1 Introductions
Motivated with the repeated appearance of the expression
$$ x(x + \mathtt{k}) (x + 2\mathtt{k})\cdots \bigl(x + (n -1)\mathtt{k} \bigr) $$
in the combinatorics of creation and annihilation operators [
13,
14] and the perturbative computation of Feynman integrals (see [
12]), a generalization of the well-known Pochhammer symbols is given in [
15] as
$$ (x)_{n,\mathtt{k}}:=x(x + \mathtt{k}) (x + 2\mathtt{k})\cdots \bigl(x + (n -1) \mathtt{k} \bigr), $$
for all
\(\mathtt{k}>0\), calling it the Pochhammer k-symbol. Closely associated functions that have relation with the Pochhammer symbols are the gamma and beta functions. Hence it is useful to recall some facts about the
k
-gamma and
k
-beta functions. The
k
-gamma function, denoted as
\(\Gamma_{\mathtt{k}}\), is studied in [
15] and defined by
$$\begin{aligned} \Gamma_{\mathtt{k}}(x):= \int_{0}^{\infty }t^{x-1}e^{-\frac{t^{ \mathtt{k}}}{\mathtt{k}}}\,dt \end{aligned}$$
(1.1)
for
\(\operatorname {Re}(x)>0\). Several properties of the
k
-gamma functions and applications in generalizing other related functions like
k
-beta and
k
-digamma functions can be found in [
15,
27,
28] and references therein.
The
k-digamma functions defined by
\(\Psi_{\mathtt{k}}:= \Gamma_{\mathtt{k}}'/\Gamma_{\mathtt{k}} \) are studied in [
28]. These functions have the series representation
$$\begin{aligned} \Psi_{\mathtt{k}}(t):=\frac{\log (\mathtt{k})-\gamma_{1}}{\mathtt{k}}- \frac{1}{t} +\sum _{n=1}^{\infty }\frac{t}{n\mathtt{k}(n\mathtt{k}+t)}, \end{aligned}$$
(1.2)
where
\(\gamma_{1}\) is the Euler–Mascheroni constant.
A calculation yields
$$\begin{aligned} \Psi_{\mathtt{k}}'(t)=\sum _{n=0}^{\infty } \frac{1}{(n\mathtt{k}+t)^{2}}, \quad \mathtt{k}>0 \mbox{ and } t>0. \end{aligned}$$
(1.3)
Clearly,
\(\Psi_{\mathtt{k}}\) is increasing on
\((0, \infty )\).
The Bessel function of order
p given by
$$ \mathtt{J}_{p}(x):= \sum_{k=0}^{\infty } \frac{(-1)^{k}}{\Gamma { ( k +p+1 ) } \Gamma { ( k+1 ) }} \biggl( \frac{x}{2} \biggr) ^{2k+p} $$
(1.4)
is a particular solution of the Bessel differential equation
$$\begin{aligned} x^{2} y''(x)+ x y'(x)+ \bigl(x^{2}-p^{2} \bigr)y(x)= 0. \end{aligned}$$
(1.5)
Here Γ denotes the gamma function. A solution of the modified Bessel equation
$$\begin{aligned} x^{2} y''(x)+ x y'(x)- \bigl(x^{2}+{\nu }^{2} \bigr)y(x)= 0, \end{aligned}$$
(1.6)
is the modified Bessel function
$$ \mathtt{I}_{\nu }(x):= \sum_{k=0}^{\infty } \frac{1}{\Gamma { ( k +\nu +1 ) } \Gamma { ( k+1 ) }} \biggl( \frac{x}{2} \biggr) ^{2k+\nu }. $$
(1.7)
The Bessel function has several generalizations (see, e.g., [
9,
10]) and is notably investigated in [
1,
17]. In [
1], a generalized Bessel function is defined in the complex plane, and sufficient conditions for it to be univalent, starlike, close-to-convex, or convex are obtained. This generalization is given by the power series
$$\begin{aligned} \mathcal{W}_{p, b, c}(z)= \sum_{k=0}^{\infty } \frac{(-c)^{k} ( \frac{z}{2} ) ^{2k+p+1}}{\Gamma ( k+1 ) \Gamma ( k+p+ \frac{b+2}{2} ) },\quad p, b, c \in \mathbb{C}. \end{aligned}$$
(1.8)
In this paper, we consider the function defined by the series
