Page 1
arXiv:cond-mat/0507388v1 [cond-mat.stat-mech] 16 Jul 2005
Teaching the Principles of Statistical Dynamics
Kingshuk Ghosh and Ken A. Dill∗
Department of Biophysics, University of California, San Francisco CA 94143
Mandar M. Inamdar, Effrosyni Seitaridou, and Rob Phillips1†
Division of Engineering and Applied Science and
California Institute of Technology, Pasadena, CA 91125
1Kavli Nanoscience Institute,
We describe a simple framework for teaching the principles that underlie the dynamical laws of
transport: Fick’s law of diffusion, Fourier’s law of heat flow, the Newtonian viscosity law, and mass-
action laws of chemical kinetics. In analogy with the way that the maximization of entropy over
microstates leads to the Boltzmann law and predictions about equilibria, maximizing a quantity
that E. T. Jaynes called “Caliber” over all the possible microtrajectories leads to these dynamical
laws.The principle of Maximum Caliber also leads to dynamical distribution functions which
characterize the relative probabilities of different microtrajectories. A great source of recent interest
in statistical dynamics has resulted from a new generation of single-particle and single-molecule
experiments which make it possible to observe dynamics one trajectory at a time.
PACS numbers: 51.10.+d 05.40.-a 05.70.Ln
I.INTRODUCTION
We describe an approach for teaching the principles that underlie the dynamical laws of transport: of particles
(Fick’s law of diffusion), energy (Fourier’s law of heat flow), momentum (the Newtonian law for viscosity),1and
mass-action laws of chemical kinetics.2Recent experimental advances now allow for studies of forces and flows at the
single-molecule and nanoscale level, representative examples of which may be found in the references.3,4,5,6,7,8,9,10For
example, single-molecule methods have explored the packing of DNA inside viruses,7and the stretching of DNA and
RNA molecules.9,10Similarly, video microscopy now allows for the analysis of trajectories of individual submicron
size colloidal particles11, and the measurement of single-channel currents has enabled the kinetic studies of DNA
translocation through nanopores.3,6
One of the next frontiers in biology is to understand the “small numbers” problem: how does a biological cell
function, given that most of its proteins and nucleotide polymers are present in numbers much smaller than Avogadro’s
number?12For example, one of the most important molecules, a cell’s DNA, occurs in only a single copy. Also, it
is the flow of matter and energy through cells that makes it possible for organisms to maintain a relatively stable
form.13Hence, in order to function, cells always have to be in this state far from equilibrium. Thus, many problems of
current interest involve small systems that are out of equilibrium. Our interest here is two-fold: to teach our students
a physical foundation for the phenomenological macroscopic laws, which describe the properties of averaged forces and
flows, and to teach them about the dynamical fluctuations, away from those average values, for systems containing
small numbers of particles.
In this article, we describe a very simple way to teach these principles. We start from the “principle of Maximum
Caliber”, first described by E. T. Jaynes.14It aims to provide the same type of foundation for dynamics of many-
degree-of-freedom systems that the second law of thermodynamics provides for equilibria of such systems. To illustrate
the principle, we use a slight variant of one of the oldest and simplest models in statistical mechanics, the Dog-Flea
Model, or Two-Urn Model.15,16Courses in dynamics often introduce Fick’s law, Fourier’s law, and the Newtonian-fluid
model as phenomenological laws, rather than deriving them from some deeper foundation. Here, instead, we describe a
simple unified perspective that we have found useful for teaching these laws from a foundation in statistical dynamics.
In analogy with the role of microstates as a basis for the properties of equilibria, we focus on microtrajectories as the
basis for predicting dynamics. One argument that might be leveled against the kind of framework presented here is that
in the cases considered here, it is not clear that it leads to anything different from what one obtains using conventional
nonequilibrium thinking. On the other hand, often, restating the same physical result in different language can provide
a better starting point for subsequent reasoning. This point was well articulated by Feynman in his Nobel lecture17
who noted: “Theories of the known, which are described by different physical ideas may be equivalent in all their
predictions and are hence scientifically indistinguishable. However, they are not psychologically identical when trying
to move from that base into the unknown. For different views suggest different kinds of modifications which might be
made and hence are not equivalent in the hypotheses one generates from them in one’s attempt to understand what
is not yet understood.”
We begin with the main principle embodied in Fick’s Law. Why do particles and molecules in solution flow from
Page 2
2
regions of high concentration toward regions of low concentration? To keep it simple, we consider one-dimensional
diffusion along a coordinate x. This is described by Fick’s first law of particle transport,1,2which says that the average
flux, ?J?, is given in terms of the gradient of average concentration, ∂?c?/∂x, by
?J? = −D∂?c?
∂x
(1)
where D is the diffusion coefficient. In order to clearly distinguish quantities that are dynamical averages from those
that are not, we indicate the averaged quantities explicitly by brackets, ?...?. We describe the nature of this averaging
below, and the nature of the dynamical distribution functions over which the averages are taken. But first, we briefly
review the standard derivation of the diffusion equation. Combining Fick’s first law with particle conservation,
∂?c?
∂t
= −∂?J?
∂x,
(2)
gives Fick’s second law, also known as the diffusion equation:
∂?c?
∂t
= D∂2?c?
∂x2.(3)
Solving Eq. (3) subject to two boundary conditions and one initial condition gives both ?c(x,t)?, the average con-
centration in time and space, and the average flux ?J(x,t)?, when no other forces are present. The generalization to
situations involving additional applied forces is the Smoluchowski equation.2
A simple experiment shows the distinction between averaged quantities vs. individual microscopic realizations.
