Journal of Technical Physics, J. Tech. Phys., 48, 1, 3947, 2007.

Polish Academy of Sciences, Institute of Fundamental Technological Research, Warszawa.

Military University of Technology, Warszawa.

THE RELATIVISTIC BOLTZMANN EQUATION

MATHEMATICAL AND PHYSICAL ASPECTS
M. T. D U D Y S K I1

and

M. L. E K I E L J E E W S K A2

1 Modern

Technologies and Filtration Ltd.

Mokotowska 40/21, 00-543 Warsaw, Poland

2 Institute

of Fundamental Technological Research

Polish Academy of Sciences
wietokrzyska 21, 00-049 Warsaw, Poland
e-mail: mekiel@ippt.gov.pl

Mathematical structure of the linearized Boltzmann equation is outlined. Physical interpretation

of the assumed relativistic cross-sections is discussed.

1. Introduction

The relativistic Boltzmann equation, one of the basic equations used by statistical physicists, is currently a subject of interest also for mathematicians. In particular,
asymptotic stability of the relativistic Maxwellian [1] and existence of global solutions
to the relativistic Boltzmann equation with near-vacuum data [2] have been recently
proved. Therefore in this paper some basic properties of the linearized relativistic Boltzmann equation are outlined, valid for a wide class of physically meaningful interactions
between the relativistic particles.
Some time ago, in the theory of relativistic hydrodynamics a deep confusion appeared, due to various formulations of the basic postulates and, consequently, various
sets of both the hydrodynamical variables and equations. Among the best known approaches, one should mention Eckart [3], Landau [4], Israel [5], Israel and Stewart [68], Mller [9] and van Kampen [10] versions of relativistic hydrodynamics.
There is still no definite experimental evidence in favour of any of these hydrodynamical theories. On the other hand, certain relativistic hydrodynamical equations are
needed to be applied both for a description of the quark plasma formed in heavy ion
collisions [1115] and in many models used in astrophysics [16].
To choose the adequate equations in a mathematically rigorous justification of relativistic hydrodynamics on the ground of microscopic theory is of great importance.
However, microscopic theory of relativistic gases and fluids has not been formulated in a
satisfactory way because of the well-known conceptual difficulties [17]. Although the relativistic Boltzmann equation has been extensively investigated and applied [16, 1921],

40

M. T. Dudyski and M. L. EkielJeewska

but even for this well-known microscopic equation the complete mathematical theory
has not been constructed.
Thus in this paper we shall give an outline of our mathematical analysis of the linearized relativistic Boltzmann equation [22, 23], based on the well-known non-relativistic
procedure [2427]. If the connection between the long time asymptotics of the Boltzmann equation and the hydrodynamical description of the system is assumed to hold
in the relativistic theory as well as it holds in the non-relativistic one, then our approach gives a possibility to choose one of the existing formulations of the relativistic
hydrodynamics.
Apart from this physical motivation, mathematical structure of the relativistic Boltzmann equation is interesting by itself. As we shall show, the linearized relativistic Boltzmann equation has a unique, global in time, non-increasing in norm, causal solution in
L2 (r, p) for a wide class of collisional cross-sections.
2. Assumptions

We consider one-component classical relativistic gas of massive, uncharged particles

in the flat space-time. We assume that before collisions, there are no particle correlations
and thus we use a one-particle distribution function and the Boltzmann equation to
determine the state of the system. Moreover, the gas is close to the global equilibrium,
thus its evolution is described by the linearized relativistic Boltzmann equation, which
for an arbitrary inertial observer has the form:
cp f
f
+
= L[f ],

t
p0 r

(2.1)1

with the collision operator L given by:

1/2

(2.1)2

L[f ] =

cf0
p0

d3 p1

d sin (g, )
"
#
f1
gs1/2
f
f1
f
f10
+ 1/2 1/2 1/2 ,
1/2
p10
f10
f0
f10
f0

where
(2.2)

f0 =

n exp (p U /kB T )
4m2 ckB T K2 (mc2 /kB T )
1/2

is the Jttner relativistic equilibrium distribution function and f0 f is the linearized

distribution function.
In Eq. (2.1), the collision between particles with momenta: p and p1 before the
collision, and p , p1 after the collision is described in c.m. frame by:
the total energy:
(2.3)

cs1/2 := c |p1 + p | ,

the relative momentum:

(2.4)

2g := |p1 p | ,

The relativistic Boltzmann equation

41

the angle of scattering:

1
,
cos := 1 2 (p p1 ) (p p ) 4m2 c2 s

(2.5)

and the differential scattering cross-section (g, ).