$$\begin{aligned} \mathtt{W}_{\nu , c}^{\mathtt{k}} (x) :=\sum _{r=0}^{\infty }\frac{(-c)^{r} }{ \Gamma_{\mathtt{k}}(r \mathtt{k}+ \nu + \mathtt{k}) r!} \biggl( \frac{x}{2} \biggr) ^{2r+\frac{ \nu }{\mathtt{k}}}, \end{aligned}$$
(1.9)
where
\(\mathtt{k}>0\),
\(\nu >-1\), and
\(c \in \mathbb{R}\). As
\(\mathtt{k}\to 1\), the
k
-Bessel function
\(\mathtt{W}_{ \nu , 1}^{1}\) is reduced to the classical Bessel function
\(J_{\nu }\), whereas
\(\mathtt{W}_{\nu , -1}^{1}\) coincides with the modified Bessel function
\(I_{\nu }\). Thus, we call the function
\(\mathtt{W}_{\nu , c} ^{\mathtt{k}}\) the generalized
k
-Bessel function. Basic properties of the
k
-Bessel and related functions can be found in recent works [
8,
19‐
21].
Turán [
30] proved that the Legendre polynomials
\(P_{n}(x)\) satisfy the determinantal inequality
$$\begin{aligned} \biggl\vert \textstyle\begin{array}{@{}c@{\quad }c@{}} P_{n}(x) & P_{n+1}(x) \\ P_{n+1}(x) & P_{n+2}(x) \end{array}\displaystyle \biggr\vert \leq 0, \quad -1 \le x \le 1, \end{aligned}$$
(1.10)
where
\(n = 0, 1, 2, \ldots \) , and the equality occurs only for
\(x = \pm 1\). The inequalities similar to (
1.10) can be found in the literature [
2,
3,
5,
11,
16,
25] for several other functions, for example, ultraspherical polynomials, Laguerre and Hermite polynomials, Bessel functions of the first kind, modified Bessel functions, and the polygamma function. Karlin and Szegö [
24] named determinants in (
1.10) as Turánians. More details about Turánians can be found in [
5,
11,
18,
22,
23,
29].
The aim of this paper is to investigate the influence of the
\(\Gamma_{\mathtt{k}}\) functions on the properties of the
k
-Bessel function defined in (
1.9). It is shown that the properties of the classical Bessel functions can be extended to the
k
-Bessel functions. Moreover, we investigate the effects of
\(\Gamma_{\mathtt{k}}\) instead of Γ on the monotonicity and log-convexity properties and related inequalities of the k-Bessel functions. The outcomes of our investigation are presented as follows.
In Section
2, we derive representation formulae and some recurrence relations for
\(\mathtt{W}_{\nu , c}^{\mathtt{k}}\). More importantly, the function
\(\mathtt{W}_{\nu , c}^{\mathtt{k}}\) is shown to be a solution of a certain differential equation of second order, which contains (
1.5) and (
1.6) for the particular case
\(\mathtt{k}=1\) and for particular values of
c. At the end of Section
2, we give two types of integral representations for
\(\mathtt{W}_{\nu , c}^{\mathtt{k}}\).
Section
3 is devoted to the investigation of monotonicity and log-convexity properties of the functions
\(\mathtt{W}_{\nu , c} ^{\mathtt{k}}\) and to relation between two
k
-Bessel functions of different order. As a consequence, we deduce Turán-type inequalities.
In Section
4, we give concluding remarks and list two tables for the zeroes of
\(\mathtt{W}^{\mathtt{k}}_{\nu , c}\), leading to an open problem for future studies.
3 Monotonicity and log-convexity properties
This section is devoted to discuss the monotonicity and log-convexity properties of the modified k
-Bessel function \(\mathtt{W} _{\nu , -1}^{\mathtt{k}}=\mathtt{I}_{\nu }^{\mathtt{k}}\). As consequences of those results, we derive several functional inequalities for \(\mathtt{I}_{\nu }^{\mathtt{k}}\).
