Classical model of bosons and fermions - CERNcds.cern.ch/record/260481/files/P00022042.pdf ·...
Transcript of Classical model of bosons and fermions - CERNcds.cern.ch/record/260481/files/P00022042.pdf ·...
P @@2E@4E OCR Output
~I\\\|\\\\\||\\\||\\\l\\\|\\|\\l\|\\||\\\|\\l|lllllQERN LIBRHRIES. BENEVQ
particles or of degree of antisymmetrisation or symmetrisation of the wave
to the parameter rc the meaning of degree of indistinguishability of identical
statistical distributions different from the FD and BE ones. We attribute
diate value, except zero, between the extreme values -1 and +1, we obtain
for rz = 1 the Bose—Einstein (BE) distribution. When rc assumes an interme
(FD) distribution, for is : 0 the Maxwell-Boltzmann (MB) distribution and
ourselves to brownian particles, we derive exactly for A : -1 the Fermi—Dirac
tion, depending on the value we give to the parameter rz. When we limit
equation we obtain a general expression of the stationary distribution func
ance the effect of the full or partial validity of EIP. After deriving the kinetic
contains a parameter rs, whose values range between -1 and +1 and can bal
transition rates by means of an inhibition or an enhancement factor, which
(EIP). This exclusion or inclusion principle is introduced into the classical
function of particles obeying a generalized exclusion-inclusion Pauli principle
D—dimensional continuous or discrete space, in order to study the distribution
(EP). In the present work, on the same grounds, we extend this kinetics to
tion for the distribution function of particles obeying an exclusion principle
In a previous work we have derived a non-linear 1—dimensional kinetic equa
Abstract
in coothsé or- eu sue:-vc rm! mo mw; . {.>g;\,·_ g
(March 10, 1994)
Istituto Nazionale di F isica Nueleare, Sezioni di Torino e di Cagliari
Consorzio Interuniversitario Nazionale di Fisica della Materia, Unitd del Politecnico di Torino
Dipartimento di Fisica del Politecnico, Corso Duca degli Abruzzi 24 10129 Torino, Italy
G. Kaniadakis , P. Quarati
Classical model of bosons and fermions
/ S Us B =~i· A '%
fat swosa
Typeset using REVTEX OCR Output
02.50.Ga,, 05.30.Fk, 05.40.+], 05.20.Dd
function ofthe particle system.
Let us clarify the above arguments with this example. lf we consider only two quantum OCR Output
entering the same state.
first fermion entering the state. But for n : 1 a second fermion is strictly inhibited from
indistinguishability requirements. lf n = O there is no inhibition of the probability for the
smaller by an inhibition factor of l~—n than it would be if there were no quantum mechanical
of fermions in a quantum state is n, the probability of one more fermion joining them is
quantum mechanical indistinguishability requirements. lf the occupation number of a system
joining them is larger by an enhancement factor of 1 + n than it would be if there were no
functions. If there are already n bosons in a quantum state, the probability of one more
for deriving classical and quantum Fermi—Dirac (FD) and Bose-Einstein (BE) distribution
known, at thermal equilibrium, where the detailed balancing postulate is the starting point
relations between the quantum transition rates and the classical transition rates are well
For quantum indistinguishable particles, bosons or fermions, we can remark that the
be in that state.
ular quantum state very delinitely influences the chance that another identical particle will
the effect of indistinguishability of quantum particles is that the presence of one in a partic
that another identical and distinguishable particle will be in that state. On the other hand,
guishable particle in some particular energy state in no way inhibits or enhances the chance
It is well known that, at thermal equilibrium, the presence of one classical and distin
probability rates and therefore, as a consequence, in the kinetic equation.
of a system of indistinguishable particles; the quantum effects are contained in the transition
particularly we shall derive and solve a non-linear kinetic equation describing the evolution
indistinguishability of identical particles, force significant changes in the classical procedures;
identical particles. In this work we shall see how quantum considerations, particularly
identical particles in a system, non classical effects arise from the indistinguishability of
ln quantum physics, if there is appreciable overlapping of the wave functions of two
by means of a factor which contains,fully or partially, the inhibition or the enhancement OCR Output
deriving from the indistinguishability of identical particles, we give an expression of them
we consider first of all the transition rates and, taking into account the quantum effects
paper, which was limited to 1—dimensional fermionic particles At the basis of this work
In this work we follow and generalize the approach already explained by us in a previous
Our approach is quite different.