In Eq. (2.2) K2 is the Bessel function of the second kind of index two; n, T and U are
respectively: density, temperature and velocity of the system in the global equilibrium
state.
In the following we perform the analysis of the linearized relativistic Boltzmann
equation (2.1), called LRBE, in the Hilbert space L2 (r, p) of real-valued functions with
the scalar product given by:
Z
(2.6)
(f |g) := d3 r d3 p f g.
Functions from this space have a simple physical interpretation as describing systems
with finite entropy1) .
We assume that there exist constants , , , B and B such that the scattering
cross-section (g, ) obeys the condition:


(2.7)
(g, ) 6 Bg + B g sin ,

with > 2, 0 < < min (4, 4 + ) and 0 6 < + 2.

As it will be discussed below, the class of functions (g, ) defined by (2.7) seems
to be broad enough to contain all those physically reasonable cross-sections, for which
the Boltzmann equation may be applicable.
For small momentum g, the scattering cross-section (g, ) should of course describe
non-relativistic interactions. It is easy to notice that (g, ) satisfying condition (2.7),
fulfills also the non-relativistic restriction on the scattering cross-section (g, ) given
by Drange [25]:
(V, ) 6 V (),

(2.8)

where is a non-negative function defined on [0, ], bounded in [1 , 2 ] 1 , 2 (0, )

and satisfying the condition:
() < B ,

when

0,

() < B( ) ,

when

with 4 < < + 2 and > 2.

In the non-relativistic physics, the condition (2.7) (or 2.8) can be interpreted in
terms of a simplified intermolecular potential [24]:
(2.9)

.
rn

The classical cross-section corresponding to (2.9) has the form:

(2.10)
1) Entropy

i.e.

(V, ) = V (),

for the linearised Boltzmann equation is to be understood as the linearised entropy [28],
d 3r d 3p f 2.

42

M. T. Dudyski and M. L. EkielJeewska

with = 4/n and () having a non-integrable singularity for grazing collisions, i.e.
for 0. However, the singularity is absent from the quantum treatment [29]. It can
be easily checked [29] that the condition (2.7) is satisfied by all interactions via the
quantum inverse power potential (2.9) with n > 2. However, for n 6 2 collective effects
in the system, making the Boltzmann equation inadequate, cannot be excluded [30, 31].
Thus for small momenta g, the class of adopted functions (g, ) (2.7) seems to be wide
enough.
Equation (2.7) is a simple, but nontrivial generalization of (2.8) for large momenta2) g. Physical interpretation of (2.7) in terms of any potential of course fails in
the relativistic regime. But according to the experimental data on the hadron-hadron
scattering, the elastic cross-section is slowly increasing for momenta g large enough [32].
Field theoretical analysis puts an upper bound (so-called Froissarts restriction) [33]
on the large momentum behaviour of :
Z
2
for
g ,
(2.11)
d sin (g, ) < C[log(g/g0 )]
where C and g0 are constants.
Of course all cross-sections satisfying (2.11) fulfill also condition (2.7).
With all the above listed assumptions we are ready to analyse the LRBE (2.1).
3. Properties of the Equation

Under the assumptions made in the previous section, the LRBE (2.1) may be rewritten in the form [22]:
cp f
f
+
= (p) + K[f ],

t
p0 r

(3.1)

where (p) is the collision frequency:

(3.2)

c
(p) =
p0

d p1

Z
0

d sin

gs1/2
(g, ) f10 ,
p10

and the operator K acts as:

(3.3)

K [f (r, p, t)]
1/2

cf0
=
p0

d p1

Z
0

gs1/2
d sin
(g, ) f10
p10

"

f1
1/2
f10

f
f0 1/2

f1
1/2

f10

Similarly as in the non-relativistic theory [24, 25], to analyse the properties of

the LBRE (3.1) first we must examine the structure of the K operator and the behaviour of the collision frequency . The results are given in the following theorems:
2) Note that using the relativistic relative velocity 4cgs1/2 instead of the relativistic relative momentum 2g to generalize (2.8), we would get a condition dierent than (2.7).

