CITE THIS NOTEBOOK: Wolfram R&D LIVE: Everything to Know about Mellin-Barnes Integrals - Part I by Oleg Marichev. Wolfram Community FEB 24 2023.

Introduction

What is Mellin-Barnes integrals (MBintegrals)

Pincherle S . (1888), Mellin H . (1910), Barnes E . W . (1908), Dixon A . L ., Ferrar W . L . (1936)
In Slater' type notations the MBintegral is (
s->
1
s
sometimes)
MBintegral=
1
2π
γ+∞
∫
γ-∞
A
∏
k=1
Gamma[
a
k
+
α
k
s]
B
∏
k=1
Gamma[
b
k
-
β
k
s]
C
∏
k=1
Gamma[
c
k
+
γ
k
s]
D
∏
k=1
Gamma[
d
k
-
δ
k
s]
-s
z
s​​(*
1
2π
γ+∞
∫
γ-∞
A
∏
k=1
Gamma[
a
k
-
α
k
s]
B
∏
k=1
Gamma[
b
k
+
β
k
s]
C
∏
k=1
Gamma[
c
k
-
γ
k
s]
D
∏
k=1
Gamma[
d
k
+
δ
k
s]
s
z
sSimmetry*)
α
k
>0∧
β
k
>0∧
γ
k
>0∧
δ
k
>0
Music, easy to remember: {A, B, C, D}, {a, b, c, d}, {α, β, γ, δ}
A+D-B-C>or<or==0(*direction'schoice
z
z
*)
A+B-C-D>or==0(*sectorexisting
z

*)
ℒ={γ-∞,γ+∞}(*ℒverticalorleftloop
ℒ
-∞
orrightloop
ℒ
∞
*)
In[]:=

s
z
,
-s
1
z
/.{z->-2,s->Random[Complex]}
In[]:=
TableWith{z=Random[]Exp[PiIkk/16],s=Random[Complex]},
s
z

-s
1
z
,{kk,0,34}//Chop
Out[]=
{1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,-0.107116-0.0279068,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.,1.}
See https://functions.wolfram.com/ElementaryFunctions/Power/16/05/01/0018/
In[]:=
PowerExpand[
b
(
a
z
)
,Assumptions->True]
2bπFloor
1
2
-
Im[aLog[z]]
2π


ab
z
Out[]=
True
1
2π
γ+∞
∫
γ-∞
-
-s
z
2
Gamma[-s]
Gamma[1+s]
Gamma[1-s]
s/;-1<γ<0
In[]:=
ResourceFunction["MeijerGForm"][Log[1+z],z]
In[]:=
System`MeijerGDump`SlaterForm[MeijerG[{{1,1},{}},{{1},{0}},z],s]
Gamma[1+s]==ComplexInfinity/;s=-1,-2,-3,…
Re[s+1]>0
In[]:=
-s
z
Gamma[s]Gamma[-s]==-
-s
z
2
Gamma[-s]
Gamma[1+s]
Gamma[1-s]
//FullSimplify
Out[]=
True
-1<γ<0

Contour ℒ for MB integrals

Out[]=

Example with Exp for {A,B,C,D}={1,0,0,0}

Poles of gamma function :
Gamma[s]==ComplexInfinity/;a->-k&&k∈Integers&&k>=0
In[]:=
Series[Gamma[s],{s,-k,1},Assumptions->{k∈Integers&&k>=0}]
Out[]=
k
(-1)
k!(s+k)
+
k
(-1)
PolyGamma[0,1+k]
k!
+
k
(-1)
(
2
π
+3
2
PolyGamma[0,1+k]
-3PolyGamma[1,1+k])(s+k)
6k!
+
2
O[s+k]
1
2π
γ+∞
∫
γ-∞
Gamma[s]
-s
z
s
-z

1
2π
γ+∞
∫
γ-∞
Gamma[
a
1
+
α
1
s]
-s
z
s
-
1
α
1
z

a
1
α
1
z
α
1
In[]:=
Solve[
a
1
+
α
1
s==-k,s]
Out[]=
s
-k-
a
1
α
1

plus Residue :
In[]:=
ResidueGamma[
a
1
+
α
1
s]
-s
z
,s,
-k-
a
1
α
1
,Assumptions->{k∈Integers&&k>=0}
Out[]=
k
(-1)
k
α
1
+
a
1
α
1
z
k!
α
1
In[]:=
Sum[%,{k,0,∞}]
Out[]=
-
1
α
1
z

a
1
α
1
z
α
1
Formally with 3 steps for left SIMPLE poles (for right poles we need to multiply on -1):
In[]:=
Gamma1[
a
1
+
α
1
s]
-s
z
/.s->
-k-
a
1
α
1

Out[]=
-
-k-
a
1
α
1
z
Gamma1[-k]
In[]:=
%/.Gamma1[-k]->
k
(-1)
α
1
k!
Out[]=
k
(-1)
-
-k-
a
1
α
1
z
k!
α
1
In[]:=
Sum[%,{k,0,∞}]
Out[]=
-
1
α
1
z

a
1
α
1
z
α
1

Example with Exp for {A,B,C,D}={0,1,0,0}


Important literature for beginners


Connection with FourierTransform

https://functions.wolfram.com/GeneralIdentities/11/
FourierTransform[f[t],t,z]
1
2π
∞
∫
-∞
f[t]
tz

t
FourierTransform
∞
∫
-∞
f[t-τ]g[τ]τ,t,z
2π
FourierTransform[f[t],t,z]FourierTransform[g[t],t,z]
Finally

General MBintegrals defining MeijerG & FoxH functions

MeijerG & generalized MeijerG https://www.wolframalpha.com/input?i=MeijerG

https://functions.wolfram.com/HypergeometricFunctions/MeijerG/
Meijer C . S . (1936, 1941, 1946)
https://functions.wolfram.com/HypergeometricFunctions/MeijerG1/

Craziness in definition of MeijerG

The reason is hiding in the complimentary formula for gamma function:

FoxH https://www.wolframalpha.com/input?i=FoxH

https://en.wikipedia.org/wiki/Fox_H-function
https://mathworld.wolfram.com/FoxH-Function.html

Evaluation of MBintegrals for MeijerG function

Main formulas for Gamma https://functions.wolfram.com/GammaBetaErf/Gamma/

From Marichev’s book 1978

THEOREM 17 (Slater’s theorem [67, book 1966, page 57]). Let
where
then for these s we have
From pages 62 - 63 of Marichev' s book 1978 (1983)
Bearing in mind the equalities
which are deduced by using the formulae
we easily obtain the following estimate of the integrand as Re(s)∞
By applying this estimate and the first of the formulae we obtain the equality