obtains distributions close to the FD one, however a full agreement is not at all reached.
a way that fermions do not collapse to the lowest level By means of this method one
treated by some authors inserting a repulsive potential into the diffusion equations, in such
rates from one occupancy site to another one. In the case of fermions the problem has been
task. These equations should contain particular constraints on the transition probability
evolution in time towards equilibrium from classical kinetic equations is a very stimulating
any time towards equilibrium. The research of quantum statistical distribution and of their
numerical solution of the FP equation, the distribution functions of fermions and bosons at
has ever been set in the transition probability rates to obtain, directly from the analytical or
derivation of this equation.To our knowledge, up to now, no inhibition or enhancement factor
distribution function n(t,v) and tl1e transition probability rates are at the basis of the
The usual, linear Fokker-Planck (FP) equation describes the time dependent, classical
distributions in a very simple way.
to derive the classical Maxwell—Boltzmann (MB) distribution and the quantum FD and BE
principle. The application of these relations to the detailed balancing postulate allows us
zero to one, the inhibition factor ranges from one to zero, in agreement with the exclusion
classical particles multiplying the classical rates by 1 — it inhibition factor; n ranges from
principle). The fermionic transition rates r{2,rg1 can be expressed in terms of the rates for
one to ever larger values (in agreement, let us say, with a principle that we call inclusion
1+11. As n ranges from zero to larger and larger values, the enhancement factor ranges from
obtain their values simply multiplying the classical rates r§2, rg, by the enhancement factor
states 1 and 2 and the two bosonic transition rates rl{2,rg1 between the two states, we can
The research of statistical distributions interpolating between the FD and BE ones and OCR Output
of the exclusion Pauli principle and of the inclusion principle.
existence is also related to the research of small violations of the FD and BE statistics, i.e.
addition to the FD an the BE distributions, also a family of different distributions, whose
an arbitrary dimensional space. In contrast to the above quantum case, we can deduce, in
and fermion systems are possible. The non—linear kinetic equations we use are defined in
space. In one and two dimensions it is shown that also intermediate cases between boson
straint. This is however only the case if the particles move in a three or higher dimensional
a natural way from the formalism, without having to be introduced as an additional con
The restriction on wave functions, to be either symmetric or antisymmetric then appears in
restriction, corresponding to the symmetrization postulate, is put on the state functions.
the particles is taken into account in the definition of the configuration space, no additional
straints on the wave function need to be postulated. ln fact, since the indistinguishability of
This fact is used in the quantized theory and has, as a consequence, that no symmetry con
is not the Cartesian product of the single particle spaces, but rather an identification space.
fected by indistinguishability of the particles. The configuration space of a N—particle system
of identical particles and shown that the classical description of a many-particle system is af
subject by Leinaas and Mirheim They have examined the classical configuration space
lating between FD and BE statistics. We wish to recall here the early work related to this
In the last years a great effort has been spent in studying intermediate statistics interpo
and +1, except the value zero.
the parameter rc contained in the inhibition or enhancement factor has values between -·1
see, we may obtain also statistical distributions different from the FD and BE ones, when
could derive numerical solutions for the time dependent distribution functions. As we shall
the equilibrium, as time goes to infinite, can be deduced in a simple way; in other cases one
kinetic equation. In some case, like in the brownian particle case, analytical solutions at
whose values range between -1 and +1. Then, as a consequence, we derive a non—linear
effect. The graduality of the inhibition or the enhancement is assured by a parameter rc
' ` T
the degree of antisymmetrisation or of symmetrisation, respectively. Depending on the value OCR Output
has the meaning of degree of indistinguishability of fermions or of bosons, corresponding to
cases when A assumes one of the intermediate values between -1 and +1. The parameter rc
and the FD case rz : -1, in addition to the classical MB case rc : O and to all intermediate
therefore, as we shall see, varying the value of rc, we can consider both the BE case rc : +1
distribution function n(t,v). The values of the parameter is ranges between -1 and +1,
As a consequence the kinetic equation becomes non-linear because the factor contains the
the inhibition or enhancement factor 1 + 1m(t,v) into the transition probabilities rates.
