Next Article in Journal
Symmetric Toeplitz Matrices for a New Family of Prestarlike Functions
Next Article in Special Issue
Closed-Form Derivations of Infinite Sums and Products Involving Trigonometric Functions
Previous Article in Journal
Distance Antimagic Product Graphs
Previous Article in Special Issue
Diagonals of Rational Functions: From Differential Algebra to Effective Algebraic Geometry
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

An Asymptotic Expansion for the Generalized Gamma Function

1
Mathematics Department, Faculty of Science, King Abdulaziz University, P.O. Box 80203, Jeddah 21589, Saudi Arabia
2
Mathematics Department, Faculty of Science, University of Jeddah, P.O. Box 80327, Jeddah 21589, Saudi Arabia
3
Mathematics Department, Faculty of Science, Mansoura University, Mansoura 35516, Egypt
*
Author to whom correspondence should be addressed.
Symmetry 2022, 14(7), 1412; https://doi.org/10.3390/sym14071412
Submission received: 11 June 2022 / Revised: 3 July 2022 / Accepted: 7 July 2022 / Published: 9 July 2022

Abstract

:
The symmetric patterns that inequalities contain are reflected in researchers’ studies in many mathematical sciences. In this paper, we prove an asymptotic expansion for the generalized gamma function Γ μ ( v ) and study the completely monotonic (CM) property of a function involving Γ μ ( v ) and the generalized digamma function ψ μ ( v ) . As a consequence, we establish some bounds for Γ μ ( v ) , ψ μ ( v ) and polygamma functions ψ μ ( r ) ( v ) , r 1 .

1. Introduction

Stirling’s formula is given by
Γ ( r + 1 ) 2 π r r e r , r
where Γ is the classical Gamma function [1]. An elementary and complete proof of this formula is available at [2]. Moreover, many mathematicians have used the logarithm of gamma function to deduce several useful properties of the gamma function, and their powerful tool for such investigations was the digamma function
ψ ( v ) = d d v ln Γ ( v ) = r = 0 1 v + r 1 r + 1 γ , v > 0
where γ = lim r k = 1 r 1 k log r 0.5772156649 is Euler–Mascheroni’s constant. For more details on bounds of the functions Γ ( v ) and d r d v r ψ ( v ) , please refer to [3,4,5,6,7] and the references therein. Many of such bounds deduced from the monotonicity properties of some functions involving Γ or ψ . An infinitely differentiable real valued function M defined on v > 0 is said to be CM if ( 1 ) r M ( r ) ( v ) 0 for all r 0 on v > 0 . For more details about CM functions and their applications, we refer to [8,9,10,11]. According to Bernstein theorem [12], function M is CM if and only if M ( v ) = 0 e v u d ν ( u ) , where ν ( u ) is a non-negative measure on u 0 such that the integral converges for v > 0 .
In 2007, Alzer and Batir [13] studied the completely monotonicity of the function
