306 Boltzmann Equation (Classical and Quantum)

Boltzmann Equation (Classical and Quantum)

M Pulvirenti, Università di Roma ‘‘La Sapienza,’’ As fundamental features of eqn [1], we have the
Rome, Italy conservation in time of the following five quantities
ª 2006 Elsevier Ltd. All rights reserved.
Z Z
dx dv f ðx; v; tÞv ½4

with  = 0, 1, 2, expressing conservation of the

Introduction probability, momentum, and energy.
R R
From now on we shall set = R3 for notational
Ludwig Boltzmann (1872) established an evolution simplicity.
equation to describe the behavior of a rarefied gas, Moreover, Boltzmann introduced the (kinetic)
starting from the mathematical model of elastic balls entropy defined as
and using mechanical and statistical considerations. Z Z
The importance of this equation is twofold. First, it Hðf Þ ¼ dx dv f log f ðx; vÞ ½5
provides a reduced description (as well as the
hydrodynamical equations) of the microscopic and proved the famous H-theorem asserting the
world. Second, it is also an important tool for the decreasing of H(f (t)) along the solutions to eqn [1].
applications, especially for dilute fluids when the Finally, in the case of bounded domains or
hydrodynamical equations fail to hold. homogeneous solutions (f = f (v; t) is independent of
The starting point of the Boltzmann analysis is to x), the distribution defined for some  > 0,  > 0,
abandon the study of the gas in terms of the detailed and u 2 R 3 by
motion of molecules which constitute it because of
 2
their large number. Instead, it is better to investigate Mðx; vÞ ¼ 3=2
eð=2Þjvuj ½6
a function f (x, v), which is the probability density of ð2=Þ
a given particle, where x and v denote its position called Maxwellian distribution, is stationary for the
and velocity. Actually, f (x, v)dx dv is often confused evolution given by eqn [1]. In addition, M minimizes
with the fraction of molecules falling in the cell of H among all distributions with given total mass ,
the phase space of size dx dv around x, v. The two given mean velocity u, and mean energy. The
concepts are not exactly the same, but they are parameter  is interpreted as the inverse
asymptotically equivalent (when the number of temperature.
particles is diverging) if a law of large numbers holds. In conclusion, Boltzmann was able to introduce
The Boltzmann equation is the following: not only an evolutionary equation with the remark-
ð@t þ v  rx Þf ¼ Qðf ; f Þ ½1 able properties expressing mass, momentum, and
energy conservation, but also the trend to the
where Q, the collision operator, is defined by eqn [2]: thermal equilibrium. In other words, he tried to
Z Z
conciliate the Newton’s laws with the second
Qðf ; f Þ ¼ dv1 dnðv  v1 Þ  n principle of thermodynamics.
R3 S2þ

 ½ f ðx; v Þf ðx; v01 Þ  f ðx; vÞf ðx; v1 Þ

0
½2
The Boltzmann Heuristic Argument
and
Thus, we want to find an evolution equation for the
v0 ¼ v  n½n  ðv  v1 Þ quantity f (x, v; t). The molecular system we are
½3
v01 ¼ v1 þ n½n  ðv  v1 Þ considering consists of N identical particles of
diameter r in the whole space R3 . We denote by
Moreover, n (the impact parameter) is a unitary x1 , v1 , . . . , xN , vN a state of the system, where xi and
vector and S2þ = {njn  (v  v1 )  0}. vi indicate the position and the velocity of the
Note that v0 , v01 are the outgoing velocities after a particle i. The particles cannot overlap (i.e., the
collision of two elastic balls with incoming velocities centers of two particles cannot be at a distance
v and v1 and centers x and x þ rn, r being the smaller than the particle diameter r).
diameter of the spheres. Obviously, the collision The particles are moving freely up to the first
takes place if n  (v  v1 )  0. Equations [3] are a instance of contact, that is, the first time when two
consequence of the conservation of total energy, particles (say particles i and j) arrive at a distance r.
momentum, and angular momentum. Note also that Then the pair interacts when an elastic collision
r does not enter in eqn [1] as a parameter. occurs. This means that they change instantaneously
 Boltzmann Equation (Classical and Quantum) 307

