A simple model for Brownian motion leading to the Langevin equation

Bart G. de Grooth

Citation: Am. J. Phys. 67, 1248 (1999); doi: 10.1119/1.19111

View online: http://dx.doi.org/10.1119/1.19111
View Table of Contents: http://ajp.aapt.org/resource/1/AJPIAS/v67/i12
Published by the American Association of Physics Teachers

Related Articles
MicroReviews by the Book Review Editor: Beyond the Black Box: The Forensics of Airplane Crashes: George
Bibel
Phys. Teach. 50, 511 (2012)
MicroReviews by the Book Review Editor: Particle or Wave: The Evolution of the Concept of Matter in Modern
Physics: Charis Anastopoulos
Phys. Teach. 50, 511 (2012)
MicroReviews by the Book Review Editor: The Rise of Nuclear Fear: Spencer R. Weart
Phys. Teach. 50, 511 (2012)
Everyday Engineering: Putting the E in STEM Teaching and Learning: Richard H. Moyer and Susan A. Everett
Phys. Teach. 50, 511 (2012)
MicroReviews by the Book Review Editor: The Little Brown Book
Phys. Teach. 50, 511 (2012)

Additional information on Am. J. Phys.

Journal Homepage: http://ajp.aapt.org/
Journal Information: http://ajp.aapt.org/about/about_the_journal
Top downloads: http://ajp.aapt.org/most_downloaded
Information for Authors: http://ajp.dickinson.edu/Contributors/contGenInfo.html

Downloaded 19 Nov 2012 to 141.117.79.62. Redistribution subject to AAPT license or copyright; see http://ajp.aapt.org/authors/copyright_permission
A simple model for Brownian motion leading to the Langevin equation
Bart G. de Grooth
Department of Applied Physics, University of Twente, P.O. Box 217, 7500AE Enschede, The Netherlands
共Received 1 March 1999; accepted 14 July 1999兲
A simple one-dimensional model is presented for the motion of a Brownian particle. It is shown how
the collisions between a Brownian particle and its surrounding molecules lead to the Langevin
equation, the power spectrum of the stochastic force, and the equipartition of kinetic energy.
© 1999 American Association of Physics Teachers.

