Theorem 3 we can prove that

0<arctanxξx<0.0055530...

for all x > 0; also, by (3.8), we have that for all x > 0,

0<1ξxarctanx<1π223π2+2+2=0.0081747....

Thus it can be seen that the maximum absolute error and maximum relative error estimating arctan x by ξ (x) are less than 0.0056 and 0.82%.

Remark 17

The families of bounds A(2,0)(x; a) and A(0,2)(x; a) for arctan x also contain several rational bounds listing in (3.9) and (4.7), that are, for x > 0,

3xx2+3<arctanx,2xπx+2<arctanx<π2x2πx+4.

Remark 18

Due to the identity

arctanx=π2arctan1x (5.2)

for x > 0, if arctan x < (>) A(x) for all x > 0, then there must be

arctanx><π2A1x

for all x > 0. Thus the inequalities (5.1) can be extended as

maxA3,0x,A1,2x,π2A2,11x,π2A0,31x<arctanx<minA2,1x,A0,3x,π2A3,01x,π2A1,21x.

And, all four lower (upper) bounds given in the above inequalities are not comparable.

Remark 19

Similarly, inequalities (3.8) can be extended as

maxξx,π2ξ1x<arctanx<minηx,π2η1x,

where ξ (x) and η (x) are defined by (3.13) and (4.12), respectively.

Remark 20

Although the bound A(3,0)(x) for arctan x is inferior to others in the sense that both of the maximum absolute errors and maximum relative errors are minimal on (0, ∞), but it is a priority selection in the applications of engineering because that for 0 < x < 1,

0<arctanxA3,0x<π4465916=0.000156...,0<arctanxA3,0xarctanx<146594π=0.000199...,

and for x > 1,

0.000156...=465916π4<arctanxπ2+A3,01x<0,0.000199...=1+46594π<arctanxπ/2+A3,01/xarctanx<0,

where the first and third inequalities follow from Theorem 3, the second follows from Theorem 1, and the last follows from Theorem 4.

Theorem 4

The ratio

R0x=π/28/3x+80/3+25x2arctanx

is strictly decreasing from (0, ∞) onto (1, ∞). In particular, the double inequality

1<π/28/3x+80/3+25x2arctanx<246594π=1.000199...

holds for x > 1.

Proof

Making a change of variable x = (4 tan t)/ 15 for t ∈ (0, π/2) yields

R0x=π/28/3x+80/3+25x2arctanx=π/2215cost/3sint+5arctan4tant/15:=ptqt,

where

pt=π2215cost3sint+5, qt=arctan4tant/15.

Differentiation yields

ptqt=123+5sint15+sin2t5+3sint2,ptqt=1527+sint1sint2cost5+3sint3>0.

This together with q > 0 yields Hp,q = (p/q)q > 0. Since

Hp,qt=ptqtqtpt=123+5sint15+sin2t5+3sint2×arctan4tant/15π2215cost3sint+50

as tπ/2, we achieve that Hp,q(t) < limt→(π/2) Hp,q(t) = 0 for t ∈ (0, π/2), which in combination with q > 0 gives (p/q) = (q/q2) Hp,q < 0 for t ∈ (0, π/2). That is, R0 is strictly decreasing on (0, ∞). Clearly, R0(0+) = ∞, R0(1) = 2 – ( 465 – 9) /(4π), R0(∞) = 1, which completes the proof.□

Acknowledgement

This work was supported by the Fundamental Research Funds for the Central Universities (No. 2015ZD29) and the Higher School Science Research Funds of Hebei Province of China (No. Z2015137).

