Home Sharp Approximations for the Generalized Elliptic Integral of the First Kind
Article Open Access

Sharp Approximations for the Generalized Elliptic Integral of the First Kind

  • Zai-Yin He , Yue-Ping Jiang and Miao-Kun Wang EMAIL logo
Published/Copyright: March 31, 2023
Become an author with De Gruyter Brill

Abstract

For a ∈ (0, 1/2], r ∈ (0, 1), let 𝒦a(r) (𝒦(r)) be the generalized (complete) elliptic integral of the first kind. In the article, we prove some monotonicity properties of certain combination of functions involving 𝒦a(r), and thus establish its two sharp inequalities, which extend and improve some well-known results of 𝒦(r).

2020 Mathematics Subject Classification: 33E05; 26D07

1. Introduction

For real numbers a, b and c with c ≠ 0, −1, −2,…, the Gaussian hypergeometric function [1, 7] is defined by

F(a,b;c;x)=n=0(a)n(b)n(c)nxnn!,x(1,1),

where (a)0 = 1 for a ≠ 0, (a)n = a(a + 1)(a + 2)…(a + n − 1) = Γ(a + n)/Γ(a) is the Pochhammer (shifted factorial) function, and Γ(x)=0tx1etddt (x>0) is the Euler gamma function (cf. [1, 30]). If a + b = c, then F(a, b; c; x) is called zero-balanced.

As a special case of the Gaussian hypergeometric function, for a ∈ (0, 1) and r ∈ (0, 1), the generalized elliptic integral of the first kind [4, 18] is defined as

1.1 {Ka=Ka(r)=πF(a,1a;1;r2)/2,Ka(0)=π/2, Ka(1)=+.

When a = 1/2, 𝒦1/2(r) = 𝒦(r) (cf. [2, 3]) is the Legendre complete elliptic integral of the first kind. By the symmetry of (1.1), we shall assume that a ∈ (0, 1/2] throughout this article.

It is well known that the above special functions F(a, b; c; x) and 𝒦a(r)(𝒦(r)) have been widely applied in mathematics and physics. A lot of problems in geometric function theory, theory of mean values, number theory and theory of elasticity all depend on these functions [5, 8, 11, 13, 25, 35, 36, 38]. During the past few decades, many mathematicians have expanded a great deal of effort in studying asymptotic properties and establishing sharp bounds for some combinations of functions involving F(a, b; c; x), 𝒦a(r) and 𝒦(r). As applications, a lot of distortion estimates in quasi-conformal mappings and analytic properties of modular functions in Ramanujan modular equation have been derived (cf. [9, 14, 15, 17, 1922, 3033, 37, 39]).

Early at the beginning of the 20-th century, Ramanujan established the asymptotic behavior of the zero-balanced Gaussian hypergeometric function (cf. [7: 1.48] or [12: pp. 33–34])

1.2 B(a,b)F(a,b;a+b;x)+log(1x)=R(a,b)+O[(1x)log(1x)]

as x ↦ 1, where B(a, b) = Γ(a)Γ(b)/Γ(a + b) is the beta function, R(a, b) = −Γ′(a)/Γ(a) − Γ′(b)/Γ(b) − 2γ, and γ=limn(k=1n1klogn)=0.5772 denotes the Euler-Mascheroni constant. Estimates for zero-balanced hypergeometric functions are given in [7] whereas the estimates for the non-zero balanced hypergeometric functions are discussed in [23]. See also [10]. Let

1.3 B(a)=B(a,1a)=Γ(a)Γ(1a)=πsin(πa)

and

1.4 R(a)=R(a,1a)=Γ(a)Γ(a)Γ(1a)Γ(1a)2γ.

Then from (1.2) we obtain

1.5 B(a)F(a,1a;1;x)+log(1x)=R(a)+O[(1x)log(1x)],x1.

Using (1.1)–(1.5), it is not difficult to get the asymptotic formulas for generalized (complete) elliptic integral of the first kind: for r ↦ 1,

2Ka(r)sin(πa)+2logr=R(a)+O[(1r2)log(1r2)],

or

Ka(r)sin(πa)log(eR(a)/2r)

and

K(r)+logr=log4+O[(1r2)log(1r2)],

or

K(r)log(4r),

respectively. Here and in what follows we set r=1r2.

In [6: Theorem 1.48], Anderson, Vamanamurthy and Vuorinen showed that the quotient function r𝒦(r)/log(4/r′) is strictly decreasing from (0, 1) onto (1, π/(4 log 2)), from which it follows that

1.6 log(4r)<K(r)<π4log2log(4r),r(0,1).

Later, Qiu, Vamanamurthy and Vuorinen [26: Theorem 1.5] proved that r𝒦(r)/ log[αr′ + (4/r′)] is piecewise monotone on (0, 1) with α = eπ/2 − 4. Consequently, they obtained the following refinement

1.7 K(r)>log(αr+4r),r(0,1).

Furthermore, the authors [26] pointed out that, for c ∈ (−∞, ∞), α is the best possible constant such that 𝒦(r) > log(cr′ + 4/r′) holds for all r ∈ (0, 1).

