LECTURE NOTES ON MATHEMATICAL PHYSICS

Department of Physics, University of Bologna, Italy

URL: www.fracalmo.org

THE EULERIAN FUNCTIONS

Francesco MAINARDI
Department of Physics, University of Bologna, and INFN,
Via Irnerio 46, I–40126 Bologna, Italy.
[email protected] [email protected]

Contents (pp. 1 – 14)

1. THE GAMMA FUNCTION : Γ(z)

Euler’s Integral Representation . . . . . . . . . . . . . . . . . . . . . . . . p. 1
Domain of Analyticity . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 1
Difference Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 1
Integer Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 1
Complex Factorial Function . . . . . . . . . . . . . . . . . . . . . . . . . p. 1
Gaussian Integral Representation . . . . . . . . . . . . . . . . . . . . . . . p. 1
Semi-Integer Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 2
Analytical Continuation by the Recurrence Formula . . . . . . . . . . . . . . p. 2
Analytical Continuation by the Cauchy-Saalscütz Representation . . . . . . . . . p. 2
Analytical Continuation by the Mixed Representation . . . . . . . . . . . . . . p. 2
Reflection or Complementary Formula . . . . . . . . . . . . . . . . . . . . p. 3
Multiplication Formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 3
Pochhammer’s Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 3
Graphical Representation of the Gamma Function on the Real Axis . . . . . . . p. 4
Analytical Continuation by Hankel’s Integral Representations . . . . . . . . . . p. 5
Notable Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 8
Asymptotic Formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 9
2. THE BETA FUNCTION : B(p, q)
Euler’s Integral Representation . . . . . . . . . . . . . . . . . . . . . . . . p. 11
Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 11
Trigonometric Integral Representation . . . . . . . . . . . . . . . . . . . . p. 11
Relation with the Gamma Function . . . . . . . . . . . . . . . . . . . . . . p. 11
Integral Representation on [0, ∞) . . . . . . . . . . . . . . . . . . . . . . . p. 12
A further Integral Representation on [0, 1] . . . . . . . . . . . . . . . . . . . p. 12
Convolution Representation . . . . . . . . . . . . . . . . . . . . . . . . . p. 12
Convolution between Gel’fand-Shilov Functions . . . . . . . . . . . . . . . . p. 12
Some Applications of the Beta Function . . . . . . . . . . . . . . . . . . . . p. 13
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 14
ii Francesco MAINARDI

c 1996-2000-2007-2010 Prof. Francesco Mainardi - Bologna - Italy

 THE EULERIAN FUNCTIONS 1

1. THE GAMMA FUNCTION : Γ(z)

The Gamma function is the most widely used of all the special functions: it is usually
discussed first since it appears in almost every integral or series representation of the other
advanced mathematical functions. This function, denoted by Γ(z), can be defined by
Euler’s Integral Representation (1)
Z ∞
Γ(z) = e−u uz−1 du , Re(z) > 0 . (G.1)
0

This representation is the most common for Γ(z), even if it is valid only in the right half-plane
of C; later, the analytic continuation to the left half-plane will be considered to obtain its
Domain of Analyticity

DΓ = C − {0, −1, −2, . . . , } . (G.2)

Using integration by parts, (G.1) shows that, at least for Re(z) > 0, Γ(z) satisfies the simple
Difference Equation (2)

Γ(z + 1) = z Γ(z) , (G.3)

which can be iterated to yield

Γ(z + n) = z (z + 1) . . . (z + n − 1) Γ(z) , n ∈ IN . (G.4)

The recurrence formulas (G.3-4) can be extended to any z ∈ DΓ . In particular, being Γ(1) = 1
we get for (non negative)
Integer Values

Γ(n + 1) = n! , n = 0, 1, 2, . . . . (G.5)

As a consequence Γ(z) can be used to define the

Complex Factorial Function

z! := Γ(z + 1) . (G.6)