I. INTRODUCTION spectrum is related to the power spectra of the position and

velocity of a Brownian particle for a particle in a harmonic
In the last several years many techniques have become potential. Finally, in Sec. VI we summarize our results.
available that enable the study of individual molecules. Ex-
amples are imaging techniques such as the scanning tunnel- II. SIMPLE ONE-DIMENSIONAL MODEL LEADING
ing microscopes1 and the atomic force microscope 共AFM兲,2 TO THE LANGEVIN EQUATION
direct measurement of the forces between individual mol-
ecules using AFM3 and optical trapping techniques,4 the We consider a one-dimensional system consisting of a
measurement of the fluorescence of single molecules using relatively large particle 共the Brownian particle兲 with mass M
confocal laser scanning microscopes,5 and near-field scan- which is hit from both sides by molecules of mass m, with
ning optical microscopy.6 These techniques have made it M Ⰷm. The collisions are assumed to be elastic. We first
possible to study the dynamical behavior of individual func- consider a single collision and designate the velocity of the
tional systems such as motor proteins7 and DNA transcrip- Brownian particle before and after the collision by V and V ⬘ ,
tion enzymes.8 respectively. Similarly, the velocity of a molecule before and
These techniques also confront us with the fundamental after the collision is v and v ⬘ . If we combine the equations
limits due to thermal fluctuations. The motion of a cantilever for conservation of momentum and energy, we can write the
of an AFM due to its interaction with the surrounding mol- velocities after the collision in terms of the velocities before
ecules limits the accuracy by which we can measure the the collision:
forces between the tip and the sample.9 Attempts to obtain
DNA sequence information by measuring the rupture forces M ⫺m 2m
V ⬘⫽ V⫹ v, 共1兲
upon unzipping the strands of a single DNA molecule are M ⫹m M ⫹m
hampered by the thermal motions of the two single strands
formed.10 m⫺M 2M
v ⬘⫽ v⫹ V. 共2兲
To obtain a better physical understanding of the dynamical M ⫹m M ⫹m
behavior of individual biomolecules,11 a good understanding
of thermal fluctuations is needed. The usual starting point is We first show that the behavior of the system satisfies the
the Langevin equation, whose form is assumed in most text- equipartition theorem provided that 共i兲 v is independent of
books without a discussion of how the collisions between a V, and 共ii兲 the time between successive collisions is indepen-
Brownian particle and its much lighter surrounding mol- dent of V. In a system consisting of a very large number of
ecules give rise to a dissipative and a stochastic force. molecules of mass m that behave as a thermal bath, the first
Gillespie has shown recently that a simple Markov process condition is fulfilled. The second condition is not obviously
leads to a dissipative force.12 The goal of this article is to satisfied, because we might expect that the time between
discuss an even simpler model for the dynamics of a Brown- collisions would be smaller for a fast moving Brownian par-
ian particle and to show in detail how it leads to the essential ticle. However, if v ⰇV, the number of collisions per unit
features of more realistic systems. time is dominated by the movement of the molecules.13 The
The model consists of a relatively heavy particle 共the second condition implies that the time average of a quantity
Brownian particle兲 moving in one dimension subject to ran- equals the average over a large number of collisions:
dom collisions with the surrounding molecules. We will
冕
N⫺1
show that a straightforward analysis leads to two forces on 1 T 兺 i⫽0 t i V 2i
具 V 2 典 ⫽ lim V 共 t 兲 2 dt⫽ lim N⫺1
the Brownian particle: a dissipative force proportional to the T→⬁ T 0 N→⬁ 兺 i⫽0 ti
velocity of the Brownian particle and a random fluctuating
force with zero average. Because both terms are the conse- N⫺1 2
¯t 兺 i⫽0 N⫺1
1Vi
quence of the collisions between the Brownian particle and ⫽ lim N⫺1 ⫽ lim
兺 t
i⫽0 i N 兺
i⫽0
V 2i ⫽V 2 , 共3兲
the surrounding molecules, we obtain a direct relation be- N→⬁ N→⬁
tween the two forces leading to the fluctuation dissipation where V i is the velocity of the Brownian particle between the
theorem.
ith and (i⫹1)th collision, t i is the time interval between
In Sec. III we show how the model can be used to obtain
an algorithm for doing numerical simulations of a Brownian these collisions, and N is the number of collisions during the
particle in the presence of arbitrary forces. In Sec. IV the time interval T. The brackets denote a time average and the
model is used to determine the power spectrum of the sto- bar denotes averages over collisions. The average time inter-
chastic force, and in Sec. V it is shown how this power val between collisions is given by t̄ ⫽ 兺 i⫽0
N⫺1
t i /N. We have

1248 Am. J. Phys. 67 共12兲, December 1999 © 1999 American Association of Physics Teachers 1248

Downloaded 19 Nov 2012 to 141.117.79.62. Redistribution subject to AAPT license or copyright; see http://ajp.aapt.org/authors/copyright_permission
used condition 共ii兲 to obtain the third equality in Eq. 共3兲.14 n⌬t⫺1
We take the square of Eq. 共1兲 and average over the N ⌬ P N ⫽2m 兺
i⫽0
v i ⫺2mnV 共 t 兲 ⌬t. 共12兲
collisions. In the limit N→⬁, we obtain:
We divide both sides of Eq. 共12兲 by ⌬t and obtain a formal
共 M ⫺m 兲 2 2 4m 2
V ⬘2⫽ V ⫹ v2 expression for the time derivative of the Brownian particle’s
共 M ⫹m 兲 2 共 M ⫹m 兲 2 velocity:

