MATHEMATICAL PERGAMON COMPUTER MODELLING Mathematical and Computer Modelling 36 (2002) 889-907 www.elsevier.com/locate/mcllr Stochastic Continuum Mechanics- A Thermodynamic-Limit-Free Alternative to Statistical Mechanics: Equilibrium of Isothermal Ideal Isotropic Uniform Fluid E. MAMONTOV AND M. WILLANDER Laboratory of Physical Elect.ronics and Photonics. IvlC2 School of Physics and Engineering Physics Gothenburg University and Chalmers Ur1iversit.y of Technology SE-412 96 Gothenburg, Sweden J. WEILAND Transport Theory Group, Department of Electromagnetics Chalmers University of Technology SE-412 96 Gotheburg, Sweden (Received and accepted September 2001) Abstract-“We do not know how to study finite systems in any clean way; that is, the thermody- namic limit is inevitable.” The lack of the clean way stressed in this sentence of Wsibois and de Leener means the following. To study finite systems, statistical mechanics and kinetic theory offer nothing but the formalism based on the thermodynamic limit (TDL) which in essence disagrees with notion of finite system. The present work proposes the model (almost-equilibrium in a certain sense) enabling one to construct continuous equilibrium descriptions of fluids, discrete multiparticle systems, with no application of TDL. For simplicity, the fluids are assumed to be isothermal, (rheologrcally) ideal, isotropic and uniform The core of the model is nonlinear Ito’s stochastic differential equation (ISDE) for the fluid-particle velocity The continuous equilibrium description is based on the stationary prob- ability density corresponding to this equation. The construction is described as a simple analytical recipe formulated in terms of quadratures and includes the velocity (or momentum) relaxation times which can be determined theoretically, experimentally, or as the results of numerical simulations and depending on the specific nature of the fluid. The recipe can be applied to an extremely wide range of fluids of the above class. It is illustrated by the derivations of the Maxwell-Boltzmann and Fermi-Dirac descriptions for the classical and fermion fluids m arbitrary space domain, bounded or unbounded. In the particular, TDL case, the derived results are in a complete agreement with those of statistical mechanics. @ 2002 Elsevier Science Ltd. All rights reserved. Keywords-Isothermal ideal isotropic uniform fluid, Thermodynamic-limit-free modelling, Non- linear Ito’s stochastic differential equation. Relaxation time, The Maxwell-Boltzmann and Fermi- Dirac descriptions. 1. INTRODUCTION “We do not know how to study finite systems in any clean way; that is, the thermodynamic limit is inevitable” [l, p. 3141. The lack of the clean way stressed in this citation means the following. To study finite systems, statistical mechanics and kinetic theory offer nothing but the formalism 0895.7177/02/$ - see front matter @ 2002 Elsevier Science Ltd. All rights reserved. Typeset by &~S-TEX PII: SO895-7177(02)00235-2
19
Embed
Stochastic Continuum Mechanics- A Thermodynamic-Limit …
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
MATHEMATICAL
PERGAMON
COMPUTER MODELLING
Mathematical and Computer Modelling 36 (2002) 889-907 www.elsevier.com/locate/mcllr
Stochastic Continuum Mechanics- A Thermodynamic-Limit-Free Alternative
to Statistical Mechanics: Equilibrium of Isothermal Ideal Isotropic Uniform Fluid
E. MAMONTOV AND M. WILLANDER Laboratory of Physical Elect.ronics and Photonics. IvlC2
School of Physics and Engineering Physics Gothenburg University and Chalmers Ur1iversit.y of Technology
SE-412 96 Gothenburg, Sweden
J. WEILAND Transport Theory Group, Department of Electromagnetics
Chalmers University of Technology SE-412 96 Gotheburg, Sweden
(Received and accepted September 2001)
Abstract-“We do not know how to study finite systems in any clean way; that is, the thermody- namic limit is inevitable.” The lack of the clean way stressed in this sentence of Wsibois and de Leener means the following. To study finite systems, statistical mechanics and kinetic theory offer nothing but the formalism based on the thermodynamic limit (TDL) which in essence disagrees with notion of finite system. The present work proposes the model (almost-equilibrium in a certain sense) enabling one to construct continuous equilibrium descriptions of fluids, discrete multiparticle systems, with no application of TDL. For simplicity, the fluids are assumed to be isothermal, (rheologrcally) ideal, isotropic and uniform The core of the model is nonlinear Ito’s stochastic differential equation (ISDE) for the fluid-particle velocity The continuous equilibrium description is based on the stationary prob- ability density corresponding to this equation. The construction is described as a simple analytical recipe formulated in terms of quadratures and includes the velocity (or momentum) relaxation times which can be determined theoretically, experimentally, or as the results of numerical simulations and depending on the specific nature of the fluid. The recipe can be applied to an extremely wide range of fluids of the above class. It is illustrated by the derivations of the Maxwell-Boltzmann and Fermi-Dirac descriptions for the classical and fermion fluids m arbitrary space domain, bounded or unbounded. In the particular, TDL case, the derived results are in a complete agreement with those of statistical mechanics. @ 2002 Elsevier Science Ltd. All rights reserved.