By the substitution u = v 2 in (G.1) we get the

Gaussian Integral Representation
Z ∞ 2
Γ(z) = 2 e−v v 2z−1 dv , Re(z) > 0 , (G.7)
0

which can be used to obtain Γ(z) when z assumes positive semi-integer values, as follows.
2 Francesco MAINARDI

Semi-Integer Values (3)

 
1
Z +∞ 2 √
Γ = e−v dv = π ≈ 1.77245 , (G.8)
2 −∞

1
 Z +∞ 2
 
1 (2n − 1)!! √ (2n)!
Γ n+ = e−v v 2n dv = Γ = π 2n , n ∈ IN . (G.9)
2 −∞ 2 2n 2 n!
The formula (G.4), proven for Re(z) > 0 using integration by parts in (G.1), can be used to
obtain the Domain of Analyticity DΓ by means of the so-called
Analytical Continuation by the Recurrence Formula (4)

Γ(z + n)
Γ(z) = . (G.10)
(z + n − 1) (z + n − 2) . . . (z + 1) z
In fact the numerator at the R.H.S of (G.10) is analytic for Re(z) > −n; hence, the L.H.S.
is analytic for Re(z) > −n except for simple poles at z = 0, −1, . . . , (−n + 2), (−n + 1) .
Since n can be arbitrarily large, we deduce that Γ(z) is analytic in the entire complex plane
except the points zn = −n (n = 0, 1, . . .), which turn out to be simple poles with residues
Rn = (−1)n /n!. The point at the infinity, being accumulation point of poles, is an essential
non isolated singularity. Thus Γ(z) is a transcendental meromorphic function.
The integration by parts in the basic representation (G.1) provides the
Analytical Continuation by the Cauchy-Saalschütz Representation
" #
∞ u2 un
Z
z−1
Γ(z) = u −u
e −1+u− + . . . + (−1)n+1 du , (G.10′ )
0 2! n!
which holds for any integer n ≥ 0 with −(n + 1) < Re(z) < −n. The representation can be
understood by iterating the first step. In fact for −1 < Re(z) < 0:
∞ 1 ∞ 1
Z Z
z−1 −u
uz e−u du =
 
u e − 1 du = Γ(z + 1) = Γ(z) .
0 z 0 z
Finally another instructive manner to obtain the Domain of Analyticity is to use the so-called
Analytical Continuation by the Mixed Representation (5)

∞
(−1)n ∞
Z
e−u uz−1 du ,
X
Γ(z) = + z ∈ DΓ . (G.11)
n=0
n!(z + n) 1

This representation can be obtained splitting the integral in (G.1) into 2 integrals, the former
over the interval 0 ≤ u ≤ 1 which is then developed in series, the latter over the interval
1 ≤ u ≤ ∞, which, being uniformly convergent inside C, provides an entire function. The
terms of the series (uniformly convergent inside DΓ ) provide the principal parts of Γ(z) at
the corresponding poles zn = −n .
 THE EULERIAN FUNCTIONS 3

Reflection or Complementary Formula (6)

π
Γ(z) Γ(1 − z) = . (G.12)
sin πz
This formula, which shows the relationship between the Γ function and the trigonometric
function, is of great importance together with the recurrence formula (G.3). It can be proven
in several manners; the simplest proof consists in proving (G.12) for 0 < Re(z) < 1 and
extend the result by analytic continuation to C except the points 0, ±1, ±2, . . .
The reflection formula shows that Γ(z) has no zeros. In fact, the zeros cannot be in
z = 0, ±1, ±2, . . . and, if Γ(z) vanished for a not integer z, because of (G.12) this zero be a
pole of Γ(1 − z), that cannot be true.