Using a microfluidics chip like that shown in Fig. (1a), it is possible to create a small fluid chamber divided into
two regions by control valves. The chamber is filled on one side with a solution containing a small concentration
of micron-scale colloidal particles. The other region contains just water. The three control valves on top of that
microfluidic chamber serve two purposes: The two outer ones are used for isolation so that no particles can diffuse
in and out of the chamber, while the middle control valve provides the partition between the two regions. The time
evolution of the system is then monitored after the removal of the partition (see Fig. (1b)). The time-dependent
particle density is determined by dividing the chamber into a number of equal-sized boxes along the long direction
and by computing histograms of the numbers of particles in each slice as a function of time. This is a colloidal solution
analog of the gas diffusion experiments of classical thermodynamics. The corresponding theoretical model usually
used is the diffusion equation. Fig. (1c) shows the solution to the diffusion equation, as a function of time, for the
geometry of the microfluidics chip. The initial condition is a step function in concentration at x = 200µm at time
t = 0.
We use this simple experiment to illustrate one main point. The key distinction is that the theoretical curves are
very smooth, while there are very large fluctuations in the experimentally observed dynamics of the particle densities.
The fluctuations are large because the number of colloidal particles is small, tens to hundreds. The experimental data
shows that the particle concentration c(x,t) is a highly fluctuating quantity. It is not well described by the standard
smoothed curves that are calculated from the diffusion equation. Of course, when averaged over many trajectories
or when particles are at high concentrations, the experimental data should approach the smoothed curves that are
predicted by the classical diffusion equation. Our aim here is to derive Fick’s Law and other phenomenological
transport relations at a microscopic level, so that we can consider both the behaviors of average properties and the
fluctuations, i.e., the dynamical distribution functions, and to illustrate the Maximum Caliber approach.
II. THE EQUILIBRIUM PRINCIPLE OF MAXIMUM ENTROPY
We are interested here in dynamics, not statics. However, our strategy follows so closely the Jaynes derivation of
the Boltzmann distribution law of equilibrium statistical mechanics,2,18that we first show the equilibrium treatment.
To derive the Boltzmann law, we start with a set of equilibrium microstates i = 1,2,3,...N that are relevant to the
problem at hand. We aim to compute the probabilities piof those microstates in equilibrium. We define the entropy,
S, of the system as
S({pi}) = −kB
N
?
i=1
pilnpi,(4)
Page 3
3
where kB is the Boltzmann constant. The equilibrium probabilities, pi = p∗
entropy to be maximal, subject to two constraints:
iare those values of pi that cause the
N
?
i=1
pi= 1,(5)
which is a normalization condition that insures that the probabilities pi’s sum to one, and
?E? =
?
i
piEi, (6)
which says that the energies, when averaged over all the microstates, sum to the macroscopically observable average
energy.
Using Lagrange multipliers λ and β to enforce the first and second constraints, respectively, leads to an expression
for the values p∗
ithat maximize the entropy:
?
i
[−1 − lnp∗
i− λ − βEi] = 0. (7)
The result is that
p∗
i=e−βEi
Z
, (8)
where Z is the partition function, defined by
Z =
?
i
e−βEi.(9)
After a few thermodynamic arguments, the Lagrange multiplier β can be shown to be equal to 1/kBT.18This
derivation, first given in this simple form by Jaynes,18identifies the probabilities that are both consistent with the
observable average energy and that otherwise maximize the entropy. Jaynes justified this strategy on the grounds that
it would be the best prediction that an observer could make, given the observable, if the observer is ignorant of all else.
In this case, the observable is the average energy. While this derivation of the Boltzmann law is now quite popular,
its interpretation as a method of prediction, rather than as a method of physics, is controversial. Nevertheless, for
our purposes here, it does not matter whether we regard this as a description of physical systems or as a strategy for
making predictions.
Now, we switch from equilibrium to dynamics, but we use a similar strategy. We switch from the Principle of
Maximum Entropy to what Jaynes called the Principle of Maximum Caliber.14In particular, rather than focusing on
the probability distribution p(Ei) for the various microstates, we seek p[{σi(t)}], where σi(t) is the ithmicroscopic
trajectory of the system. Again we maximize an entropy-like quantity, obtained from p[{σi(t)}], to obtain the predicted
distribution of microtrajectories. If there are no constraints, this maximization results in the prediction that all the
possible microtrajectories are equally likely during the dynamical process. In contrast, certain microtrajectories will
be favored if there are dynamical constraints, such as may be specified in terms of the average flux.
In the following section, we use the Maximum Caliber strategy to derive Fick’s Law using the Dog-Flea model,
which is one of the simplest models that contain the physics of interest.
III.FICK’S LAW FROM THE DOG-FLEA MODEL
We want to determine the diffusive time evolution of particles in a one-dimensional system. The key features of this
system are revealed by considering two columns of particles separated by a plane, as shown in Fig. (2). The left-hand
column (1) has N1(t) particles at time t and the right-hand column (2) has N2(t) particles. This is a simple variant of
the famous “Dog-Flea” model of the Ehrenfest’s introduced in 1907.15,16Column (1) corresponds to Dog (1), which
has N1fleas on its back at time t, and column (2) corresponds to Dog (2), which has N2fleas at time t. In any time
interval between time t and t + ∆t, any flea can either stay on its current dog, or it can jump to the other dog. This
model has been used extensively to study the Boltzmann H-theorem and to understand how the time asymmetry of
diffusion processes arises from an underlying time symmetry in the laws of motion.15,16,19Our model is used for a
slightly different purpose. In particular, our aim is to take a well-characterized problem like diffusion and to reveal
how the Principle of Maximum Caliber may be used in a concrete way. We follow the conventional definition of flux,
Page 4
4
of a number of particles transferred per unit time, and per unit area. For simplicity, we take the cross-sectional area
to be unity.