Complete antisymmetrisation means complete indistinguishability. We shall introduce
know that pairs of fermions are not completely bosons, but only approximately so.
particle-particle pairs of nucleons can be coupled to angular momentum two, although we
of fermions, the fact that we cannot antisymmetrize completely all the nucleons and that
heavy nuclei. However IBM contains fundamental problems as for instance the bosonization
(IBM) describes the main properties (i.e. energy levels, electric quadrupole moments) of
free neutral hydrocarbon aromatic radicals. In nuclear physics the independent boson model
wave function studied by Brovetto and Bussetti [8] to analyze the hyperfine structure of the
functions it is possible to describe the physical properties of systems. One example is the trial
are used being characterized by a partial antisymmetrisation: only with this kind of wave
symmetrisation. In some physical systems of many—idcntical fermions trial wave functions
wave functions. Let us introduce the concept of degree of antisymmetrisation or degree of
Indistinguishability is strictly related to antisymmetrisation and to symmetrisation of
nuclear, particle and condensed matter physics and in astrophysics and cosmology.
deviations from the Pauli principle. Implications of such deviations are reported in atomic,
fermions [5-7]. Experimental high precision methods have been developed to verify small
and annihilation operators obey to commutation relations interpolating between bosons and
and BE statistics has developed many studies on the intermediate statistics, where creation
early work by Gentile in 1940 Very recently the search for small violations of the FD
allowing that N fermions, in place of one alone, occupy a single state, started with the
The evolution equation can be written in the form: OCR Output
the transition probability rr(t,v —> u) from the point v to the point u in the phase space.
particle distribution at any point of the phase space. The particle kinetics is described by
We call n(t, v) the occupational number, it is a scalar Held which describes univocally the
Let us consider the kinetics of particles in a continuous space of arbitrary dimension D.
II. PARTICLE KINETICS IN THE PHASE SPACE
Conclusions are reported in Sec.VI.
interval between two close states. The distribution functions are also derived.
velocity (or discrete energy) space. We may deduce the quantum of energy which is the
In Sec.V, the kinetics of the previous section is studied on a lattice i.e. on a discrete
rc == :1:1, to the BE and FD distribution and when zz = 0 to the MB one.
to innnite we may derive the expression of the distribution function which coincides, when
tional to the particle velocity and the diffusion coeflicient is a constant. As the time goes
In Sec.IV, we develop the kinetics for Brownian particles: the drift coefficient is propor
is derived whose expression is valid for any value of the parameter re, between -1 and +1.
second order momentum of the transition rate different from zero. A stationary distribution
In Sec.III, we limit ourselves to consider nearest neighbors interactions and only first and
equation.
dimensional space that, in the case of rc = O, becomes the well—known, linear Fokker-Planck
In Sec.II, we introduce a general, non—linear kinetic equation in an arbitrary D
of intermediate statistics.
through inhibition or enhancement factors. We send to ref.[9] for a classical kinetic model
within the framework of classical, non—linear equations of motion containing quantum effects
be possible to develop interesting models for atomic, nuclear and particle quantum dynamics
of particles under consideration. The result obtained in this paper holds promise that it may
of rt we may choose the degree of indistinguishability or the degree of classicity of the system
t — : —M »V UN ml 5 <>OCR Output(—1)"‘ @'"¢l¤(i»V)l ——-——-————;-—·— Q O, Om »~ » M
The function 1/¤[n(t,v — u)] also, in the second integral of Eq.(3), can be written as:
u(,,uc,,...uO,m . (4)r(t, v + u, u)ep[n(t, v + u)] : go E,1 am r ¢ v)]}
We obtain:
expanded in a Taylor series as a function of v + u, in an interval around v, being u < 11.
first integral of the right hand side of Eq.(3) the quantity r(t, v + u, u) ¢p[n(t, V+ u)] can be
In this paper we consider physical systems evolving very slowly. This means that in the
<p[n(t,v)] [ r(t,v,u) zf»[n(t,v — u)] duu
,= 1b[n(t,v)] /r(t,v + u,u) ¤p[n(t,v + u)] dun8 r, ngV)
assumes the form:
The evolution equation, taking into account the expression of the transition probability,
degree, the transition rates, from site to site.