Keywords-Isothermal ideal isotropic uniform fluid, Thermodynamic-limit-free modelling, Non- linear Ito’s stochastic differential equation. Relaxation time, The Maxwell-Boltzmann and Fermi- Dirac descriptions.
1. INTRODUCTION
“We do not know how to study finite systems in any clean way; that is, the thermodynamic limit
is inevitable” [l, p. 3141. The lack of the clean way stressed in this citation means the following.
To study finite systems, statistical mechanics and kinetic theory offer nothing but the formalism
0895.7177/02/$ - see front matter @ 2002 Elsevier Science Ltd. All rights reserved. Typeset by &~S-TEX
PII: SO895-7177(02)00235-2
890 E. MAMONTOV et al
based on the thermodynamic limit (TDL) ( see Section 3.2) which in essence disagrees with notion
of finite system. There were many of attempts to eliminate this, generally speaking, inadmissible
contradiction (see Section 3.3). The present work is one more endeavor in this direction. In so
doing, it concentrates on the topic which was not in the research focus before. It proposes the
model (almost-equilibrium in a certain sense) enabling one to construct continuous equilibrium
descriptions of fluids, discrete multiparticle systems, with no application of TDL. For simplicity,
the fluids are assumed to be isothermal, (rheologically) ideal, isotropic and uniform. The core of
the model is nonlinear Ito’s stochastic differential equation (ISDE) for the fluid-particle velocity
studied by the authors before in connection with another problem. The continuous equilibrium
description is presented with the analogue of the one-particle distribution function formed as the
product of the stationary probability density corresponding to this ISDE and the fluid concentra-
tion also provided by this density. This construction is described with a simple analytical recipe
formulated in terms of quadratures and includes the velocity (or momentum) relaxation times
which can be determined theoretically, experimentally, or as the results of numerical simulations
and depending on the specific nature of the fluid. The recipe can be applied to an extremely wide
range of fluids of the above class. It is illustrated by the derivations of the Maxwell-Boltzmann
and Fermi-Dirac descriptions for the classical and fermion fluids, respectively, in the arbitrary
space domain. In the particular, TDL case, i.e., when the domain coincides with the whole phys-
ical space and the number of the particles in it is infinite, the derived results are in a complete
agreement with those of statistical mechanics.
Section 2 concerns some characteristics of fluids common in statistical mechanics (SM) and
kinetic theory (KT), the sciences based on TDL. The section presents such reading of these
characteristics which does not use TDL, thereby illustrating that their physical interpretation is
independent. Section 3 focuses on TDL, considers the limitations originated from it and lists some
possible ways to avoid it. The TDL-free almost-equilibrium model for the continuous equilibrium
distributions of the fluid-particle velocity is proposed in Section 4. This section also includes the
examples of application of the model and discusses its details. Section 5 summarizes the practical
procedure to use the proposed model, considers the general picture of the model, and presents
the concluding remarks.
2. INTERPRETATION OF COMMON FLUID CHARACTERISTICS WITHOUT THE THERMODYNAMIC LIMIT
This work considers an isotropic uniform fluid which occupies a domain (i.e., open connected
set) a(t) C R3 with piecewise smooth boundary do(t) where R = (-00,co) and t E R is the
time. The domain can be bounded or unbounded, in particular, can coincide with the whole
space R3. Points in this space are denoted with x E R3. The fluid is assumed to be ideal in the
rheological sense (e.g., [2, Sections 4.1.1 of Vol. I]). This means (e.g., [2, Sections 4.1.2 of Vol. I])
that the matrix of the components of the fluid stress tensor is -II1 where II is the fluid pressure
and I is the 3 x 3 identity matrix.
2.1. The Fluid Distribution Function and the Corresponding Probabilistic Picture
Fluids are multiparticle systems. They are usually modelled in probabilistic terms. In SM
and KT, the corresponding characteristic which, in the simplest approximation, describes the
fluid behavior is its so-called one-particle distribution function. The present section exemplifies
how the distribution function and the related quantities can be introduced in terms of probability
theory and continuum fluid mechanics (CFM), with no involvement of SM or KT.
Continuum fluid mechanics applies a general mathematical definition of the fluid concentration,
i.e., number of the fluid particles per unit volume (e.g., [2, Section 3.1 of Vol. I]). According to
it! the concentration is by definition the limit of the ratio of the number of the particles to the
volume occupied by the part of the fluid containing these particles as this volume tends to zero.