First, consider the equilibrium state of our Dog-Flea model. The total number of ways of partitioning the (N1+N2)
fleas (particles) is W(N1,N2),
W(N1,N2) =(N1+ N2)!
N1!N2!
. (10)
The state of equilibrium is that for which the entropy, S = kBlnW is maximal. A simple calculation shows that the
entropy is maximal when the value N1= N∗
dogs will have approximately the same number of fleas, in the absence of any bias.
Our focus here is on the dynamics – on how the system reaches that state of equilibrium. We discretize time into a
series of intervals ∆t. We define a dynamical quantity p, which is the probability that a particle (flea) jumps from one
column (dog) to the other in any time interval ∆t. Thus, the probability that a flea stays on its dog during that time
interval is q = 1 − p. We assume that p is independent of time t, and that all the fleas and jumps are independent of
each other.
In equilibrium statistical mechanics, the focus is on the microstates. However, for dynamics we focus on processes,
which, at the microscopic level, we call the microtrajectory. Characterizing the dynamics requires more than just
information about the microstates; we must also consider the processes. Let m1 represent the number of particles
that jump from column (1) to (2), and m2is the number of particles that jump from column (2) to (1), between time
t and t+∆t. There are many possible different values of m1and m2: it is possible that no fleas will jump during the
interval ∆t, or that all the fleas will jump, or that the number of fleas jumping will be in between those limits. Each
one of these different situations corresponds to a distinct microtrajectory of the system in this idealized dynamical
model. We need a principle to tell us what number of fleas will jump during the time interval ∆t at time t. Because
the dynamics of this model is so simple, the implementation of the caliber idea is reduced to nothing more than a
simple exercise in enumeration and counting using the binomial distribution.
1is as nearly equal to N2= N∗
2as possible. In short, at equilibrium, both
A. The Dynamical Principle of Maximum Caliber
The probability, Wd(m1,m2|N1,N2), that m1particles jump to the right and that m2particles jump to the left in
a discrete unit of time ∆t, given that there are N1(t) and N2(t) fleas on the dogs at time t is
?
?
Wdis a count of microtrajectories in dynamics problems in the same vein that W counts microstates for equilibrium
problems. In the same spirit that the Second Law of Thermodynamics says to maximize W to predict states of
equilibrium, now for dynamics, we maximize Wd over all the possible microtrajectories (i.e. over m1 and m2) to
predict the fluxes of fleas between the dogs. This is the implementation of the Principle of Maximum Caliber within
this simple model. Maximizing Wdover all the possible processes (different values of m1and m2) gives our prediction
(right flux m1= m∗
Since the jumps of the fleas from each dog are independent, we find our predicted macroscopic dynamics by
maximizing Wd1and Wd2separately, or for convenience their logarithms:
????
Note that applying Stirling’s approximation to Eq. (11) Wdgives:
Wd(m1,m2|N1(t),N2(t)) =pm1qN1−m1
N1!
m1!(N1− m1)!
??
?
?
Wd1
?
?
pm2qN2−m2
N2!
m2!(N2− m2)!
??
?
?
Wd2
.(11)
1and left flux m2= m∗
2) for the macroscopic flux that we should observe in experiments.
∂ lnWdi
∂mi
N,mi=m∗
i
= 0,i = 1,2. (12)
lnWdi= milnp + (Ni− mi)lnq + NilnNi− milnmi− (Ni− mi)ln(Ni− mi).
We call C= lnWdthe caliber. Maximizing C with respect to m gives
∂ lnWdi
∂mi
(13)
= lnp − lnq − lnm∗
i+ ln(Ni− m∗
i) = 0.(14)
This result may be simplified to yield
ln
?
m∗
i
Ni− m∗
i
?
= ln
?
p
1 − p
?
,(15)
Page 5
5
which implies that the most probable jump number is simply given by
m∗
i= pNi.(16)
But since our probability distribution Wd is nearly symmetric about the most probable value of flux, the average
number and the most probable number are approximately the same. Hence, the average net flux to the right will be,
?J(t)? =m∗
1− m∗
∆t
2
= p
?N1(t) − N2(t)
∆t
?
≈ −p∆x2
∆t
∆c(x,t)
∆x
which is Fick’s law, in this simple model, and where the diffusion constant is given by D = p∆x2/∆t. We have
rewritten N1− N2= −∆c∆x.
Hence we have a simple explanation for why there is a net flux of particles diffusing across a plane down a concen-
tration gradient: more microscopic trajectories lead downhill than uphill. It shows that the diffusion constant D is a
measure of the jump rate p. This simple model does not make any assumptions that the system is “near-equilibrium”,
i.e., utilizing the Boltzmann distribution law, for example, and thus it indicates that Fick’s Law ought also to apply
far from equilibrium. For example, we could have imagined that for very steep gradients, Fick’s Law might have been
only an approximation and that diffusion is more accurately represented as a series expansion of higher derivatives of
the gradient. But, at least within the present model, Fick’s Law is a general result which emerges from counting up
microtrajectories. On the other hand, we would expect Fick’s law to break down when the particle density becomes so
high that the particles start interacting with each other thus spoiling the assumption of independent particle jumps.