On the other hand, in 2000, Qiu [24: Theorem 2.3] extended (1.6) to the zero-balanced Gaussian hypergeometric function and proved that:

1.8 1B(a,b)log(eR(a,b)1x)<F(a,b;a+b;x)<1R(a,b)log(eR(a,b)1x)

for a, b > 0 with R(a, b) > 0 and x ∈ (0, 1). Letting b = 1 − a and r=x, then from (1.1), (1.3), (1.4) and (1.8), we have

1.9 sin(πa)log(eR(a)/2r)<Ka(r)<πR(a)log(eR(a)/2r),r(0,1).

Motivated by (1.6), (1.7) and (1.9), one of the main purposes of this paper is to find analogue of the inequality (1.7) for the generalized elliptic integral of the first kind 𝒦a(r).

For r ∈ (0, 1), let arth(r) denote the inverse of the hyperbolic tangent function. Very recently, with the combination of the arithmetic-geometric mean and some other classical bivariate means, Wang et al. [34: Theorems 3.5 and 4.2] proved that the function r ↦ [4r𝒦(r)/π − arth(r) − 3r/(2 + r′)]/[arth(r)/2 + r/2 − 3r/(2 + r′)] is strictly increasing from (0, 1) onto (17/44, 8/π − 2). Consequently, the following sharp inequality has been derived: for r ∈ (0, 1),

1.10 π2[1744(3arth(r)4r+14)+(11744)(arth(r)2r+32(2+r))]<K(r)<π2[(8π2)(3arth(r)4r+14)+(38π)(arth(r)2r+32(2+r))].

Another purpose of this paper is to extend the above result to the generalized elliptic integral of the first kind 𝒦a(r).

We now state our main results.

Theorem 1.1

Let a ∈ (0, 1/2], c ∈ (−∞, +∞),

λa=eπ/[2sin(πa)][sin(πa)a(1a)π]/[2sin(πa)]

and

1.11 Fc(r)=reKa(r)/sin(πa)eR(a)/2cr2,r(0,1).

Then we have the following conclusions:

  1. The function Fc(r) is strictly decreasing if and only if c ≤ λa, in which case the range of Fc(r) on (0, 1) is (0,eπ/[2sin(πa)]eR(a)/2c). Consequently, the inequality

    1.12 sin(πa)log(cr+eR(a)/2r)<Ka(r)<sin(πa)log(cr+eπ/[2sin(πa)]cr)

    holds for all r ∈ (0, 1);

  2. If c > λa, then there exists an r0 ∈ (0, 1), such that Fc(r) is strictly increasing on (0, r0) and strictly decreasing on (r0, 1). Moreover, the inequality

    1.13 Ka(r)>sin(πa)log(cr+eR(a)/2r)

    holds for all r ∈ (0, 1) if and only if cc0=eπ/[2sin(πa)]eR(a)/2.

Theorem 1.2

Let a ∈ (0, 1/2] and

G(r)=2rKa(r)/π+(6a26a+1)arth(r)6(3a23a+1)r/(2+r)arth(r)6r/(2+r)+r,r(0,1).

Then G(r) is strictly increasing from (0, 1) onto ((45a4 − 90a3 + 96a2 − 51a + 11)/11, 2 sin(πa)/π + 6a2 − 6a + 1). Consequently, the inequality

1.14 15π22a(a1)(3a23a1)arth(r)r+(4511a4+9011a36311a2+1811a)3π2+r+π2(4511a49011a3+9611a25111a+1)<Ka(r)<sin(πa)arth(r)r+(3a2+3a2sin(πa)π)3π2+r+π2(2sin(πa)π+6a26a+1)

holds for all r ∈ (0, 1).

Remark 1

If we take a = 1/2, then the inequalities (1.13) and (1.14) reduce to (1.7) and (1.10), respectively.

2. Lemmas

In order to prove our main results, we shall need several formulas and lemmas which we present in this section. First of all, for a ∈ (0, 1) and r ∈ (0, 1), the generalized elliptic integral of the second kind [4, 18] is defined as

2.1 Ea=Ea(r)=π2F(a1,1a;1;r2),Ea(0+)=π2,Ea(1)=sin(πa)2(1a).

In the particular case a = 1/2, then a reduces to the complete elliptic integral of the second kind (cf. [2, 7]). We also need the following derivative formulas (cf. [4: Theorem 4.1]):

dKadr=2(1a)(Ear2Ka)rr2,dEadr=2(a1)(KaEa)r,

d(Ear2Ka)dr=2arKa,d(KaEa)dr=2(1a)rEar2.