Stochastic Continuum Mechanics 891
In what follows, we consider the fluid characteristics at t 2 to where to E R is the initial time
point. The above definition means that the fluid concentration n = n(t,x) is nonnegative, i.e.,
n(t,x) 2 0, t 2 to, 5 E R(t), (2.1)
and number N(t) of the fluid particles in domain O(t) is expressed as
N(t) J n(t, x) dx > 0, t 2 to. n(t)
(2.2)
REMARK 2.1. We emphasize that both volume J&t) dx of domain n(t) and number N(t) in (2.2)
need not be small or large in any sense which is beyond the above-mentioned definition, in
particular, in any sense associated with statistical issues. If R(t) is bounded, then N(t) is finite.
Relation (2.2) shows that quantity n(t, x) can be presented as follows:
46 x) = N(t)eit, x:), t >to, x E R(t), (2.3)
where
e(4 xl 2 0, t 2 to, x 6 Q(t), (2.4)
J e(t,x) dx = 1, t > to. (2.5)
n(t)
This means that quantity e(t,x), as a function of x at every fixed t > to, is the probability
density of the particle position y (represented with variable x of function Q). This means that y
is a stochastic process, i.e., y = ~(5, t) where [ E Z is an elementary event and E is the space of
elementary events.
It follows from (2.3) that if 0 < N(t) < cm at t 2 to, then e(t,x) = n(t,x)/N(t), x E R(t).
Expression (2.3) points out the probabilistic meaning of the concentration: it is the product of
number N(t) of the fluid particles in the domain and the corresponding particle-position prob-
ability density e(t, x). This probabilistic picture is well known in statistical mechanics (e.g., [3>
(3.82), (3.73); 4, (6.4.13)]).
REMARK 2.2. Importantly, the notion or even availability of probability density ,Q of the particle
position in domain Cl(t) d oes not presume that the number of the particles in the domain at
time t is nonzero, i.e., N(t) > 0. A simple example illustrating this fact is the homogeneous
linear diffusion equation with constant coefficients. In general, the models for p when N(t) = 0 can be obtained as the limit cases of the models for N(t) > 0 as N(t) + 0 (uniformly in t > to).
However, this limit involves a nontrivial mathematical problem.
Similarly, to the particle position y, velocity u of the fluid particle is also assumed to be a
stochastic process, i.e., u = v(<, t), ,( E Z, t > to. The rest of this section focuses on joint
probability density of random variables y(., t) and v(., t) and the related quantities.
Function f of properties
f(GX>U) 2 0, t 2 to, (x,u) E R(t) x R3,
n(t,x) = J
f(t> 5, u) du, t 2 to, x E Wt), R3
(2.6)
(2.7)
where u E R3 represents random velocity w is called the fluid (one-particle) distribution function.
Relations (2.7) and (2.1) h s ow that j(t,x,u) can be presented as follows:
At, x> u) = 46 x:)dt, 5, u), t 2 to, (x,u) E Cl(t) x R3, (2.8)
892 E. MAMONTOV et al.
where function p is such that
This means that p(t, 5, u), as a function of u at every fixed t 2 to and z E 0(t), is a probability
density. Its meaning in terms of the particle velocity is revealed below (see the text below (2.14)).
It follows from (2.8) that if 0 < n(t, CC) < co at t 2 to, z E R(t), then p(t, z, U) = f(t, z, u)/n(t, z),
u E R3. Expression (2.8) points out the probabilistic meaning of the distribution function: it is
the product of the concentration n(t, X) of the fluid particles in the domain and the corresponding
probability density p(t, 2, u). This probabilistic picture corresponds to that in SM or KT (e.g.. [5,
Section 3.3.10]). It follows from (2.8) and (2.3) that
We denote the mass of the fluid particle and the fluid absolute temperature with m and T.
respectively. Mass m is assumed to be independent parameter.
REMARK 2.3. Equality (2.11) explicitly shows that the distribution function f is a scaled density
in the six-dimensional position-velocity space. According to the well-known Heisenberg uncer-
tainty principle (e.g., [6, Section ?I), the smallest volume in this space is (h/m)3 where h is the
Planck constant. Hence, f(t, 2, U) N (m/h)3. Moreover, N(t) - (2s + 1) where s is the spin of
the fluid particle (s = 0 for spineless particle). Thus, one obtains f(t,rc,~) - (2s + l)(m/h)3.
Equalities (2.4), (2.5), (2.9), and (2.10) show that quantity
rl(C 5, u) = e(t, z)p(t, 5, u), t >to, (x,u) E R(t) x R3, (2.12)
is of properties
and, as a function of vector (z,u) E R(t) x R3 at every fixed t > to, is the joint probability
density of random variables y(., t) and v(., t) or, that is the same, probability density of random
variable
(2.13)
Densities Q, p and random variable v(., t) are characterized in terms of (2.13) or (2.12) (cf. (2.15)
and (2.16) below) in the following way.