B.Fluctuations in Diffusion
Above, we have shown that the most probable number of fleas that jump from dog (1) to dog (2) between time t
and t+∆t is m∗
1= pN1(t). The model also tells us that sometimes we will have fewer fleas jumping during that time
interval, and sometimes we will have more fleas. These variations are a reflection of the fluctuations resulting from
the system following different microscopic pathways.
We focus now on predicting the fluctuations. To illustrate, let us first make up a table of Wd, the different numbers
of possible microtrajectories, taken over all the values of m1and m2(Table (I)). To keep this illustration simple, let
us consider the following particular case: N1(t) = 4 and N2(t) = 2. Let us also assume p = q = 1/2. Here, then, are
the multiplicities of all the possible routes of flea flow. A given entry tells how many microtrajectories correspond to
the given choice of m1and m2.
Notice first that the table confirms our previous discussion. The dynamical process for which Wdis maximal (12
microtrajectories, in this case), occurs when m∗
compute the probability of that particular flux by dividing Wd = 12 by the sum of entries in this table, which is
26= 64 the total number of microtrajectories. The result, which is the fraction of all the possible microtrajectories
that have m∗
2= 1, is 0.18. We have chosen an example in which the particle numbers are very small,
so the fluctuations are large; they account for more than 80 percent of the flow. In systems having large numbers of
particles, the relative fluctuations are much smaller than this.
Now look at the top right corner of this table. This entry says that there is a probability of 1/64 that both fleas on
dog (2) will jump to the left while no fleas will jump to the right, implying that the net flux, for that microtrajectory,
is actually backwards, relative to the concentration gradient. We call these “bad actor” microtrajectories. In those
cases, particles flow to increase the concentration gradient, not decrease it. Traditionally, “Maxwell’s Demon” was an
imaginary microscopic being that was invoked in similar situations in heat flow processes, i.e., where heat would flow
from a colder object to heat up a hotter one, albeit with low probability.20In particular, the Demon was supposed
to capture the bad actor microtrajectories. At time t, there are 4 fleas on the left dog, and 2 on the right. At the
next instant in time, t +∆t, all 6 fleas are on the left dog, and no fleas are on the right-hand dog. Notice that this is
not a violation of the Second Law, which is a tendency towards maximum entropy, because the Second Law is only a
statement that the average flow must increase the entropy; it says nothing about the fluctuations.
Similarly, if you look at the bottom left corner of the table, you see a regime of “superflux”: a net flux of 4 particles
to the right, whereas Fick’s Law predicts a net flow of only 2 particles to the right. This table illustrates that Fick’s
Law is only a description of the average or most probable flow, and it shows that Fick’s Law is not always exactly
correct at the microscopic level. However, such violations of Fick’s Law are of low probability, a point that we will
make more quantitative below. Such fluctuations have been experimentally measured in small systems.21
We can further elaborate on the fluctuations by defining the “potencies” of the microtrajectories. We define the
potency to be the fraction of all the trajectories that lead to a substantial change in the macrostate. The potencies of
trajectories depend upon how far the system is from equilibrium. To see this, let us continue with our simple system
1= pN1 = 1/2 × 4 = 2, and m∗
2= pN2 = 1/2 × 2 = 1. You can
1= 2 and m∗
Page 6
6
having 6 particles. The total number of microscopic trajectories available to this system at each instant in our discrete
time picture is 26= 64. Suppose that at t = 0 all 6 of these particles are in dog (1). The total number of microscopic
trajectories available to the system can be classified once again using m1 and m2, where in this case m2= 0 since
there are no fleas on dog (2) (See Table. (II)).
What fraction of all microtrajectories changes the occupancies of both dogs by more than some threshold value,
say ∆Ni> 1? In this case, we find that 57 of the 64 microtrajectories cause a change greater than this to the current
state. We call these potent trajectories.
Now, let us look at the potencies of the same system of 6 particles in a different situation, N1 = N2 = 3 when
the system is in macroscopic equilibrium (see Table. (III)). Here, only the trajectories with (m1,m2) pairs given by
(0,2),(0,3),(1,3),(2,0),(3,0), and (3,1) satisfy our criterion. Summing over all of these outcomes shows that just 14
of the 64 trajectories are potent in this case.
There are two key observations conveyed by these arguments. First, for a system far from equilibrium, the vast
majority of trajectories at that time t are potent, and move the system significantly away from its current macrostate.
Second, when the system is near equilibrium, the vast majority of microtrajectories leave the macrostate unchanged.
Let us now generalize from the tables above, to see when fluctuations will be important.
1.Fluctuations and Potencies
A simple way to characterize the magnitude of the fluctuations is to look at the width of the Wd distribution.2
It is shown in standard statistics texts that for a binomial distribution such as ours, for which the mean and most
probable value both equal m∗
i= Nipq. The variance characterizes the width. Moreover, if
Niis sufficiently large, a binomial distribution can be well-approximated by a Gaussian distribution
i= Npi, the variance is σ2
P(mi,Ni) =
1
√2πNipqexp
?
−(mi− Nip)2
2Nipq
?
, (17)
an approximation we find convenient since it leads to simple analytic results. However, this distribution function is
not quite the one we want. We are interested in the distribution of flux, P(J) = P(m1−m2), not the distribution of
right-jumps m1or left-jumps m2alone, P(m).
However, due to a remarkable property of the Gaussian distribution, it is simple to compute the quantity we want.