Multiplication Formulas (7−8)

Gauss proved the following Multiplication Formula (7)
n−1
k
Γ(nz) = (2π)(1−n)/2 nnz−1/2
Y
Γ(z + ), n = 2, 3, . . . , (G.13)
k=0
n

which reduces, for n = 2, to Legendre’s Duplication Formula (8)

1 1
Γ(2z) = √ 22z−1/2 Γ(z) Γ(z + ) , (G.14)
2π 2
and, for n = 3 to the Triplication Formula
1 3z−1/2 1 2
Γ(3z) = 3 Γ(z) Γ(z + ) Γ(z + ) . (G.15)
2π 3 3

Pochhammer’s Symbols (9)

Pochhammer’s symbols (z)n are defined for any non negative integer n as

Γ(z + n)
(z)n := z (z + 1) (z + 2) . . . (z + n − 1) = , n ∈ IN . (G.16)
Γ(z)

with (z)0 = 1. In particular, for z = 1/2 , we obtain from (G.9)

1 Γ(n + 1/2) (2n − 1)!!
 
:= = .
2 n Γ(1/2) 2n
Here, we take the liberty of extending the above notation to negative integers, defining
Γ(z + 1)
(z)−n := z (z − 1) (z − 2) . . . (z − n + 1) = , n ∈ IN . (G.17)
Γ(z − n + 1)
4 Francesco MAINARDI

Graphical Representation of the Gamma Function on the Real Axis (10)

One can have an idea of the graph of the Gamma function on the real axis using the formulas

Γ(x)
Γ(x + 1) = xΓ(x) , Γ(x − 1) = ,
x−1
to be iterated starting from the interval 0 < x ≤ 1, where Γ(x) → +∞ as x → 0+ and
Γ(1) = 1 . For x > 0 Euler’s integral representation (G.1) yields Γ(x) > 0 and Γ′′ (x) > 0
since (10) Z ∞ Z ∞
Γ(x) = e−u ux−1 du , Γ′′ (x) = e−u ux−1 (log u)2 du .
0 0
As a consequence, on the positive real axis Γ(x) turns out to be positive and convex so that
it is either a monotonic decreasing function, or it first decreases and then increases exhibiting
a minimum value. Since Γ(1) = Γ(2) = 1, we must have a minimum at some x0 , 1 < x0 < 2.
It turns out to be x0 = 1.4616 . . . where Γ(x0 ) = 0.8856 . . .; hence x0 is quite close to the
√
point x = 1.5 where Γ attains the value π/2 = 0.8862 . . ..
On the negative real axis Γ(x) exhibits vertical asymptotes at x = −n (n = 0, 1, 2, . . .); it
turns out to be positive for −2 < x < −1, −4 < x < −3, . . ., and negative for −1 < x < 0,
−3 < x < −2, . . ..
Plots of Γ(x) (continuous line) and 1/Γ(x) (dashed line) are shown for −4 ≤ x ≤ 4 in Fig. 1,
and for 0 ≤ x ≤ 3 in Fig. 2.

6
Γ(x)

1/Γ(x)

−2

−4

−6
−4 −3 −2 −1 0 1 2 3 x 4

Fig. 1
Plots of Γ(x) (continuous line) and 1/Γ(x) (dashed line) for −4 ≤ x ≤ 4
 THE EULERIAN FUNCTIONS 5

2.5
Γ(x)

1/Γ(x)
1.5

1
X: 1.462
Y: 0.8856

0.5

0
0 0.5 1 1.5 2 2.5 3 3.5 4
x

Fig. 2
Plots of Γ(x) (continuous line) and 1/Γ(x) (dashed line) for 0 < x ≤ 3

Analytical Continuation by Hankel’s Integral Representations (11−12−13)

Hankel (1864) provided a complex integral representation of the function 1/Γ(z) valid for
unrestricted z; it reads (11)