S ρ ( v ) = ln Γ ( v ) + 1 2 ψ ( v + ρ ) + v 1 2 ln ( 2 π ) v ln v , ρ 0 ; v > 0
and deduced the following double inequality:
2 π exp 1 2 ψ ( a + v ) v < v v Γ ( v ) < 2 π exp 1 2 ψ ( b + v ) v , v > 0
with the constants a = 1 3 and b = 0 being the best possible constants. In 2008, Batir [14] modified S ρ ( v ) and deduced some bounds for Γ ( v ) in terms of digamma and polygamma functions.
Euler [15] originally defined gamma function as Γ ( v ) = lim μ Γ μ ( v ) , where
Γ μ ( v ) = μ v μ ! ( μ + v ) ( v + 2 ) ( v + 1 ) v , v > 0 , μ = 1 , 2 ,
which satisfies the following recurrence relation.
Γ μ ( r + v ) = v μ r Γ μ ( v ) v + n + μ s = 1 r 1 s + v s + μ + v , v > 0 , r N .
In 2010, Krasniqi and Shabani [16] presented the strictly CM property of function ψ μ on ( 0 , ) , where the following is the case.
ψ μ ( v ) = ln μ i = 0 μ 1 v + i .
Krasniqi and Merovci [17] introduced the following integral representations for ψ μ and its derivatives:
ψ μ ( v ) = ln μ + 0 e ( μ + 1 ) u 1 1 e u e v u d u , v > 0 , μ N
and
ψ μ ( r ) ( v ) = ( 1 ) r 0 e ( μ + 1 ) u 1 1 e u u r e v u d u , v > 0 , μ , r N .
Nantomah, Prempeh and Twum [18] presented the following generalization of the functions Γ and ψ :
Γ μ , k ( v ) = k μ + 1 ( 1 + μ ) ! ( μ k ) v k 1 v ( v + k ) ( v + 2 k ) ( v + μ k ) = 0 μ u v u 1 u k μ k μ d u , v , k > 0 , μ N
and
ψ μ , k ( v ) = 1 k ln ( k μ ) i = 0 μ 1 v + i k = 1 k ln ( k μ ) + 0 e k ( μ + 1 ) u 1 1 e k u e v u d u .
Nantomah, Merovci and Nasiru [19] presented bounds
1 ( k + v + μ k ) v 1 ψ μ , k ( v ) k 1 ln v k μ v + k + μ k 0 , μ N , v , k > 0
and
1 ( k + μ k + v ) 2 1 v 2 ψ μ , k ( v ) + k 1 v 1 1 ( v + k + μ k ) 0 , μ N , v , k > 0 .
Recently, several inequalities involving the generalized gamma function have been presented [17,18,19].
Research involving Gamma function is conducted by many authors currently. Results concerning extensions of Gamma function involving Mittag–Leffler function are presented in [20]. Other extensions of Gamma function are investigated in [21] using the two-parameter Mittag–Leffler matrix function and some important properties of these extended matrix functions are proved. A new series representation of the extended k gamma function is provided in [22] and particular cases involving the original gamma function are discussed as corollaries.
In the following, we will present the asymptotic expansion Γ μ ( r + 1 ) μ r + 1 r r + 1 2 e μ + 1 μ ! ( r + μ + 1 ) r + μ + 3 2 for large values of r and we will discuss the property of completely monotonicity of the function
S α , μ ( v ) = ln Γ μ ( v ) + 1 / 2 ψ μ ( α + v ) v ln μ v v + μ + 1 + ( μ + 1 ) ln ( v + μ + 1 ) ln μ ! μ e μ + 1 ,
for μ N , v > 0 and different values of α 0 . As a consequence, we establish some bounds for Γ μ ( v ) , ψ μ ( v ) and ψ μ ( r ) ( v ) , r 1 .