If you have two Gaussian distributions, one with mean ?x1? and variance σ2
variance σ2
2, then the distribution function, P(x1− x2) for the difference will also be a Gaussian distribution, having
mean ?x1? − ?x2? and variance σ2= σ2
For our binomial distributions, the means are m∗
σ2
2= N2pq, so the distribution of the net flux, J = m1− m2is
1
?2π(pqN)exp
where N = N1+ N2.
Figure (3) shows an example of the distributions of fluxes at different times, using p = 0.1, and starting from
N1= 100,N2= 0. We update each time step using an averaging scheme, N1(t+∆t) = N1(t)−N1(t)p+N2(t)p. The
figure shows how the mean flux is large at first and decays towards equilibrium, J = 0. This result could also have
been predicted from the diffusion equation. However, equally interesting are the wings of the distributions, which
show the deviations from the average flux, and these are not predictable from the diffusion equation. One measure
of the importance of the fluctuations in a given dynamical problem is the ratio of the standard deviation σ, to the
mean,
√σ2
J
1, and the other with mean ?x2? and
1+ σ2
2.
1= pN1 and m∗
2= pN2 and the variances are σ2
1= N1pq and
P(J) =
?
−(J − p(N1− N2))2
2pqN
?
,(18)
=
√Npq
(N1− N2)p.
(19)
In the limit of large N1, the above reduces to,
√σ2
J
=
√Npq
(N1− N2)p∼ N−1/2. (20)
In a typical bulk experiment, particle numbers are large, of the order of Avogadro’s number 1023. In such cases, the
width of the flux distribution is exceedingly small and it becomes overwhelmingly probable that the mean flux will
Page 7
7
be governed by Fick’s law. However, within biological cells and in applications involving small numbers of particles,
the variance of the flux can become significant. It has been observed that both rotary and translational single motor
proteins sometimes transiently step backwards, relative to their main direction of motion.22
As a measure of the fluctuations, we now calculate the variance in the flux. It follows from Eq. (18) that ?J2? = Npq,
where N = N1+ N2. Thus, we can represent the magnitude of the fluctuations as δ,
δ =
?
?(∆J)2?
?J?2
=
√Npq
pfN
∝1
f
?q
pN−1,
where the quantity N = N1+ N2is the total number of fleas and f = (N1− N2)/(N), the normalized concentration
difference. The quantity δ is also a measure of the degree of backflux. In the limit of large N, δ goes to zero. That is,
the noise diminishes with system size. However, even when N is large, δ can still be large (indicating the possibility
of backflux) if the concentration gradient, N1− N2, is small.
Let us look now at our other measure of fluctuations, the potency. Trajectories that are not potent should have
|m1− m2| ≈ 0 which corresponds to a negligible change in the current state of the system as a result of a given
microtrajectory. In Fig. (4), the impotent microtrajectories are shown as the shaded band for which m1≈ m2. To
quantify this, we define impotent trajectories as those for which |m1− m2| ≤ h, (h ≪ N). In the Gaussian model,
the fraction of impotent trajectories is
Φimpotent ≈
?h
1
2
−h
?
dJ
1
√2πNpqexp
?h + (N1− N2)p
?−(J − (N1− N2)p)2
?
2Npq
?h − (N1− N2)p
?
(21)
=
erf
√2Npq
+ erf
√2Npq
??
, (22)
and corresponds to summing over the subset of trajectories that have a small flux. To keep it simple, we did a
computation, taking p = q = 1/2, and for which the expression for the probability distribution for the microscopic
flux m1− m2is given by Eq. (18). The choice of h is arbitrary, so let us just choose h to be one standard deviation,
?N/4. Fig. (5) shows potencies for various values of N1and N2. When the concentration gradient is large, most
gradient is small, most trajectories have little effect on the macrostate.
As another measure of fluctuations, let us now consider the “bad actors” (see Fig. (6)). If the average flux is in the
direction from dog (1) to dog (2), what is the probability you will observe flux in the opposite direction (bad actors)?
Using Eq. (18) for P(J), we get
trajectories are potent, leading to a statistically significant change of the macrostate, whereas when the concentration
Φbadactors ≈
?0
1
2
−∞
?
1
√2πNpqexp
?(N1− N2)p
?−(J − (N1− N2)p)2
??
2Npq
?
(23)
=
1 − erf
√2Npq
,N2> N1
(24)
which amounts to summing up the fraction of trajectories for which J ≤ 0. Figure (7) shows the fraction of bad
actors for p = q = 1/2. Bad actors are rare when the concentration gradient is large, and highest when the gradient
is small. The discontinuity in the slope of the curve in Fig. (7) at N1/N = 0.5 is a reflection of the fact that the mean
flux abruptly changes sign at that value.
IV. FOURIER’S LAW OF HEAT FLOW
While particle flow is driven by concentration gradients, according to Fick’s law, ?J? = −D∂c
by temperature gradients, according to Fourier’s law1:
∂x, energy flow is driven
?Jq? = −κ∂T
∂x.
Here, Jq is the energy transferred per unit time and per unit cross-sectional area by heat flow and ∂T/∂x is the
temperature gradient that drives it, indicated here for the one-dimensional case. κ, the thermal conductivity,1plays
the role that the diffusion coefficient plays in Fick’s Law.