1 1 et
Z
= z
dt , z ∈ C, (G.18a)
Γ(z) 2πi Ha−t

where Ha− denotes the Hankel path defined as a contour that begins at t = −∞ − ia (a > 0),
encircles the branch cut that lies along the negative real axis, and ends up at t = −∞ + ib
(b > 0). Of course, the branch cut is present when z is not integer because t−z is a multivalued
function; in this case the contour can be chosen as in Fig. 3 where
+π , above the cut,

arg (t) =
−π , below the cut.
When z is an integer, the contour can be taken to be simply a circle around the origin,
described in the counterclockwise direction.
An alternative representation is obtained assuming the branch cut along the positive real
axis; in this case we get

1 1 e−t
Z
=− z
dt , z ∈ C, (G.18b)
Γ(z) 2πi Ha+(−t)
6 Francesco MAINARDI

where Ha+ denotes the Hankel path defined as a contour that begins at t = +∞ + ib (b > 0),
encircles the branch cut that lies along the positive real axis, and ends up at t = +∞ − ia
(a > 0). When z is not integer the contour can be chosen as in Fig. 4 where
0, above the cut,

arg (t) =
2π , below the cut.
When z is an integer, the contour can be taken to be simply a circle around the origin,
described in the counterclockwise direction.
We note that Ha− → Ha+ if t → t e−iπ , while Ha+ → Ha− if t → t e+iπ .

Fig. 3
The left Hankel Contour Ha−

Fig. 4
The right Hankel Contour Ha−
 THE EULERIAN FUNCTIONS 7

The advantage of Hankel’s representations (G.18) over Euler’s integral representation (G.1)
is that they converge for all complex z and not just for Re(z) > 0. As a consequence 1/Γ(z)
is a transcendental entire function (of maximum exponential type); the point at infinity is
an essential non isolated singularity, which is an accumulation point of zeros (zn = −n , n =
0, 1, . . .). Since 1/Γ(z) is entire, Γ(z) does not vanish in C.
The formulas (G.18) are very useful for deriving integral representations in the complex plane
for several special functions.
Furthermore, using the reflection formula (G.12), we can get the integral representations of
Γ(z) itself in terms of the Hankel paths (referred to as Hankel’s Integral Representations
for Γ(z)), which turn out to be valid in the whole Domain of Analyticity DΓ (12) These
representations, which provide the required analytical continuation of Γ(z), are
using the path Ha−

1
Z
Γ(z) = et tz−1 dt , z ∈ DΓ ; (G.19a)
2i sin πz Ha−

using the path Ha+

1
Z
Γ(z) = − e−t (−t)z−1 dt , z ∈ DΓ . (G.19b)
2i sin πz Ha+

Finall, from (G.19b) we get the formula (13)

G(z)
Z
Γ(z) = 2πiz , z ∈ DΓ ; G(z) := e−t tz−1 dt (G.20)
e −1 Ha+

where G(z) has the same integrand as Euler’s integral representation of Γ(z).

Notable Integrals (14−16)

∞ Γ(α + 1)
Z
e−st tα dt = , Re(s) > 0 , Re(α) > −1 . (G.21)
0 sα + 1
This formula (14) provides the Laplace transform of the power function tα ; in particular it
shows that the Laplace transform of this function exists for Re(α) > −1 with abscissa of
convergence σc = 0 .
8 Francesco MAINARDI

∞ β Γ(1 + 1/β)
Z
e−at dt = , Re(a) > 0 , β > 0. (G.22)
0 a1/β
This formula (15) provides for fixed a and β = 2 the Gauss integral. Therefore, for fixed a,
the l.h.s. of (G.22) may be referred to as the generalized Gauss integral. Plots of the integral
for a = 1 versus β are reported in Fig. 5, from which we note that the minimum value is
attained at β0 = 2.16638 . . . and holds I(β0 ) = 0.8856 . . .