The following result of Biernacki and Krzyż [
7] will be required.
The lemma still holds when both f and g are even or both are odd functions.
We now state and prove our main results in this section. Consider the functions
$$\begin{aligned} \mathcal{I}_{\nu }^{\mathtt{k}}(x):= \biggl( \frac{2}{x} \biggr) ^{\frac{ \nu }{\mathtt{k}}}\Gamma_{\mathtt{k}} (\nu +\mathtt{k}) \mathtt{I} _{\nu }^{\mathtt{k}}(x)=\sum_{r=0}^{\infty }f_{r}( \nu ) x^{2r}, \end{aligned}$$
(3.1)
where
$$\begin{aligned} \begin{aligned} & \mathtt{I}_{\nu }^{\mathtt{k}}(x)= \mathtt{W}_{\nu , -1}^{\mathtt{k}} (x) =\sum_{r=0}^{\infty }\frac{1 }{ \Gamma_{\mathtt{k}}(r \mathtt{k}+ \nu + \mathtt{k}) r!} \biggl( \frac{x}{2} \biggr) ^{2r+\frac{\nu }{ \mathtt{k}}} \quad \mbox{and} \\ &f_{r}(\nu )= \frac{\Gamma_{ \mathtt{k}}{(\nu +\mathtt{k})}}{\Gamma_{\mathtt{k}}{(r{\mathtt{k}}+ \nu +\mathtt{k})}4^{r} r!}. \end{aligned} \end{aligned}$$
(3.2)
Then we have the following properties.
Our final result is based on the Chebyshev integral inequality [
26, p. 40], which states the following: suppose
f and
g are two integrable functions and monotonic in the same sense (either both decreasing or both increasing). Let
\(q: (a, b) \to \mathbb{R}\) be a positive integrable function. Then
$$\begin{aligned} \biggl( \int_{a}^{b} q(t) f(t)\,dt \biggr) \biggl( \int_{a}^{b} q(t) g(t)\,dt \biggr) \leq \biggl( \int_{a}^{b} q(t)\,dt \biggr) \biggl( \int_{a}^{b} q(t) f(t) g(t)\,dt \biggr) . \end{aligned}$$
(3.6)
Inequality (
3.6) is reversed if
f and
g are monotonic in the opposite sense.
The following function is required:
$$\begin{aligned} \mathcal{J}_{\nu }^{\mathtt{k}}(x):= \biggl( \frac{2}{x} \biggr) ^{\frac{ \nu }{\mathtt{k}}}\Gamma_{\mathtt{k}} (\nu +\mathtt{k}) \mathtt{J} _{\nu }^{\mathtt{k}}(x)=\sum_{r=0}^{\infty }g_{r}( \nu ) x^{2r}, \end{aligned}$$
(3.7)
where
$$\begin{aligned} \begin{aligned} &\mathtt{J}_{\nu }^{\mathtt{k}}(x)= \mathtt{W}_{\nu , 1}^{\mathtt{k}} (x) =\sum_{r=0}^{\infty }\frac{(-1)^{r} }{ \Gamma_{\mathtt{k}}(r \mathtt{k}+ \nu + \mathtt{k}) r!} \biggl( \frac{x}{2} \biggr) ^{2r+\frac{ \nu }{\mathtt{k}}} \quad \text{and} \\ &g_{r}(\nu )= \frac{(-1)^{r} \Gamma_{\mathtt{k}}{(\nu +\mathtt{k})}}{\Gamma_{\mathtt{k}}{(r{\mathtt{k}}+ \nu +\mathtt{k})}4^{r} r!}. \end{aligned} \end{aligned}$$
(3.8)
4 Conclusion
It is shown that the generalized k-Bessel functions \(W^{k}_{\nu ,c}\) are solutions of a second-order differential equation, which for \(k=1\) is reduced to the well-known second-order Bessel differential equation. It is also proved that the generalized modified k-Bessel function \(\mathcal{I}_{\nu }^{\mathtt{k}}\) is decreasing and log-convex on \((-\mathtt{k}, \infty )\) for each fixed \(x >0\). Several other inequalities, especially the Turán-type inequality and reverse Turán-type inequality for \(\mathcal{I}_{\nu }^{\mathtt{k}}\) are established.
Furthermore, we investigate the pattern for zeroes of \(\mathcal{W} _{\nu }^{\mathtt{k}, 1}\) in two ways: (i) with respect to fixed k
and variation of ν and (ii) with respect to fixed ν and variation of k
.
From the data in Table
1 and Table
2, we can observe that the zeroes of
\(\mathtt{W}_{ \nu , 1}^{\mathtt{k}}\) are increasing in in both cases. However, we have no any analytical proof for this monotonicity of the zeroes of
\(W^{k}_{ \nu ,1}\). As there are several works on the zeroes of the classical Bessel functions, the zeroes of
\(\mathtt{W}_{ \nu , 1}^{\mathtt{k}}\) would be an interesting topic for future investigations. The monotonicity of the zeroes of
\(\mathtt{W}_{ \nu , c}^{\mathtt{k}}\) with respect to
c and fixed
k
,
ν will be another open problem for further studies.
Table 1
Positive zeroes of \(\mathtt{W}_{ \nu , 1}^{\mathtt{k}}\) for fixed ν and different k
ν = −0.4 and c = 1 |
1st zero | 0.662422 | 1.75098 | 2.42334 | 2.95334 | 3.40423 |
2nd zero | 2.96686 | 4.87852 | 6.24148 | 7.3588 | 8.32849 |
3rd zero | 5.2018 | 8.01663 | 10.0812 | 11.7913 | 13.2836 |
ν = 0.5 and c = 1 |
1st zero | 2.70943 | 3.14159 | 3.55493 | 3.93277 | 4.28026 |
2nd zero | 4.96077 | 6.28319 | 7.38858 | 8.35255 | 9.21757 |
3rd zero | 7.19373 | 9.42478 | 11.2315 | 12.7879 | 14.1752 |
Table 2
Positive zeroes of \(\mathtt{W}_{ \nu , 1}^{\mathtt{k}}\) for different ν and k
k = 0.5 and c = 1
|
1st zero | 0.662422 | 0.97534 | 1.70047 | 2.70943 | 3.63143 | 4.51146 | 5.36577 | 6.20238 |
2nd zero | 2.96686 | 3.21271 | 3.90328 | 4.96077 | 5.95189 | 6.90209 | 7.82393 | 8.72471 |
3rd zero | 5.2018 | 5.43751 | 6.11911 | 7.19373 | 8.21647 | 9.20314 | 10.1629 | 11.1017 |
k = 1 and c = 1
|
1st zero | 1.75098 | 1.92285 | 2.40483 | 3.14159 | 3.83171 | 4.49341 | 5.13562 | 5.76346 |
2nd zero | 4.87852 | 5.04213 | 5.52008 | 6.28319 | 7.01559 | 7.72525 | 8.41724 | 9.09501 |
3rd zero | 8.01663 | 8.17785 | 8.65373 | 9.42478 | 10.1735 | 10.9041 | 11.6198 | 12.3229 |
k = 1.5 and c = 1
|
1st zero | 2.42334 | 2.55767 | 2.9453 | 3.55493 | 4.13426 | 4.69286 | 5.2362 | 5.76774 |
2nd zero | 6.24148 | 6.37291 | 6.76069 | 7.38858 | 7.9979 | 8.5923 | 9.1744 | 9.74613 |
3rd zero | 10.0812 | 10.2116 | 10.5986 | 11.2315 | 11.8513 | 12.4599 | 13.0587 | 13.6488 |
k = 2 and c = 1
|
1st zero | 2.95334 | 3.06754 | 3.40094 | 3.93277 | 4.44288 | 4.93703 | 5.41885 | 5.8908 |
2nd zero | 7.3588 | 7.47176 | 7.80657 | 8.35255 | 8.88577 | 9.40825 | 9.92154 | 10.4269 |
3rd zero | 11.7913 | 11.9037 | 12.2382 | 12.7879 | 13.3286 | 13.8616 | 14.3875 | 14.907 |
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.