The relativistic Boltzmann equation

43

Theorem 1
The K operator is compact in L2 (p).
Proof is given by us in Ref. [22].
Theorem 2A
Let us assume that > 2, 0 6 < + 2, B > 0 and c0 > 0 so that
(3.4)

(g, ) > B

g +1
sin .
c0 + g

Then the collision frequency obeys:

(3.5)

(p) > 0 [p0 /mc]

/2

Theorem 2B
Let us assume that > 2, 0 6 < 4 and B > 0 so that
(g, ) < B g sin .

(3.6)
Then:
(3.7)

(p) < 0 [p0 /mc]

/2

6 0 ,

where
(3.8)

,
2,

for 0 < < 3,

for 3 < < 4,

+ 1,

for = 3,

and

0 < < 1.

Proofs of theorems 2A and 2B are given in Ref. [22].

The analysis of the collision frequency (p) performed in Theorems 2A and 2B allows
to discriminate between relativistic soft and hard interactions, defined as satisfying
(3.6) or (3.4), respectively. The basic properties of the LRBE (such as existence and
uniqueness of solutions, provided by Theorem 3) can be obtained for the whole class
of adopted interactions (2.7). But from the physical point of view one expects that for
arbitrary initial values f (r, p) L2 (r, p), the solution should decay in time in such a
way that asymptotically the hydrodynamical approximation holds. This property can
be proved (see Theorem 4) for hard interactions only [23].
Theorem 3
Assume that (g, ) satisfies (2.7).
Then the operator B = p/p0 + L generates a strongly continuous contraction
semigroup on L2 (r, p) given explicitly as:
Z
1
d3 k exp(ik r) exp(Bk t) f (k p),
(3.9)
exp(tB)f (r, p) =
(2)3/2
where Bk = ikp/p0 +L is in general an unbounded operator on L2 (p) with the domain:
(3.10)

D(Bk ) = {f (p) L2 (p) : (p)f (p) L(p)}.

44

M. T. Dudyski and M. L. EkielJeewska

Proof of this theorem is given in Ref. [22].

The Cauchy problem for the LBRE is then solved by the linear Boltzmann semigroup as:
(3.11)

f (t) = exp(tB) f (t = 0).

We see immediately from Theorem 3 that f (t) is unique and that:

(3.12)

||f (t)||L2 (r,p) 6 ||f (0)||L2 (r,p) .

As we have shown in [34, 35], the solution is causal.

Theorem 4
Assume that (g, ) satisfies (2.7) and (3.4). Then there exist , 1 , 2 > 0 such that
for any f D(Bk ), the following properties are satisfied :
i) For any k, |k| 6 ,
(3.13)

exp[tBk ]f =

5
X

exp[tj (k)]hej (k)|f iL2 (p) ej (k)

j=1

+ exp [tAk ] f + exp [t1 ] Z1 (k, t)f,

where j and ej are eigenvalues and eigenvectors of Bk and for |k| < they have
analytical expansion in k as:
(3.14)

j (k) =

3
X

n
aj,n (i|k|) + o |k|9 ,

3
X



n
9
ej,n (k/|k|) (i|k|) + o |k| ,

n=1

(3.15)

ej (k) =

n=0

aj,n are constants with aj,2 > 0, ej,n are functions of (k/|k|) and
(3.16)

hei (k)|ej (k)iL2 (p) = i,j

i, j = 1, . . . , 5.

ii) For any k, |k| > ,

(3.17)

exp[tBk ]f = exp[tAk ]f + exp[t2 ]Z2 (k, t)f,

where
(3.18)

(3.19)

Ak = (p) + ikp/p0 ,

1
Zj (k, t)f = lim
2

Zi

exp(it) Z(j + i; k) f d,

The relativistic Boltzmann equation

(3.20)

Z(, k) = ( Ak )

and
(3.21)

45

1
1
K( Ak )
I K( Ak )1

|Zj (k, t)f |L2 (p) 6 C|f |L2 (p) ,

with C independent of k and t.