their velocities, according to the conservation of that we have to integrate over the hemisphere
the energy and linear and angular momentum. S2þ = {(v2  v)  n > 0}:
More precisely, the velocities after a collision Z
with incoming velocities v and v1 are those given 2
G ¼ ðN  1Þr dv2
by formula [3]. After the first collision, the Z
system evolves by iterating the procedure. Here  dn f2 ðx; v; x þ nr; v2 Þjðv2  vÞ  nj ½11
we neglect triple collisions because they are Sþ
unlikely. The evolution equation for a tagged
Summing G and L, we get
particle is then of the form
Z
ð@t þ v  rx Þf ¼ Coll ½7 Coll ¼ ðN  1Þr2 dv2
Z
where Coll denotes the variation of f due to the  dn f2 ðx; v; x þ nr; v2 Þðv2  vÞ  n ½12
collisions.
We have which, however, is not a very useful expression
Coll ¼ G  L ½8 because the time derivative of f is expressed in terms
of another object, namely f2 . An evolution equation
where L and G (the loss and gain terms, respectively) for f2 will imply f3 , the joint distribution of three
are the negative and positive contributions to the particles, and so on, up to we include the total
variation of f due to the collisions. More precisely, particle number N. Here the basic main assumption
L dx dv dt is the probability of the test particle to of Boltzmann enters, namely that two given particles
disappear from the cell dx dv of the phase space are uncorrelated if the gas is rarefied, namely
because of a collision in the time interval (t, t þ dt)
and Gdx dv dt is the probability to appear in the f ðx; v; x2 ; v2 Þ ¼ f ðx; vÞf ðx2 ; v2 Þ ½13
same time interval for the same reason. Let us
Condition [13], referred to as the propagation of
consider the sphere of center x with radius r and a
chaos, seems contradictory at first sight: if two
point x þ rn over the surface, where n denotes the
particles collide, correlations are created. Even though
generic unit vector. Consider also the cylinder with
we could assume eqn [13] at some time, if the test
base area dS = r2 dn and height jVjdt along the
particle collides with particle 2, such an equation
direction of V = v2  v.
cannot be satisfied anymore after the collision.
Then a given particle (say particle 2) with velocity
Before discussing the propagation of chaos
v2 can contribute to L because it can collide with the
hypothesis, we first analyze the size of the collision
test particle in the time dt, provided it is localized in
operator. We remark that, in practical situations
the cylinder and if V  n  0. Therefore, the contri-
for a rarefied gas, the combination Nr3  104 cm3
bution to L due to the particle 2 is the probability of
(i.e., the volume occupied by the particles) is very
finding such a particle in the cylinder (conditioned to
small, while Nr2 = O(1). This implies that G = O(1).
the presence of the first particle in x). This quantity is
Therefore, since we are dealing with a very large
f2 (x, v, x þ nr, v2 ) j (v2  v)  njr2 dn dv2 dt, where f2
number of particles, we are tempted to perform the
is the joint distribution of two particles. Integrating in
limit N ! 1 and r ! 0 in such a way that
dn and dv2 , we obtain that the total contribution to
r2 = O(N1 ). As a consequence, the probability that
L due to any predetermined particle is
Z Z two tagged particles collide (which is of the order of
the surface of a ball, i.e., O(r2 )) is negligible.
r2 dv2 dn f2 ðx; v; x þ nr; v2 Þjðv2  vÞ  nj ½9
S2 However, the probability that a given particle
performs a collision with any one of the remaining
where S2 is the unit hemisphere (v2  v)  n < 0. N  1 particles (which is O(Nr2 ) = O(1)) is not
Finally, we obtain the total contribution multiplying negligible. Therefore, condition [13] is referring to
by the total number of particles: two preselected particles (say particles 1 and 2), so
Z
that it is not unreasonable to conceive that it holds
L ¼ ðN  1Þr2 dv2 in the limiting situation in which we are working.
Z However, we cannot insert [13] in [12] because
 dn f2 ðx; v; x þ nr; v2 Þjðv2  vÞ  nj ½10 this latter equation refers to instants before and after
S
the collision and, if we know that a collision took
The gain term can be derived analogously by place, we certainly cannot invoke eqn [13]. Hence, it
considering that we are looking at particles which is more convenient to assume eqn [13] in the loss
have velocities v and v2 after the collisions so term and work over the gain term to keep advantage
308 Boltzmann Equation (Classical and Quantum)