3
I(β)

0
0 1 2 3 4 5 6 7 8 9 10
β

1
I(β)

0.95

0.9

0.85
1 2 3
β

Fig. 5
Plots of the function I(β) = Γ(1 + 1/β) versus β ,
TOP: for 0 < β ≤ 10, BOTTOM: zoom for 1 ≤ β ≤ 3.
 THE EULERIAN FUNCTIONS 9

∞ µ 1 Γ(ν/µ) 1 Γ(1 + ν/µ)

Z
e−zt tν − 1 dt = = , Re(z) > 0 , µ > 0 , Re(ν) > 0 . (G.23)
0 µ z ν/µ ν z ν/µ

This formula (16) contains (G.21-22); it reduces to (G.21) for {z = s , µ = 1 , ν = α + 1}, and
to (G.22) for {z = a , µ = β , ν = 1}.

Asymptotic Formulas

√ −z z−1/2

1 1

Γ(z) ≃ 2π e z 1+ + + ... ; z → ∞, |arg z| < π . (G.24)
12 z 288 z 2

This asymptotic expression is usually referred to as Stirling’s Formula, originally given for
n! . The accuracy of the formula is surprisingly very good on the positive real axis also for
moderate values of z = x > 0 , as it can be noted from the following exact formula,
√
2 π e(−x+ 12x ) xx+1/2 ;
θ
x! = x > 0, 0 < θ < 1. (G.25)

In Fig. 6 we show the comparison between the plot of the Gamma function (continuous line)
with that provided by the first term of the Stirling approximation (in dashed line), in the
range 0 ≤ x ≤ 4

5
Γ(x)
4

0 1 2 3 4
x
Fig. 6
The Γ(x) (continuous line) compared with its first order Stirling aproximation (dashed line)
10 Francesco MAINARDI

In Fig. 7 we show the relative error of the first term approximation with respect to the exact
value in the range 1 ≤ x ≤ 10 ; we note that this error decreases from less than 8 % to less
than 1 % .

10
102ε
9

0
1 2 3 4 5 6 7 8 9 10
x

Fig. 7
Plot of the relative error for the first Stirling approximation to Γ(x) for 1 ≤ x ≤ 10 .

The two following asymptotic expressions provide a generalization of the Stirling formula.
√
Γ(az + b) ≃ 2 π e−az (az)az+b−1/2 ; z → ∞, |arg z| < π , a > 0. (G.26)

Γ(z + a) (a − b) (a + b − 1)
 
≃ z a−b 1 + + ... , (G.27)
Γ(z + b) 2z
as z → ∞ along any curve joining z = 0 and z = ∞ providing z 6= −a, −a − 1, . . . , and
z 6= −b, −b − 1, . . . .
 THE EULERIAN FUNCTIONS 11

THE BETA FUNCTION : B(p, q)

Euler’s Integral Representation
Z 1
B(p, q) = up−1 (1 − u)q−1 du , Re(p) > 0 , Re(q) > 0 . (B.1)
0

This representation is the standard one for the Beta function, which is also referred to as the
Euler integral of the first kind, while the integral representation (G.1) for Γ(z) is referred to
as the Euler integral of the second kind. The Beta function is therefore a complex function
of two complex variables whose analyticity properties can be deduced later, as soon as the
relation with the Gamma function will be established.

Symmetry

B(p, q) = B(q, p) . (B.2)

This property is a simple consequence of the definition (B.1).

Trigonometric Integral Representation

Z π/2
B(p, q) = 2 (cos ϑ)2p−1 (sin ϑ)2q−1 dϑ , Re(p) > 0 , Re(q) > 0 . (B.3)
0

This noteworthy representation follows from (B.1) by setting u = (cos ϑ)2 .

Relation with the Gamma Function

Γ(p) Γ(q)
B(p, q) = . (B.4)
Γ(p + q)
This relation is of fundamental importance. Furthermore, it allows us to obtain the analytical
continuation of the Beta function. The proof of (B.4) can be easily obtained by writing the
product Γ(p) Γ(q) as a double integral which is to be evaluated introducing polar coordinates.
In this respect we must use the Gaussian representation (G.7) for the Gamma function and
the trigonometric representation (B.3) for the Beta function. In fact,
2 2
Γ(p) Γ(q) = 4 0∞ 0∞e−(u +v ) u2p−1 v 2q−1 du dv
R R
2 R π/2
= 4 0∞e−ρ ρ2(p+q)−1 dρ 0 (cos ϑ)2p−1 (sin ϑ)2q−1 dϑ
R

= Γ(p + q) B(p, q).