Proof is given in Ref. [23].

Theorem 4 has a very important physical interpretation. From (3.13) and (3.17) it is
clear that for long times, the solution of the Boltzmann equation is well approximated
by its hydrodynamical part given as:
(3.22)

fH =

5
X

exp [tj (k)] hej (k)|f iL2 (p) ej (k).

j=1

For small k one can perform the perturbation expansion of (3.22) and then use it to
obtain a set of hydrodynamical equations [23].
4. Conclusions

After tedious estimations [22] the basic mathematical structure of the linearised relativistic Boltzmann equation turned out to be the same as that of the non-relativistic
one [24]. Namely, for a physically reasonable class of scattering cross-sections, the collision operator was proved to have a form of the multiplication operator (p) plus the
compact in L2 (p) perturbation K. However, the analysis of the collision frequency (p)
led to a discrimination between the relativistic soft and hard intermolecular interactions
different than in the non-relativistic case. Namely, there is a large class of functions ,
which corresponds to scattering cross-sections for relativistic soft, but non-relativistic
hard interactions [22].
The basic structure of the collision operator was used to prove existence, uniqueness
and causality of the solution to the linearized relativistic Boltzmann equation in L2 (r, p)
[22, 34, 35].
For the relativistic hard interactions the solution was proved to converge asymptotically in time to its hydrodynamical part [23], similarly as it had been done in the
non-relativistic theory [26, 27]. This gives a possibility to derive rigorously a set of
relativistic hydrodynamical equations from the microscopic equation, using the perturbation expansion in small wave number k. Note that relativistic perturbation expansion
in small wave number k is analytical, unlike the non-relativistic one. This difference can
be easily understood, because the relativistic perturbation operator ikp/p0 is bounded
in p, contrary to its non-relativistic counterpart ikp.
References
1. L. Hsiao, H.J. Yu, Asymptotic stability of the relativistic Maxwellian, Math. Method Appl. Sci.,
29, 14811499, 2006.
2. R.T. Glassey, Global Solutions to the Cauchy Problem for the Relativistic Boltzmann Equation
with Near-Vacuum Data, Commun. Math. Phys., 264, 705724, 2006.
3. C. Eckart, The Thermodynamics of Irreversible Processes. III. Relativistic Theory of the Simple
Fluid, Phys. Rev., 58, 919924, 1940.