To explore Fourier’s law, we return to the Dog-Flea model as described in Sec. III. Now, columns (1) and (2) can
differ not only in their particle numbers, N1(t) and N2(t), but also in their temperatures, T1(t) and T2(t). To keep
Page 8
8
it simple here, we assume that each column is at thermal equilibrium and that each particle that jumps carries with
it the average energy, ?mv2/2? = kBT/2 from the column it left. Within this simple model, all energy is transported
by hot or cold molecules switching dogs. Although in general, heat can also flow by other mechanisms mediated by
collisions, for example, our aim here is just the simplest illustration of principle. The average heat flow at time t is
?Jq? =m∗
1
∆t(kBT1/2) −m∗
2
∆t(kBT2/2) =pkB
∆t[N1T1− N2T2](25)
where m∗
particle numbers are identical, N/2 = N1= N2, then
1and m∗
2are, as defined in Sec. IIIA the numbers of particles jumping from each column at time t. If the
?Jq? =pkBN
∆t
(T1− T2) = −κ∆T
∆x,
which is Fourier’s Law for the average heat flux, within this two-column model. The model predicts that the thermal
conductivity is κ = (pkBN∆x)/(∆t), which can be expressed in a more canonical form as κ = pkBnvav∆x, when
written in terms of the particle density n = N/∆x and the average velocity, vav= ∆x/∆t. Our simple model gives
the same thermal conductivity as given by the kinetic theory of gases,1κ = (1/2)kBnvavl, where l is the mean free
path, if p∆x in our model corresponds to l/2, half the mean-free path. Hence, this simple model captures the main
physical features of heat flow, again by appealing to the idea of summing over the weighted microtrajectories available
to the system.
V. NEWTONIAN VISCOSITY
Another phenomenological law of gradient-driven transport is that of Newtonian viscosities,1
τ = ηdvy
dx,
where τ is the shear stress that is applied to a fluid, dvy/dx is the resultant shear rate, and the proportionality
coefficient, η, is the viscosity of a Newtonian fluid. Whereas Fick’s law describes particle transport and Fourier’s law
describes energy transport, this Newtonian law describes the transport (in the x-direction, from the top moving plate
toward the bottom fixed plate) of linear momentum that acts in the y-direction (parallel to the plates) (see Fig. (8)).
Returning to the Dog-Flea model of Sec. III, suppose each particle in column (1) carries momentum mvy1along the
y-axis, and that m∗
1particles hop from column (1) to (2) at time t, carrying with them some linear momentum. As
before, we consider the simplest model that every particle carries the same average momentum from the column it
leaves to its destination column.
The flux, Jp, is the amount of y-axis momentum that is transported from one plane to the next in the x-direction,
per unit area:
?Jp? =m∗
1
∆t(mvy1) −m∗
2
∆t(mvy2) =pm
∆t[N1vy1− N2vy2].
If the number of particles is the same in both columns, N/2 = N1= N2, this simplifies to
?Jp? =pmN
∆t
[vy1− vy2] = η∆vy
∆x,
which is the Newtonian law of viscosity in this two-column model. The viscosity is predicted by this model to be
η = (pmN∆x)/(∆t). Converting this to a more canonical form gives η = pmnvav∆x, where n = N/∆x is the particle
density, and vav= ∆x/∆t is the average velocity. This is equivalent to the value given by the kinetic theory of gases,1
η = (1/3)mnlvav, if p∆x from our model equals (1/3)l, one-third of the mean-free path length. Note that this simple
model based upon molecular motions will clearly not be applicable to complex fluids where the underlying molecules
possess internal structure.
VI.CHEMICAL KINETICS WITHIN THE DOG-FLEA MODEL
Let us now look at chemical reactions using the Dog-Flea model. Chemical kinetics can be modeled using the
Dog-Flea model when the fleas have preference for one dog over the other. Consider the reaction
A
kf
⇋
krB.
Page 9
9
The time-dependent average concentrations, [A](t) and [B](t) are often described by chemical rate equations,2
d[A]
dt
d[B]
dt
= −kf[A] + kr[B]
= kf[A] − kr[B] (26)
where kf is the average conversion rate of an A to a B, and kris the average conversion rate of a B to an A. These
rate expressions describe only average rates; they do not tell us the distribution of rates. Some A’s will convert to B’s
faster than the average rate kf[A] predicts, and some will convert more slowly. Again, we use the Dog-Flea model as
a microscopic model for this process. We use it to consider both the average concentrations and the fluctuations in
concentrations.
Now, dog (1) represents chemical species A and dog (2) represents chemical species B. The net chemical flux from
1 to 2 is given by Jc= m∗
situations is that now the intrinsic jump rate from column 1 (species A), p1, is different than the jump rate from
column 2, p2. This simply reflects the fact that a forward rate can differ from a backward rate in a chemical reaction.
Now, fleas have a different escape rate from each dog. Fleas escape from Dog (1) at rate p1and fleas escape from Dog
(2) at rate p2. Maximizing Wdgives m∗
as the most probable flux because of the approximately symmetric nature of the binomial distribution) at time t is,
1−m∗
2. What is different about our model for these chemical processes than in our previous
1= N1p1and m∗
2= N2p2, so the average flux (which is the almost the same
?J? = N1p1− N2p2= kf[A] − kr[B],
which is just the standard mass-action rate law, expressed in terms of the mean concentrations. The mean values
satisfy detailed balance at equilibrium ( ?J?=0 ⇒ N2/N1= p1/p2= kf/kr).
More interesting than the well-known behavior of the mean chemical reaction rate is the fluctuational behavior.
For example, if the number of particles is small, then even when kf[A] − kr[B] > 0, indicating an average conversion
of A’s to B’s, the reverse can happen occasionally instead. When will these fluctuations be large? As in Sec. III, we
first determine the probability distribution of the flux J. In this case, the probability distribution becomes:
P(J) =
1
?2π(p1q1N1+ p2q2N2)exp
?