of the factorization property which will be assumed a two-body interaction V = V(r), the resulting
only before the collision. Boltzmann equation is eqn [1], with
Coming back to eqn [11] for the outgoing pair Z Z
 
velocities v, v2 (satisfying the condition (v2  v)  n > 0), Qðf ; f Þ ¼ dv1 dn Bðv  v1 ; nÞ f 0 f10  ff1 ½17
we make use of the continuity property S2þ
  where we are using the usual shorthand notation:
f2 ðx; v; x þ nr; v2 Þ ¼ f2 x; v0 ; x þ nr; v02 ½14
 
where the pair v0 , v02 is pre-collisional. On f2 f 0 ¼ f ðx; v0 Þ; f10 ¼ f x; v01 ; f ¼ f ðx; vÞ;
½18
expressed before the collision, we can reasonably f1 ¼ f ðx; v1 Þ
apply condition [13] and obtain and B = B(v  v1 ; n) is a suitable function of the
Z Z relative velocity v  v1 and the impact parameter n,
2
G  L ¼ ðN  1Þr dv2 dnðv  v2 Þ  n which is proportional to the cross section relative to
S2þ
  the potential V. Another equivalent, sometimes
 ½f ðx; v Þf x  nr; v02
0
more convenient, way, to express eqn [17] is
 f ðx; vÞf ðx þ nr; v2 Þ ½15 Z Z Z
0
 
Qðf ; f Þ ¼ dv1 dv dv01 W v; v1 jv0 ; v01
after a change n ! n in the gain term, using the
 0 0 
notation S2þ for the hemisphere {nj = (v2  v)  n  0}. f f1  ff1 ½19
This transforms the pair v0 , v02 from a pre-collisional
to a post-collisional pair. with
Finally, in the limit N ! 1, r ! 0, Nr2 = 1 , we  
W v; v1 jv0 ; v01
find    
¼ w v; v1 jv0 ; v01   v þ v1  v0  v01
 
ð@t þ v  rx Þf 2  2
Z Z   12 v2 þ v21  ðv0 Þ  v01 ½20
¼ 1 dv2 dnðv  v2 Þ  n
Sþ where w is a suitable kernel. All the qualitative
  properties, such as the conservation laws and the
 ½f ðx; v0 Þf x; v02  f ðx; vÞf ðx; v2 Þ ½16
H-theorem, are obviously still valid.
The parameter , called mean free path, represents,
roughly speaking, the typical length a particle can
cover without undergoing any collision. In eqns [1] Consequences
and [2], we just chose  = 1. The Boltzmann equation provoked a debate involving
Equation [16] (or, equivalently, eqns [1] and [2]) is Loschmidt, Zermelo, and Poincaré, who outlined
the Boltzmann equation for hard spheres. Such an inconsistencies between the irreversibility of the equa-
equation has a statistical nature, and it is not tion and the reversible character of the Hamiltonian
equivalent to the Hamiltonian dynamics from which dynamics. Boltzmann argued the statistical nature of
it has been derived. Indeed, the H-theorem shows that his equation and his answer to the irreversibility
such an equation is not reversible in time as expected paradox was that ‘‘most’’ of the configurations behave
of any law of mechanics. as expected by the thermodynamical laws. However,
This concludes the heuristic preliminary analysis of he did not have the probabilistic tools for formulating
the Boltzmann equation. We certainly know that the in a precise way the statements of which he had a
above arguments are delicate and require a more precise intuition.
rigorous and deeper analysis. If we want the Boltzmann Grad (1949) stated clearly the limit N ! 1,
equation not to be a phenomenological model, derived r ! 0, Nr2 ! const:, where N is the number of
by ad hoc assumptions and justified only by its particles and r is the diameter of the molecules, in
practical relevance, but rather that it is a consequence which the Boltzmann equation is expected to hold.
of a mechanical model, we must derive it rigorously. In This limit is usually called the Boltzmann–Grad limit
particular, the propagation of chaos should be not a (B–G limit in the sequel).
hypothesis but the statement of a theorem. The problem of a rigorous derivation of the
Boltzmann equation was an open and challenging
problem for a long time. Lanford (1975) showed that,
Beyond the Hard Spheres
although for a very short time, the Boltzmann equation
The heuristic arguments we have developed so far can be derived starting from the mechanical model of the
can be extended to different potentials than that of hard-sphere system. The proof has a deep content but is
the hard-sphere systems. If the particles interact via relatively simple from a technical viewpoint.
 Boltzmann Equation (Classical and Quantum) 309