References

[1] Shafer R.E., Elementary problems: E 1867, Amer. Math. Monthly, 1966, 73, 309.10.1080/00029890.1966.11970755Search in Google Scholar

[2] Shafer R.E., Grinstein L.S., Marsh D.C.B., Konhauser J.D.E., Problems and solutions: Solutions of elementary problems: E1867, Amer. Math. Monthly, 1967, 74, 726-727.10.2307/2314285Search in Google Scholar

[3] Qi F., Zhang S.-Q., Guo B.-N., Sharpening and generalizations of Shafer’s inequality for the arc tangent function, J. Inequal. Appl., 2009, 2009:930294.10.1155/2009/930294Search in Google Scholar

[4] Chen C.-P., Cheung W.-S., Wang W., On Shafer and Carlson inequalities, J. Inequal. Appl., 2011, 2011:840206.10.1155/2011/840206Search in Google Scholar

[5] Mortici C., Srivastava H.M., Estimates for the arctangent function related to Shafer’s inequality, Colloq. Math., 2014, 136, 263-270.10.4064/cm136-2-8Search in Google Scholar

[6] Malešević B., Rašajski M., Lutovac T., Refined estimates and generalizations of inequalities related to the arctangent function and Shafer’s inequality, Math. Probl. Eng., 2018, 2018:4178629.10.1155/2018/4178629Search in Google Scholar

[7] Shafer R.E., Analytic inequalities obtained by quadratic approximation, Publ. Elektroteh. Fak. Univ. Beogr. Ser. Mat. Fiz., 1977, 96-97, 577-598.Search in Google Scholar

[8] Zhu L., On a quadratic estimate of Shafer, J. Math. Inequal., 2008, 2, 571-574.10.7153/jmi-02-51Search in Google Scholar

[9] Alirezaei G., A sharp double inequality for the inverse tangent function, arXiv:1307.4983v1.Search in Google Scholar

[10] Nishizawa Y., Refined quadratic estimations of Shafer’s inequality, J. Inequal. Appl., 2017, 2017(40).10.1186/s13660-017-1312-4Search in Google Scholar PubMed PubMed Central

[11] Sun J.-L., Chen C.-P., Shafer-type inequalities for inverse trigonometric functions and Gauss lemniscate functions, J. Inequal. Appl., 2016, 2016:212.10.1186/s13660-016-1157-2Search in Google Scholar

[12] Qiao Q.-X., Chen C.-P., Approximations to inverse tangent function, J. Inequal. Appl., 2018, 2018:141.10.1186/s13660-018-1734-7Search in Google Scholar PubMed PubMed Central

[13] Gasull A., Utzet F., Approximating Mills ratio, J. Math. Anal. Appl., 2014, 420, 1832-1853.10.1016/j.jmaa.2014.05.034Search in Google Scholar

[14] Tian J.-F., Triple Diamond-Alpha integral and Hölder-type inequalities, J. Inequal. Appl., 2018, 2018:111.10.1186/s13660-018-1704-0Search in Google Scholar PubMed PubMed Central

[15] Wang M.K., Zhang W., Chu Y.M., Monotonicity, convexity and inequalities involving the generalized elliptic integrals, Acta Math. Sci. Ser. B (Engl. Ed.), 2019, 39, 1440-1450.10.1007/s10473-019-0520-zSearch in Google Scholar

[16] Tian J.-F., Zhu Y.-R., Cheung W.-S., N-tuple Diamond-Alpha integral and inequalities on time scales, Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat., 2019, 113, 2189-2200.10.1007/s13398-018-0609-6Search in Google Scholar

[17] Tian J.-F., Ha M.-H., Wang C., Improvements of generalized Hölder’s inequalities and their applications, J. Math. Inequal., 2018, 12, 459-471.10.7153/jmi-2018-12-34Search in Google Scholar

[18] Yang Z.-H., Tian J.-F., A class of completely mixed monotonic functions involving the gamma function with applications, Proc. Amer. Math. Soc., 2018, 146, 4707-4721.10.1090/proc/14199Search in Google Scholar

[19] Yang Z.-H., A new way to prove L’Hospital monotone rules with applications, arXiv:1409.6408 [math.CA].Search in Google Scholar

[20] Yang Z.-H., Chu Y.-M., A monotonicity property involving the generalized elliptic integral of the first kind, Math. Inequal. Appl., 2017, 20, 729-735.10.7153/mia-2017-20-46Search in Google Scholar

[21] Yang Z.-H., Chu Y.-M., Wang M.-K., Monotonicity criterion for the quotient of power series with applications, J. Math. Anal. Appl., 2015, 428, 587-604.10.1016/j.jmaa.2015.03.043Search in Google Scholar