−(J − (N1p1− N2p2))2
2(p1q1N1+ p2q2N2)
?
, (27)
Again, let us use this flux distribution function to consider the fluctuations in the chemical reaction. The relative
variance in the flux is
√N1p1q1+ N2p2q2
N1p1− N2p2
As before, the main message is that when the system is not yet at equilibrium (i.e., the denominator is non-zero),
macroscopically large systems will have negligibly small fluctuations. The relative magnitude of fluctuations scales
approximately as N−(1/2). Let us also look at the potencies of microtrajectories as another window into fluctuations.
Using Eq. (22) with p1and p2gives the fraction of trajectories that are impotent as
?(∆J)2?
(?J?)2
=.
Φimpotent ≈
?h
1
2
−h
?
dJ
1
?2π(N1p1q1+ N2p2q2)exp
h + (N1p1− N2p2)
?2(N1p1q1+ N2p2q2)
?−(J − (N1p1− N2p2))2
?
2(N1p1q1+ N2p2q2)
?
(28)
=
erf
?
?
+ erf
h − (N1p1− N2p2)
?2(N1p1q1+ N2p2q2)
??
,(29)
Using N1+ N2= N = 100, and p1= 0.1 and p2= 0.2, Φpotent= 1 − Φimpotentas a function of N1/N is shown in
Fig. 9.
VII.DERIVING THE DYNAMICAL DISTRIBUTION FUNCTION FROM MAXIMUM CALIBER
Throughout this paper, we have used the binomial distribution function, Wd, as the basis for our treatment of
stochastic dynamics. The Maximum Caliber idea says that if we find the value of Wdthat is maximal with respect to
the microscopic trajectories, this will give the macroscopically observable flux. Here, we now restate this in a more
general way, and in terms of the probabilities of the trajectories.
Page 10
10
Let P(i) be the probability of a microtrajectory i during the interval from time t to t + ∆t. A microtrajectory is
a specific set of fleas that jump; for example microtrajectory i = 27 might be the situation in which fleas number 4,
8, and 23 jump from dog (1) to (2). We take as a constraint the average number of jumps, ?m?, the macroscopic
observable. The quantity mi = 3 in this case indicates that trajectory i involves 3 fleas jumping. We express the
caliber C as
C =
i
?
P(i)lnP(i) − λ
?
i
miP(i) − α
?
i
P(i) (30)
where, λ is the Lagrange multiplier that enforces the constraint of the average flux and α is the Lagrange multiplier
that enforces the normalization condition that the P(i)’s sum to one. Maximizing the caliber gives the populations
of the microtrajectories,
P(i) = exp(−α − λmi). (31)
Note that the probability P(i) of the ithtrajectory depends only on the total number miof the jumping fleas. Also,
all trajectories with the same mi have same probabilities. Now, in the same way that it is sometimes useful in
equilibrium statistical mechanics to switch from microstates to energy levels, we now express the population P(i) of a
given microtrajectory, instead, in terms of ρ(m) the fraction of all the microtrajectories that involve m jumps during
this time interval,
ρ(m) = g(m)Q(m), (32)
where, g(m) = N!/[m!(N − m)!] is the “density of trajectories” with flux m (in analogy with the density of states
for equilibrium systems), and Q(m) is the probability P(i) of the microtrajectory i with mi= m. In other words, i
denotes a microtrajectory (specific set of fleas jumping) while m denotes a microprocess (the number of fleas jumping).
The total number of i’s associated with a given m is precisely g(m). It can also be easily seen that,
?
i
P(i) =
N
?
?
m=0
g(m)Q(m) =
N
?
m=0
ρ(m) = 1 (33)
?m? =
i
miP(i) =
N
?
m=0
mg(m)Q(m) =
N
?
m=0
mρ(m). (34)
Thus, the distribution of jump-processes written in terms of the jump number, m, is
ρ(m) =
N!
m!(N − m)!exp(−α)exp(−λm). (35)
The Lagrange multiplier α can be eliminated by summing over all trajectories and requiring that?N
N
?
Combining Eqs. (35) and (36) gives
mi=0ρ(mi) = 1,
i.e.,
exp(α) =
m
g(m)e−λm=
N
?
m
N!
m!(N − m)!e−λm= (1 + e−λ)N. (36)
ρ(m) =
N!
m!(N − m)!
exp(−λm)
(1 + exp(−λ))N. (37)
If we now let
p =
exp(−λ)
1 + exp(−λ), (38)
then we get
pm=
exp(−λm)
(1 + exp(−λ))m,(39)
Page 11
11
and
(1 − p)N−m=
1
(1 + exp(−λ))N−m.(40)
Combining Eqs. (37), (39), and (40) gives the simple form
ρ(m) =
N!
m!(N − m)!pm(1 − p)(N−m),
that we have used throughout this paper in Eq. (11).
VIII.SUMMARY AND COMMENTS
We have shown how to derive the phenomenological laws of nonequilibrium transport, including Fick’s law of
diffusion, the Fourier law of heat conduction, the Newtonian law of viscosity, and mass-action laws of chemical
kinetics, from a simple physical foundation that can be readily taught in elementary courses. We use the Dog-Flea
model, originated by the Ehrenfests, for describing how particles, energy, or momentum can be transported across a
plane. We combine that model with the Principle of Maximum Caliber, a dynamical analog of the way the Principle
of Maximum Entropy is used to derive the laws of equilibrium. In particular, according to Maximum Entropy, you
maximize the entropy S(p1,p2,...,pN) with respect to the probabilities piof N microstates, subject to constraints,
such as the requirement that the average energy is known. That gives the Boltzmann distribution law. Here, for
dynamics, we focus on microtrajectories, rather than microstates, and we maximize a dynamical entropy-like quantity,
subject to an average flux constraint. In this way, maximizing the caliber is the dynamical equivalent of minimizing a
free energy for predicting equilibria. A particular value of this approach is that it also gives us fluctuation information,
not just averages. In diffusion, for example, sometimes the flux can be a little higher or lower than the average value
expected from Fick’s Law. These fluctuations can be particularly important for biology and nanotechnology, where
the numbers of particles can be very small, and therefore where there can be significant fluctuations in rates, around
the average.