⫹2 冉 M ⫺m
M ⫹m 冊冉 2m
M ⫹m
Vv. 冊 共4兲 M
dV
dt
⫽F s ⫺ ␥ V, 共13兲

For a stationary system we have V ⬘ 2 ⫽V 2 , and condition 共i兲 where the stochastic force F s is defined as
implies that V v ⫽V̄ v̄ ⫽0. Hence, we obtain the result 1
n⌬t⫺1

M V 2 ⫽m v 2 . 共5兲
F s⫽
⌬t 兺
i⫽0
2m v i , 共14兲

We see that given these two conditions, the average of the and the damping constant ␥ is given by
kinetic energy of the Brownian particle is the same as that of ␥ ⫽2mn. 共15兲
the surrounding molecules, in agreement with the equiparti-
tion theorem. Equation 共13兲 has the form of the Langevin equation with 16

We now proceed to obtain the Langevin equation. We use explicit expressions for the damping and stochastic forces in
the assumption M Ⰷm to write: the model. From Eq. 共15兲 we see that the damping constant ␥

冉冉 冊 冊
is proportional to the number of collisions per second and the
M ⫺m m m 2
mass of the surrounding molecules. In the above derivation
⬇1⫺2 ⫹O , 共6兲 we assumed that ⌬t contains a large number of collisions.
M ⫹m M M
For this reason dV/dt in Eq. 共13兲 should be interpreted with
M
M ⫹m
m
⬇1⫺ ⫹O
M 冉冉 冊 冊 m
M
2
, 共7兲
caution. It is referred to as a ‘‘coarse grained time
derivative.’’ 17

冉冉 冊 冊
The equipartition theorem can be used to arrive at the well
2
m m m known relation between the fluctuations of the position of a
⬇ ⫹O . 共8兲 Brownian particle 共in the absence of an external force兲 and
M ⫹m M M
␥:16
With these approximations Eq. 共1兲 can be rewritten as:

冉 冊
2k B T
2m 2m 具 ⌬x 2 典 ⫽ t, 共16兲
V ⬘ ⫽ 1⫺ V⫹ v. 共9兲 ␥
M M
where k B is Boltzmann’s constant and T is the absolute tem-
From Eq. 共9兲 we see that the change in momentum of the perature. Equation 共16兲 is an example of the fundamental
Brownian particle due to a single collision equals relation between fluctuations and dissipation. We can under-
stand the basis of this relation from Eq. 共15兲. A large damp-
⌬ P⫽2m v ⫺2mV. 共10兲
ing constant implies many collisions per second, which re-
Equation 共10兲 shows that the change of momentum of a sults in a reduction of the persistent motion of the Brownian
Brownian particle due to collisions with its surrounding mol- particle.
ecules results in two contributions. The momentum change Dissipation would also occur if the collisions with the
due to the first term is positive or negative, but on the aver- molecules were not randomly distributed, but occurred at a
age this contribution is zero because collisions from the left regular interval of 1/n. In that case the motion of the particle
and right have the same probability. The second contribution would be damped 共dissipation兲, but would not fluctuate and
tends to reduce the speed of the Brownian particle and is a hence Eq. 共16兲 would not be applicable. The reason for the
damping term proportional and opposite to the velocity of relation between dissipation and fluctuation is that the time
the Brownian particle.15 between collisions is a random variable.
Let us look at what happens during a time interval ⌬t
which is small enough that the velocity of the Brownian
particle does not change appreciably, but because M Ⰷm, we III. NUMERICAL SIMULATIONS OF BROWNIAN
still have a large number of collisions. From Eq. 共10兲 we can MOTION
write the momentum change of the Brownian particle due to
To obtain a better understanding of Brownian motion un-
N collisions as:
der different conditions, computer simulations can be very
N⫺1 N⫺1 helpful. We start with the Langevin equation 共13兲 and divide
⌬ P N ⫽2m
i⫽0
兺 v i ⫺2m 兺
i⫽0
Vi . 共11兲 the time into q intervals so that t⫽q⌬t. Using Eq. 共13兲, the
new velocity V q⫹1 can be expressed in terms of the previous
Because the velocity of the Brownian particle is assumed to velocity V q as
not change appreciably during ⌬t, we can approximate the Vq
second sum in Eq. 共11兲 by 2mNV⫽2mnV(t)⌬t, where V(t) V q⫹1 ⫽V q ⫺ ␥ ⌬t⫹⌬V s , 共17兲
M
is the velocity of the Brownian particle at time t, and n is the
mean number of collisions per second so that N⫽n⌬t. where ⌬V s is the velocity change due to the stochastic term
Hence, we write in Eq. 共13兲. The damping term poses no difficulty as long as