[22] Yang Z.-H., Zhang W., Chu Y.-M., Sharp Gautschi inequality for parameter 0 < p < 1 with applications, Math. Inequal. Appl., 2017, 20, 1107-1120.10.7153/mia-2017-20-71Search in Google Scholar

[23] Anderson G.D., Vamanamurthy M., Vuorinen M., Monotonicity rules in calculus, Amer. Math. Monthly, 2016, 113, 805-816.10.1080/00029890.2006.11920367Search in Google Scholar

[24] Biernacki M., Krzyz J., On the monotonicity of certain functionals in the theory of analytic functions, Ann. Univ. Mariae Curie-Sklodowska, 1995, 9, 135-145.Search in Google Scholar

[25] Abramowitz M., Stegun I. A., Handbook of Mathematical Functions with Formulas, Graphs and Mathematical Tables, Dover, New York, 1965.10.1063/1.3047921Search in Google Scholar

[26] Mortici C., The natural approach of Wilker-Cusa-Huygens inequalities, Math. Inequal. Appl., 2011, 14, 535-541.10.7153/mia-14-46Search in Google Scholar

[27] 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, 2018:90.10.1186/s13662-018-1545-7Search in Google Scholar PubMed PubMed Central

[28] Nenezic M., Zhu L., Some improvements of Jordan-Steckin and Becker-Stark inequalities, Appl. Anal. Discr. Math., 2018, 12, 244-256.10.2298/AADM1801244NSearch in Google Scholar

[29] Malesevic B., Makragic M., A method for proving some inequalities on mixed trigonometric polynomial functions, J. Math. Inequal., 2016, 10, 849-876.10.7153/jmi-10-69Search in Google Scholar

Received: 2018-12-29
Accepted: 2019-09-15
Published Online: 2019-12-13

© 2019 Yang et al., published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 Public License.

Articles in the same Issue

  1. Regular Articles
  2. On the Gevrey ultradifferentiability of weak solutions of an abstract evolution equation with a scalar type spectral operator of orders less than one
  3. Centralizers of automorphisms permuting free generators
  4. Extreme points and support points of conformal mappings
  5. Arithmetical properties of double Möbius-Bernoulli numbers
  6. The product of quasi-ideal refined generalised quasi-adequate transversals
  7. Characterizations of the Solution Sets of Generalized Convex Fuzzy Optimization Problem
  8. Augmented, free and tensor generalized digroups
  9. Time-dependent attractor of wave equations with nonlinear damping and linear memory
  10. A new smoothing method for solving nonlinear complementarity problems
  11. Almost periodic solution of a discrete competitive system with delays and feedback controls
  12. On a problem of Hasse and Ramachandra
  13. Hopf bifurcation and stability in a Beddington-DeAngelis predator-prey model with stage structure for predator and time delay incorporating prey refuge
  14. A note on the formulas for the Drazin inverse of the sum of two matrices
  15. Completeness theorem for probability models with finitely many valued measure
  16. Periodic solution for ϕ-Laplacian neutral differential equation
  17. Asymptotic orbital shadowing property for diffeomorphisms
  18. Modular equations of a continued fraction of order six
  19. Solutions with concentration and cavitation to the Riemann problem for the isentropic relativistic Euler system for the extended Chaplygin gas
  20. Stability Problems and Analytical Integration for the Clebsch’s System
  21. Topological Indices of Para-line Graphs of V-Phenylenic Nanostructures
  22. On split Lie color triple systems
  23. Triangular Surface Patch Based on Bivariate Meyer-König-Zeller Operator
  24. Generators for maximal subgroups of Conway group Co1
  25. Positivity preserving operator splitting nonstandard finite difference methods for SEIR reaction diffusion model
  26. Characterizations of Convex spaces and Anti-matroids via Derived Operators
  27. On Partitions and Arf Semigroups
  28. Arithmetic properties for Andrews’ (48,6)- and (48,18)-singular overpartitions
  29. A concise proof to the spectral and nuclear norm bounds through tensor partitions
  30. A categorical approach to abstract convex spaces and interval spaces
  31. Dynamics of two-species delayed competitive stage-structured model described by differential-difference equations
  32. Parity results for broken 11-diamond partitions
  33. A new fourth power mean of two-term exponential sums
  34. The new operations on complete ideals
  35. Soft covering based rough graphs and corresponding decision making
  36. Complete convergence for arrays of ratios of order statistics
  37. Sufficient and necessary conditions of convergence for ρ͠ mixing random variables
  38. Attractors of dynamical systems in locally compact spaces
  39. Random attractors for stochastic retarded strongly damped wave equations with additive noise on bounded domains
  40. Statistical approximation properties of λ-Bernstein operators based on q-integers
  41. An investigation of fractional Bagley-Torvik equation
  42. Pentavalent arc-transitive Cayley graphs on Frobenius groups with soluble vertex stabilizer
  43. On the hybrid power mean of two kind different trigonometric sums
  44. Embedding of Supplementary Results in Strong EMT Valuations and Strength
  45. On Diophantine approximation by unlike powers of primes
  46. A General Version of the Nullstellensatz for Arbitrary Fields
  47. A new representation of α-openness, α-continuity, α-irresoluteness, and α-compactness in L-fuzzy pretopological spaces
  48. Random Polygons and Estimations of π
  49. The optimal pebbling of spindle graphs
  50. MBJ-neutrosophic ideals of BCK/BCI-algebras
  51. A note on the structure of a finite group G having a subgroup H maximal in 〈H, Hg
  52. A fuzzy multi-objective linear programming with interval-typed triangular fuzzy numbers
  53. Variational-like inequalities for n-dimensional fuzzy-vector-valued functions and fuzzy optimization
  54. Stability property of the prey free equilibrium point
  55. Rayleigh-Ritz Majorization Error Bounds for the Linear Response Eigenvalue Problem
  56. Hyper-Wiener indices of polyphenyl chains and polyphenyl spiders
  57. Razumikhin-type theorem on time-changed stochastic functional differential equations with Markovian switching
  58. Fixed Points of Meromorphic Functions and Their Higher Order Differences and Shifts
  59. Properties and Inference for a New Class of Generalized Rayleigh Distributions with an Application
  60. Nonfragile observer-based guaranteed cost finite-time control of discrete-time positive impulsive switched systems
  61. Empirical likelihood confidence regions of the parameters in a partially single-index varying-coefficient model
  62. Algebraic loop structures on algebra comultiplications
  63. Two weight estimates for a class of (p, q) type sublinear operators and their commutators
  64. Dynamic of a nonautonomous two-species impulsive competitive system with infinite delays
  65. 2-closures of primitive permutation groups of holomorph type
  66. Monotonicity properties and inequalities related to generalized Grötzsch ring functions
  67. Variation inequalities related to Schrödinger operators on weighted Morrey spaces
  68. Research on cooperation strategy between government and green supply chain based on differential game
  69. Extinction of a two species competitive stage-structured system with the effect of toxic substance and harvesting
  70. *-Ricci soliton on (κ, μ)′-almost Kenmotsu manifolds
  71. Some improved bounds on two energy-like invariants of some derived graphs
  72. Pricing under dynamic risk measures
  73. Finite groups with star-free noncyclic graphs