arrival site u. These two functions can inhibit or enhance, fully or partially, at a certain
bution at the initial state v while 1,b[n(t, u)] depends on the occupational distribution of the
number to be fully expressed: cp[n(t, v)] is a function depending on the occupational distri—
v — u of the particle state during the transition, <,0 and rb are functions of the occupational
where r(t, v, v-—u) is the transition rate, depending on the initial state v and on the variation
(2)1r(t,v —> u) = r(t,v,v — u) gp[n(t,v)] 1/¤[n(t, u)] ,
the following way:
Let us define the transition probability of a particle from the state v to the state u in
point of the space.
arriving in v from any point of the space and the particles leaving v and going towards any
The variation in time of n(t,v) is due exclusively to the difference between the particles
(1)= [¢r(t,u —> v) — 1r(t,v —> u)] duu6 ¢ T,g-?-2 /
equation given by Eq.(7) becomes: OCR Output
with the occupation number of the starting site. Taking into account Eq.s (9), the evolution
high number of particles,the probability that one of them leaves the state increases linearly
the arrival site increases. The relation ¤p(n) = n means that if the initial state contains a
xc is equal to 1 and the probability transition is ever larger as the occupational number n in
if n = O, we have = 1 and transitions shall take place. In the case particles are bosons
site n = 1, the inhibition factor is zero and transitions do not take place; on the other hand,
case are bosons. In the Hrst case, must represent an inhibition factor: if in the arrival
where it is a parameter which has the value -1 in case the particles are fermions and +1 in
cp(n) = n , : 1 + rm , (9)
¤p(n) and are linear functions of n, we have:
impose that its population affects the transition probability. In the more simple case where
site if, in it, particles are absent. If, on the contrary, the arrival site is populated, we can
the second condition imposes that the transition probability does not depend on the arrival
The hrst condition means we do not have transitions when the initial site is empty, while
@(0) = 0 , 1/»(0) = I (8)
initial or arrival site is empty:
We observe that the functions <p(n) and must obey the two conditions, in the case the
(7)m + (-1) +1C¤,¤2...¤.,.(% V)<Pl"(#»V)l )@"‘¤/¤l¤(( V)l
Bt 7,,:0 l 30,,, 3110,2 . . .0vOm¤(¢»V) Z i (1/][n(, v]@”‘{C¤m...¤m(¢»V)<Pl¤('i»V)l}
equation:
the integro-differential Eq.(3) can be transformed into the following infinite order differential
(6)u §O,,,,,___,,m(t,v) : r(t,v,u)u,,,uc,,...u,,md u ,1 RT/
After the introduction of the mm order momentum of the transition rate
10 OCR Output
which is the continuity equation and takes into account the local balance in the phase space.
Ga+ VJUN) = 0 » (15)
t 6n( iv)
Eq_.(13) has the expression:
the exclusion—inclusion principle. After introducing the particle current given by Eq.(14),
is a drift current depending on §¤(t,v), O§(,6(t,v)/Ovp and on the parameter rt related to
The last term on the right hand side of Eq.(l4) is the Fick current, while the first term
(14). J¤(*»V) = —lC¤(i»V) + ····g·—l¤(¢»V)l1 + ~¤(f,V)l — C¤¤(tV)··y Un U:BQ, (t,v) L 3n(t,v)
We define the particle current j0,(t,v) by means of the relation:
8% 6%t, l»tt, Gt, -—-————. 13 l¤( V)l 4 r¤( V)l +C ¤( V) } ( )——-——=—— qt, Gt 3Un{lC( V) +
3(_"¤p(t,v) 3n(t,v)0n(t,v) 3
Eq.(l0) can be written as it follows:
(12)g`O,p(t,v) : r(t,v,u)uO,up duué/
(Il)§O,(t,v)= /r(t,v,u}u(, duu ,
momentum of the transition rate are different from zero; we have:
In this section we consider the Eq.(10) in the case only the first and the second order
III. NEAREST NEIGHBOURS INTERACTION
In the case rc = 0 we obtain the well known, linear Fokker—Planck equation.
of how much the particle kinetics is affected by the particle population of the arrival site.