1249 Am. J. Phys., Vol. 67, No. 12, December 1999 Bart G. de Grooth 1249

Downloaded 19 Nov 2012 to 141.117.79.62. Redistribution subject to AAPT license or copyright; see http://ajp.aapt.org/authors/copyright_permission
the velocity change during the interval ⌬t is small compared
to 冑k B T/M . Our main problem is to find a useful expression
for the stochastic term in Eq. 共17兲. If the time between itera-
tions were sufficiently small, the procedure would be
straightforward. For each iteration we determine from a ran-
dom drawing whether a collision has taken place or not. The Fig. 1. The stochastic force is approximated by a constant force f during a
probability of a collision is such that on the average, there collision time ␶ c and is otherwise zero.
are 1/n collisions with an equal chance of a collision from
the left or the right. If there is no collision, the new velocity
can be calculated from Eq. 共17兲 with ⌬V s ⫽0. If a collision An interactive computer program which uses this algo-
occurs, the stochastic term in Eq. 共17兲 is given by 关see Eq. rithm to simulate a Brownian particle in a single or double
共10兲兴: potential well 共the Kramers problem20兲 can be downloaded
2m v 2 v 2 v from the author’s Web site.21
⌬V s ⫽ ⫽ 冑m 2 v 2 ⫽ 冑mk B T
M M 兩v兩 M 兩v兩

⫽
1 v
M 兩v兩
冑␥ 2
k T.
n B
共18兲
IV. POWER SPECTRUM OF THE STOCHASTIC
FORCE
We have used Eq. 共15兲 and have set the kinetic energy of the Because the statistical behavior of the motion of a Brown-
molecules to be 21 k B T. For simplicity, we have assumed that ian particle is related to 兩 F̃ s ( ␻ ) 兩 2 , the power spectrum of the
all molecules have the same kinetic energy. stochastic force F s 共see Sec. V兲, we now derive an expres-
The difficulty is that in practice the number of collisions
necessary to change the velocity of the Brownian particle sion for 兩 F̃ s ( ␻ ) 兩 2 . From the Wiener–Khintchine theorem
appreciably is so high that the above algorithm would take 兩 F̃ s ( ␻ ) 兩 2 is determined from the autocorrelation function of
too much computing time. Therefore, we have to find an F s : 22
expression for the contribution of a relatively large number
of collisions p, where pⰇ1. For a realistic Brownian particle
of 1 ␮m, the velocity change of p⫽1000 collisions with the
兩 F̃ s 共 ␻ 兲 兩 2 ⫽2 冕⫺⬁
⬁
e ⫺i ␻ t 具 F s 共 t ⬘ 兲 F s 共 t ⬘ ⫹t 兲 典 dt, 共23兲
surrounding molecules would still be very small. Because the
collisions occur randomly from the left and right, the effect where F̃ s ( ␻ ) is the Fourier transform of F s (t). The brackets
of p collisions follows a binomial distribution with no aver- indicate time averaging over t ⬘ .
age velocity change and a half width equal to the square root To obtain an expression for the autocorrelation function of
of p/2 times the magnitude of the velocity change due to a F s , we first consider a single collision whose duration is ␶ c .
single collision. For p sufficiently large, the binomial distri- The momentum change of the Brownian particle from the
bution can be approximated by a Gaussian distribution. stochastic force due to a single collision is according to Eq.
Therefore, we can write for the velocity change after p col- 共10兲 given by
lisions:
⌬ P s ⫽2m v . 共24兲
⌬V p ⫽w q 冑 p
2
⌬V 1 ⫽
wq
M
冑 p
␥ k T⫽
n B
wq
M
冑␥ k B T⌬t, 共19兲 Hence, the average force during the collision is