2. An Asymptotic Expansion for Γ μ ( v )

Theorem 1.
For all μ N ,
Γ μ ( r + 1 ) μ r + 1 r r + 1 2 e 1 + μ μ ! ( r + μ + 1 ) 3 2 + μ + r , r .
Proof. 
For μ = 1 , we have
lim r Γ 1 ( 1 + r ) r r + 1 2 e 2 ( r + 2 ) r + 5 2 = 1 e 2 lim r r ! ( r + 2 ) ! ( r + 2 ) 2 1 + 2 r 1 1 2 + r = 1 .
Similarly, for μ = 2 , we have lim r Γ 2 ( r + 1 ) 2 r + 2 r r + 1 2 e 3 ( r + 3 ) r + 7 2 = 1 . Now, for μ 3 , taking logarithm for both sides of (4) at v = 1 yields the following.
ln Γ μ ( r + 1 ) = ln μ 2 ( 1 + μ ) ( 2 + μ ) + s = 2 r ln μ s μ + 1 + s .
The function A μ ( v ) = μ v μ + v + 1 is strictly increasing on v > 0 and, hence, A μ ( v ) > A μ ( 2 ) 1 for all μ 3 . Using the relation between the integral and the Riemann sums, we have
2 r ln v μ μ + v + 1 d v < s = 2 r ln s μ μ + s + 1 < 2 r + 1 ln μ v μ + v + 1 d v
and hence, we have the following:
ln u μ ( r ) < ln Γ μ ( 1 + r ) < ln u μ ( 1 + r ) ,
where
u μ ( r ) = r μ 1 + r + μ r ( 3 + μ ) μ + 3 4 ( 1 + r + μ ) 1 + μ ( μ + 1 ) ( μ + 2 ) .
Sequence u μ ( r ) satisfies
u μ ( r + 1 ) = 1 r + μ + 1 + 1 ( r + 2 + μ ) r μ r + μ + 1 r 1 + 1 r r 1 + 1 u μ ( r )
and using inequality ( r 1 + 1 ) r < e < ( r 1 + 1 ) r + 1 for r N , we have the following.
u μ ( r + 1 ) < 2 μ r μ + r + 1 u μ ( r ) .
Substituting (13) into (12) yields
Γ μ ( 1 + r ) = u μ ( r ) θ μ ( r ) ,
where
1 < θ μ ( r ) < 2 μ r r + 1 + μ .
Let
a μ ( r ) = Γ μ ( r + 1 ) ( r μ ) 2 r + 1 2 μ + r + 1 μ r 3 2 , r , μ N .
Using (4) at v = 1 , we have
ln a μ ( r ) = ( r + 1 / 2 ) ln r + ln μ 1 + μ + s = 1 r ln s s + 1 + μ + 3 / 2 + μ + r ln ( μ + 1 + r )
and then
ln a μ ( r ) ln a μ ( 1 + r ) = H ( r ) H ( 1 + μ + r ) ,
where H ( v ) = ( v + 1 2 ) ln 1 + 1 v , v > 0 is decreasing function on v > 0 and consequently ( ln a μ ( r ) ) r N is decreasing. Using (14), we obtain the following.
ln a μ ( 1 ) ln a μ ( r ) = m = 1 r 1 [ H ( m ) H ( m + μ + 1 ) ] .
Using well-known series ln 1 + y 1 y = 2 y + y 3 3 + y 5 5 + , | y | < 1 and letting y = 1 2 v + 1 , we obtain the following.
H ( v ) 1 = τ = 1 1 1 + 2 τ 1 2 v + 1 2 τ < τ = 1 1 3 1 2 v + 1 2 τ = 1 12 v ( 1 + v ) .
As H ( v ) is decreasing, we have the following.
H ( v ) > lim v H ( v ) = 1 .
Substituting (16) and (17) into (15) produces
ln a μ ( 1 ) ln a μ ( r ) < 1 12 τ = 1 r 1 1 τ 1 ( τ + 1 ) < 1 12
and then ( ln a μ ( r ) ) r N is bounded from below by ( ln a μ ( 1 ) 1 12 ) . Hence, ( ln a μ ( r ) ) r N is convergent to some constant and depends on μ . Then, the following is obtained.
Γ μ ( 1 + r ) C μ ( r μ ) r + 1 2 ( μ + r + 1 ) μ + r + 3 / 2 .
Using Stirling’s Formula (1), we obtain lim r 2 4 r ( r ! ) 4 π r [ ( 2 r ) ! ] 2 = 1 and lim r π ( r + μ ) [ ( 2 r + 2 μ ) ! ] 2 2 4 ( r + μ ) [ ( r + μ ) ! ] 4 = 1 .
Multiplying these two limits produces the following.
lim r r ! ( μ + r ) ! 4 ( 2 μ + 2 r ) ! ( 2 r ) ! 2 = 2 4 μ .
Letting v = 1 in (4) and using relation ( 2 r + 2 μ ) ! = ( 2 r + μ + 1 ) ! s = 2 + μ 2 μ ( 2 r + s ) , we have
Γ μ ( 2 r + 1 ) = ( 2 r ) ! μ 1 + 2 r μ ! s = μ + 2 2 μ ( s + 2 r ) ( 2 r + 2 μ ) ! .
Now, using (4) at v = 1 and inserting (20) into (19) yields
lim r Γ μ ( r + 1 ) 4 Γ μ ( 2 r + 1 ) 2 ( r + μ + 1 ) 4 s = μ + 2 2 μ ( 2 r + s ) 2 = 2 4 μ μ 2 ( μ ! ) 2
and using (18), we obtain the following:
C μ 2 lim r 2 r + μ + 1 2 r + 2 μ + 2 4 r s = μ + 2 2 μ s + 2 r r + 1 + μ 2 = ( μ ! ) 2 2 2 μ 2 μ
which leads to
C μ = μ μ ! e μ + 1 , μ 3 ,
which completes the proof. □
In the next part, we provide a double inequality involving Γ μ ( v + 1 ) .

Some Bounds for the Function Γ μ ( v )