Z
Existence 1 3 1
v2 M dv ¼ T þ u2 ½25
The mathematical study of the Boltzmann equation 2 2 2
starts with the problem of proving the existence of Moreover, the only solution to the equation
the solutions. One would like to be able to show that, Z
for all (or at least for a physically significant family hðvÞQðf ; f Þ dv ¼ 0 ½26
of) initial distributions (which are positive and
summable functions) with finite momentum, energy, is any linear combination of the quantities (1, v, v2 ),
and entropy, there exists a unique solution to eqn [1] called collision invariants. The last property
with the same mass, momentum, and energy as of the obviously corresponds to the mass, momentum,
initial distribution. Moreover, the entropy should and energy conservation.
decrease and the solution should approach the right With this in mind, consider a change of
Maxwellian as t ! 1. The problem, in such a variables in the Boltzmann equation [1], passing
generality, is still unsolved, but several results in this from microscopic to macroscopic variables,
direction have been achieved since the pioneering x ! "x, t ! "t. Here " is a small scale parameter
works due to Carleman (1933) for the homogeneous expressing the ratio between the typical inter-
equation. Actually, there are satisfactory results for particle distances and the typical distances over
some special situations, such as the homogeneous which the macroscopic equations are varying.
solutions (independent of x) close to the equilibrium, Such a change yields
to the vacuum, or to homogeneous data. The most
1
general result we have up to now is, unfortunately, ð@t þ v  rx Þf" ¼ Qðf" ; f" Þ ½27
not constructive. This is due to Di Perna and Lions "
(1989), who showed the existence of suitable weak We need to allow the small parameter " (mean free
solutions to eqn [1]. However, we still do not know path or the Knudsen number) to tend to zero. In
whether such solutions, which preserve mass and order to eliminate the singularity on the right-hand
momentum, and satisfy the H-theorem, are unique side of [27], we multiply both sides by the collision
and also preserve the energy. invariants v with  = 0, 1, 2; and obtain the five
equations:
Z
dv v ð@t þ v  rx Þf" ¼ 0 ½28
Hydrodynamics
The derivation of hydrodynamical equations from On the other hand, if f" converges to f, as " ! 0,
the Boltzmann equation is a problem as old as the necessarily Q(f , f ) = 0 and hence f = M. Therefore,
equation itself and, in fact, it goes back to Maxwell we expect that in the limit " ! 0,
and Hilbert. Preliminary to the discussion of the Z
hydrodynamic limit, we establish a few properties of dv v ð@t þ v  rx ÞM ¼ 0 ½29
the collision kernel.
It is a well-known fact that the only solution to Equation [29] fixes a relation among the fields , u, T
the equation as functions of x and t. A standard computation gives
us the Euler equations for compressible gas
Qðf ; f Þ ¼ 0 ½21
@t  þ divðuÞ ¼ 0 ½30
is a local Maxwellian, namely

f ðx; vÞ :¼ Mðx; vÞ 1
@t u þ ðu  rÞu þ rp ¼ 0 ½31

ðxÞ 2
¼ 3=2
ejvuðxÞj =2TðxÞ ½22
ð2TðxÞÞ
@t T þ ðu  rÞT þ 23Tru ¼ 0 ½32
where the local parameters , u, and T satisfy the
where the pressure p is related to the density  and
relations
the temperature T by the perfect gas law
Z
M dv ¼  ½23 p ¼ T ½33

Z In order to make the above arguments rigorous,

Hilbert (1916) developed a useful tool, called the
vM ¼ u ½24
Hilbert expansion, to control the limiting procedure.
310 Boltzmann Equation (Classical and Quantum)