In view of (2.10), the particle-position probability density ~(t, .) is the corresponding marginal
probability density with respect to joint density q(t, ., .). Random variable ZJ(., t) is described
with the particle-velocity marginal probability density
q(t, x, u) dx = s e(t, X)P(L 2, u) dxc, t 2 to n(t)
(2.14)
Probability density p(t, 2, ,) at every fixed t 2 to and IC E R(1) is the conditional probability
density of the particle-velocity random variable v(t, .) under the condition that y(<, t) = IC.
Let v(E, t, ,) be the random variable corresponding to this density. Then, this variable is a
random field, i.e.,
v = v(<, t, X)> f L to, z E R(t). (2.15)
Stochastic Continuum Mechanics 893
At every fixed t 2 to and x E O(t), it presents the velocity vector of the fluid particle under
the condition that its position y(<, t) at time t is equal to Z, i.e., y(<, t) = x. Subsequently,
we shall call velocity v = v(<, t, x) and conditional probability density p conditional velocity of
the fluid particle under the above condition and probability density of this conditional velocity,
respectively. Note, that the particle velocities u([: t) and v(<, t, x) are described with different
densities, marginal density s and conditional density p, and coupled according to equality
w(,c,t) = J 4E, 6 xc)dt, x) dx, t 2 to, (2.16) n(t)
which stems from (2.14).
2.2. Expectations and the Related Quantities
Analyzing one or another characteristic of the fluid, say. quantity C(t,x,u), one is usually
interested in its expectation. In connection with function f of property (2.11), the following two
expectations can be considered: expectation
J-W& ., .)I = L,,;,,, <(t> x: u)v(t, 2, u) dx du
=J ((6 x! u)e(t, x)p(t, x, ~1 dx du, t ‘1 to,
n(t) x R”
and conditional expectation
e[C(t, 5, ,) I XI = JR3 C(t, 5, u)p(t, 5, u) du, t 2 to, x E 0(t). (2.17)
In so doing, the former is expressed with the help of the latter similarly to (2.16), namely,
E[C(t, ., .)I = E[e(<(t, x! .) I x)1 = l,,, &(t, 5, .I I 4dt. 2) dx. t 2 to.
One of the most important conditional expectations corresponds to quantity <(t, x;u) E u and is
the fluid-particle macroscopic velocity (cf. [5, (3.13a) on p. 1521)
(2.18)
which is the expectation of the conditional velocity of the particle at time t under the condition
that the particle-position y at t is equal to x, i.e., y(<, t) = x.
The two quantities related to conditional expectation (2.17) are
J&X 5, .)I = n(t, x)e[C(t, 5, .) I 4, t L to, x E n(t), (2.19)
= Teq.MB(5)1 + exp{[U,q(u) - p(S)]/(KT)}’ (x,u) E R,, x R3.
(4.17)
PROOF. The proof follows from (3.9), (4.8), and (4.9). I
One can also check that relaxation time (4.17) passes the test described in Remark 4.1.
Since the only difference of the fermion fluid described with the FD density (3.9) from the
classical one is that fermions obey the Pauli exclusion principle (see the text between (3.8)
and (3.9)), quantity (4.17) can be represented as
Teq FD,k’&~~2, =
1
Teq. MB(Z) + (x,u) E Re, x R3, (4.18)
where the subscript “Pe” stands for the Pauli exclusion and the last term on the right-hand side
of (4.18) is due to this phenomenon alone. It follows from (4.17) and (4.18) that
r eq. pe [cc, (W(U))“] = 7eq. MB(x) exp { ‘ueq(;&;(z)‘} , (? u) E %, x R3. (4.19)
Stochastic Continuum Mechanics 903
This expression for the Pauli-exclusion relaxation time is exact in the present approach and, to our
knowledge, is new. Inequality Tag. p,[~(w(u))~] 2 ~eq.MB(Z)exP[-~(z)/(KT)], (z,u) E f&XR3, shows that contribution of relaxation time (4.19) into (4.18) is zero in the classical limit case (i.e..
as ,u(z)/(KT) + -oo). The fact that the Pauli exclusion is of no importance for the classical
fluid is well known in SM or KT. Relations (4.18) and (4.19) also show that relaxation time (4.19)
dominates the relaxation time r eq.~s(~) for vectors u inside the Fermi sphere when P(Z) > 0)
i.e., under the conditions close to the quantum limit (see Section 3.1).
In general, the idea to allow for the Pauli exclusion principle in the ISDE-modelling is not
new. It goes back to at least [35]. However, the purpose, spirit, and techniques of the present
work are altogether different from those of [35]. The Pauli exclusion principle in connection with
stochastic phenomena is also investigated experimentally (e.g., [36]).
