GAMMA-POLYLOGARITHM A BEAUTIFUL IDENTITY

Andrés L. Granados M., 30/Nov/2018, Rev.01/Dic/2020

In modern mathematics, the polylogarithm (also known as Jonquière’s function, for Alfred Jonquière
[7]) is a special function Lis (z) of order s and argument z. Only for special values of s does the polylogarithm
reduce to an elementary function such as the natural logarithm or rational functions. In quantum statistics,
the polylogarithm function appears as the closed form of integrals of the Fermi-Dirac distribution and the
Bose-Einstein distribution, and is also known as the Fermi-Dirac integral or the Bose-Einstein integral.
In quantum electrodynamics, polylogarithms of positive integer order arise in the calculation of processes
represented by higher-order Feynman diagrams [3].
The polylogarithm function is equivalent to the Hurwitz zeta function [8] (either function can be
expressed in terms of the other) and both functions are special cases of the Lerch transcendent Φ function
[9]. Polylogarithms should not be confused with polylogarithmic functions nor with the offset logarithmic
integral (Eulerian logarithmic integral) which has a similar notation [4].
Recall that the gamma function is
 ∞
Γ(s) = ts−1 e−t dt Γ(s + 1) = s Γ(s) (1)
0

and the polylogarithm function [3] is defined by

∞
  z
zk Lis (t) ∂Lis (z) Lis−1 (z)
Lis (z) = Lis+1 (z) = dt = (2)
ks 0 t ∂z z
k=1

a power series in z, which is also a Dirichlet series in s. This definition also includes recurrent expressions for
integrals (2.b) and derivatives (2.c). This definition is valid for arbitrary complex order s and for all complex
arguments z with |z| < 1; it can be extended to |z| ≥ 1 by the process of analytic continuation. The special
case s = 1 involves the ordinary natural logarithm,
 ∞

z
ln(1 − t) z 2k+1 1
Li1 (z) = − ln(1−z) Li2 (z) = − dt χν (z) = = 2 [ Liν (z)−Liν (−z) ] (3)
0 t (2k + 1)ν
k=0

while the special cases s = 2 and s = 3 are called the dilogarithm (also referred to as Spence’s function) and
trilogarithm, respectively. The name of the function comes from the fact that it may also be defined as the
repeated integral of itself. Chi function χν (z) is the antisymmetrical part of Liν (z). The primitive integral
developed by series

ln(a + bx) bx b2 x2 b3 x3 b4 x4
dx = (ln a) ln x + − 2 2 + 2 3 − 2 4 + ··· [b2 x2 < a2 ]
x a 2 a 3 a 4 a
[ln(bx)]2 a a2 a3 a4
= − + 2 2 2 − 2 3 3 + 2 4 4 − ··· [b2 x2 > a2 ] (4)
2 bx 2 b x 3 b x 4 b x
 
bx bx
= (ln a) ln x + Φ − , 2, 1 [a > 0]
a a

from tables of integral [5] and [6] may be used to integrate (3.b).
The following general identities
 ∞  ∞
ts−1 ts−1
Lis (z) Γ(s) = dt − Lis (−z) Γ(s) = dt (5)
0 et /z − 1 0 et /z + 1

1
may be obtained by making the change of variables t = k(u−ln z) and application of the formula for the sum of
∞
terms from a infinite geometric progression k=1 rk = r/(1 − r) (|r| < 1), with ratio r = z/eu = exp(ln z − u)
(initial term a = r).
The deduction of the identity (5.a) follows after introduction of the mentioned change of variable into
gamma expresión (1.a). Thus
 ∞
ks
Γ(s) = k us−1 e−k(u−ln z) du (6)
z 0

Hence  ∞
zk 1 z
s
= us−1 rk du r= = eln z−u (7)
k Γ(s) 0 eu
Substitution of this in the series (2.a) gives
∞   ∞ 
1  ∞
s−1 k 1 ∞
s−1

k 1 ∞
r
Lis (z) = u r du = u r du = us−1 du (8)
Γ(s) 0 Γ(s) 0 Γ(s) 0 1−r
k=1 k=1

and finally, reversing r substitution, results the sought identity (integrated in u instead t). The restriction of
this integral identity is the same as the geometric infinite progression, ie. |r| = |z e−u | < |z| < 1.
Particularly, the next equivalences of functions

Lis (1) = ζ(s) Lis (−1) = −η(s) (9)

are taken into account for the incoming results. Indeed, the definition of the η(s) Dirichlet function [2] is
followed from the series
∞
 ∞
 ∞

1 (−1)k−1 1
ζ(s) = = ζ(s, 1) η(s) = ζ(s, α) = (10)
ks ks (k + α)s
k=1 k=1 k=0

related to the ζ(s) Riemann Function [1] as

η(s) = (1 − 21−s ) ζ(s) (11)

The following particular identities may be simply obtained from the general identities (5), with the
equivalence (9),
 ∞ s−1  ∞ s−1
x x
ζ(s) Γ(s) = dx η(s) Γ(s) = dx (12)
0 e −1
x
0 ex + 1
and with this next change of variable

x = e−u u = − ln x
(13)
dx = −e−u du du = −dx/x

the following definite integrals are resolved

 1  ∞
ln x u π2
dx = du = ζ(2) Γ(2) = (14)
0 x−1 0 eu −1 6
 1  ∞
ln x u π2
dx = − du = −η(2) Γ(2) = − (15)
0 x+1 0 eu + 1 12
obtaining known results, where Γ(2) = 1.

2
 A more general identity yet comes from the use of the Lerch trascendent Φ function given by [9]

∞
 zk
Φ(z, s, α) = (16)
(k + α)s
k=0

being Riemann zeta ζ(s) [1], Dirichlet eta η(s) [2], polylogarithm Lis (z) [3], Hurwitz zeta ζ(s, α) [8], Lerch
zeta L(λ, α, s) [9] and Legendre chi χν (z) [13] particular cases

ζ(s) = ζ(s, 1) = Lis (1) = Φ(1, s, 1) Lis (z) = z Φ(z, s, 1) L(λ, α, s) = Φ(e2πiλ , s, α)
(17)
η(s) = −Lis (−1) = Φ(−1, s, 1) ζ(s, α) = Φ(1, s, α) χν (z) = 2−ν z Φ(z 2 , ν, 1/2)

Therefore the more general identity is

 ∞
ts−1 e−αt
Φ(z, s, α) Γ(s) = dt (18)
0 1 − z e−t

that includes (5) and (12), as particular cases. The identities (5) and (18) are the Mellin transforms M[f ](s) =
 ∞ s−1
0 t f (t) dt (see (19) below) of the functions f (t) = (et /z − 1)−1 and f (t) = [eαt (1 − z/et)]−1 , respectively.
In this case the deduction is similar to identity (5), but the change of variables is t = (k + α) u − k ln z and

the formula for the sum of terms from a infinite geometric progression is ∞ k=0 a r = a/(1 − r) (|r| < 1),
k
−αu
with the same ratio r = z/e = exp(ln z − u) and a different initial term a = e
u
.
Various of the preceding identities are the result of the definition of the Mellin Transform [10] (Robert
Hjalmar Mellin, Finland 1854-1933)
 ∞  c+i∞
1
M[f ](s) = g(s) = xs−1 f (x) dx M−1 [g](x) = f (x) = x−s g(s) ds (19)
0 2πi c−i∞

(The notation in the inverse implies this is a line integral taken over a vertical line in the complex plane,
whose real part c is arbitrary, provided that it meets certain conditions given in the Mellin invertion theorem)
applied to the Ramanujan’s Master Theorem when

∞
 h(k)
f (x) = (−x)k =⇒ M[f ](s) = Γ(s) h(−s) (20)
k!
k=0

(see [11] with several examples) where f (x) is a complex-valued analytic function expansion in a infinite
series. Example of this may be seen in [12] for the Hurwitz zeta function (identity (18) with z = 1). Review
[14] for a more general treatment of this subject.

REFERENCES
[1] https://en.wikipedia.org/wiki/Riemann zeta function
[2] https://en.wikipedia.org/wiki/Dirichlet eta function
[3] https://en.wikipedia.org/wiki/Polylogarithm
[4] https://en.wikipedia.org/wiki/Logarithmic integral function
[5] Dwight, H. B. Tables of Integrals and Other Mathematical Data, 3rd Edition. The MacMillan
Company (New York), 1957. Integral 621.1, p.137.
[6] Gradshteyn, I. S.; Ryzhik, I. M. Table of Integrals Series and Products, 7th Edition. Elsevier-
Academic Press (Amsterdam), 2007. Integral 2.728, 2◦ , p.239.
∞ n
[7] Jonquière, A.,Note sur la sèrie n=1 xns , Bull. de la S.M.F., Vol.17, pp.142-152, (1889).

3
 [8] https://en.wikipedia.org/wiki/Hurwitz zeta function
[9] https://en.wikipedia.org/wiki/Lerch zeta function
[10] https://en.wikipedia.org/wiki/Mellin transform
[11] https://en.wikipedia.org/wiki/Ramanujan’s master theorem
[11a] Amdeberhan, T.; Gonzalez, I.; Harrison, M.; Moll, V. H.; Straub, A. “Ramanujan’s Master Theorem”.
The Ramanujan Journal, Vol.29, pp.103-120, (2012).
[12] Espinosa, O.; Moll, V. “On some definite integrals involving the Hurwitz zeta function: Part 1/2”.
The Ramanujan Journal, Vol.6, pp.159-188/449-468, (2002).
[13] https://en.wikipedia.org/wiki/Legendre chi function
[14] Kim, N. “Ramanujans Integral Identities of The Riemann Ξ-Function and The Lerch Transcendent”.
Journal of Number Theory, Vol.168, pp.292-305, (2016). doi:10.1016/j.jnt.2016.04.006