which describes a non—linear kinetics. We note that the parameter rc gives a quantification
u
1 ’ ( 0)t +""( "’) g O"‘[§., (t v)n(t v)] 8"‘n(t v) 102 crm 7 7 _1m+] Ot O OI fi 7 i av¤1ar,2...0t¤m fl l C 1 m( "’)avmat.,2...av,,mlGt ,:0 314,1 5210,2 . . . Hvdm
8n(t,v) @2 8"‘[§m,,2___L,m(t, v)n(t, v)]
11 OCR Output
variation dv of the position of a particle along the direction oz; we have:
local balance of the particles along the generic direction cx. Let us consider an infinitesimal
that is exactly the Eq.(15). This equation can be obtained also directly, by means of a
6v,,Gt Ovc,A _ ¤]¤(f»v)l1 + ~¤(¢»v)l + D.<t,v>¤i . uv= V{[J(¢.v) +
OD., t, ( v) 6 t, i’@3 t, "( V)
current into Eq.(10) we obtain:
In Eq. (18) repeated indices are not summed over. Substituting this expression of the
. ],;,(t,v) = —[JO,(t,v) + -—-———]n(t,v)[1 + zm(t,v)] — Dc,(i,v)——dv,. Ov, (18)GDC, t, LY2 3 t, ?iv>
mix with each other; we have:
consequence that during the transitions the various components of the current vector do not
Let us remark that the second order momentum is a diagonal matrix. This fact has the
Eq.(17) represents the conditions to be satisfied only for the nearest neighbors interaction.
Coq¤;...r.vm(t;v) : 0 m > 2
(c,p(t,v) : D,,,(t,v)6a,6 , (17)
g`O,(t,v) : JO,(t,v) ,
mornenta:
the transition rate and using Eq.(6) we have, for Av ——> 0 the following expressions for the
where D¤(t,v) and J,,(t,v) are the diffusion and the drift coefficients. With this choice of
(16)+lD¤(i»V) —§J¤(i,V)Avl<$(¤¤ + A10} »
(pT(¢,v,u) Z -5%; g]! 6(u,,){[n(,,(t,v) + gJC,(r,v)At]a(u(, - Av)
rate assumes the following form:
only one component of the position vector u can change of a quantity Av, the transition
tions among the nearest neighbors. If we make the hypothesis that during a single transition
We demonstrate that the kinetics described by Eq.(15) takes into account only interac
12 OCR Output
equation j(t, v) = O, when J,,(v) and DQ(v) are related by means of the following expression:
t —> oo the particle current is equal to zero. It is easy to integrate the first order differential
The stationary solutions n(v) can be obtained directly by observing that at the limit of
tn(v) = lim n(i, v) (28)
tD(,,(v) = lim D,(t,v) , (27)
tJO,(v) = liin J(,,(i,v) , (26)
We pose now our attention to the steady states of Eq.(19). We define:
consequently Eq.(19).
If we expand this expression, up to the first order in dv, we obtain immediately Eq.(18) and
(25)
jQ(t,v)dv = r(t,v,dv)n(t,v)[1+ :m(t,v + dv)] — r(t,v + dv, -—dv)n(t,v + dv)[1+ »m(t,v)]
and explicitly we have:
j,,(t,v)dv: 7r(t,v——> v+dv)—1r(t,v+dv -—>v) , (24)
The particle current along the direction oz is consequently given by:
1r(t, v + dv —> v) : r(t, v + dv, —dv)n(t,v + dv)[1 + rm(t,v)] (23)
(22)¢r(t,v —+ v + dv) : r(t,v,dv)n(t,v)[1+ ;m(t, v + dv)] ,
following expressions:
the transition probabilities of the two transitions v —> v + dv and v + dv —> v assume the
(21)r(t,v, ;l;dv) = D,,,(t,v) fc §.]¤(t,v)dv ,
Defining the transition rate by the relation:
(20)(dv),6 : 6,8,,,dv
13 OCR Output
diffusion coefficient is a constant:
this type of particles, the drift coefficient is proportional to the particle velocity while the
In this Section we consider the brownian particle kinetics in a 3—dimensional space; for
IV. BOSONS AND FERMIONS IN A CONTINUOUS SPACE
framework of validity of the nearest neighbors interaction hypothesis.