Theorem 2.
Assume that μ N and v ( 0 , ) . Then, we have
exp λ ( μ + 1 + v ) + λ ( v ) < Γ μ ( 1 + v ) μ 1 + v v 1 2 + v e μ + 1 μ ! ( μ + v + 1 ) μ + v + 3 / 2 < exp β ( μ + v + 1 ) + β ( v ) ,
where β ( v ) = 1 12 v and λ ( v ) = β ( v ) 1 360 v 3 1 120 v 4 .
Proof. 
Let
W μ ( v ) = H ( v + μ + 1 ) + H ( v ) + Δ β ( v + μ + 1 ) β ( v ) ,
where Δ g ( v ) = g ( v ) g ( v + 1 ) . Then W μ ( v ) = ( 1 + μ ) ( 2 + μ + 2 v ) D μ ( v ) 6 v 3 ( 1 + v ) 3 ( 1 + μ + v ) 3 ( 2 + μ + v ) 3 < 0 , where
D μ ( v ) = 4 + 12 μ + 13 μ 2 + 6 μ 3 + μ 4 + ( 14 + 29 μ + 19 μ 2 + 4 μ 3 ) v + ( 19 + 23 μ + 7 μ 2 ) v 2 + ( 12 + 6 μ ) v 3 + 3 v 4 .
Hence, W μ ( v ) is decreasing on v > 0 with lim v W μ ( v ) = 0 ; thus, W μ ( v ) > 0 for all v > 0 .
Hence, W μ ( v ) is increasing on v > 0 with lim v W μ ( v ) = 0 . Then, we have the following.
H ( v ) H ( v + μ + 1 ) < Δ β ( v ) β ( v + μ + 1 ) .
Similarly, we obtain the following.
Δ λ ( v ) λ ( v + μ + 1 ) < H ( v ) H ( v + μ + 1 ) .
Combining (23) with (24) provides the following:
Δ λ ( v ) λ ( v + μ + 1 ) < H ( v ) H ( v + μ + 1 ) < Δ β ( v ) β ( v + μ + 1 )
and using (14), we obtain the following.
exp Δ λ ( v ) λ ( v + μ + 1 ) < a μ ( v ) a μ ( v + 1 ) < exp Δ β ( v ) β ( v + μ + 1 ) .
It follows that
a μ ( v + 1 ) exp λ ( v + 2 + μ ) λ ( 1 + v ) < a μ ( v ) exp λ ( v + μ + 1 ) λ ( v )
and consequently, function
T μ ( v ) = a μ ( v ) exp λ ( v + μ + 1 ) λ ( v )
is strictly decreasing on v > 0 with lim v T μ ( v ) = C μ . Thus, we have the following.
a μ ( v ) > C μ exp λ ( v ) λ ( 1 + v + μ ) .
In a similar way, function
L μ ( v ) = a μ ( v ) exp β ( v + μ + 1 ) β ( v )
is strictly increasing on v > 0 with lim v L μ ( v ) = C μ . Then,
a μ ( v ) < C μ exp β ( v ) β ( μ + v + 1 ) .
Combining (26) with (27), we obtain
exp λ ( v ) λ ( v + μ + 1 ) < a μ ( v ) C μ < exp β ( v ) β ( v + μ + 1 ) ,
which completes the proof. □
Corollary 1.
Γ μ ( v ) μ v v v 1 2 e μ + 1 μ ! ( v + 1 + μ ) 1 / 2 + v + μ exp λ ( v ) λ ( v + μ + 1 ) , v
where λ ( v ) = 1 12 v 1 360 v 3 1 120 v 4 .
Proof. 
Using inequality (22) and relation (4) at r = 1 , we obtain the following.
1 < Γ μ ( v ) μ v v v 1 2 e 1 + μ μ ! ( 1 + v + μ ) v + μ + 1 2 exp λ ( v ) λ ( 1 + v + μ )
< exp 1 360 v 3 + 1 120 v 4 1 360 ( 1 + v + μ ) 3 1 120 ( 1 + v + μ ) 4 .
Hence, we have
lim v Γ μ ( v ) μ v v v 1 2 e 1 + μ μ ! ( v + 1 + μ ) 1 / 2 + μ + v exp λ ( v ) λ ( v + 1 + μ ) = 1 .
This completes the proof. □
Corollary 2.
Let μ and s be positive integers. Then,
ln Γ μ ( v ) ( μ + 1 ) ln v + μ + 1 + 2 v 1 2 ln μ v v + 1 + μ + ln μ μ ! e 1 + μ + λ ( v ) λ ( v + 1 + μ ) ,
ψ μ ( v ) 1 2 1 v 1 v + 1 + μ + ln μ v v + 1 + μ + λ ( v ) λ ( v + 1 + μ ) , v
and
ψ μ ( s ) ( v ) ( 1 ) s 1 ( s 1 ) ! 1 v s 1 ( μ + 1 + v ) s ( 1 ) s s ! 2 1 v s + 1 1 ( μ + 1 + v ) s + 1
+ λ ( s + 1 ) ( v ) λ ( s + 1 ) ( μ + 1 + v ) , v
where
λ ( s ) ( v ) = ( 1 ) s s ! 12 v 1 + s ( 2 + s ) ! 720 v 3 + s ( 3 + s ) ! 720 v 4 + s s N .
In the next section, we will generalize some results presented by Alzer and Batir [13].