2m v
where we used Eq. 共18兲 and ⌬t⫽p/n. The variable w q is f⫽ . 共25兲
sampled randomly from a Gaussian distribution with zero ␶c
average and a variance equal to 2. Such a random variable
can be obtained by calculating the sum: The simplest model for the interaction is to assume a con-
stant force f during the collision time ␶ c as shown in Fig. 1.

冉 冊
12 Due to the stochastic nature of F s , the correlation between
1
w q ⫽& 兺
i⫽1
r i⫺
2
, 共20兲 F s (t ⬘ ) and F s (t ⬘ ⫹t) differs from zero only if 兩 t 兩 ⬍ ␶ c . To
obtain the time average of F s (t ⬘ )F s (t⫹t ⬘ ), it is sufficient to
where r i is uniformly distributed in the unit interval. average this function during the mean time between colli-
These considerations lead to the following algorithm for sions, ␶ ⫽1/n. If 兩 t 兩 ⬍ ␶ c , the average correlation is given by

冓冕 冔
the simulation of the Langevin equation:18 ␶
1
具 F s 共 t ⬘ 兲 F s 共 t ⬘ ⫹t 兲 典 ⫽ F s 共 t ⬘ 兲 F s 共 t ⬘ ⫹t 兲 one collision dt ⬘
Vq w q⫹1 F ␶
V q⫹1 ⫽V q ⫺ ␥ ⌬t⫹ 冑␥ k B T⌬t⫹ ext ⌬t, 共21兲
0
M M M
关 ␶ c⫺兩t兩 兴
⫽ 具 f 2典 , 共26兲
X q⫹1 ⫽X q ⫹V q⫹1 ⌬t. 共22兲 ␶

An additional external force F ext has been added for gener- whereas if 兩 t 兩 ⬎ ␶ c , the correlation is zero 共see Fig. 2兲. Using
ality. The algorithm in Eqs. 共21兲 and 共22兲 is primarily in- Eqs. 共25兲 and 共15兲, the relation ␶ ⫽1/n, and writing 21 k B T for
tended to illustrate the physical origin of the stochastic term. the average kinetic energy of the molecules, we can rewrite
More efficient algorithms can be found in the literature.19 Eq. 共26兲 as:

1250 Am. J. Phys., Vol. 67, No. 12, December 1999 Bart G. de Grooth 1250

Downloaded 19 Nov 2012 to 141.117.79.62. Redistribution subject to AAPT license or copyright; see http://ajp.aapt.org/authors/copyright_permission
 Fig. 4. The power spectrum of the stochastic force given by Eq. 共32兲 is
independent of the frequency. In reality, it drops to zero for frequencies
above the inverse of the collision time.

which means that frequencies above 1/␶ c are absent 共see

Fig. 4兲.
Fig. 2. To calculate the time average of the autocorrelation function of F s ,
we can average during a single ‘‘cycle,’’ the time between two collisions
1/n. The autocorrelation is indicated by the area of the dashed square. V. POWER SPECTRUM OF THE DISPLACEMENT
AND THE VELOCITY OF A BROWNIAN
PARTICLE
关 ␶ c⫺兩t兩 兴 具 4m 2 v 2 n 典 Because the behavior of a particle subjected to Brownian
具 F s 共 t ⬘ 兲 F s 共 t ⬘ ⫹t 兲 典 ⫽ 具 f 2 典 ⫽ 关 ␶ c⫺兩t兩 兴 motion is stochastic, we have to look at its statistical prop-
␶ ␶ 2c
erties to obtain a quantitative description. From experiments
关 ␶ c⫺兩t兩 兴 it is easy to determine the power spectra of the square of
⫽2 ␥ k B T 共27兲 position and velocity. By analyzing these spectra we can
␶ 2c deduce the relevant parameters.
for 兩 t 兩 ⬍ ␶ c . We consider a one-dimensional system consisting of a
In summary we have: particle of mass M in a harmonic potential 共with force con-