rc can vary continuously between -1 and +1: all kind of statistics can be derived, in the
and D,,,(v,,) the integration can be analytically or numerically accomplished. The parameter
motion along the various directions is decoupled. Depending on the expressions of J,,,(vO,)
Eq.(33) is the most general expression for the stationary distribution function n(v) when the
BD —·“1 Xplg, f `5QT·75lJ<~("¤) + J.,L,ld”¤— W) · Kn(v) : ———N (33)
In this case the steady state given by Eq.(30) assumes the expression:
D¤(V) = D¤(v¤) (32)
J¤(V) Z J¤(v¤) » (31)
the phase space is completely decoupled:
The condition given by Eq.(29) is satisfied if the motion along the various directions of
that expresses the conservation of the particle number.
where Hp is the integration constant and can be calculated from the normalization condition
a1exp[;f h,,,,(vO,)dvO, — Bp] —- rc
71(V) = (30)
where h,,(vO,) is an arbitrary function. We obtain with easy calculations:
011,,,J0,(v) : h0,(v(,)Dc,(v) (29)
OD, (V)
14
Gt+ <I>(t,v) = O (41) OCR Output
6N z, ( U)
the equation describing the evolution of the particles inside the sphere E., is:
*I>(i,v) =i(i»v)¤d¤» (40)/ EI}
If we indicate with <I>(t, v) the flux of particles outcoming the sphere Ev at time t we have:
N (t, oo) = N (39)
and therefore we have the condition:
Let us call N the total number of particles. This number must be conserved at any time
(38)N(t,v) = n(t,v)d3v[3
having the center at the origin and the radius equal to v; we have:
velocity §_ v. Of course these are the particles lying, in the velocity space, in a sphere E.,
We now indicate with N(t, 11) the number of particles having at time t the modulus of
E == §m(v$ + 11;+ vg)
The kinetic energy of a particle is given by:
we will show that Eq.(36) can be written as a continuity equation in the energy variable.
Let us remark that Eq.(36) is a continuity equation, written in the velocity space. Later on
t(36)1 V{cvn(t,v)[1+ Ȣn(t,v)] + c-/;5Vn(t, v)}3 t, ngV)
The kinetics equation assumes the form:
(35)j : —cvn(t,v)[1 + rm(t,v)] —— %—Vn(i,v)
where c is an arbitrary constant. The current is given by:
(34)JO,(t,v) := cv, , D(,(t,v) = , B = ff ,
15 OCR Output
course, the particle current in the energy axis and assumes the following form:
The flux of particles <I>(t, E) outcoming the sphere EU of radius v = (/2E/m represents, of
m
(48)a = 4c7r(;)3/2
We introduce the constant:
(47)pn, E) : 2¤(;)3/A/1T§n(t, E)
From Eq.(46) it follows that the function p(t, E) is given by:
(46)p(i,E)dE : 4rrv2n(t,E)dv
interval [11,21 + dv]; we therefore have:
shell delimited by the two spheres E,,+d,, and EU, having the modulus of the velocity in the
particles that lie in the interval [E, E + dE] is equal to the number of particles lying in the
space: we introduce the occupation probability p(t,E) at energy E, then the number of
We now want to interpret Eq.(45) as a continuity equation in the one-dimensional energy
8n(t,E) _ 2c G 3/2 l 3n(t,E)
and, after the introduction of the energy as the variable, we have:
at U20v -—--—-——- Hm gv }) »2 = ———— (v {v¤(i»V)l1+~¤(i»V)l+On(t,v) c 8 1 3n(t,v)
Then Eq.(42) assumes the following form:
<I>(t,v) = 4vrvvj(t,v) (43)
The values of <l>(t, v) can easily be calculated by Eq.(40) which assumes the expression:
Ot 41rv2 61242 ( )6‘n(t,v) 1 3<I>(t,v) I
Eq.(41) in the form:
We differentiate Eq.(4l) with respect to v and take into account Eq.(38); we can write
16 OCR Output
all the other values of rc a family of different statistical distributions.