3. Study of a CM Function Involving Γ μ and ψ μ Functions

Theorem 3.
Suppose that μ N and v > 0 . Then, the function
S α , μ ( v ) = ln Γ μ ( v ) + 1 / 2 ψ μ ( α + v ) v ln μ v v + μ + 1 + ( μ + 1 ) ln ( v + μ + 1 ) ln μ μ ! e 1 + μ , α 0
is CM on v > 0 if and only if α 1 3 . Moreover, S α , μ ( v ) is CM on v > 0 if and only if α = 0 .
Proof. 
From (6), (7) and identity ln ( h d ) = 0 e d t e h t t d t for h , d > 0 (see [1]), we have
S α , μ ( v ) = ψ μ ( v ) ln μ v v + 1 + μ + 1 2 ψ μ ( α + v ) = 0 e ( μ + v + 1 ) u u ( e u 1 ) φ ( u ) d u ,
where
φ ( u ) = e ( 2 + μ ) u + 1 e ( 1 + μ ) u e u u e ( 2 + μ ) u e u 1 2 u 2 e u e ( 2 + μ ) u e α u .
Let α 1 3 , then we obtain
φ ( u ) e ( 2 + μ ) u e ( 1 + μ ) u e u + 1 u e ( 2 + μ ) u e u + 1 2 u 2 e ( μ + 5 3 ) u e 2 3 u
and hence,
φ ( u ) r = 3 f μ ( r ) ( r + 2 ) ! u r + 2 < 0 ,
where
f μ ( r ) = ( μ + 1 ) 2 + r 1 + ( μ + 2 ) 2 + r + 1 ( μ + 2 ) 1 + r ( 2 + r ) + μ + 5 / 3 r 2 / 3 r ( 1 + r ) ( 2 + r ) 2 = s = 1 r ( r + 2 ) s 1 + r + s 2 + r + s r 2 / 3 r s ( 2 + r ) ( 1 + r ) 2 ( 1 + μ ) s = s = 1 r 3 r s ( 2 + r ) ( 1 + r ) ( 2 + r s ) s r 3 r s + 2 r s + 2 r s 1 ( r s ) ( μ + 1 ) s = s = 1 r 3 r s ( 1 + r ) ( 2 + r ) ( 2 + r s ) s r l = 2 r s l r s 2 r s l ( 1 + μ ) s .
Consequently, S α , μ ( v ) is CM on ( 0 , ) for α 1 3 . Thus, S α , μ ( v ) is decreasing and using asymptotic (29) and (30), we have lim v S α , μ ( v ) = 0 and then S α , μ ( v ) > 0 . Then, S α , μ ( v ) is a CM function on v > 0 for α 1 3 . Conversely, if function S α , μ ( v ) is CM, then we obtain for v > 0 , μ N that
v 2 ( 1 + μ ) S α , μ ( v ) = v 2 ( 1 + μ ) ln Γ μ ( v ) v v 1 2 μ v e 1 + μ μ ! ( v + 1 + μ ) 1 / 2 + μ + v 1 2 ln μ v v + 1 + μ + 1 2 ψ μ ( α + v ) > 0 .
From (29), we have
lim v v 2 ( 1 + μ ) ln Γ μ ( v ) μ v v v 1 2 e 1 + μ μ ! ( μ + v + 1 ) μ + v + 1 2 = 1 12 .
Using asymptotic (30), we have lim v v 2 ( 1 + μ ) ln v μ μ + 1 + v ψ μ ( v ) = 1 2 and using (5), we obtain lim v v 2 ( 1 + μ ) ψ μ ( v ) ψ μ ( α + v ) = α .
Hence, we conclude that
lim v v 2 ( μ + 1 ) ln μ v v + μ + 1 ψ μ ( v + α ) = 1 2 α .
From (32), we conclude that 1 12 1 2 ( 1 2 α ) 0 and then α 1 3 . Now, for α = 0 , we obtain the following.
S 0 , μ ( v ) = 0 e ( v + 1 + μ ) u u ( e u 1 ) r = 2 s = 1 r ( r + 2 ) ( r + 1 ) ( r s ) 2 ( 2 + r s ) s r ( μ + 1 ) s u r + 2 ( r + 2 ) ! d u
Therefore, S 0 , μ ( v ) is CM function on u > 0 . Thus, S 0 , μ ( v ) is an increasing function on v > 0 with lim v S 0 , μ ( v ) = 0 and hence, S 0 , μ ( v ) < 0 . Then, S 0 , μ ( v ) is CM on v > 0 . Conversely, if we assume that S α , μ ( v ) is CM on v > 0 with α > 0 , then S α , μ ( v ) < 0 on v > 0 . However, this contradicts lim v 0 S α , μ ( v ) = ; hence, α = 0 . □
Corollary 3.
Let μ N . Then
lim v 0 v 1 + r ψ μ ( r ) ( v ) = ( 1 ) 1 + r r ! , r = 0 , 1 , 2 ,
and
lim v 0 v r ψ μ ( r ) ( b + v ) = 0 , r N , b > 0 .