Namely, he expressed a formal solution to eqn [27] the upstream and the downstream values of the
in the form of a power series expansion: densities, mean velocities, and temperatures. Such
X relations are known in gas dynamics as the
f" ¼ fj "j ½34 Rankine–Hugoniot conditions. A solution of this
j0
problem has been found by Caflisch and Nikolaenko
where f0 is the local Maxwellian, with the para- (1983) in case of a weak shock (namely, when Mþ
meters , u, T satisfying the Euler equations. All the and M are close) by using Hilbert expansion
other coefficients fj of the developments can be techniques. More recently, Liu and Yu (2004)
determined by recurrence, inverting suitable opera- established also stability and positivity of this
tors. However, the series is not expected to be solution.
convergent, so that the way to show the validity of
the hydrodynamical limit rigorously is to truncate
the expansion and to control the remainder. The Quantum Kinetic Theory
first result in this direction was obtained by Caflisch
(1980). However, this approach is based on the Uehling and Uhlembeck (1933) introduced the
regularity of the solutions to the Euler equations, following kinetic equation for describing a large
which is known to hold only for short times since system of weakly interacting bosons or fermions:
Z Z Z
shocks can be formed. How to approximate the 0
 
shocks in terms of a kinetic description is still a ð@t þ v  rx Þf ¼ dv1 dv dv01 W v; v1 jv0 ; v01
difficult and open problem.  fð1 f Þð1 f1 Þf 0 f10
Note that the hydrodynamical picture of the 0
 
Boltzmann equation just means that we are looking  ð1 fÞ 1 f10 ff1 g ½36
at the solutions of this equation at a suitable Here the þ/ sign, stand for bosons/fermions,
macroscopic scale. The rarefaction hypothesis respectively, and
underlying the Boltzmann description is reflected in  
the law of perfect gas, which states that the W v; v1 jv0 ; v01
 
particles, in the local thermal equilibrium, are free. ^ 0  v1 ÞÞ2  v þ v1  v0  v0
^ 0  vÞ  Vðv
¼ ðVðv 1
  2 
  12 v2 þ v21  ðv0 Þ2  v01 ½37
Stationary Problems Moreover,
Stationary non-Maxwellian solutions to the Z
^
VðpÞ ¼ 4 dx eipx ½38
Boltzmann equation should describe stationary
nonequilibrium states exhibiting nontrivial flows.
In spite of the physical relevance of these problems, where V is the interaction potential. Note that eqn
not many complete mathematical results are, at the [37] is the expression of the cross section of a
moment, available. Among them, there is the quantum scattering in the Born approximation.
traveling-wave problem, which can be formulated The unknown f = f (x, v; t) in eqn [37] is the expected
in the following way. We look for a solution number of molecules falling in the unit (quantum) cell
f = f (x  ct, v), f : R  R 3 ! Rþ , constant in form of the phase space. This function is proportional to the
but traveling with a constant velocity c > 0, to one-particle Wigner function, introduced by Wigner
(1932) to handle kinetic problems in quantum
ðv1  cÞf 0 ¼ Qðf ; f Þ ½35 mechanics, and defined as (setting h = 1):
0
where v1 is the first component of v and f denotes Z
1  
the spatial derivative of f. Equation [35] must be 3
dy eiyv  x þ 12 y; x  12 y
ð2Þ
complemented by the boundary conditions which
are f ! M , as x ! 1, where M are the right where (x; z) is the kernel of a one-particle density
and left Maxwellians, namely two prescribed equili- matrix. Basically, the Wigner function is an equiva-
brium situations at infinity. The parameters (density, lent way to describe a state of a quantum system.
mean velocity, and temperature) of the Maxwel- For instance, eqn [40] below expresses the equili-
lians, however, cannot be chosen arbitrarily. Indeed, brium distributions for bosons and fermions in
the conservations of the mass, momentum, and terms of Wigner functions. In general, the Wigner
energy (which are properties of Q) imply the functions, due to the uncertainty principle, are real
conservations (in x) of the fluxes of these quantities. but not necessarily positive; however, the integral
Hence, we have to impose five equations that relate with respect to x and v gives the probability
 Boltzmann Equation (Classical and Quantum) 311