再
stant k), damping ␥, subject to a stochastic force F s . For this
2 ␥ k B T 关 ␶ c ⫺ 兩 t 兩 兴 / ␶ 2c for 兩 t 兩 ⬍ ␶ c
具 F s 共 t ⬘ 兲 F s 共 t ⬘ ⫹t 兲 典 ⫽ 0 for 兩 t 兩 ⬎ ␶ . system we can write the Langevin equation in Eq. 共13兲 as
c
共28兲 M ẍ⫹ ␥ ẋ⫹kx⫽F s 共 t 兲 . 共33兲
This function is drawn in Fig. 3. We can express the statistical behavior of x and v of this
If we assume that the duration of the collisions is very particle by taking the Fourier transform of Eq. 共33兲. We
short, we can approximate the autocorrelation function by a write

冕
delta function. In order to obtain the correct normalization, ⬁
we integrate Eq. 共28兲 with respect to t which gives: F̃ 共 ␻ 兲 ⫽ F s 共 t 兲 e ⫺i ␻ t dt,

冕
⫺⬁
␶c
共34兲
⫺␶c
具 F s 共 t ⬘ 兲 F s 共 t ⬘ ⫹t 兲 典 dt⫽2 ␥ k B T,

where we have used:

共29兲
F s共 t 兲 ⫽
1
2␲
冕⫺⬁
⬁
F̃ 共 ␻ 兲 e i ␻ t d ␻

冕 冕
␶c ␶c and a similar expression for x̃( ␻ ), and we obtain for the
关 ␶ c ⫺ 兩 t 兩 兴 dt⫽2 共 ␶ c ⫺t 兲 dt⫽ ␶ 2c . 共30兲 Fourier transform of Eq. 共33兲:24
⫺␶c 0
x̃ 共 ␻ 兲关 k⫺i ␥ ␻ ⫺M ␻ 2 兴 ⫽F̃ s 共 ␻ 兲 . 共35兲
It follows that the form
If we multiply Eq. 共35兲 by its complex conjugate, we find:
具 F s 共 t ⬘ 兲 F s 共 t ⬘ ⫹t 兲 典 ⫽2 ␥ k B T ␦ 共 t 兲 共31兲
具 兩 F s共 ␻ 兲兩 2典
is consistent with Eq. 共29兲. Equation 共31兲, which gives a 具 兩 x̃ 共 ␻ 兲 兩 2 典 ⫽ . 共36兲
direct relation between the spectral properties of the stochas- 共 M ␻ 2 ⫺k 兲 2 ⫹ ␥ 2 ␻ 2
tic force and the dissipation of the system, is sometimes re- We see that the power spectrum of x is directly related to
ferred to as the fluctuation dissipation theorem.23 Using Eq. that of the stochastic force. Using Eq. 共32兲 for the stochastic
共23兲, we obtain the power spectrum of F s : force originating from random collisions, we obtain:
兩 F̃ s 共 ␻ 兲 兩 2 ⫽4 ␥ k B T. 共32兲 4 ␥ k BT
具 兩 x̃ 共 ␻ 兲 兩 2 典 ⫽ . 共37兲
We see that the power spectrum is independent of the fre- 共 M ␻ 2 ⫺k 兲 2 ⫹ ␥ 2 ␻ 2
quency, that is, there is a white noise spectrum. In reality we Similarly, we find for the power spectrum of the squared
know that there are correlations during the collision time velocity:
4 ␻ 2␥ k BT
具 兩 ṽ共 ␻ 兲 兩 2 典 ⫽ . 共38兲
共 M ␻ 2 ⫺k 兲 2 ⫹ ␥ 2 ␻ 2
Equations 共37兲 and 共38兲 can be used to fit the experimentally
measured power spectra from which the values of the param-
eters M , k and ␥ can be deduced.25
Fig. 3. The average autocorrelation function of the stochastic force as a We have approximated the power spectrum of F s ( ␻ ) by a
function of the time interval t. constant function, which means that all frequencies are as-