4.4. Accounting a Few Relaxation Mechanisms
The right-hand side of expression (4.16) includes only one relaxation mechanism, namely, that
in the MB fluid. The right-hand side of expression (4.18) takes into account two relaxation
mechanisms, one of them is the same as that in (4.16), i.e., associated with the relaxation in the
MB fluid, whereas the other one is due to the Pauli exclusion. In the real fluids, relaxation time
reeq[~, (w(v))~] in ISDE (4.1) ’ g is enerally contributed by a few, say, T 2 1 mechanisms. They can
be described with respective relaxation times
Teq, kb-, bW21 > 0, (T u) E Gq x R3, k=l,...,r-. (4.20)
Thus, the generalization of expressions (4.16) or (4.18) is (cf., [37! (6f.l)], [17, (C.1.8)])
1
G&G (w(u))21 = (2,~) E 02,, x R3. (4.21)
Each of the relaxation terms summed on the right-hand side of (4.21) can in turn be represented
with a similar sum depending on the specific nature of the fluid and the specific modelling purpose.
So, in general, one has a hierarchy of the relaxation mechanisms. This hierarchy always includes
at least one type of the relaxations, the interparticle one. (If, in so doing, no other types of the
relaxation are present, the fluid is called simple.) The corresponding examples are discussed in
Sections 4.2 and 4.3. More complex representations can be obtained by means of generalization
of more simple ones. Thus, the resulting hierarchy can be fairly extensive. For example, the
classical-fluid and Pauli-exclusion terms in (4.18) for the electron fluid can be accompanied by
the electron-electron scattering due to the Coulomb interaction. In this case, the corresponding
term should be added to the right-hand side of (4.21). Accounting the extra phenomena, for
instance, the electron-photon interaction leads to the additional terms.
Fortunately, in many problems, not all relaxation mechanisms are equally important. So, it is
in many cases reasonable to keep in the sum in (4.21) only the dominating terms.
The following crucial role of function req in our almost-equilibrium model (4.1), (4.4) should
be noted.
REMARK 4.2. The equilibrium properties of the fluid are completely described with equilib-
rium probability density (4.8) ( see also (4.9)) which in turn depends on quantity (3.7) and
relaxation-time function req on the left-hand side of (4.21). Thus, the overall adequacy of ISDE
model (4.1),(4.4) is determined by the physical relevance and numerical accuracy of the summed
terms on the right-hand side of (4.21). The relaxation times in these terms are in our model
assigned to depend not only on the specific properties of the fluid but also on the shape and size
of domain R,, and the conditions at its boundary a&,. There is, in contrast to the SM or KT
formalisms, no need to involve limitation R,, = R3 associated with TDL.
904 E. MAMONTOV et al.
The summed terms in (4.21) can be obtained in different ways. One of them is theoretical
derivations of the terms. Many examples of the derivations can be found in [37] or [38]. Another
option is to extract the expressions for the above terms from the numerical simulations, for
instance, those based on the molecular-dynamics or Monte-Carlo approaches. The extraction
can be applied to the data on the covariance of the stationary solution of nonlinear ISDE (4.1)
mentioned in Theorem 1. In so doing, one can apply the approximate analytical-numerical
technique to evaluate the covariance in case of ISDE (4.1) developed in [17] (see the summary in
Section 4.10.4 therein) on the basis of the so-called deterministic-transition approximation. The
analytical part of this technique is formulated in quadratures. (In so doing, the modification
concerning the passing discussed in the text around (4.24) below should be incorporated.) The
covariance can also be obtained experimentally, for example, by means of the well known diffusing-
wave spectroscopy (e.g., [39]).
4.5. Equilibrium Concentration
It follows from definition of the fluid pressure (e.g.. (2.25)), the physical meaning of chemical
potential (e.g., [20, p. 100)) and the fact noted in Remark 3.3 that relation
(4.22)
is valid. Under the TDL condition, this equality is well-known in SM (e.g., [20, p. 1011). The
pressure in (4.22) is expressed by means of the concentration as is shown (4.13). Thus, rela-
tion (4.22) presents the ordinary differential equation (ODE) for equilibrium concentration n,,-,.
The concentration is deterrnined from this equation under initial condition (3.8). In particu-
lar, one can check with the help of the properties of FD functions (e.g., [16]) that initial value
problem (4.22),(3.8) in the FD case leads to (3.10).
Concentration neq(z) and probability density pes(z, u) provide the equilibrium version
fecI(x, u) = %q(Xhq(X, u), (x, ~1 E fL, x R3, (4.23)
of distribution function (2.8). Concentration neq(z) also enables one to pass from the equilib-
rium versions of the e-expectations (see (2.17)) to the equilibrium versions of the &expectations
(see (2.20)) by means of (2.19). F or example, standard deviation oeq(z) in (3.11) is obtained
from a,,(x) in (4.12) ( see also (4.11)) with the help of equality
(4.24)
which stems from the equilibrium version of (2.27).