46

M. T. Dudyski and M. L. EkielJeewska

4. L.D. Landau, E.M. Lifshitz, Fluid Mechanics, Pergamon Press, Oxford 1959.
5. W. Israel, Relativistic Kinetic Theory of a Simple Gas, J. Math. Phys., 4, 1163, 1963.
6. W. Israel, J.M. Stewart, Thermodynamics of nonstationary and transient eects in a relativistic gas, Phys. Letters, 58A, 213215, 1976.
7. W. Israel, J. M. Stewart, Transient relativistic thermodynamics and kinetic theory, Ann. Phys.
(N.Y.), 118, 341372, 1979.
8. W. Israel, J. M. Stewart, On Transient relativistic thermodynamics and kinetic theory. II,
Proc. R. Soc. London, A 365, 4352, 1979.
9. I. Shih Liu, I. Mller and T. Ruggeri, Relativistic thermodynamics of gases, Ann. Phys., 169,
191219, 1986.
10. N. G. van Kampen, Chapman-Enskog as an application of the method for eliminating fast variables, J. Stat. Phys., 46, 709727, 1987.
11. U. Heinz, Kinetic Theory for Plasmas with Non-Abelian Interactions, Phys. Rev. Lett., 51, 351
354, 1983.
12. D. D. Holm, B. A. Kupershmidt, Relativistic chromohydrodynamics and Yang-Mills Vlasov
plasma, Phys. Letters A, 105, 225228, 1984.
13. K. Kajantie, Hydrodynamics and approach to equilibrium in quark-gluon plasma, Nuclear
Physics A, 418, 4155, 1984.
14. G. Baym, [in:] Quark Matter 84, K. Kajantie [Ed.], Lecture Notes in Physics, Vol. 221, Springer,
Berlin 1985.
15. O. K. Kalashnikov, [in:] Quantum Field Theory and Quantum Statistics, [Ed.] I.A. Batalin et al.,
Vol. 1, p. 499, Adam Hilger, Bristol 1987.
16. D. Mihalas, B. W. Mihalas, Foundations of radiation hydrodynamics, Oxford University Press,
1984.
17. P. G. Bergmann, Generalized statistical mechanics, Phys. Rev., 84, 10261033, 1951.
18. P. Havas, [in:] Statistical mechanics of equilibrium and non-equilibrium, J. Meixner [Ed.], North
Holland, Amsterdam 1965.
19. J. L. Synge, The relativistic gas, North Holland, Amsterdam 1957.
20. J. M. Stewart, Non-equilibrium relativistic kinetic theory. Lecture Notes in Physics, Vol. 10.
Springer, Berlin 1971.
21. S. R. de Groot, W. A. van Leeuwen, Ch. G. van Weert, Non-Equilibrium relativistic kinetic
theory-principles and applications. North Holland, Amsterdam 1980.
22. M. Dudyski, M. L. Ekiel-Jeewska, On the linearized relativistic Boltzmann equation, Commun. Math. Phys., 115, 607, 1988.
23. M. Dudyski, On the linearized relativistic Boltzmann equation. II. Existence of hydrodynamics,
J. Stat. Phys., 57, 199, 1989.
24. H. Grad, Asymptotic theory of the Boltzmann equation, II, [in:] Third international rareed gas
symposium, J.A. Lauremann [Ed.], Academic Press, 1, p. 26, New York 1963.
25. H. B. Drange, The linearized boltzmann collision operator for cut-o potentials, SIAM J. Appl.
Math., 29, 665676, 1975.
26. R. S. Ellis, M. A. Pinsky, The rst and second uid approximations to the linearized Boltzmann
equation, J. Math. Pures et Appl., 54, 125156, 1975, and The projection of the Navier-Stokes
equations upon the Euler equations, J. Math. Pures et Appl., 54, 157182, 1975.
27. T. Nishida, Fluid dynamical limit of the nonlinear Boltzmann equation to the level of the compressible Euler equation, Commun. Math. Phys., 61, 119148, 1978.
28. H. Grad, [in:] Rareed Gas Dynamics, F. M. Devienne [Ed.], Pergamon Press, London 1960.

The relativistic Boltzmann equation

47

29. L. D. Landau, E. M. Lifshitz, Quantum MechanicsNon-Relativistic Theory, Part 1, Sec. 106

and 108, Pergamon Press, Oxford 1965.
30. M. Klaus, The linear Boltzmann operator-spectral properties and short-wavelength limit, Helv.
Phys. Acta, 48, 99129, 1975 and Boltzmann collision operator without cut-o, Helv. Phys. Acta,
50, 893903, 1977.
31. Young-Ping Pao, Boltzmann collision operator with inverse-power intermolecular potentials,
I, Comm. Pure Appl. Math., 27, 407428, and Boltzmann collision operator with inverse-power
intermolecular potentials, II Comm. Pure Appl. Math., 559581, 1974.
32. G. Siacomelli, Total cross sections and elastic scattering at high energies, Phys. Reports, 23,
123235, 1976.
33. B.H. Bransden, D. Evans, J.V. Major, The fundamental particles, Van Nostrand Reinhold
Company, London 1973.
34. M. Dudyski, M.L. Ekiel-Jeewska, Causality of the linearized relativistic Boltzmann equation, Phys. Rev. Lett., 55, 28312834, 1985.
35. M. Dudyski, M.L. Ekiel-Jeewska, Causality problem in the relativistic kinetic theory, [in:]
Recent developments in nonequilibrium thermodynamics: Fluids and related topics, J. Casas
Vzquez et. al. [Ed.], Lecture Notes in Physics, Vol. 253. Springer, Berlin 1986.

Received October 10, 2006; revised version February 6, 2007.