Henceforth, we shall exhibit other integral representations for B(p, q), all valid for Re(p) > 0
and Re(q) > 0.
12 Francesco MAINARDI

Integral Representations on [0, ∞)

∞ xp−1 ∞ xq−1 1 ∞ xp−1+ xq−1

Z Z Z
B(p, q) = dx = dx = dx . (B.5)
0 (1 + x)p+q 0 (1 + x)p+q 2 0 (1 + x)p+q

The first representation follows from (B.1) by setting u = x/(1 + x) ; then, the other two are
easily obtained by using the symmetry property of B(p, q) .

A further Integral Representation on [0, 1]

Z 1 y p−1
+ y q−1
B(p, q) = dy . (B.6)
0 (1 + y)p+q
This representation is obtained from the first integral in (B.5) as a sum of two contributions,
on [0, 1] and [1, ∞) .
The Beta function plays a fundamental role in the Laplace convolution of power functions. We
recall that the Laplace convolution is the convolution between causal functions (i.e. vanishing
for t < 0), Z +∞ Z t
f (t) ∗ g(t) = f (τ ) g(t − τ ) dτ = f (τ ) g(t − τ ) dτ .
−∞ 0
The convolution satisfies both the commutative and associative properties in that
f (t) ∗ g(t) = g(t) ∗ f (t) , f (t) ∗ [g(t) ∗ h(t)] = [f (t) ∗ g(t)] ∗ h(t) .

It is straightforward to show by setting in (B1) u = τ /t , the

Convolution Representation
Z t
tp−1 ∗ tq−1 = τ p−1 (t − τ )q−1 dτ = tp+q−1 B(p, q) . (B.7)
0

Introducing the causal Gel’fand-Shilov function

tλ−1
+
Φλ (t) := , λ ∈ C,
Γ(λ)
(where the suffix + just denotes the causality property of vanishing for t < 0), we can write
the previous result in the following interesting form

Convolution between Gel’fand-Shilov Functions

Φp (t) ∗ Φq (t) = Φp+q (t) . (B.8)

In fact, dividing by Γ(p) Γ(q) the L.H.S of (B.7), and using (B.4), we just obtain (B.8).
 THE EULERIAN FUNCTIONS 13

Some Applications of the Beta Function

The results (B.7-8) show that the convolution integral between two (causal) functions, which
are absolutely integrable in any interval [0, t] and bounded in every finite interval that does
not include the origin, is not necessarily continuous at t = 0, even if a theorem ensures
that this integral turns out to be continuous for any t > 0 , see G. Doetsch, Introduction to
the Theory and Application of the Laplace Transformation, Springer-Verlag, Berlin 1974, pp.
47-48. In fact, considering two arbitrary real numbers α, β greater than −1, we have

 +∞
 if − 2 < α + β < −1 ,
α β α+β+1
Iα,β (t) := t ∗ t = B(α+1, β+1) t ⇒ lim+ Iα,β (t) = c(α) if α + β = −1 ,
t→0  0 if α + β > −1 ,


where c(α) = B(α + 1, −α) = Γ(α + 1) Γ(−α) = π/ sin(−απ).