Acknowledgments
It is a pleasure to acknowledge the helpful comments and discussions with Dave Drabold, Mike Geller, Jan´ e Kondev,
Stefan M¨ uller, Hong Qian, Darren Segall, Pierre Sens, Jim Sethna, Ron Siegel, Andrew Spakowitz, Zhen-Gang Wang,
and Paul Wiggins. We would also like to thank Sarina Bromberg for the help with the figures. KAD and MMI would
like to acknowledge support from NIH grant number R01 GM034993 . RP acknowledges support from NSF grant
number CMS-0301657, the Keck foundation, NSF NIRT grant number CMS-0404031, and NIH Director’s Pioneer
Award grant number DP1 OD000217 .
∗Electronic address: dill@maxwell.ucsf.edu
†Electronic address: phillips@pboc.caltech.edu
1F. Reif, Fundamentals of statistical and thermal physics (McGraw-Hill, NY, 1965).
2K. Dill and S. Bromberg, Molecular driving forces: statistical thermodynamics in chemistry and biology (Garland Science,
New York, 2003).
3J. Kasianowicz, E. Brandin, D. Branton, and D. Deamer, Proc. Nat. Acad. Sci. USA 93, 13770 (1996).
4H. P. Lu, L. Xun, and X. S. Xie, Science 282, 1887 (1998).
5M. Reif, R. S. Rock, A. D. Mehta, M. S. Mooseker, R. E. Cheney, and J. A. Spudich, Proc. Nat. Acad. Sci. 97, 9482 (2000).
6A. Meller, L. Nivon, and D. Branton, Phys. Rev. Lett. 86, 3435 (2001).
7D. E. Smith, S. J. Tans, S. B. Smith, S. Grimes, D. L. Anderson, and C. Bustamante, Nature 413, 748 (2001).
8H. Li, W. A. Linke, A. F. Oberhauser, M. Carrion-Vazquez, J. G. Kerkvliet, H. Lu, P. Marszalek, and J. M. Fernandez,
Nature 418, 998 (2002).
9J. Liphardt, S. Dumont, S. B. Smith, I. Tinocho, and C. Bustamante, Science 296, 1832 (2002).
10C. Bustamante, Z. Bryant, and S. B. Smith, Nature 421, 423 (2003).
11E. R. Dufresne, D. Altman, and D. G. Grier, Europhys. Lett. 53, 264 (2001).
12B. Alberts, D. Bray, A. Johnson, J. Lewis, M. Raff, K. Roberts, and P. Walter, Essential Cell Biology: An Introduction to
the Molecular Biology of the Cell (Garland Publishing, New York, 1997).
Page 12
12
13D. Kondepudi and I. Prigogine, Modern thermodynamics:From heat engines to dissipative structures (John Wiley and sons,
1998).
14E. T. Jaynes, in E. T. Jaynes: Papers on probability, statistics and statistical physics, edited by R. D. Rosenkrantz (Kluver
academic publishers, 1980), chap. 14.
15M. Klein, Physica 22, 569 (1956).
16G. G. Emch and C. Liu, The logic of thermostatical physics (Springer-Verlag, NY, 2002).
17R. P. Feynman, in Nobel Lectures in Physics: 1901–1995 (World Scientific Publishing Co., 1998).
18E. T. Jaynes, in E. T. Jaynes: Papers on probability, statistics and statistical physics, edited by R. D. Rosenkrantz (Kluver
academic publishers, 1957), chap. 1.
19M. Kac, Amer. Math. Monthly 54, 369 (1947).
20G. Gamow, One Two Three...Infinity (Dover Publications, NY, 1988).
21G. M. Wang, E. M. Sevick, E. Mittag, D. J. Searles, and D. J. Evans, Phys. Rev. Lett. 89, 050601 (2002).
22J. Howard, Mechanics of motor proteins and the cytoskeleton (Sinauer Associates Inc., 2001).
Page 13
13
FIG. 1: Colloidal free expansion setup to illustrate diffusion involving small numbers of particles. (a) Schematic of experimental
setup (see text for details.) (b) Several snapshots from the experiment. (c) Normalized histogram of particle positions during
the experiment. The solution to the diffusion equation for the microfluidic “free expansion” experiment is superposed for
comparison.
Page 14
14
FIG. 2: Schematic of the simple dog-flea model. (a) State of the system at time t, (b) a particular microtrajectory in which
two-fleas jump from the dog on the left and one jumps from the dog on the right, (c) occupancies of the dogs at time t + ∆t.
Page 15
15
FIG. 3: Schematic of the distribution of fluxes for different time points as the system approaches equilibrium.
FIG. 4: Schematic of which trajectories are potent and which are impotent. The shaded region corresponds to the impotent
trajectories for which m1 and m2 are either equal or approximately equal and hence make relatively small change in the
macrostate. The unshaded region corresponds to potent trajectories.
Download full-text