Our discussion in this note will center on the evaluation of integrals of the form
∞
∫
0
ℋ
1
x
t
ℋ
2
(t)
t
t
ℋ(x),
where the
ℋ
i
are given functions of a single, real variable, and x > 0. Defining the Mellin transform,
*
ℋ
, of ℋ via
*
ℋ
(s)
∞
∫
0
s-1
τ
ℋ(τ)τ,
it can be shown that that the value ℋ(x) of eq.1 is related to the input functions
ℋ
i
by
*
ℋ
1
(s)
*
ℋ
2
(s)
*
ℋ
(s).
Assuming that the required Mellin transforms
*
ℋ
i
can be computed (an assumption which is justified for a large class of functions that we will later define), our task of evaluating the original integral reduces to finding the ℋ(x) corresponding to
*
ℋ
(s)
, a task which can be accomplished, for a large class of functions, using a result known as Slater's theorem to be described now.
The conditions of Slater's theorem are as follows. Let
*
ℋ
(s)=Γ
(a)+s
(b)-s
(c)+s
(d)-s
≡
A
∏
k=1
Γ(
a
k
+s)
B
∏
k=1
Γ(
b
k
-s)
C
∏
k=1
Γ(
c
k
+s)
D
∏
k=1
Γ(
d
k
-s)
where Γ is the gamma function, and the a, b, c, d, A, B, C, and D are parameters.
represent generalized members of the family of functions known as "hypergeometric". They are defined in terms of so-called "Mellin-Barnes integrals" as follows:
m,n
G
p,q
z
a
1
,…,
a
n
,
a
n+1
,…,
a
p
b
1
,…,
b
m
,
b
m+1
,…,
b
q
1
2π
∫
ℒ
m
∏
k=1
Γ(s+
b
k
)
n
∏
k=1
Γ(1-
a
k
-s)
p
∏
k=n+1
Γ(s+
a
k
)
q
∏
k=m+1
Γ(1-
b
k
-s)
-s
z
s
and
m,n
H
p,q
z
(
a
1
,
α
1
),…,(
a
n
,
α
n
)
,
(
a
n+1
,
α
n+1
),…,(
a
p
,
α
p
)
(
b
1
,
β
1
),…,(
b
m
,
β
m
)
,
(
b
m+1
,
β
m+1
),…,(
b
q
,
β
q
)
1
2π
∫
ℒ
m
∏
j=1
Γ
b
j
+
β
j
s
n
∏
i=1
Γ(1-
a
i
-
α
i
s)
p
∏
i=n+1
Γ(
a
i
+s
α
i
)
q
∏
j=m+1
Γ1-
b
j
-
β
j
s
-s
z
s
Notethat
m,n
H
p,q
reducesto
m,n
G
p,q
whenthe
α
i
areallequaltoone.
Much of the utility of these functions stems from the fact that a huge group of algebraic and special functions can be written as special cases of them when appropriate choices of the parameters are taken.