5. SUMMARY OF THE PRACTICAL PROCEDURE, DISCUSSION ON THE PROPOSED MODEL AND CONCLUDING REMARKS
This section considers some general aspects of the SCFM model proposed in Section 4.
5.1. Summary of the Model
The almost-equilibrium fluid model in Section 4 can serve as the basis of the practical procedure
to evaluate the equilibrium probability density and the corresponding standard deviation of the
fluid-particle velocity as well as the equilibrium fluid concentration. This procedure includes the
following six steps.
(1) Determine number T of the relaxation mechanisms which should be taken into account
and the corresponding relaxation times (4.20) to be used in (4.21) (see the discussion in
Stochastic Continuum Mechanics 905
Section 4.4). In so doing, the results of Sections 4.2 and 4.3 can be applied. For example,
the relaxation time for the MB (or classical) fluid is independent of the particle velocity
(see (4.16)). Th e relaxation time for the FD (or fermion) fluid is described with (4.17).
The relaxation time corresponding to the Pauli exclusion principle alone is (4.19).
(2) Determine quantity rees[zrr (W(V))“] by means of (4.21).
(3) Make sure that the hypothesis of Theorem 1 holds for function rees resulting from Step (2).
Regarding the globality pointed out in the hypothesis. check at least that this function
passes the test described in Remark 4.1.
(4) Determine the particle-velocity equilibrium probability density pes with (4.9) and (4.8).
(5) Evaluate ces according to (4.12) and then determine the particle-velocity standard devi-
ation geq by means of (4.24).
(6) Evaluate concentration neq from initial value problem (4.22),(3.8) (see also (4.13)) and
distribution function fes from (4.23).
The above procedure is not very complex and does not involve TDL (3.13). The domain R,,
occupied by the fluid may be bounded or unbounded. In the latter case, it may or may not
coincide with the whole physical space R3.
5.2. Discussion on the Proposed Model
The core equation in the almost-equilibrium model is ISDE (4.1) coupled with ISDE (4.6) by
means of (4.7). The dimensionless time scale 0 in (4.6) is abstract. The world in this time scale
is not very interesting. For example, it includes only one fluid, namely, the one with the almost-
equilibrium behavior described with ISDE (4.6). Th e individual features of a specific fluid are
associated only with the actual time t and manifest themselves only in the passing to it by means
of (4.7). In so doing, all these features are represented with a single scaling parameter, relaxation-
time function reTes which, however, can have a fairly complex structure discussed in Section 4.4.
The corresponding expression (4.21) enables one to take into account various physical relaxation
phenomena. Depending on a specific content of the right-hand side of (4.21), one obtains different
equilibrium probability densities as is described in the procedure in Section 5.1.
The core equation (4.1) is, strictly speaking, an assumption. However, it is not of uncertain
features. Indeed, it admits a clear physical reading (see the part of Section 4 above Theorem 1)
and enables one to derive (see Sections 4.2 and 4.3) the characteristics well known in SM or
KT but in the TDL-free way. Moreover, ISDE (4.1) turned out to be helpful even in the cases
which are not very well suited for application of SM or KT. For example (see [17, Sections 4.9
and 4.10]), it underlies the approximate analytical derivation on the long, nonexponential asymp-
totic representation for the particle-velocity covariance in the hard-sphere fluid. This derivation
is fully-time domain (i.e., does not involve any time-space Fourier-like frequency techniques ap-
plicable to only certain, very special responses and under very special conditions), and is valid
regardless of whether the domain is bounded or unbounded. At present, we cannot point out any
result of this kind in SM or KT.
5.3. Concluding Remarks
Various models available in SM or KT are also derived independently, for instance: within
the CFM (or “phenomenological”) approach (e.g., [2]) or the extended thermodynamic theory
(e.g., [40]). In so doing, there is usually no need to involve TDL. An interesting example of
the phenomenological derivation of the kinetic equations is presented in the survey [41j devoted
to tumor dynamics in competition with immune system (see also, recent developments in the
framework of mean field theory in (421). Th us, the resulting models are not a distinguishing
feature of the SM/KT paradigm.
Its cornerstone is the imperative requirement to describe the phenomena from first principles.
This feature, in conjunction with the severely limited set of the employed mathematical models,
906 E. MALIONTO\~ et al.
makes it necessary at certain stages of constructing the comprehensive theory to apply such
assumptions as TDL.
The SCFM model developed in Section 4: summarized in Section 5.1 and discussed in Sec-
tion 5.2, is not concentrated on first principles. The notion of first principles is related to the
feature “to be absolute”. However, allowing for the present state of physics where many major
theoretical fields include a series of still unsolved fundamental difficulties, one has no sufficient
ground to apply the term “first principles” (or the equivalent ones).