Eq.(52) gives exactly the FD (rc :: -1), MB (rc = 0) and BE (rc = +1) statistics and for
(55)/°° .—[}y, CXP Z _ K`,.>”“ mdz Z N
the condition given by Eq.(53) assumes the form:
542W12 N = i —-— 3/
constant. After the introduction of the constant N, defined by:
V and h is a constant with the right dimensions which can be identified by the Planck
dn = d(mv;)dmi/h. N is the total number of particles of the system occupying the volume
where dr is the elementary volume in the phase space and is given by dr = dn dT2d·r;», with
n(E)d*r = N , (53)
The integration constant ;i can be determined from the normalization condition:
52 ( )E : —-————¥——————— fl <=><pB(E—#)—*<
After integration we have:
+ + : 0 (51)
bution at the limit t —> oo must satisfy by the first order differential equation:
If we set the particle current <I>(t, E) equal to zero we obtain that the stationary distri
Ot ' OE(50)
3p(t, E) I 3<I>(t,E)
space:
Eq.(45) assumes finally the form of a continuity equation in the one—dimensional energy
H OE(49)<I>(t, E) = —aE3/2{n(t, E)[1 + »m(t,E)] +
E
‘ T F _r·r··r
17 OCR Output
can write the expression for the particle current :
and are given by Eq.(9) and remember that now the velocity is a discrete variable, we
We consider the definition of the transition probability of Eq.(2), where the functions g0(n)
2D,(t) : r(t, 1);+1, ——Av) + r(t,v,+1, —Av) (61)
Ji(t) Av = r(t,v,, —l-Av) — r(t,v;, —Av) , (60)
transition rates:
We now introduce the drift J,(t) and the diffusion Di(t) coefficients by means of the
dt Av
d , t A Q t
nuity equation in a discrete space:
The Pauli master equation given by Eq.(56) assumes an expression analogous to the conti
Ajit) =J;(t)-J;-1(i) (58)
and call the currentvariation between two adjacent sites:
Mi) = lr(i,ve —> var) ·¢r(i,v¢+r —> v¢)lAv » (57)
occupation probability of the site i. We define the particle current as:
where rr(t, v, ——·> vj) is the transition probability from the site i to the site j and n;(t) is the
d= 7K'(t,'U;-1 —> 1),)+ rr(t,v,+1 ——> vi) — 7l'(t,Ug —> 11,+1) —— 7r(t,v; —> 11,-1) , (56)n; )
in the following way:
not change during the transition. The Pauli master equation for this process can be written
transitions to the nearest neighbors are allowed and that the direction of the particles does
interval between close levels is represented by a quantum of energy. Let us assume that only
particles. The results for this case can be useful when energies are discretized and the
In this section we wish to discuss, in a discrete velocity space, the kinetics of brownian
V. BOSONS AND FERMIONS IN A LATTICE
18
e = ——·—— ; Q +1 1 _ BAE? 69 ( ) OCR Output
Eq.(67) become:
fh ·· *6TL; =- —-—— , (68)
Introducing the variable qi:
1 + BAEK11;; = n +1 67 ( )
(I —
To deduce ni, let us write the recurrence relation of Eq.(66) in the form:
substitution if we remember the definition of the transition probability given by Eq.(2).
tute + with the quantity n,(1 + mm,1). It is straightforward to justify this
equation given by Eq.(51). On the other hand, when we discretize Eq.(5l) we must substi
In the limit to the continuum we can transform this recurrence relation into the differential
m(1 + mai) + (66)gkgmm
current equal to zero:
To obtain the stationary distribution of the system, it is enough to pose the particle
Mi) = -iAE{¤;(i)l1+ #<m+1(f)l + (65)Z,··§é}Qthe expression of the current assumes the form:
Av ·· "mimr » (64)
Considering the energy as the variable instead of the velocity and taking into account that:
(63)J;(t) = cv; , D;(t) :
have the following expressions:
We are considering brownian motion by imposing that the drift and the diffusion coefficients
Av Av(62). Je(i) = —{lJi(i) + —·l¤¢(¢)[1 + ~m+1(¢)l + De(*)·—}A6AD; i l2 AH; I Q
19
€Xp¤r»»r>—~ll i Z: ( 1 OCR Output
Eq.(59):
the stationary distribution given by Eq.(70) becomes the continuous distribution given by
AE—•Olim iAE = E , (75)
Let us remark that if we calculate the limit to the continuum and consider that
which express the detailed balance principle and is valid for any value of is E [-1, +1].