distributions of the velocity and the position, limit, which consists in scaling space and time and the
respectively. In the kinetic regime, in which we are interaction potential as
interested, the scales are mesoscopic, namely the pffiffiffi
typical quantum oscillations are on a scale much x ! "x; t ! "t; ! " ½43
smaller than the characteristic scales of the problem,
where "1 = N 1=3 is a parameter diverging when the
so that we expect that f should be a genuine
number of particles N tends to infinity.
probability distribution, since the Heisenberg
We mention, incidentally, that under such a
principle does not play an essential role. However,
scaling, a classical system is described by a transport
the interaction occurs on a microscopic scale, so that
equation, called Fokker–Planck–Landau equation,
we expect that the statistics play a role in addition with a diffusion operator in the velocity space.
to the quantum rules for the scattering. The B–G limit considered for classical particle
In this framework, the entropy functional is
systems is different from that considered here
Z Z
for weakly interacting quantum systems. It is actually
Hðf Þ ¼ dx dv ½ f ðx; vÞ log f ðx; vÞ equivalent to rescaling space and time according to
ð1 f ðx; vÞÞ logð1 f ðx; vÞÞ ½39 x ! "x; t ! "t ½44
It is decreasing along the solutions to eqn [35] and it is leaving the interaction unscaled but, in order to
also minimized (among the distributions with given control the total interaction, we make the density
mass, momentum, and energy) by the equilibria diverging gently as "1 = N 1=2 .
z A quantum system under such a scaling is expected to
MðvÞ ¼ 2 ½40 be described by a Boltzmann equation [1] with the
eð=2Þjvuj z collision operator Q computed with the full quantum
namely the Bose–Einstein and the Fermi–Dirac cross section. Now we do not have any effect of the
distributions, respectively. Here  > 1 and z > 0 statistics because in this rarefaction limit these correc-
are the inverse temperature and the activity, respec- tions disappear. On the other hand, the cross section is
tively. Note that, for the Bose–Einstein distribution, that arising from the analysis of the quantum scattering.
z < 1. This creates, in a sense, an inconsistency with Since we do not rescale the interaction, all the other
eqn [36]. Indeed, assuming u = 0 and an initial terms in the Born expansion of the cross section play a
distribution f = f0 (v) with the density larger than the role. This kind of Boltzmann equation is a good
maximal density allowed by eqn [40], namely description of a rarefied gas in which quantum effects
Z are not negligible.
1
c :¼ dv ð=2Þv2 ½41
e 1 See also: Adiabatic Piston; Evolution Equations: Linear
and Nonlinear; Gravitational N-Body Problem (Classical);
it cannot converge to any equilibrium. In order to Interacting Particle Systems and Hydrodynamic
overcome this difficulty related to the Bose con- Equations; Kinetic Equations; Multiscale Approaches;
densation, one can enlarge the definition of the Nonequilibrium Statistical Mechanics: Dynamical
equilibria family by setting Systems Approach; Quantum Dynamical Semigroups.

1
MðvÞ ¼ þ ðvÞ ½42
eð=2Þv2 1
Further Reading
to take care of excess of mass by means of a condensate
Balesku R (1978) Equilibrium and Nonequilibrium Statistical
component. However, it is not clear whether eqn Mechanics. Moscow: Mir (distributed by Imported Publica-
[36] can actually describe the Bose condensation tions, Chicago, Ill).
since its derivation from the Schrödinger equation Caflisch RE (1980) The fluid dynamical limit of the nonlinear
requires, just from the very beginning, the existence of Boltzmann equation. Communications of Pure and Applied
bosonic quasifree states which can be constructed only Mathematics 33: 651–666.
Caflisch RE and Nicolaenko B (1983) Shock waves and the
if the density is moderate. Further analyses are certainly Boltzmann equation. Nonlinear partial differential equations.
needed to clarify the situation. A rigorous derivation of Contemporary Mathematics 17: 35–44.
the Uehling and Uhlembeck equation is, up to now, far Carleman T (1933) Sur la théorie de l’équation intégro-differentielle
from being obtained even for short times; nevertheless, de Boltzmann. Acta Mathematica 60: 91–146.
such an equation is extensively used in the applications. Cercignani C (1998) Ludwig Boltzmann. The Man Who Trusted
Atoms. Oxford: Oxford University Press.
Equation [36] concerns a weakly interacting gas of Cercignani C, Illner R, and Pulvirenti M (1994) The Mathema-
quantum particles. From a mathematical viewpoint, it tical Theory of Dilute Gases. Springer Series in Applied
is expected to be valid in the so-called weak-coupling Mathematics, vol. 106. New York: Springer.
312 Bose–Einstein Condensates