The Landau-Lifshitz (LL) stochastic fluid model was proposed in [43] and described in detail
in [44, Section 88 of Part 21. Its key component is a system of stochastic partial differential
equations (SPDEs). The model was developed under the TDL assumption. Hence, it can only be
used within the TDL-based treatments and applied to the problems relevant to TDL. The linear
form of equation [44, (88.11) in Part 21 underlying the derivation (see also, [44, Section 122 or
Part I]) is in fact an assumption. Moreover, this equation is involved heuristically.
Our model is based on nonlinear ISDE (4.1) which is beyond the frames of SM or KT. The
model does not lead to TDL. If necessary, it enables one to derive the TDL-based descriptions (see
Corollaries 1 and 2). In so doing, it provides the corresponding exact expressions for the relaxation
times for the fermion fluid (see (4.17)) and due to the Pauli exclusion alone (see (4.19)). The
model is applicable to the fluids in both unbounded and bounded domains containing arbitrary
large or small (up to the quantum corrections) numbers of the fluid particles. The corresponding
boundary conditions can be accounted as is noted in Remark 4.2. The model provides considerable
flexibility (see Section 4.4) to different user communities (from the basic-aspects researchers to
engineers in industry), to construct the relaxation-time function 7eq depending on the specific
problems, specific purposes of the analysis, admissible complexity, desirable accuracy! and other
issues. The future application of the model will give the material to improve it.
REFERENCES
1. P. Resibois and M. deLeener, Classzcal Kinetzc Theory of Fhids, John Wiley & Sons, New York, (1977).
2 L.I. Sedov, A Course in Continuum Mechanzcs, Vols. I-IV, Wolters-Noordhoff, Groningen, (1971-1972).
3. S.B. Pope, PDF methods for turbulent reactive flows, Pray. Energy Cornbust. Sci. 11 (l), 119-192 (1985). 4. C.W. Gardiner, Handbook of Stochastic Methods for Physacs, Chemistry and the Natural Sczences, Springer-
Verlag, New York, (1994).
5 R.L. Liboff, Kinetzc Theory: Classzcal, Quantum and Relatzvzstzc Descrzptrons, John Wiley & Sons. New
York. (1998).
6. I. Bialynicki-Birula, M. Cieplak and J Kaminski, Theory of Quanta, Oxford Univ. Press, New York. (1992)
7 N. Hashitsume, A statistical theory of linear dissipative systems II, Progr. Theor. Phys. 15 (4). 369-413 (1956).
8. B.B. Kadomtsev. Fluctuation in gases, Sovzet Physpcs JETP 5 (4), 771-772 (1957). 9. H. Ueyama, Fluctuating hydrodynamic equations of mixed and of chemically reacting gases, J. Stat. Phys
23 (4). 463-482 (1980). 10. H. Ueyama, The Langevin-equation approach to dynamics of dense fluids, Progr. Theor. Phys. 66 (6).
1926-1946 (1981).
11. Yu.L. Klimontovich, Kinetic Theory of Nonideal Gases and Nonideal Plasmas, Pergamon Press, Oxford. (1982).
12. D.Ya. Petrina, Methods of derivation of the stochastic Boltzmann hierarchy, Ukrain. Math. Zh. 52 (4), 474-491 (2000).
13. L. Arnold, Stochastic Difierential Equations: Theory and Applications, John Wiley & Sons, New York,
(1974). 14 J. Honerkamp, Statzstical Physzcs: An Advanced Approach wath Applications, Springer-Verlag, Berlin,
(1998). 15. A.H Marshak and C.M. vanvliet, Electrical current and carrier density in degenerate materials with
nonuniform band structure, Proc. IEEE 72 (2), 148-164 (1984). 16. J.S. Blakemore, Approximations for Fermi-Dirac integrals, especially the function F,,z(q) used to describe
electron density in a semiconductor, Solid-State Electron. 25 (ll), 1067-1076 (1982) 17. Y.V. Mamontov and M. Willander, High-Dzmensaonal Nor&near Diffusion Stochastic Processes. Modellrng
for Engzneerzng Applzcations, World Scientific, Singapore, (2001). 18 N.G. van Kampen. Stochastzc Processes zn Physzcs and Chemzslry, North-Holland, Amsterdam, (1981).
Stochastic Continuum Mechanics 907
19. N. Bellomo, A. Palczewski and G. Toscani, Mathematical Topics in Nonlinear Kinetzc Theory, World Scientific, Singapore, (1988).
20. N.N. Bogolubov and N.N. Bogolubov Jr., An Introductzon to Quantum Slatistical Mechanics, Gordon and Breach, Lausanne, (1994).