MU + mar) = €><p[—f’(E¢+i - Ei)l¤e+1(1+ me) » (74)
Once we consider relation (72), the Eq.(66) can be written in the following form:
where k is the Boltzmann constant.
may derive that the relation between w and T is given by eu : (kT/h) log[1/(1 —— AE/kT)],
lf we assume that E is proportional to w through the relation S = hw, from Eq.(72) we
ln conclusion the quantum E coincides with the lattice interval AE only when ,8 —-—> O (T ~—>
(73)2 e AE + gmaiz)2 + §52(AE)3 +.
of the system. Let us expand the energy E in power series of AE, we have:
two close states, is different from the lattice interval AE. This is due to the finite temperature
Let us remark that the quantum of energy E, which represents the energy interval between
H 1 _ ME8 = —l 0g 72 ( )1 1
and E has the expression:
Ei = ii , (71)
which is the discretized distribution. ln Eq.(70) the energy E, is given by:
€><1:>5(E¢—u) —·<TL; = )
We can easily calculate q, and then finally ni:
20 OCR Output
we must recall that in recent works [10] it has been shown that the quantum average fluc
tions. We have limited ourselves in considering explicitly only systems of brownian particles;
quantum dynamics is equivalent to classical dynamics with the inclusion of quantum fluctua
The results obtained in this work should not come as a surprise; in fact we know that
a small violation of the exclusion or inclusion principle.
whose wave functions must not be completely antisymmetrised or to physical processes with
sation or symmetrisation, therefore these intermediate statistics can be applied to systems
We give to rc the meaning of degree of indistinguishability or of degree of antisymmetri—
of different statistics.
classical MB (rc = 0, no quantum effects are considered). When A gé 0, :k1 we have a family
When the particles are brownian we find exactly the FD (rc ·: -1), the BE (rs : 1) and the
of distribution functions of many-particle systems which is the main interest in this work.
of transitions caused by a small violation of the Pauli principle rather than to the study
Green and by other authors in the past Their interest was mainly devoted to the study
The statistics which differ in a discrete way from the usual ones were introduced by
of the Pauli exclusion principle and of BE statistics.
Greenberg and collaborators [6,7] in view of considering the possibility of small violations
The exact validity of BE and FD statistics was questioned recently in many papers by
account, i.e. the particle wave functions are only partially antisymmetrised or symmetrised.
are derived in which the exclusion or the inclusion principle is only partially taken into
quantum effects are not considered, in all the other cases different statistical distributions
ple when rc = +1, while the case rc = 0 means that the distribution is a classical one because
and 1. We take fully into account the exclusion principle when rc = -1, the inclusion princi
ment factor. The factor contains a parameter KZ whose value ranges continuously from -1
effects introduced into the transition probabilities by means of an inhibition or an enhance
In this work we introduce a non—linear classical kinetic equation, containing quantum
21 OCR Output
We thank A. Erdas for critical reading of the manuscript.
ACKNOWLEDGMENTS
a dynamics very close to the quantum dynamics of a system of identical particles.
conclude that the non—linear kinetic equation we have introduced in this paper can describe
tuations of the quantum dynamics coincides with the brownian fluctuations. Therefore we
22
sitaf di Pavia, (1992) and references therein.
10 M.RoncadeI1i, A.Defendi, I cammini di Fei nman, uaderni di Fisica. Teorica, Univer- J
together with the present work.
[9] G.Kaniadakis, Classical intermediate statistics, submitted for publication to Phys.Rev.E
[8] P.Brovetto and G.Bussetti, Nuovo Cimento 64 B, 284 (1969).
[7] O.W.Greenberg, Phys.Rev.D 43, 4111 (1991).
[6] O.W.Greenberg, R.N.Mohapatra., Phys.Rev.D 39, 2032 (1989).
H.S.Grcon, Phys.Rev. 90, 270 (1953).[5]
G.Gonti1o, Nuovo Cimonto 17,493 (1940).[4]
J.M.Leinaas,J.Myrheim, Nuovo Cimonto 37 B, 1 (1977).[3]
G.Kaniadakis and P.Quara.ti, Phys.Rev.E 48,4263 (1993).[2]
C.Dorso, S.Duarte, J.Ramdrup, Phys.Lett.B 287,1988 (1987).[1]
REFERENCES