Some Sharp Bounds for Γ μ and ψ μ ( r ) Functions

Now, we will present some sharp bounds of Γ μ and ψ μ ( r ) depending on Theorem (3).
Corollary 4.
Let two real numbers a , b 0 . For μ N and v > 0 , we have
v v μ 1 2 + v μ ! e 1 + μ ( v + 1 + μ ) v + 1 + μ exp 1 2 ψ μ ( a + v ) < Γ μ ( v ) < v v μ 1 2 + v μ ! e 1 + μ ( v + 1 + μ ) v + 1 + μ exp 1 2 ψ μ ( b + v )
with the constants a = 1 3 and b = 0 are best possible.
Proof. 
In inequality (36), the left-hand side is equivalent v 2 ( μ + 1 ) S a , μ ( v ) > 0 , which leads to a 1 3 as stated in the proof of Theorem (3). Using the increasing property of the function ψ μ ( v ) on v > 0 , we have e 1 2 ψ μ ( v + a ) e 1 2 ψ μ ( v + 1 3 ) for a 1 3 . Then, a = 1 3 is the best possible constant in (36). Moreover, Theorem (3) proves the right-hand side of the inequality (36) at b = 0 . If there exist b > 0 such that the upper bound of Γ μ ( v ) in (36) is valid for v ( 0 , ) , then we would have
lim v 0 Γ μ ( v ) μ e μ + 1 μ ! ( μ + 1 ) μ + 1 exp 1 2 ψ μ ( b ) lim v 0 v v + μ + 1 v
and hence,
lim v 0 Γ μ ( v ) μ μ ! e μ + 1 ( μ + 1 ) μ + 1 exp 1 2 ψ μ ( b )
which contradicts with lim v 0 Γ μ ( v ) = . Then, b = 0 in (36) is the best possible constant. □
Remark 1.
If we let μ in (36), then we obtain (3).
Corollary 5.
Assume that a , b [ 0 , ) and μ N . Then, for all v ( 0 , ) , we have
1 2 ψ μ ( a + v ) < ψ μ ( v ) + ln v μ v + 1 + μ < 1 2 ψ μ ( b + v ) ,
with constants a = 1 3 and b = 0 being the best possible.
Proof. 
In inequality (37), the left-hand side is equivalent
v 3 ( 1 + μ ) S a , μ ( v ) = v 3 ( 1 + μ ) σ μ ( v ) + v 3 2 ( 1 + μ ) ψ μ ( a + v ) ( 1 + μ ) v ( v + 1 + μ ) < 0 ,
where
σ μ ( v ) = ψ μ ( v ) ln μ v v + 1 + μ + ( 1 + μ ) 2 v ( v + 1 + μ ) .
Using asymptotic expansions (30) and (31), we have
lim v v 3 ( μ + 1 ) σ μ ( v ) = 1 6 and lim v v 3 ( 1 + μ ) ψ μ ( v ) ( 1 + μ ) ( v + 1 + μ ) v = 1 .
Using relation (5), we have
lim v v 3 ( 1 + μ ) ψ μ ( v + a ) ψ μ ( v ) = 2 a .
Hence, we conclude from inequality (38) that 1 6 + 1 2 2 a + 1 0 or a 1 3 .
Then, the left-hand side of the inequality (37) is satisfied only if a 1 3 . Using the decreasing property of ψ μ ( v ) on ( 0 , ) , we obtain a = 1 3 in (37), and it is the best possible constant. Moreover, Theorem (3) proves the right-hand side of the inequality (37) for b = 0 . If there exist b > 0 such that the upper bound of (37) valid for v > 0 , then we obtain
lim v 0 v ln μ v v + μ + 1 ψ μ ( v ) < 1 2 lim v 0 v ψ μ ( v + b )
which leads to
lim v 0 v ψ μ ( v ) < 1 2 lim v 0 v ψ μ ( v + b ) .