21. R. Calinon, S. Johannesen, D. Merlini and R. Tartini, The thermodynamic limit with a few particles: The one-dimensional Metha-dyson plasma, In Proceedings of the Conference Held in Ascona and Locarno, July 4-9, 1988, (Edited by S. Albeverio, G. Casati, U. Cataneo, D. Merlini and R. Moresi), World Scientific.
Teaneck (NJ, U.S.A.), pp. 269-276, (1990). 22. R. Kutner, R. Przenioslo and M. Kwiatkowski, Quantum statistics and discreteness. Differences between
the canonical and grand canonical ensembles for a fermionic lattice gas, Ann. Physik 4 (7), 646~667 (1995). 23. S.-C. Kim, Ensemble dependence of confined hard-rod fluids, J. Chem. Phys. 110 (24), 12265-12666 (1999). 24. F.L. RomBn, A. GonzBlez, J.A. White and S. Velasco, Fluctuations in the number of particles of the ideal
gas: A simple example of explicit finite-size effects, Am. J. Phys. 67 (12), 1149-1151 (1999). 25. J.-M. Bismut, Me’canique Al&toire, Springer-Verlag, Berlin, (1981). 26. Ph. Blanchard, Ph. Combe and W. Zheng, Mathematzcal and Physical Aspects of Stochastzc Mechanzcs,
Springer-Verlag, Berlin, (1987). 27. Y.E. Gliklikh, Ordinary and Stochastic Differentzal Geometry as a Tool for Mathematical Physics. Kluwer,
Dordrecht, (1996). 28. H. Haken, Advanced Synergetics. Instability Hierarchaes of Self-Organizing Systems and Devices, Springer-
Verlag, Berlin, (1987). 29. R. Serra, M. Andretta, M. Compiani and G. Zanarini, Introduction to the Physics of Complex Systems.
The Mesoscopic Approach to Fluctuations, Nonlznearity and Self-Organzzation, Pergamon Press. Oxford,
(1986). 30. E. Suraud, On kinetic equations in heavy-ion physics, Ann. Phys. FT. 21 (5), 461-502 (1996). 31. P.L. Bhatnagar, E.F. Gross and M. Krook, A model for collisional processes in gases. I. Small amplitude
processes in charged and neutral one-component systems, Phys. Rev. 94 (3), 511-525 (1954). 32. P. Welander, On temperature jump in a rarefied gas, Arkiv fiir Fysik 7 (44), 507~553 (1954). 33. Yu.V. Prohorov and Yu.A. Rozanov, Probability Theory, Springer-Verlag, Berlin, (1969). 34. X. Mao, Stochastic Di%ferential Equations and their Applications, Horwood, Chichester, (1997). 35. S.Yu. Kun, Relaxation time in the exciton model, Sov. J. Nucl. Phys. 44 (l), 28-30 (1986).
36. C. Kurdak, C.-J. Chen, D.C. Tsui, J.P. Lu, M. Shayegan, S. Parihar and S.A. Lyon, Experimental investi- gation of the Pauli exclusion principle using noise in coherent electron transport, Surface Science 361/362, 705-708 (1996).
37. K. Seeger, Semiconductor Physics, Springer-Verlag, Wien, (1973). 38. B.K. Ridley, Quantum Processes in Semiconductors, Clarendom Press, Oxford, (1988). 39. D.A. Weitz, J.X. Zhu, D.J. Durian, H. Gang and D.J. Pine, Diffusing-wave spectroscopy: The technique
and some applications, Physica Scripta T49B, 610-621 (1993). 40. A.M. Anile, V. Roman0 and G. Russo, Extended hydrodynamical model of carrier transport in semicon-
ductors, SIAM J. Appl. Math. 61 (l), 74-101 (2000). 41. N. Bellomo and L. Preziosi, Modelling and mathematical problems related to tumor evolution and its
interaction with the immune system, Math. Comp. Modelling 32 (3/4), 413-452 (2000).
42. E. de Angelis and L. Mesin, Modelling of the immune response: Conceptual frameworks and applications. Math. Models Meth. Appl. Sci. 11 (9). 1609-1630 (2001).
43. L.D. Landau and E.M. Lifshitz, Hydrodynamic fluctuations, Soviet Physzcs JETP 5 (3), 512-513 (1957). 44. E.M. Lifshitz and L.P. Pitaevskii, Statistical Physzcs, Parts I-2, Pergamon Press, Oxford, (1980). 45. A.M. Anile and S. Pennisi, Thermodynamic derivation of the hydrodynamical model for charge transport
in semiconductors, Phys. Rev. B 46 (20), 13186-13193 (1992). 46. L. Streit, Stochastic differential equations. A pedagogical random walk, In Nonlinear Klein-Gordon and
Schriidinger Systems: Theory and Applications, (Edited by L. VBzquez, L. Streit and V.M. Pkrez-Garcia), World Scientific, River Edge, NJ, U.S.A., pp. 87-108, (1996).