Lemma 2.1. (cf. [7: Theorem 1.25])

Let −∞ < a < b < ∞, f, g: [a, b] → ℝ be continuous on [a, b] and differentiable on (a, b), and g′(x) ≠ 0 on (a, b). If f′(x)/g′(x) is increasing (decreasing) on (a, b), then so are the functions

f(x)f(a)g(x)g(a),f(x)f(b)g(x)g(b).

If f′(x)/g′(x) is strictly monotone, then the monotonicity in the conclusion is also strict.

Lemma 2.2. (cf. [16] or [23: Lemma 2.1])

Suppose that the power series f(x)=n=0anxn and g(x)=n=0bnxn both converge for |x| < R, and bn > 0 for all n ∈ {0, 1, 2,…}. If the non-constant sequence {an/bn}n=0 is increasing (decreasing), then h(x) = f(x)/g(x) is strictly increasing (decreasing) on (0, R).

Lemma 2.2 is basically due to [16] and in this form with a general setting was stated in [23] along with many applications which were later adopted by a number of researchers.

Lemma 2.3

Let a ∈ (0, 1/2]. Then the following statements are true:

  1. The function ϕ1(r) = (ar2𝒦a/r2 is strictly increasing from (0, 1) onto (πa/2, sin(πa)/[2(1 − a)]);

  2. The function ϕ2(r) = rc𝒦a is strictly decreasing from (0, 1) onto (0, π/2) if and only if c ≥ 2a(1 − a). Moreover, rrKa is strictly decreasing for each a ∈ (0, 1/2];

  3. The function ϕ3(r) = 𝒦a/sin(πa) + log ris strictly decreasing from (0, 1) onto (R(a)/2, π/(2 sin(πa)));

  4. The function ϕ4(r)=[(1+r2)sin(πa)2(1a)(Ear2Ka)]/r2(1+a)(2a)/3 is strictly decreasing from (0, 1) onto (0, 2 sin(πa)).

Proof

Parts (1), (2) and (3) can be found in [4: Lemmas 5.2(1), 5.4(1) and 5.5(1)].

For part (4), let ϕ41(r)=(1+r2)sin(πa)2(1a)(Ear2Ka) and ϕ42(r) = r2(1 + a)(2 − a)/3. Simple computations lead to that

2.2 ϕ4(r)=ϕ41(r)ϕ42(r),ϕ41(1)=ϕ42(1)=0

and

2.3 ϕ41(r)ϕ42(r)=3(1+a)(2a)[r2(a2a+1)/3sin(πa)+2a(1a)r2(a2a+1)/3Ka].

It is not difficult to verify that r2(a2a + 1)/3𝒦a is strictly decreasing on (0, 1) from part (2), so that ϕ41(r)/ϕ42(r) is strictly decreasing on (0, 1) by (2.3). Combining with (2.2) and applying Lemma 2.1, we derive that ϕ4(r) is strictly decreasing on (0, 1).

Part (1) implies that ϕ4(0+) = 2 sin(πa), and by (2.2), (2.3) and L’Hospital’s rule we get ϕ4(1) = 0. Therefore, part (4) is valid. □

Lemma 2.4

For a ∈ (0, 1/2], define ψ on (0, 1) by

2.4 ψ(r)=(1a)(Ear2Ka)2+sin(πa)r2[ar2Ka(Ear2Ka)]r4.

Then φ(r) is strictly increasing from (0, 1) onto (πa2(1a)[π+sin(πa)]/4,[sin2(πa)]/[4(1a)]).

Proof

Let ψ1(r)=(1a)(Ear2Ka)2+sin(πa)r2[ar2Ka(Ear2Ka)]. Then differentiating ψ1 gives

ψ1(r)=2r[2a(1a)Ka(Ear2Ka)+(a2+a+1)sin(πa)(Ear2Ka)asin(πa)r2Ka],

so that

2.5 r5ψ(r)=rψ1(r)4ψ1(r)=2r2[2a(1a)Ka(Ear2Ka)+(a2+a+1)sin(πa)(Ear2Ka)asin(πa)r2Ka]4(1a)(Ear2Ka)24sin(πa)r2[ar2Ka(Ear2Ka)]=4(1a)(Ear2Ka)[ar2Ka(Ear2Ka)]+2a(1a)sin(πa)r2(Ear2Ka)2sin(πa)r2[ar2Ka(Ear2Ka)]4sin(πa)r2[ar2Ka(Ear2Ka)]=2a(1a)sin(πa)r2(Ear2Ka)×{11a(1a)sin(πa)[r2Ear2Ka]×[(1+r2)sin(πa)2(1a)(Ear2Ka)r23(1+a)(2a)]×[r23(1+a)(2a)(ar2Ka(Ear2Ka))r4]}.

Let ψ2(r)=r23(1+a)(2a)[ar2Ka(Ear2Ka)]/r4. Then from (1.1) and (2.1) one has

ψ2(r)=π[a2(1a)/4]r23(1+a)(2a)F(1+a,2a;3;r2).