1251 Am. J. Phys., Vol. 67, No. 12, December 1999 Bart G. de Grooth 1251

Downloaded 19 Nov 2012 to 141.117.79.62. Redistribution subject to AAPT license or copyright; see http://ajp.aapt.org/authors/copyright_permission
sumed to be present 共white noise兲. As we have noted, in scanning optical microscopy,’’ Science 262, 1422–1425 共1993兲.
7
reality, frequencies above the inverse of the collision time K. Svoboda, C. F. Schmidt, B. J. Schnapp, and S. M. Block, ‘‘Direct
observation of kinesin stepping by optical trapping interferometry,’’ Na-
are absent. We can now see that this approximation does not
ture 共London兲 365, 721–727 共1993兲.
change the frequency dependence of the power spectra for all 8
M. D. Wang, M. J. Schnitzer, H. Yin, R. Landick, J. Gelles, and S. M.
practical cases. The numerator in Eq. 共36兲 goes to zero for Block, ‘‘Force and velocity measured for single molecules of RNA poly-
␻ ⬎k/ ␥ , which in realistic systems occurs at frequencies that merase,’’ Science 282, 902–907 共1998兲.
are much smaller than 1/␶ c . In other words, whether or not 9
F. Gittes and C. F. Schmidt, ‘‘Thermal noise limitations on micromechani-
the stochastic force contains frequencies higher than 1/␶ c is cal experiments,’’ Eur. Biophys. J. 27, 75–81 共1998兲.
10
U. Bockelmann, B. Essevaz-Roulet, and F. Heslot, ‘‘Molecular stick-slip
not important for the behavior of the system because the motion by opening DNA with piconewton forces,’’ Phys. Rev. Lett. 79,
system is much too slow to respond to such high frequencies. 4489–4492 共1997兲; R. E. Thompson and E. D. Siggia, ‘‘Physical limits on
Finally, it is reassuring to find that the derived expressions the mechanical measurement of the secondary structure of bio-
for v and x are in agreement with the equipartition theorem. molecules,’’ Europhys. Lett. 31, 335–340 共1995兲.
11
To see this we use Parseval’s theorem M. O. Magnasco, ‘‘Forced thermal ratchets,’’ Phys. Rev. Lett. 71, 1477–
1481 共1993兲.