Di Perna RJ and Lions P-L (1989) On the Cauchy problem for the Liu T-P and Yu S-H (2004) Boltzmann equation: micro–macro
Boltzmann equations: Global existence and weak stability. decompositions and positivity of shock profiles. Communica-
Annals of Mathematics 130: 321–366. tions in Mathematical Physics 246(1): 133–179.
Grad H (1949) On the kinetic theory of rarified gases. Spohn H (1994) Quantum kinetic equations. In: Fannes M, Maes C,
Communications in Pure and Applied Mathematics and Verbeure A (eds.) On Three Levels: Micro, Meso and Macro
2: 331–407. Approaches in Physics. New York: Plenum.
Hilbert D (1916) Begründung der Kinetischen Gastheorie. Uehling EA and Uhlembeck GE (1933) Transport phenomena in
Mathematische Annalen 72: 331–407. Einstein–Bose and Fermi–Dirac gases. I. Physical Reviews
Lanford OE III (1975) Time evolution of large classical systems. 43: 552–561.
In: Ehlers J, Hepp K, and Weidenmüller HA (eds.) Lecture Wigner EP (1932) On the quantum correction for thermodynamic
Notes in Physics, vol. 38, pp. 1–111. Berlin: Springer. equilibrium. Physical Reviews 40: 749–759.

Bose–Einstein Condensates
F Dalfovo, L P Pitaevskii, and S Stringari, general ground, one can start with the definition
Università di Trento, Povo, Italy of the one-body density matrix
ª 2006 Elsevier Ltd. All rights reserved. ^ y ðrÞðr
^ 0Þ
nð1Þ ðr; r 0 Þ ¼  ½1
The quantities  ˆ
^ y (r) and (r) are the field operators
Introduction which create and annihilate a particle at point r,
In 1924 the Indian physicist S N Bose introduced a new respectively; they satisfy the bosonic commutation
statistical method to derive the blackbody radiation law relations
in terms of a gas of light quanta (photons). His work, ^
½ðrÞ; ^ y ðr 0 Þ ¼ ðr  r 0 Þ;
 ^
½ðrÞ; ^ 0 Þ ¼ 0
ðr ½2
together with the contemporary de Broglie’s idea of
matter–wave duality, led A Einstein to apply the same If the system is in a pure state described by the
statistical approach to a gas of N indistinguishable N-body wave function (r 1 , . . . , r N ), then the
particles of mass m. An amazing result of his theory was average [1] is taken following the standard rules of
the prediction that below some critical temperature a quantum mechanics and the one-body density
finite fraction of all the particles condense into the matrix can be written as
lowest-energy single-particle state. This phenomenon,
named Bose–Einstein condensation (BEC), is a conse- nð1Þ ðr; r 0 Þ
quence of purely statistical effects. For several years, Z
such a prediction received little attention, until 1938, ¼ N dr 2 dr N  ðr;r 2 ; ...; r N Þðr 0 ; r 2 ;.. .;r N Þ ½3
when F London argued that BEC could be at the basis of
the superfluid properties observed in liquid 4 He below involving the integration over the N  1 variables
2.17 K. A strong boost to the investigation of Bose– r 2 , . .., r N . In the more general case of a statistical
Einstein condensates was given in 1995 by the observa- mixture of pure states, expression [3] must be
tion of BEC in dilute gases confined in magnetic traps averaged according to the probability for a system
and cooled down to temperatures of the order of a few to occupy the different states.
nK. Differently from superfluid helium, these gases Since n(1) (r, r 0 ) = (n(1) (r 0 , r)) the quantity n(1) ,
allow one to tune the relevant parameters (confining when regarded as a matrix function of its indices
potential, particle density, interactions, etc.), so to make r and r 0 , is Hermitian. It is therefore always possible
them an ideal test-ground for concepts and theories on to find a complete orthonormal basis of single-
BEC. particle eigenfunctions, ’i (r), in terms of which the
density matrix takes the diagonal form
X
What Is BEC? nð1Þ ðr; r 0 Þ ¼ ni ’i ðrÞ’i ðr 0 Þ ½4
i
In nature, particles have either integer or half-
integer spin. Those having half-integer spin, like P ni are subject to the normal-
The real eigenvalues
electrons, are called fermions and obey the Fermi– ization condition i ni = N and have the meaning of
Dirac statistics; those having integer spin are occupation numbers of the single-particle states ’i .
called bosons and obey the Bose–Einstein statis- BEC occurs when one of these numbers (say, n0 )
tics. Let us consider a system of N bosons. In becomes macroscopic, that is, when n0 N0 is a
order to introduce the concept of BEC on a number of order N, all the others remaining of order 1.