Since the function x ↦ (1 − x)d F(a, b; c; x) with a, b, c > 0 and d ∈ ℝ is decreasing on (0, 1) if and only if d ≥ max{a + bc, ab/c}(cf. [27: Lemma 2.15] and also [23]), then ψ2(r) is strictly decreasing on (0, 1).

Obviously,

2.6 ψ2(0+)=a2(1a)π4andψ2(1)=0.

Furthermore, it follows from (2.5), (2.6) together with Lemma 2.3 (1) and (4) that ψ′(r) > 0 for all r ∈ (0, 1), so that ψ(r) is strictly increasing on (0, 1).

Using (2.6) and Lemma 2.3(1), we get

2.7 ψ(0+)=πa2(1a)4[π+sin(πa)]andψ(1)=sin2(πa)4(1a).

Therefore, Lemma 2.4 follows from (2.7) and the monotonicity of ψ(r). □

Lemma 2.5

For a ∈ (0, 1/2], define φ on (0, 1) by

2.8 φ(r)=eKa(r)sin(πa)sin(πa)2(1a)(Ear2Ka)/r2r.

Then φ(r) is strictly increasing from (0, 1) onto (eπ/[2 sin(πa)][sin(πa) − a(1 − a)π], +∞).

Proof

Differentiation gives

2.9 φ(r)=eKa(r)sin(πa)2(1a)(Ear2Ka)rr2sin(πa)[sin(πa)2(1a)(Ear2Ka)/r2r]+eKa(r)sin(πa)4(1a)r2[(Ear2Ka)ar2Ka]+r2[r2sin(πa)2(1a)(Ear2Ka)]r3r3=eKa(r)sin(πa)r3r3sin(πa){r4[sin(πa)]24(1a)2(Ear2Ka)4(1a)sin(πa)r2[ar2Ka(Ear2Ka)]}=r[sin(πa)]eKa(r)sin(πa)r3[14(1a)[sin(πa)]2ψ(r)],

where ψ(r), defined in (2.4), is strictly increasing on (0, 1).

By (1.1), (2.1) and Lemma 2.3 (1), (3) together with L’Hospital’s rule, we obtain that

2.10 φ(0+)=eπ2sin(πa)[sin(πa)a(1a)π]

and

2.11 φ(1)=limr1eKa(r)sin(πa)+logrlimr1sin(πa)2(1a)(Ear2Ka)/r2r2=2(1a)eR(a)2limr1ar2Ka(Ear2Ka)r4=+.

Therefore, Lemma 2.5 follows from (2.10), (2.11) and the monotonicity of φ(r). □

Lemma 2.6

For a ∈ (0, 1/2], define the function h on (0, 1) by

h(r)=r2[a(12a)r2(Ear2Ka)(a22a+2)r2(KaEa)a2(12a)r4Ka+2(KaEa)2(1a)r2Ea]r2[(1a)r2Eaa(12a)r2(Ear2Ka)r2(KaEa)].

Then h(r) is strictly decreasing from (0, 1) onto (0, a(1 − a)(7a2 − 7a + 4)/[3(5a2 − 5a + 2)]).

Proof

Let

h1(r)=r2[a(12a)r2(Ear2Ka)(a22a+2)r2(KaEa)a2(12a)r4Ka+2(KaEa)2(1a)r2Ea]

and

h2(r)=r2[(1a)r2Eaa(12a)r2(Ear2Ka)r2(KaEa)].

Then h(r) = h1(r)/h2(r) and tedious computations lead to that

2πh1(r)=r2[n=0a2(12a)(a)n(1a)n(n+1)(n!)2r2n+4+n=0(a22a+2)(a)n(1a)n+1(n+1)(n!)2r2n+4+n=0a2(12a)(a)n(1a)n(n!)2r2n+4+n=02(a)n(1a)n+1(n+1)(n!)2r2n+2+n=02(1a)(a1)n(1a)n(n!)2r2n+2]=r2[n=0a2(12a)(a)n(1a)n(n+1)(n!)2r2n+4+n=0(a22a+2)(a)n(1a)n+1(n+1)(n!)2r2n+4+n=0a2(12a)(a)n(1a)n(n!)2r2n+4+n=02(a22a+n+2)(a)n(1a)n+1(n+2)(n+1)(n!)2r2n+4]=r2n=0[a2(12a)(a)n(1a)n(n+1)(n!)2+(a22a+2)(a)n(1a)n+1(n+1)(n!)2+a2(12a)(a)n(1a)n(n!)2+2(a22a+n+2)(a)n(1a)n+1(n+2)(n+1)(n!)2]r2n+4=n=1a(a)n(1a)n[2(a2a+1)n+5a25a+2](n+2)(n+1)n[(n1)!]2r2n+4n=0a(a)n(1a)n[2(a2a+1)n+5a25a+2](n+2)(n+1)n[(n1)!]2r2n+6=n=0a(a)n(1a)n(n+3)(n+2)(n+1)(n!)2[2(a2a+1)n2+2(a42a3+6a25a+1)na(1a)(7a27a+4)]r2n+6

and

2πh2(r)=n=0a(a)n(1a)n[2(a2a+1)n+5a25a+2](n+2)(n+1)(n!)2r2n+6.

Then

2.12 h(r)=h1(r)h2(r)=n=0Anr2nn=0Bnr2n,

where

An=a(a)n(1a)n(n+3)(n+2)(n+1)(n!)2×[2(a2a+1)n2+2(a42a3+6a25a+1)na(1a)(7a27a+4)]

and

Bn=a(a)n(1a)n[2(a2a+1)n+5a25a+2](n+2)(n+1)(n!)2>0

for all n ∈ ℕ. Let Cn = An/Bn, one has

Cn=2(a2a+1)n2+2(a42a3+6a25a+1)na(1a)(7a27a+4)[2(a2a+1)n+5a25a+2](n+3)

and

Cn+1Cn=(a+1)(2a)[2(a2a+1)n+7a27a+4][2(a2a+1)n+5a25a+2](n+3)(n+4)×{2[(2a22a+3)n+(16a216a+9)](a2a+1)n+61a4122a3+113a252a+12}.

It is not difficult to verify that 61a4 − 122a3 + 113a2 − 52a + 12 > 0 for all a ∈ (0, 1/2]. Thus, Cn + 1Cn > 0, i.e. Cn is strictly increasing for all n ∈ ℕ, and h(r) is strictly decreasing on (0, 1) by Lemma 2.2 and (2.12).

Note that

2.13 h(0+)=A0B0=a(1a)(7a27a+4)3(5a25a+2),h(1)=0.

Therefore, Lemma 2.6 follows from (2.13) and the monotonicity of h(r). □

Lemma 2.7

For a ∈ (0, 1/2] and c ∈ (−∞, +∞), define gc on (0, 1) by

2.14 gc(r)=rc[(1a)r2Eaa(12a)r2(Ear2Ka)r2(KaEa)]r4.

Then the following statements are true:

  1. gc(r) is strictly decreasing on (0, 1) if and only if c ≥ 2a(1 − a)(7a2 − 7a + 4)/[3(5a2 − 5a + 2)];

  2. gc(r) is strictly increasing on (0, 1) if and only if c ≤ 0.

Proof

Differentiating gc gives

2.15 gc(r)=rc2[(1a)r2Eaa(12a)r2(Ear2Ka)r2(KaEa)]r3[2h(r)c],

where h(r) is defined in Lemma 2.6. The proof of Lemma 2.6 shows that the function r(1a)r2Eaa(12a)r2(Ear2Ka)r2(KaEa) is positive on (0, 1). Combining with (2.15), we obtain that gc(r)<0 for r ∈ (0, 1) if and only if c ≥ 2a(1 − a)(7a2 − 7a + 4)/[3(5a2 − 5a + 2)], and gc(r)>0 for r ∈ (0, 1) if and only if c ≤ 0. Therefore, Lemma 2.7 follows. □

3. Proofs of Theorems 1.1 and 1.2

Proof of Theorem 1.1. Differentiating Fc in (1.11) gives

3.1 Fc(r)=rreKa(r)sin(πa)+eKa(r)sin(πa)2(1a)(Ear2Ka)rrsin(πa)+2cr=eKa(r)sin(πa)rrsin(πa)[2(1a)(Ear2Ka)r2sin(πa)]+2cr=r[2cφ(r)sin(πa)],

where φ(r) is defined in (2.8). The assertion about monotonicity properties of Fc in parts (1) and (2) directly follows from (3.1) and Lemma 2.5. Employing (1.1) and Lemma 2.3 (3) gives

3.2 Fc(0+)=eπ/(2sin(πa))eR(a)/2c,Fc(1)=limr1[eKa(r)sin(πa)+logreR(a)2cr2]=0.

Therefore the inequality (1.12) holds.

Finally, for the inequality (1.13), according to the range of Fλa on (0, 1) in part (1) we conclude that c0=eπ/[2sin(πa)]eR(a)/2>λa for a ∈ (0, 1/2], so that Fc0 is piecewise monotone on (0, 1). This, together with Fc0(0+)=Fc0(1)=0, shows that Fc0(r)>0 for all r ∈ (0, 1). Since the inequality

Fc(r)=reKa(r)/sin(πa)eR(a)/2cr2>0,r(0,1)

is equivalent to (1.13), we see that

Ka(r)>sin(πa)log(c0r+eR(a)/2r)

holds for all r ∈ (0, 1). Furthermore, (3.2) implies that (1.13) holds for all r ∈ (0, 1) only if cc0. Until now, the optimality of c0 follows, and the proof is completed. □

Proof of Theorem 1.2. Let

G1(r)=2rKa(r)π+(6a26a+1)arth(r)6(3a23a+1)r2+r,G2(r)=arth(r)6r2+r+r,G3(r)=2[r2Ka+2(1a)(Ear2Ka)]π6(3a23a+1)r(1+2r)(2+r)2+6a26a+1,G4(r)=16r(1+2r)(2+r)2+r2,G5(r)=2a(12a)r2Ka2(1a)(KaEa)πr2+3(3a23a+1)(2+7r)r(2+r)3,G6(r)=1+3(2+7r)r(2+r)3.

Then simple computations lead to that

G(r)=G1(r)/G2(r),G1(r)/G2(r)=G3(r)/G4(r),G3(r)/G4(r)=G5(r)/G6(r),G1(0+)=G2(0+)=G3(0+)=G4(0+)=G5(0+)=G6(0+)=0

and

3.3 G5(r)G6(r)=3a23a+14(1a)3πr1/2(2+r)421r2+8r+4×r1/2[(1a)r2Eaa(12a)r2(Ear2Ka)r2(KaEa)]r4.

It is easy to check that the function rr(2 + r2)4/(21r4 + 8r2 + 4) is strictly increasing from (0, 1) onto (0, 27/11), and 1/2 > 2a(1 − a)(7a2 − 7a + 4)/[3(5a2 − 5a + 2) for a ∈ (0, 1/2]. Thus, G5(r)/G6(r) is strictly increasing on (0, 1) by (3.3) and Lemma 2.7. Applying Lemma 2.1 three times, we conclude that G(r) is strictly increasing on (0, 1). Moreover,

3.4 G(0+)=limr0+G5(r)G6(r)=45a490a3+96a251a+1111

and

3.5 G(1)=limr1F(a,1a;1;r2)+(6a26a+1)F(12,1;32;r2)3(3a23a+1)F(12,1;32;r2)2=2sin(πa)π+6a26a+1.

Therefore, Theorem 1.2 follows from (3.4) and (3.5) together with the monotonicity of G(r). □

Remark 2

(1) One of the reviewers reminded us that the function R(a) and its properties are very important, and suggested to add its graph in the paper. In fact, it has been shown in [27: Lemma 2.14 (2)] that a ↦ [R(a) − log 16]/(1/2 − a)2 is strictly decreasing from (0, 1/2] onto (14ζ(3), ∞), so that inequality R(a) ≥ 14ζ(3)(1/2 − a)2 + log 16 ≥ log 16 is valid for all a ∈ (0, 1/2], and aR(a) is also strictly decreasing from (0, 1/2] onto (log 16, +∞)(cf. Figure 1(a)), where ζ(x) is the Riemann zeta function. For series expansion and more asymptotic inequalities of R(a), readers can refer to [28, 29].

(2) Numerical experiment results demonstrate that for some given a ∈ (0, 1/2], our lower bound for 𝒦a(r) in (1.14) is particularly good for r ∈ (0, 0.8), and the upper bounds in Theorems 1.1 and 1.2 are not comparable on the whole interval (0, 1). Moreover, the difference of the two sides in (1.13) with c=c0=eπ/[2sin(πa)]eR(a)/2 tends to 0 as r tends to 0 or 1. For the above, see Figure 1(b)–(f) and Table 1. Here, for (a, r) ∈ (0, 1/2] × (0, 1),

H1(a,r)=sin(πa)log((eπ/[2sin(πa)]eR(a)/2)r+eR(a)/2r),H2(a,r)=15πa(a1)(3a23a1)22arth(r)r+(4511a4+9011a36311a2+1811a)3π2+r+π2(4511a49011a3+9611a25111a+1),H3(a,r)=sin(πa)arth(r)r+(3a3a22sin(πa)π)3π2+r+π2(2sin(πa)π+6a26a+1)

and

H4(a,r)=sin(πa)log(eπ/[2sin(πa)][sin(πa)a(1a)π]2sin(πa)r+eπ/[2sin(πa)][sin(πa)+a(1a)π]2sin(πa)r).

Figure 1
Figure 1

Table 1

Relative errors of the approximation of 𝒦a(r) by H2(a, r) with a = 1/4 and a = 1/3

r K1/4(r)H2(1/4,r)K1/4(r)×100% K1/3(r)H2(1/3,r)K1/3(r)×100%
0.1 6.02 × 10−10 1.27 × 10−9
0.2 3.98 × 10−8 3.28 × 10−8
0.3 4.87 × 10−7 3.94 × 10−7
0.4 3.01 × 10−6 2.43 × 10−6
0.5 1.31 × 10−5 1.06 × 10−5
0.6 4.71 × 10−5 3.79 × 10−5
0.7 1.52 × 10−4 1.22 × 10−4
0.8 4.83 × 10−4 3.86 × 10−4
0.9 1.74 × 10−3 1.38 × 10−3


(Communicated by Marek Balcerzak)


Funding statement: The work was supported by the Natural Science Foundation of China (Grant No. 11701176, 61673169).

Acknowledgement

The authors wish to thank the anonymous referees for helpful comments and suggestions.

References

[1] 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

[2] ALZER, H.: Sharp inequalities for the complete elliptic integral of the first kind, Math. Proc. Cambridge Philos. Soc. 124 (1998), 309–314.10.1017/S0305004198002692Search in Google Scholar

[3] ALZER, H.—QIU, S. L.: Monotonicity theorems and inequalities for the complete elliptic integrals, J. Comput. Appl. Math. 172 (2004), 289–312.10.1016/j.cam.2004.02.009Search in Google Scholar

[4] ANDERSON, G. D.—QIU, S. L.—VAMANAMURTHY, M. K.—VUORINEN, M.: Generalized elliptic integrals and modular equations, Pacific J. Math. 192 (2000), 1–37.10.2140/pjm.2000.192.1Search in Google Scholar

[5] ANDERSON, G. D.—VAMANAMURTHY, M. K.—VUORINEN, M.: Hypergeometric functions and elliptic integrals. In: Current Topics in Analytic Function Theory (H. M. Srivastava and S. Owa, eds.), World Sci. Publ., River Edge, NJ, 1992, pp. 48–85.10.1142/9789814355896_0005Search in Google Scholar

[6] ANDERSON, G. D.—VAMANAMURTHY, M. K.—VUORINEN, M.: Functional inequalities for hypergeometric functions and complete elliptic integrals, SIAM J. Math. Anal. 23 (1992), 512–524.10.1137/0523025Search in Google Scholar

[7] ANDERSON, G. D.—VAMANAMURTHY, M. K.—VUORINEN, M.: Conformal Invariants, Inequalities, and Quasiconformal Maps, John Wiley & Sons, New York, 1997.Search in Google Scholar

[8] BALASUBRAMANIAN, R.—NAIK, S.—PONNUSAMY, S.—VUORINEN, M.: Elliott’s identity and hypergeometric functions, J. Math. Anal. Appl. 271 (2002), 232–256.10.1016/S0022-247X(02)00126-9Search in Google Scholar

[9] BALASUBRAMANIAN, R.—PONNUSAMY, S.—VUORINEN, M.: Functional inequalities for the quotients of hypergeometric functions, J. Math. Anal. Appl. 218 (1998), 256–268.10.1006/jmaa.1997.5776Search in Google Scholar

[10] BALASUBRAMANIAN, R.—PONNUSAMY, S.: On Ramanujan asymptotic expansions and inequalities for hypergeometric functions, Proc. Indian Acad. Sci. Math. Sci. 108 (1998), 95–108.10.1007/BF02841543Search in Google Scholar

[11] BERNDT, B. C. : Ramanujan’s Notebooks, Part I, Springer-Verlag, Berlin, 1985.10.1007/978-1-4612-1088-7Search in Google Scholar

[12] BERNDT, B. C. : Ramanujan’s Notebooks, Part II, Springer-Verlag, Berlin, 1989.10.1007/978-1-4612-4530-8Search in Google Scholar

[13] BERNDT, B. C. : Ramanujan’s Notebooks, Part III, Springer-Verlag, Berlin, 1991.10.1007/978-1-4612-0965-2Search in Google Scholar

[14] BERNDT, B. C.—BHARGAVA, S.—GARVAN, F. G.: Ramanujan’s theories of elliptic functions to alternative bases, Trans. Amer. Math. Soc. 347 (1995), 4163–4244.10.1090/S0002-9947-1995-1311903-0Search in Google Scholar

[15] BHAYO, B. A.—VUORINEN, M.: On generalized complete elliptic integrals and modular functions, Proc. Edinb. Math. Soc. (2) 55 (2012), 591–611.10.1017/S0013091511000356Search in Google Scholar

[16] BIERNACKI, M.—KRZYZ, J.: On the monotonicity of certain functionals in the theory of analytic functions, Ann. Univ. Mariae Curie-Skłodowska 9 (1955), 135–147.Search in Google Scholar

[17] BORWEIN, J. M.— BORWEIN, P. B.: Pi and the AGM: A study in Analytic Number Theory and Computational Complexity, John Wiley & Sons, New York, 1998.Search in Google Scholar

[18] BYRD, P. F.—FRIEDMAN, M. D.: Handbook of Elliptic Integrals for Engineers and Scientists, Springer-Verlag, New York, 1971.10.1007/978-3-642-65138-0Search in Google Scholar

[19] HEIKKALA, V.—VAMANAMURTHY, M. K.—VUORINEN, M.: Generalized elliptic integrals, Comput. Methods Funct. Theory 9 (2009), 75–109.10.1007/BF03321716Search in Google Scholar

[20] KARP, D.—SITNIK, S. M.: Asymptotic approximations for the first incomplete elliptic integral near logarithmic singularity, J. Comput. Appl. Math. 205 (2007), 186–206.10.1016/j.cam.2006.04.053Search in Google Scholar