From (34) and (35), we have lim v 0 v ψ μ ( v ) = 1 and lim v 0 v ψ μ ( v + b ) = 0 , which contradict inequality (39). Hence, b = 0 is the best possible constant in (37). □
Remark 2.
Substituting k = 1 in (8), we obtain
0 ψ μ ( v ) + ln v μ v + 1 + μ .
Using the completely monotonicity property of ψ μ ( v ) and inequality (37), we have
0 < 1 2 ψ μ ( 1 / 3 + v ) < ψ μ ( v ) + ln v μ v + 1 + μ .
Then, the lower bound of (37) improves its counterpart in (8) at k = 1 for all v ( 0 , ) .
Corollary 6.
Suppose that a , b [ 0 , ) and μ N . Then, for all v ( 0 , ) and r = 2 , 3 , , we have
( 1 ) r + 1 2 ψ μ ( r ) ( a + v ) < ( r 2 ) ! ( 1 + v + μ ) r 1 ( r 2 ) ! v r 1 + ( 1 ) r ψ μ ( r 1 ) ( v ) < ( 1 ) 1 + r 2 ψ μ ( r ) ( b + v ) ,
with the constants a = 1 3 and b = 0 being the best possible.
Proof. 
In inequality (40), the left-hand side is equivalent
v r + 2 ( 1 + μ ) ( 1 ) r S a , μ ( r ) ( v ) = v 2 + r ( 1 + μ ) F μ ( v ) + v 2 + r 2 ( 1 + μ ) ( 1 ) r ψ μ ( r ) ( a + v ) + 1 v r 1 ( 1 + v + μ ) r ( r 1 ) ! > 0 ,
where
F μ ( v ) = ( 1 ) r ψ μ ( r 1 ) ( v ) ( r 1 ) ! 2 1 v r 1 ( v + 1 + μ ) r + ( r 2 ) ! ( v + 1 + μ ) r 1 ( r 2 ) ! v r 1 .
Using the asymptotic expansion (31), we have
lim v v r + 2 ( 1 + μ ) F μ ( v ) = ( 1 + r ) ! 12
and
lim v v r + 2 ( 1 + μ ) ( 1 ) r ψ μ ( r ) ( v ) + 1 v r 1 ( v + 1 + μ ) r ( r 1 ) ! = ( 1 + r ) ! 2 .
Using relation (5), we obtain
lim v ( 1 ) r v 2 + r ( 1 + μ ) ψ μ ( r ) ( a + v ) ψ μ ( r ) ( v ) = ( 1 + r ) ! a .
Then, we conclude from inequality (3) that 1 12 + 1 2 a 1 2 ( r + 1 ) ! 0 or a 1 3 . Since ψ μ ( v ) is strictly CM function on v > 0 , then for r = 0 , 1 , 2 , , function ( 1 ) r ψ μ ( r ) ( v ) is increasing on v > 0 ; hence, the best possible constant in (40) is a = 1 3 . Moreover, Theorem (3) proves the right-hand side of inequality (40) for b = 0 . If there exist b > 0 such that the upper bound of (40) is valid for v > 0 , then we obtain the following.
lim v 0 v r ( 1 ) r ψ μ ( r 1 ) ( v ) ( r 2 ) ! v r 1 ( r 2 ) ! ( v + 1 + μ ) r 1 < lim v 0 ( 1 ) r v r 2 ψ μ ( r ) ( b + v ) = 0 .
Using (34), we have
lim v 0 v r ( 1 ) r ψ μ ( r 1 ) ( v ) + ( r 2 ) ! ( v + 1 + μ ) r 1 ( r 2 ) ! v r 1 = ( r 1 ) ! , r = 2 , 3 ,
which contradicts with inequality (42); hence, b = 0 is the best possible constant. □
Remark 3.
Inserting k = 1 in (9), we obtain the following.
0 ψ μ ( v ) 1 v 1 ( 1 + v + μ ) .
Using the completely monotonicity property of ψ μ ( v ) and inequality (40), we have the following.
0 < 1 2 ψ μ ( 1 / 3 + v ) < ψ μ ( v ) 1 v 1 ( 1 + v + μ ) .
Then, for r = 2 , the lower bound of (40) improves its counterpart in (9) at k = 1 for all v ( 0 , ) .