We note that in the case α + β = −1 (and therefore −1 < α < 0) the convolution integral
attains for any t > 0 the constant value c(α) ≥ π . In particular, for α = β = −1/2 , we
obtain the minimum value for c(α) (relevant result in the problem of the tautochrone), i.e.
Z t dτ
√ √ = π. (B.9)
0 τ t−τ
The Beta function is also used to prove some basic identities for the Gamma function, like
the Complementary Formula (G.12) and the Duplication Formula (G.14).
For the Complementary Formula we know that it is sufficient to prove it for 0 < α < 1 ,
π
Γ(α) Γ(1 − α) = , 0 < α < 1.
sin πα
We note from (B.4-5) that
∞ xα−1
Z
Γ(α) Γ(1 − α) = B(α, 1 − α) = dx ,
0 1+x
and from a classical exercise in complex analysis
∞ xα−1 π
Z
dx = . (B.10)
0 1+x sin πα
For the Duplication Formula we note that it is equivalent to
Γ(1/2) Γ(2z) = 22z−1 Γ(z) Γ(z + 1/2) ,
and hence, after simple manipulations, to
B(z, 1/2) = 22z−1 B(z, z) .
This identity is easily verified for Re(z) > 0 , using the trigonometric representation (B.3)
for the Beta function and noting that
Z π/2 Z π/2 Z π/2
(cos ϑ)α dϑ = (sin ϑ)α dϑ = 2α (cos ϑ)α (sin ϑ)α dϑ , Re(α) > −1 ,
0 0 0
since sin 2ϑ = 2 sin ϑ cos ϑ .
14 Francesco MAINARDI

REFERENCES
– M. Abramowitz and I.A. Stegun (Editors) :
Hanbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables,
Dover, New York (1992); Ch. 6, pp. 253-266.
– E. Artin : The Gamma Function,
Holt, Rinehart and Winston, New York (1964); pp. 39.
– L. Amerio : Funzioni Analitiche e Trasformazione di Laplace,
Tamburini, Milano (1972); §24, pp. 114-120.
– L.C. Andrews : Special Functions of Mathematics and Engineering,
McGraw-Hill, New York (1992); Ch. 2, pp. 61-86.
– V. Barcilon : Special Functions, Ch. 7, pp. 344-377,
in C.E. Pearson (Editor), Handbook of Applied Mathematics, 2-nd edition,
Van Nostrand Reinhold, New York (1990); §7.2, pp. 348-351.
– R. Campbell : Les Intégrales Eulériennes et leurs Applications,
Dunod, Paris (1992); pp. 268.
– P.J. Davis : Leonard Euler’s Integral: A Historical Profile of the Gamma Function,
Amer. Math. Monthly, 66, 849-869 (1959).
– A. Erdélyi, W. Magnus, F. Oberhettinger and F.G. Tricomi :
Higher Transcendental Functions, Bateman Manuscript Project, 3 Voll,
McGraw-Hill, New York (1953); Vol. 1.
– O.J. Farrel and B. Ross : Solved Problems in Analysis,
Dover, New York (1971) [first published by McMillan, New York (1963)];
Ch. 1, pp. 1-49; Ch. 2, pp. 50-117.
– L. Gatteschi : Funzioni Speciali, UTET, Torino (1973); Ch. 1, pp. 1-25.
– I.M. Gel’fand and G.E. Shilov : Generalized Functions, Vol. 1,
Academic Press, New York (1964); pp. 115-122.
– P. Henrici : Applied and Computational Analysis,
Vol. II: Special Functions, Integral Transformations, Asymptotics, Continued Fractions,
Wiley, New York (1977); §8.4 - §8.7, pp. 24-61.
– W. Magnus, F. Oberhettinger and R.P. Soni : Formulas and Theorems for the Special
Functions of Mathematical Physics, Springer-Verlg, Berlin (1966); Ch. 1, pp. 1-36.
– L. Schwartz : Mathematics for the Physical Sciences,
Hermann, Paris (1966); Ch. 8, pp. 311-333.
– N. Temme : Special Functions. An Introduction to the Classical Functions
of Mathematical Physics, Wiley, New York (1996); Ch. 3, pp. 41-77.
– E.T. Wittaker and G.N. Watson : A Course of Modern Analysis,
Cambridge Univ. Press, New York (1962). Ch. 12, pp. 235-264.