[21] KARP, D.—SITNIK, S. M.: Log-convexity and log-concavity of hypergeometric-like functions, J. Math. Anal. Appl. 364 (2010), 384–394.10.1016/j.jmaa.2009.10.057Search in Google Scholar

[22] KARP, D.—SAVENKOVA, A.—SITNIK, S. M.: Series expansions for the third incomplete elliptic integral via partial fraction decompositions, J. Comput. Appl. Math. 207 (2007), 331–337.10.1016/j.cam.2006.10.019Search in Google Scholar

[23] PONNUSAMY, S.—VUORINEN, M.: Asymptotic expansions and inequalities for hypergeometric functions, Mathematika 44 (1997), 278–301.10.1112/S0025579300012602Search in Google Scholar

[24] QIU S. L.: Grötzsch ring and Ramanujan’s modular equations, Acta Math. Sin. 43 (2000), 283–290.Search in Google Scholar

[25] QIU, S. L.—MA, X. Y.—CHU, Y. M.: Sharp Landen transformation inequalities for hypergeometric functions, with applications, J. Math. Anal. Appl. 474 (2019), 1306–1337.10.1016/j.jmaa.2019.02.018Search in Google Scholar

[26] QIU, S. L.—VAMANAMURTHY, M. K.—VUORINEN, M.: Some inequalities for the growth of elliptic integrals, SIAM J. Math. Anal. 29 (1998), 1224–1237.10.1137/S0036141096310491Search in Google Scholar

[27] QIU, S. L.—VUORINEN, M.: Duplication inequalities for the ratios of hypergeometric functions, Forum Math. 12 (2000), 109–133.10.1515/form.1999.025Search in Google Scholar

[28] QIU, S. L.—MA, X. Y.—HUANG, T. R.: Some properties of the difference between the Ramanujan constant and beta function, J. Math. Anal. Appl. 446 (2017), 114–129.10.1016/j.jmaa.2016.08.043Search in Google Scholar

[29] QIU, S. L.—MA, X. Y.—HUANG, T. R.: Sharp approximations for the Ramanujan constant, Constr. Approx. 51 (2020), 303–330.10.1007/s00365-019-09464-3Search in Google Scholar

[30] RAINVILLE, E. D.: Special Functions, MacMillan, New York, 1960.Search in Google Scholar

[31] RICHARDS, K. C.—SMITH, J. N.: A concavity property of generalized complete elliptic integrals, Integral Transforms Spec. Funct. 32 (2021), 240–252.10.1080/10652469.2020.1815726Search in Google Scholar

[32] TAN, SH. Y.—HUANG, T. R.—CHU, Y. M.: Functional inequalities for Gaussian hypergeometric function and generalized elliptic integral of the first kind, Math. Slovaca 71 (2021), 667–682.10.1515/ms-2021-0012Search in Google Scholar

[33] TIAN, J. F.—YANG, Z. H.—HA, M. H.—XING, H. J.: A family of high order approximations of Ramanujan type for perimeter of an ellipse, Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat. RACSAM 115 (2021), Art. No. 85, 20 pp.10.1007/s13398-021-01021-7Search in Google Scholar

[34] WANG, M. K.–CHU, Y. M.—LI, Y. M.—ZHANG W.: Asymptotic expansion and bounds for complete elliptic integrals, Math. Inequal. Appl. 23 (2020), 821–841.10.7153/mia-2020-23-67Search in Google Scholar

[35] WANG, M. K.—CHU, Y. M.—JIANG, Y. P.: Ramanujan’s cubic transformation inequalities for zero-balanced hypergeometric functions, Rocky Mountain J. Math. 46 (2016), 679–691.10.1216/RMJ-2016-46-2-679Search in Google Scholar

[36] WANG, M. K.—CHU, Y. M.—ZHANG, W.: Monotonicity and inequalities involving zero-balanced hypergeometric function, Math. Inequal. Appl. 22 (2019), 601–617.10.7153/mia-2019-22-42Search in Google Scholar

[37] WANG, M. K. —LI, Y. M.—CHU, Y. M.: Inequalities and infinite product formula for Ramanujan generalized modular equation function, Ramanujan J. 46 (2018), 189–200.10.1007/s11139-017-9888-3Search in Google Scholar

[38] YANG, Z. H.—TIAN, J. F.: Sharp inequalities for the generalized elliptic integrals of the first kind, Ramanujan J. 48 (2019), 91–116.10.1007/s11139-018-0061-4Search in Google Scholar

[39] ZHAO, T. H.—WANG, M. K.—CHU, Y. M.: Monotonicity and convexity involving generalized elliptic integral of the first kind, Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat. RACSAM 115 (2021), Art. No. 46, 13 pp.10.1007/s13398-020-00992-3Search in Google Scholar

Received: 2021-10-05
Accepted: 2022-03-20
Published Online: 2023-03-31
Published in Print: 2023-04-01

© 2023 Mathematical Institute Slovak Academy of Sciences

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

Downloaded on 18.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/ms-2023-0032/html
Scroll to top button