4. Conclusions

The main conclusions of this paper are stated in Theorems 1–3. We deduced the following asymptotic expansions for the generalized gamma
Γ μ ( r + 1 ) μ r + 1 r r + 1 2 e 1 + μ μ ! ( r + μ + 1 ) 3 2 + μ + r , μ N ; r .
and the bounds
exp 1 360 30 v 3 v 3 v 4 + μ 30 ( μ + v + 1 ) 3 + v + 4 ( μ + v + 1 ) 4
< Γ μ ( 1 + v ) μ 1 + v v 1 2 + v e μ + 1 μ ! ( μ + v + 1 ) μ + v + 3 / 2 < exp μ + 1 12 v ( μ + v + 1 ) , μ N ; v > 0 .
Moreover, we proved that the function
S α , μ ( v ) = ln Γ μ ( v ) + ψ μ ( α + v ) 2 v ln μ v v + μ + 1 + ( μ + 1 ) ln ( v + μ + 1 ) ln μ μ ! e 1 + μ , μ N
is CM on v > 0 if and only if α 1 3 , and S α , μ ( v ) is the CM function on v > 0 if and only if α = 0 .

Author Contributions

Writing to Original draft, M.M., H.A. and H.M. All authors contributed equally to the writing of this paper. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Acknowledgments

The authors are thankful to the editor and the anonymous reviewers for their valuable comments and suggestions.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Abramowitz, M.; Stegun, I.A. Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables; Dover Publications, Inc.: New York, NY, USA, 1970. [Google Scholar]
  2. Cichon, J. Stiriling Approximation Formula. Available online: https://cs.pwr.edu.pl/cichon/Math/StirlingApp.pdf (accessed on 10 June 2022).
  3. Batir, N. Inequalities for the gamma function. Arch. Math. 2008, 91, 554–563. [Google Scholar] [CrossRef]
  4. Batir, N. An approximation formula for n! Proyecciones J. Math. 2013, 32, 173–181. [Google Scholar] [CrossRef] [Green Version]
  5. Guo, B.-N.; Qi, F. Sharp inequalities for the psi function and harmonic numbers. Anal.-Int. Math. J. Anal. Its Appl. 2014, 34, 201–208. [Google Scholar] [CrossRef] [Green Version]
  6. Qi, F.; Guo, B.-N. Necessary and sufficient conditions for functions involving the tri- and tetra-gamma functions to be completely monotonic. Adv. Appl. Math. 2010, 44, 71–83. [Google Scholar] [CrossRef] [Green Version]
  7. Qi, F.; Luo, Q.-M. Bounds for the ratio of two gamma functions: From Wendel’s asymptotic relation to Elezović-Giordano-Pečarić’s theorem. J. Inequalities Appl. 2013, 542, 20. [Google Scholar] [CrossRef] [Green Version]
  8. Alzer, H. On some inequalities for the gamma and psi function. Math. Comput. 1997, 66, 373–389. [Google Scholar] [CrossRef] [Green Version]
  9. Burić, T.; Elezović, N. Some completely monotonic functions related to the psi function. Math. Inequal. Appl. 2011, 14, 679–691. [Google Scholar]
  10. Qi, F.; Wang, S.-H. Complete monotonicity, completely monotonic degree, integral representations, and an inequality related to the exponential, trigamma, and modified Bessel functions. Glob. J. Math. Anal. 2014, 2, 91–97. [Google Scholar] [CrossRef] [Green Version]
  11. Qi, F.; Guo, S.; Guo, B.-N. Complete monotonicity of some functions involving polygamma functions. J. Comput. Appl. Math. 2010, 233, 2149–2160. [Google Scholar] [CrossRef] [Green Version]
  12. Widder, D.V. The Laplace Transform; Princeton University Press: Princeton, NJ, USA, 1946. [Google Scholar]
  13. Alzer, H.; Batir, N. Monotonicity properties of the gamma function. Appl. Math. Lett. 2007, 20, 778–781. [Google Scholar] [CrossRef] [Green Version]
  14. Batir, N. On some properties of the gamma function. Expo. Math. 2008, 26, 187–196. [Google Scholar] [CrossRef] [Green Version]
  15. Sandor, J. Selected Chapters of Geometry, Analysis and Number Theory; RGMIA Monographs, Victoria University of Technology: Victoria, Australia, 2005. [Google Scholar]
  16. Krasniqi, V.; Shabani, A. Convexity properties and inequalities for a generalized gamma function. Appl. Math. E-Notes 2010, 10, 27–35. [Google Scholar]
  17. Krasniqi, V.; Merovci, F. Some completely monotonic properties for the (p, q)-gamma function. arXiv 2014, arXiv:1407.4231. [Google Scholar]
  18. Nantomah, K.; Prempeh, E.; Twum, S.B. On a (p, k)-analogue of the gamma function and some associated inequalities. Moroc. J. Pure Appl. Anal. 2016, 2, 79–90. [Google Scholar] [CrossRef] [Green Version]
  19. Nantomah, K.; Merovci, F.; Nasiru, S. Some monotonicity properties and inequalities for the (p, k)-gamma function. Kragujev. J. Math. 2018, 42, 287–297. [Google Scholar] [CrossRef] [Green Version]
  20. Tassaddiq, A.; Alriban, A. On Modification of the Gamma Function by Using Mittag-Leffler Function. J. Math. 2021, 2021, 9991762. [Google Scholar] [CrossRef]
  21. Goyal, R.; Agarwal, P.; Oros, G.I.; Jain, S. Extended Beta and Gamma Matrix Functions via 2-Parameter Mittag-Leffler Function. Mathematics 2022, 10, 892. [Google Scholar] [CrossRef]
  22. Tassaddiq, A. A new representation of the extended k-gamma function with applications. Math. Methods Appl. Sci. 2021, 44, 11174–11195. [Google Scholar] [CrossRef]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Mahmoud, M.; Almuashi, H.; Moustafa, H. An Asymptotic Expansion for the Generalized Gamma Function. Symmetry 2022, 14, 1412. https://doi.org/10.3390/sym14071412

AMA Style

Mahmoud M, Almuashi H, Moustafa H. An Asymptotic Expansion for the Generalized Gamma Function. Symmetry. 2022; 14(7):1412. https://doi.org/10.3390/sym14071412

Chicago/Turabian Style

Mahmoud, Mansour, Hanan Almuashi, and Hesham Moustafa. 2022. "An Asymptotic Expansion for the Generalized Gamma Function" Symmetry 14, no. 7: 1412. https://doi.org/10.3390/sym14071412

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop