Physical origin of the power-law tailed statistical distributions

G. Kaniadakis
Department of Applied Science and Technology, Politecnico di Torino,
Corso Duca degli Abruzzi 24, 10129 Torino, Italy
giorgio.kaniadakis@polito.it

Starting from the BBGKY hierarchy, describing the kinetics of nonlinear particle system, we
obtain the relevant entropy and stationary distribution function. Subsequently, by employing the
arXiv:1206.2250v1 [cond-mat.stat-mech] 11 Jun 2012

Lorentz transformations we propose the relativistic generalization of the exponential and logarithmic
functions. The related particle distribution and entropy represents the relativistic extension of the
classical Maxwell-Boltzmann distribution and of the Boltzmann entropy respectively and define
the statistical mechanics presented in [Phys. Rev. E 66, 056125 (2002)] and [Phys. Rev. E 72,
036108 (2005). The achievements of the present effort, support the idea that the experimentally
observed power law tailed statistical distributions in plasma physics, are enforced by the relativistic
microscopic particle dynamics.

I. INTRODUCTION The present paper, deals with the relativistic statistical

In plasma physics, the power-law tails in the particle
population, has been systematically observed in the last
fifty years. For instance the cosmic ray spectrum
fi ∝ χ(βEi − βµ) , (1)
which extends over 13 decades in energy, from 108 eV
to 1020 eV, and spans 33 decades in particle flux, from
10−29 to 104 units, obeys the Boltzmann law of classical
statistical mechanics i.e.
χ(x) ∼ exp(−x) , (2)
x→ 0
for low energies, while for high energies this spectrum
presents power law fat tails i.e
χ(x) ∼ x−1/κ , (3)
x→ +∞
the spectral index κ being close to 0.32-0.37.
The above spectrum was approached for the first time
in 1968, by using a different distribution from the Boltz-
mann one. In his proposal Vasyliunas heuristically iden-
tified the function χ(x) with the Student distribution
function which presents power-law tails [1]. In the last
40 years a vast amount of literature has been produced,
regarding the so called kappa-plasmas, based on the Va-
syliunas distribution. Up to now, several physical mech-
anisms have been explored in order to furnish theoretical
support to the experimentally observed power-law-tailed
distribution functions. However there is currently an in-
tense debate regarding the theoretical foundations of the
non-Boltzmannian distributions.
In the last years, after noting that the power-law tails
are placed in the high energy region, and then regards rel-
ativistic particles, the question has been posed whether
the solution of the problem, i.e. the theoretic determina-
tion of the function χ(x) and consequently of the related
distribution and entropy, can be explained by invoking
the basic principles of special relativity.
theory [2–10], predicting for the function χ(x), the very
simple form i.e.
χ(x) = expκ (−x) , (4)
with
p
expκ (x) = ( 1 + κ2 x2 + κx)1/κ . (5)
The parameter κ < 1 is the reciprocal of light speed in a
dimensionless form while the function expκ (x) represents
the relativistic generalization of the ordinary exponential
which recovers in the classical limit κ → 0.
In the last few years various authors have considered
the foundations of the statistical theory based on the dis-
tribution function involving the generalized exponential
expκ (x), e.g. the H-theorem and the molecular chaos
hypothesis [11, 12], the thermodynamic stability [13, 14],
the Lesche stability [15–18], the Legendre structure of the
ensued thermodynamics [19, 20], the thermodynamics of
non-equilibrium systems [21], quantum versions of the
theory [22–24], the geometrical structure of the theory
[25], various mathematical aspects of the theory [26–31],
etc. On the other hand specific applications to physi-
cal systems have been considered, e.g. the cosmic rays
[4], relativistic [32] and classical [33] plasmas in presence
of external electromagnetic fields, the relaxation in rela-
tivistic plasmas under wave-particle interactions [34, 35],
anomalous diffusion [36, 37], non-linear kinetics [38, 39],
kinetics of interacting atoms and photons [40], particle
kinetics in the presence of temperature gradients [41],
particle systems in external conservative force fields [42],
stellar distributions in astrophysics [43–45], quark-gluon
plasma formation [46], quantum hadrodynamics models
[47], the fracture propagation [48], etc. Other applica-
tions regard dynamical systems at the edge of chaos [49–
51], fractal systems [52], field theories [53], the random
matrix theory [54, 55], the error theory [56], the game
theory [57], the information theory [58], etc. Also appli-
cations to economic systems have been considered e.g. to
study the personal income distribution [59–62], to model
deterministic heterogeneity in tastes and product differ-
entiation [63, 64] etc. Finally in [65], some historical
 2

remarks, on the theories dealing with power-law tailed
distribution functions, are reported.
In the present contribution we reconsider critically the
foundations of the statistical theory the generalized ex-
ponential (5). Our main goal is to show that (i) the
function expκ (x), can be obtained within the one-particle
relativistic dynamics, in a very simple and transparent
way, starting directly from the Lorentz transformations
(ii) the standard principles of ordinary relativistic kinet-
ics, conduct unambiguously to the relativistic general-
ization of the classical Boltzmann entropy and Maxwell-
Boltzmann distribution, without invoking any extra prin-
ciple or assumption.
Λ(1) = 0 and L(f ) = Λ(f ). After taking into account
the later relationship, Eq. (9) becomes
Λ(f ⊗f1) = Λ(f ) + Λ(f1 ) . (10)
Consequently the collision invariant Λ(f ) permits us to
determine univocally the correlation function f ⊗ f1 as
follows
 
f ⊗f1 = Λ−1 Λ(f ) + Λ(f1 ) . (11)
In relativistic kinetics, the collision invariant Λ(f ), un-
less an additive constant, is proportional to the micro-
scopic relativistic invariant I. Then we can pose

II. KINETIC EQUATION Λ(f ) = −βI + βµ , (12)

Let us consider the most general relativistic equation β and βµ being two arbitrary constants. In presence of
imposing the particle conservation during the evolution an external electromagnetic field Aν , the more general
of a many body system. That equation is the first equa- microscopic relativistic invariant I, has a form propor-
tion of the BBGKY hierarchy i.e. tional to

I = (pν + qAν/c) Uν − mc2 ,

Z 3 ′ 3
ν ν ∂f d p d p1 d3 p′1 (13)
p ∂ν f −mF = G [f ′ ⊗f1′ −f ⊗f1] ,
∂p ν p′ 0 p10 p′10
(6) Uν being the hydrodynamic four-vector velocity with
and describes, through the one-particle correlation func- U ν Uν = c2 [66]. Finally, after inversion of Eq. (12),
tion or distribution function f = f (x, p), a relativistic the stationary distribution is obtained as follows
particle system in presence of an external force field. The

f = Λ−1 − β I + β µ . (14)
streaming term as well as the Lorentz invariant integra-
tions in the collision integral, have the standard forms of It is remarkable to note that Eq. (12) follows from the
the relativistic kinetic theory [66, 67]. variational equation
The two particle correlation function [68], here denoted
by f ⊗f1 , appearing in Eq. (6), at the moment remains " #
δ
Z Z
an unknown, two variable function. We recall that in 3 3
S − β d p I f + βµ d p f = 0 , (15)
classical kinetics, the two-particle correlation function, δf
according to the molecular chaos hypothesis, is the ordi-
nary product of f and f1 i.e. f ⊗ f1 = f f1 . Therefore where the functional S, unless an arbitrary additive con-
the composition law f ⊗ f1 can be viewed as a relativis- stant, is given by
tic generalized product of f and f1 , isomorphic to the Z Z 
ordinary product. S=− d p 3
Λ(f ) df , (16)
Following standard lines of kinetic theory, we note that
in stationary conditions, the collision integral in Eq. (6) R
vanishes and then it follows that f ⊗f1 = f ′ ⊗f1′ . More Λ(f ) df being the indefinite integral of Λ(f ). The vari-
in general it holds ational equation (15) represents the maximum entropy
principle. The constants β and βµ are the Lagrange mul-
L(f ⊗f1) = L(f ′ ⊗f1′ ) , (7) tipliers while the functional S, defined though Eq. (16),
is the system entropy.
L(x) being an arbitrary function, and this relationship
We stress that the function Λ(f ) defines univocally
expresses a conservation law for the particle system. On
both the stationary distribution (14) and the entropy
the other hand a conservation law has form
(16) of the system as well as the two-particle correlation
Λ(f ) + Λ(f1 ) = Λ(f ′ ) + Λ(f1′ ) , (8) function (11). In classical statistical mechanics, it is well
known that Λ(f ) = ln(f ) so that the two particle correla-
Λ(f ) being the collision invariant of the system. There- tion function becomes f ⊗ f1 = f f1 , while (14) and (16)
fore we can pose reduces to the exponential distribution and Boltzmann
L(f ⊗f1) = Λ(f ) + Λ(f1 ) . (9) entropy respectively.
In the next section, in order to develop a relativistic
From the definition of the correlation function and tak- statistical mechanics, we will determine the function Λ(f )
ing into account of the indistinguishability of the parti- within the special relativity, starting from the Lorentz
cles, it follows that f⊗1 = 1 ⊗f = f and this implies that transformations.

III. LORENTZ TRANSFORMATIONS
In the present section we will show that the function
Λ−1 (f ) emerges within the special relativity as the rela-
tivistic generalization of the ordinary exponential of clas-
sical physics.
Let us consider in the one-dimension frame S two iden-
tical particles A and B, of rest mass m. We suppose that
the two particles move with velocity vA and vB respec-
tively. The momenta of the two particles are indicated
with pA = p (v A) and pB = p (vB ), while their energies
are indicated with EA = E (vA ) and EB = E (vB ) respec-
tively.
In classical physics, in the rest frame S ′ of particle
B, the momentum of the particle B is p′B = 0 while
the momentum of the particle A is given by the Galilei
transformation formula
p′A = pA − pB . (17)
After introducing in place of p the dimensionless mo-
mentum q = p/p∗ , we note that the exponential function
exp(q), permits us to write the Galilei additivity law (17),
in the following factorized form
exp(qA′ ) = exp(qA ) exp(−qB ) . (18)
The Galilei relativity principle, imposes the equiva-
lence of all the inertial frames. According to this prin-
ciple, the inverse Galilei transformation must have the
same structure of the direct transformation (18) except
for the substitutions qA′ ↔ qA and qB → −qB . This re-
quirement is satisfied thanks to the following property of
the exponential function
exp(x) exp(−x) = 1 . (19)
We consider now the two particles in the rest frame S ′
of particle B, within the special relativity. The velocity,
momentum and energy of the particle B are vB′ = 0,
p′B = 0 and EB′ = mc2 respectively. In S ′ the velocity of
particle A is given by the formula
vA − vB
vA′ = . (20)
1 − vA vB /c2
defining the relativistic velocity composition law, which
follows directly from the kinematic Lorentz transforma-
tions. In the same frame S ′ the momentum and energy
of particle A are given by the dynamic Lorentz transfor-
mations
p′A = γ(vB )pA − c−2 vB γ(vB )EA , (21)
EA′ = γ(vB )EA − vB γ(vB )pA , (22)
γ(vB ) = (1 − vB2 /c2 )−1/2 being the Lorentz factor. Af-
ter taking into account the expression of the momentum
pB = mvB γ(vB ) and of the energy EB = mc2 γ(vB ) of the
particle B the latter transformations become
p′A = pA EB /mc2 − EA pB /mc2 , (23)
EA′ = EA EB /mc2 − pA pB /m , (24)
3
Let us introduce in place of the dimensional variables
(v, p, E) the dimensionless variables (u, q, E) through
r
v p E
= = = |κ|c = v∗ < c . (25)
u mq mE
From its definition, it follows that κ can be viewed
as the Einstein β factor related to the velocity v∗ . The
condition v∗ < c, implies that −1 < κ < +1.
The dynamic Lorentz transformations for the dimen-
sionless momentum and energy variables become
qA′ = κ2 qA EB − κ2 qB EA , (26)
EA′ = κ2 EA EB − qA qB , (27)
while the classical limit c → ∞ is replaced by the limit
κ → 0.
For a particle at rest it results E(0) = m c2 and then
E(0) = 1/κ2 . Then 1/κ2 represents the dimensionless
rest energy of the particle. Alternatively 1/κ can be
viewed as the refractive index of a medium in which the
light speed is v∗ .
From the Lorentz invariance it follows easily the
energy-momentum dispersion relation
κ4 E 2 − κ2 q 2 = 1 . (28)
After expressing in the right hand side ofpEq. (26), the
energy in terms of the momentum E = 1 + κ2 q 2 /κ2 ,
we obtain the momentum relativistic additivity law as
follows
p p
qA′ = qA 1 + κ2 qB2 − qB 1 + κ2 qA2 , (29)
which in the classical limit κ → 0 reproduces the ordinary
additivity law (17), of classical physics.
The Galilei relativity principle, holding both in classi-
cal physics and in special relativity, imposes the equiva-
lence of all the inertial frames. According to this prin-
ciple, the inverse transformation of (29) must have the
same structure of the direct transformations (29) except
for the substitutions qA′ ↔ qA and qB → −qB i.e.
p q
qA = qA′ 1 + κ2 qB2 + qB 1 + κ2 qA′ 2 , (30)
It is easy to verify that (30) follows directly from (29)
and viceversa.
The Lorentz transformation for the relativistic mo-
menta (29), representing the momenta additivity law in
special relativity, has the important feature that the con-
tributions of the two particles, appearing in the right
hand side of the equation, are not factorized. Sponta-
neously the question emerges at this point, whether new
variables exist, able to factorize the contribution of the
two particles in the right hand side of the relativistic ad-
ditivity law (29).
It is easy to verify that the new variable is given by
function
p 1/κ
expκ (q) = 1 + κ2 q 2 + κq , (31)
 4

so that the Lorentz transformation (29) assumes the fol- with

lowing factorized form 1
γ= √ , (40)
1 − κ2
expκ (qA′ ) = expκ (−qB ) expκ (qA ) . (32)
ǫ = expκ (γ), (41)
On the other hand Galilei relativity principle imposes, the entropy (16) simplifies as
for the inverse Lorentz transformation, the following fac- Z
torized form S = − γ d3 p f lnκ (f /ǫ) . (42)
expκ (qA ) = expκ (qB ) expκ (qA′ ) . (33) It is worth stressing that the latter relationship de-
fines the relativistic entropy, as proportional to the mean
By comparison of the above direct and inverse Lorentz value of − lnκ (f /ǫ), like in the case of classical sta-
transformations, it obtains the property tistical mechanics where the Boltzmann entropy, S =
R
− d3 p f ln (f /e), is proportional to the mean value of
expκ (q) expκ (−q) = 1 , (34) − ln(f /e). Clearly the entropy (42), in the classical limit
reduces to the Boltzmann entropy. The constant γ, given
which can be verified easily, by direct inspection of the
by (40), represents the Lorentz factor related to the ve-
definition (31).
locity v∗ appearing in (25), and in the classical limit,
It is remarkable that the function expκ (q) emerges in
approaches the unity. On the other hand the constant
one-particle special relativity as the variable, able to fac-
ǫ given by (41), represent a relativistic generalization of
torize the momentum Lorentz transformation, and repre-
the Napier number e, which recovers in the classical limit.
sents the relativistic generalization of the ordinary expo-
The entropy (42) can be written explicitly as follows
nential which factorize the momentum Galilei transfor-  1−κ
f 1+κ

mation in classical physics. Clearly in the classical limit 1 f
Z
3
S= d p − , (43)
κ → 0, expκ (q) reduces to exp(q). 2κ 1−κ 1+κ
The inverse function of expκ (q) indicated by lnκ (q) and
defined through lnκ (expκ (q)) = expκ (lnκ (q)) = 1, repre- while the related stationary distribution (14) assumes the
sents the relativistic generalization of the ordinary loga- form


rithm, which recovers in the classical limit κ → 0, and is f = expκ − β I + β µ , (44)
given by
and reduces to the Maxwell-Boltzmann distribution, in
q κ − q −κ the classical κ → 0 limit.
lnκ (q) = . (35) The distribution (44), in the global rest frame where
2κ
Uν = (c, 0, 0, 0) and in absence of external forces i.e.
The property (34) of expκ (q), enforced by the Galilei rel- Aν = 0, simplifies as
ativity principle, transforms into the following property f = expκ (−β E + β µ) , (45)
of lnκ (x)
E being the relativistic kinetic energy. This distribution
lnκ (1/x) = − lnκ (x) , (36) at low energies (E → 0) reduces to the classical Maxwell-
Boltzmann distribution i.e. f ∝ exp(−β E), while at
holding also for the ordinary logarithm of classical relativistic energies (E → +∞) presents power-law tails
physics. i.e.
f ∝ E −1/κ , (46)
IV. RELATIVISTIC STATISTICAL in accordance with the experimental evidence. We recall
MECHANICS that the first experimental validation of the distribution
(45), concerns cosmic rays and has been considered in
In the previous section it has been shown that the func- ref. [4]. Recently a computer validation of the same
tions expκ (q) and lnκ (q) emerge as the relativistic gen- distribution, has been considered in refs. [34, 35], where
eralizations of the ordinary exponential and logarithm the relaxation in relativistic plasmas under wave-particle
functions of classical physics. Therefore in the following interaction, has been simulated numerically.
we pose Finally, after posing Λ(f ) = lnκ (f ), in Eq. (11), the
relativistic two-particle correlation function assumes the
Λ(x) = lnκ (x) , (37) form
Λ−1 (x) = expκ (x) . (38) f ⊗ f1 = expκ ( lnκ f + lnκ f1 ) , (47)
After taking into account of the property which in the κ → 0 classical limit, reduces to f f1 , as dic-
tated by the molecular chaos hypothesis. Consequently
d 1 the relationship (47) can be viewed as defining the rela-
x lnκ (x) = lnκ (ǫx) , (39) tivistic extension of the molecular chaos hypothesis.
dx γ
 5

V. CONCLUSIONS mum Entropy Principle, independently on the particular

values of the arbitrary constants ξ1 and ξ2 .
Let us consider the probability distribution function In kinetic theory, customarily it is posed

ξ1 = ξ2 = 1 , (50)


f = ξ1 expκ − ξ2 β [E − µ] , (48)

ξ1 and ξ2 being two arbitrary constants. In the expres-
sion of the distribution function E is the relativistic mi-
croscopic energy, while β and βµ are the Lagrange mul-
tipliers. The Maximum Entropy Principle asserts that
the distribution (48) can be obtained, by maximizing the
entropy
γ
Z
S=− d3 p f lnκ (f / ǫ ξ1 ) , (49)
ξ2
√ ξ2 = γ .
with γ = 1/ 1 − κ2 and ǫ = expκ (γ), under the con-
straints imposing the conservation of the norm of f and
of the mean value of the E. It is remarkable the above
distribution and entropy are linked through the Maxi-
in order to simplify the expression of the distribution
function and of the two particle correlation function, ap-
pearing in kinetic equation. This choice, is made natu-
rally, in the present paper, being the starting point of our
presentation the kinetic equation (6).
On the other hand, in statistical mechanics, in order
to simplify the expression of the entropy, it is posed
ξ1 = 1/ǫ , (51)
(52)
This latter choice has been made for instance in the ref.
[4, 5] where the starting point for the presentation of the
theory was the entropy functional.

[1] V.M. Vasyliunas, J. Geophys. Res. 73, 2839 (1968). Math. and Computation, 218, 2439 (2011).
[2] G. Kaniadakis, Physica A 296, 405 (2001). [31] P. Tempesta, Phys. Rev. E 84, 021121 (2011).
[3] G. Kaniadakis, Phys. Lett. A 288, 283 (2001). [32] Guo Lina, Du Jiulin, and Liu Zhipeng, Phys. Lett. A
[4] G. Kaniadakis, Phys. Rev. E 66, 056125 (2002). 367, 431-435 (2007).
[5] G. Kaniadakis, Phys. Rev. E 72, 036108 (2005). [33] Guo Lina and Du Jiulin, Phys. Lett. A 362, 368-370
[6] G. Kaniadakis, Physica A 365, 17 (2006) (2007).
[7] G. Kaniadakis, Eur. Phys. J. B 70, 3 (2009). [34] G. Lapenta, S. Markidis, A. Marocchino, and G. Kani-
[8] G. Kaniadakis, Eur. Phys. J. A 40, 275 (2009). adakis, Astrophysical Journal 666, 949-954 (2007).
[9] G. Kaniadakis, Europhys. Lett. 92, 35002 (2010). [35] G. Lapenta, S. Markidis, G. Kaniadakis, J. of Stat.
[10] G. Kaniadakis, Phys. Lett. A 375, 356 (2011). Mech., P02024 (2009).
[11] R. Silva, Eur. Phys. J. B 54, 499 (2006). [36] T. Wada, A.M. Scarfone, Eur. Phys. J. B 70, 29 (2009).
[12] R. Silva, Phys. Lett. A 352 17 (2006). [37] T. Wada, Eur. Phys. J. B 73, 287 (2010).
[13] T. Wada, Physica A 340, 126 (2004). [38] G. Kaniadakis, P. Quarati, A.M. Scarfone, Physica A
[14] T. Wada, Contin. Mechan. and Thermod. 16, 263 (2004). 305, 76 (2002).
[15] G. Kaniadakis, A.M. Scarfone, Physica A 340, 102 [39] T.S. Biro, G. Kaniadakis, Eur. Phys. J. B 50, 3 (2006)
(2004). [40] A. Rossani and A.M. Scarfone, J. Phys. A 37, 4955
[16] S. Abe, G. Kaniadakis and A.M. Scarfone, J. Phys. A: (2004).
Math. Gen. 37, 10513 (2004). [41] Guo L.N., Du J.L., Physica A 389, 47-51 (2010).
[17] J. Naudts, Physica A 316, 323 (2002). [42] J.M. Silva, R. Silva, J.A.S. Lima, Phys. Lett. A 372,
[18] J. Naudts, Rev. Math. Phys. 16, 809 (2004). 5754 (2008).
[19] A.M. Scarfone, T. Wada, Progress of Theor. Phys. Suppl. [43] J. C. Carvalho, R. Silva, J.D. do Nascimento jr., and J.
162 45 (2006). R. De Medeiros, Europhysics Letters 84, 59001 (2008).
[20] T. Yamano, Phys. Lett. A 308, 364 (2003). [44] J. C. Carvalho, J.D. do Nascimento jr., R. Silva, and J.
[21] U. Lucia, Physica A 389, 4558 (2010). R. De Medeiros, Astrophysical Journal Letters 696, L48
[22] A. Aliano, G. Kaniadakis, E. Miraldi, Physica B 325, 35 (2009).
(2003). [45] J.C. Carvalho, R. Silva, J.D.Jr. do Nascimento, B.B.
[23] A.P. Santos, R. Silva, J.S. Alcaniz, D.H.A.L. Anselmo, Soares, J.R. De Medeiros, Europhys. Lett. 91, 69002
Phys. Lett. A 375, 352 (2011). (2010).
[24] A.P. Santos, R. Silva, J.S. Alcaniz, D.H.A.L. Anselmo, [46] A.M. Teweldeberhan, H.G. Miller, and R. Tegen, Int. J.
Phys. Lett. A 375, 3119 (2011). Mod. Phys. E 12, 669 (2003).
[25] G. Pistone, Eur. Phys. J. B 70, 29 (2009). [47] F.I.M. Pereira, R. Silva, J.S. Alcaniz, Nucl. Phys. A 828,
[26] G. Kaniadakis, M. Lissia, A.M. Scarfone, Physica A 40, 136 (1009).
41 (2004). [48] M. Cravero, G. Iabichino, G. Kaniadakis, E. Miraldi,
[27] G. Kaniadakis, M. Lissia, A.M. Scarfone, Phys. Rev. E A.M. Scarfone, Physica A 340, 410-417 (2004).
71, 046128 (2005). [49] M. Coraddu, M. Lissia, R. Tonelli, Physica A 365, 252
[28] G. Kaniadakis, A.M. Scarfone, Physica A 305, 69 (2002). (2006).
[29] T. Oikonomou, G.B. Bagci, Rep. Math. Phys. 66, (2010). [50] R. Tonelli, G Mezzorani, F. Meloni, M. Lissia, M.
[30] M.S. Stankovic, S.D. Marinkovic, P.M. Rajkovic, Appl. Coraddu. Prog. Theor. Phys. 115, 23 (2006).

[51] A. Celikoglu A, U. Tirnakli, Physica A 372, 238 (2006).
[52] A.I. Olemskoi, V.O. Kharchenko, V.N. Borisyuk, Physica
A 387, 1895 (2008).
[53] A.I. Olemskoi, S.S. Borysov, I.A. Shuda, Eur. Phys. J. B
77, 219 (2010).
[54] A.Y. Abul-Magd, Phys. Lett. A 361, 450 (2007).
[55] A.Y. Abul-Magd, Eur. Phys. J. B 70, 39 (2009).
[56] T. Wada, H. Suyari, Phys. Lett. A 348, 89 (2006).
[57] F. Topsoe, Physica A 340 11 (2004).
[58] T. Wada, H. Suyari, Phys. Lett. A 368, 199 (2007).
[59] F. Clementi, M. Gallegati, G. Kaniadakis, Eur. Phys. J.
B 57, 187 (2007).
[60] F. Clementi, T. Di Matteo, M. Gallegati, G. Kaniadakis,
Physica A 387, 3201 (2008).
[61] F. Clementi, M. Gallegati, G. Kaniadakis, J. of Stat.
Mech., P02037 (2009).
6
[62] F. Clementi, M. Gallegati, G. Kaniadakis, Empirical Eco-
nomics 39, 559 (2011).
[63] D. Rajaonarison, D. Bolduc, and H. Jayet, Econ. Lett.
86, 13-20 (2005).
[64] D. Rajaonarison, Econ. Lett. 100, 396 (2008).
[65] G. Kaniadakis, M. Lissia, Physica A 340, pages XV-XIX
(2004).
[66] S.R. de Groot, W.A. van Leeuwen, and Ch.G. van Weert,
Relativistic Kinetic Theory: Principle and Applications,
North-Holland, (1980).
[67] C. Cercignani, and G. Medeiros Kremer, The relativistic
Boltzmann equation: theory and applications, Birkhauser
Verlag, (2002).
[68] Kerson Huang, Statistical Mechanics, John Wiley and
Sons, Inc.,(1987).