  74. A degree approach to relationship among fuzzy convex structures, fuzzy closure systems and fuzzy Alexandrov topologies
  75. S-shaped connected component of radial positive solutions for a prescribed mean curvature problem in an annular domain
  76. On Diophantine equations involving Lucas sequences
  77. A new way to represent functions as series
  78. Stability and Hopf bifurcation periodic orbits in delay coupled Lotka-Volterra ring system
  79. Some remarks on a pair of seemingly unrelated regression models
  80. Lyapunov stable homoclinic classes for smooth vector fields
  81. Stabilizers in EQ-algebras
  82. The properties of solutions for several types of Painlevé equations concerning fixed-points, zeros and poles
  83. Spectrum perturbations of compact operators in a Banach space
  84. The non-commuting graph of a non-central hypergroup
  85. Lie symmetry analysis and conservation law for the equation arising from higher order Broer-Kaup equation
  86. Positive solutions of the discrete Dirichlet problem involving the mean curvature operator
  87. Dislocated quasi cone b-metric space over Banach algebra and contraction principles with application to functional equations
  88. On the Gevrey ultradifferentiability of weak solutions of an abstract evolution equation with a scalar type spectral operator on the open semi-axis
  89. Differential polynomials of L-functions with truncated shared values
  90. Exclusion sets in the S-type eigenvalue localization sets for tensors
  91. Continuous linear operators on Orlicz-Bochner spaces
  92. Non-trivial solutions for Schrödinger-Poisson systems involving critical nonlocal term and potential vanishing at infinity
  93. Characterizations of Benson proper efficiency of set-valued optimization in real linear spaces
  94. A quantitative obstruction to collapsing surfaces
  95. Dynamic behaviors of a Lotka-Volterra type predator-prey system with Allee effect on the predator species and density dependent birth rate on the prey species
  96. Coexistence for a kind of stochastic three-species competitive models
  97. Algebraic and qualitative remarks about the family yy′ = (αxm+k–1 + βxmk–1)y + γx2m–2k–1
  98. On the two-term exponential sums and character sums of polynomials
  99. F-biharmonic maps into general Riemannian manifolds
  100. Embeddings of harmonic mixed norm spaces on smoothly bounded domains in ℝn
  101. Asymptotic behavior for non-autonomous stochastic plate equation on unbounded domains
  102. Power graphs and exchange property for resolving sets
  103. On nearly Hurewicz spaces
  104. Least eigenvalue of the connected graphs whose complements are cacti
  105. Determinants of two kinds of matrices whose elements involve sine functions
  106. A characterization of translational hulls of a strongly right type B semigroup
  107. Common fixed point results for two families of multivalued A–dominated contractive mappings on closed ball with applications
  108. Lp estimates for maximal functions along surfaces of revolution on product spaces
  109. Path-induced closure operators on graphs for defining digital Jordan surfaces
  110. Irreducible modules with highest weight vectors over modular Witt and special Lie superalgebras
  111. Existence of periodic solutions with prescribed minimal period of a 2nth-order discrete system
  112. Injective hulls of many-sorted ordered algebras
  113. Random uniform exponential attractor for stochastic non-autonomous reaction-diffusion equation with multiplicative noise in ℝ3
  114. Global properties of virus dynamics with B-cell impairment
  115. The monotonicity of ratios involving arc tangent function with applications
  116. A family of Cantorvals
  117. An asymptotic property of branching-type overloaded polling networks
  118. Almost periodic solutions of a commensalism system with Michaelis-Menten type harvesting on time scales
  119. Explicit order 3/2 Runge-Kutta method for numerical solutions of stochastic differential equations by using Itô-Taylor expansion
  120. L-fuzzy ideals and L-fuzzy subalgebras of Novikov algebras
  121. L-topological-convex spaces generated by L-convex bases
  122. An optimal fourth-order family of modified Cauchy methods for finding solutions of nonlinear equations and their dynamical behavior
  123. New error bounds for linear complementarity problems of Σ-SDD matrices and SB-matrices
  124. Hankel determinant of order three for familiar subsets of analytic functions related with sine function
  125. On some automorphic properties of Galois traces of class invariants from generalized Weber functions of level 5
  126. Results on existence for generalized nD Navier-Stokes equations
  127. Regular Banach space net and abstract-valued Orlicz space of range-varying type
  128. Some properties of pre-quasi operator ideal of type generalized Cesáro sequence space defined by weighted means
  129. On a new convergence in topological spaces
  130. On a fixed point theorem with application to functional equations
  131. Coupled system of a fractional order differential equations with weighted initial conditions
  132. Rough quotient in topological rough sets
  133. Split Hausdorff internal topologies on posets
  134. A preconditioned AOR iterative scheme for systems of linear equations with L-matrics
  135. New handy and accurate approximation for the Gaussian integrals with applications to science and engineering
  136. Special Issue on Graph Theory (GWGT 2019)
  137. The general position problem and strong resolving graphs
  138. Connected domination game played on Cartesian products
  139. On minimum algebraic connectivity of graphs whose complements are bicyclic
  140. A novel method to construct NSSD molecular graphs
Downloaded on 24.12.2025 from https://www.degruyterbrill.com/document/doi/10.1515/math-2019-0098/html
Scroll to top button