冕 兩 y 共 t 兲 2 dt⫽
1
2␲
冕 兩 ỹ 共 ␻ 兲 兩 2 d ␻ , 共39兲
12

13
D. T. Gillespie, ‘‘Fluctuation and dissipation in Brownian motion,’’ Am. J.
Phys. 61, 1077–1083 共1993兲.
The other extreme case, v ⫽0 共no thermal movement兲, has recently been
and perform the integration using Eq. 共37兲 or 共38兲. Note that discussed. It was shown that in this case the interactions between m and M
we have not made use of the equipartition theorem, so the lead to a deceleration proportional to V 2 . See M. I. Molina, ‘‘Body motion
outcome indicates that our model implies the equipartition of in a one-dimensional resistive medium,’’ Am. J. Phys. 66, 973–974
kinetic energy between the Brownian particle and the mol- 共1998兲. In this model the molecules of mass m are assumed to be at rest.
ecules, in agreement with our expectations at the beginning Thus collisions only occur due to the movement of M , from which it
of Sec. II. follows that the number of collisions per second are then proportional to
V. Each collision reduces the velocity of M by an amount proportional to
V 关according to Eq. 共10兲兴. Therefore the deceleration is proportional to V 2 .
VI. CONCLUSIONS 14
According to condition 共ii兲, t i is independent of V i , and hence we can
We have studied a simple one-dimensional model for the replace the product t i V 2i in Eq. 共3兲 by t j V 2i . For N→⬁, we can replace t j
interaction of a Brownian particle with its surroundings. We by the average interval time t̄ from which the third equality of Eq. 共3兲
follows.
obtained insight into how this interaction leads to the fluc- 15
Equation 共10兲 can be deduced directly by using a moving coordinate sys-
tuations and a dissipation force in the Langevin equation, the tem in which the Brownian particle is at rest. Assuming M Ⰷm, we im-
equipartition of kinetic energy, and the power spectra. It is mediately arrive at Eq. 共10兲.
hoped that this simple model can lead to a better understand- 16
The original paper of Langevin on the theory of Brownian motion was
ing and physical intuition of about real systems. published in English translation by D. S. Lemons and A. Gythiel, Am. J.
Phys. 65, 1079–1081 共1997兲.
ACKNOWLEDGMENTS
17
F. Reif, Fundamentals of Statistical and Thermal Physics 共McGraw-Hill,
NY, 1965兲, p. 572.
18
I would like to thank Kobus Kuipers for valuable discus- H. Risken, The Fokker-Planck Equation: Methods of Solution and Appli-
sions and critically reading of the manuscript and Juleon cations 共Springer, Berlin, 1989兲, 2nd ed., pp. 60–62.
19
Schins for valuable discussions. A. Greiner, W. Strittmatter, and J. Hohnerkamp, ‘‘Numerical integration
of stochastic differential equations,’’ J. Stat. Phys. 51, 95–108 共1988兲.
20
1 P. Hänggi, P. Talkner, and M. Borkovec, ‘‘Reaction-rate theory: fifty years
G. Binnig, H. Rohrer, and Ch. Gerber, ‘‘Surface studies with scanning
after Kramers,’’ Rev. Mod. Phys. 62, 251–341 共1990兲.
tunneling microscopy,’’ Phys. Rev. Lett. 49, 57–61 共1982兲. 21
2 The author’s URL is http//www.tn.utwente.nl/top/index_bft.html
G. Binnig, C. F. Quate, and Ch. Gerber, ‘‘Atomic force microscopy,’’ 22
The physical meaning of the Wiener-Khintchine theorem is clearly illus-
Phys. Rev. Lett. 56, 930–933 共1986兲.
3
V. T. Moy, E. L. Florin, and H. E. Gaub, ‘‘Intermolecular forces and trated in the physical description of Fourier transform spectroscopy. See S.
energies between ligands and receptors,’’ Science 266, 257–259 共1994兲. G. Lipson, H. Lipson, and P. S. Tannhauser, Optical Physics 共Cambridge
4
M. S. Kellermayer, S. B. Smith, H. L. Granzier, and C. Bustamante, U.P., Cambridge, MA, 1995兲, 3rd ed., pp. 303–311.
23
‘‘Folding-unfolding transitions in single titin molecules characterized with Ref. 15, pp. 560–594. See also Ref. 17, pp. 56–594.
24
laser tweezers,’’ Science 276, 1112–1116 共1997兲. The Fourier transform of ẋ and ẍ can be expressed in terms of x by
5
S. Nie, D. T. Chiu, and R. N. Zare, ‘‘Probing individual molecules with integration by parts, yielding: ˜ẋ ⫽⫺i ␻ x̃ and ˜ẍ ⫽⫺ ␻ 2 x̃.
confocal fluorescence microscopy,’’ Science 266, 1018–1021 共1994兲. 25
K. Svoboda and S. M. Block, ‘‘Biological applications of optical forces,’’
6
E. Betzig and R. J. Chichester, ‘‘Single molecules observed by near-field Annu. Rev. Biophys. Biomol. Struct. 23, 247–285 共1994兲.

1252 Am. J. Phys., Vol. 67, No. 12, December 1999 Bart G. de Grooth 1252

Downloaded 19 Nov 2012 to 141.117.79.62. Redistribution subject to AAPT license or copyright; see http://ajp.aapt.org/authors/copyright_permission