PHYS7322 - Nonequilibrium Physics - Fall 2024
Problem set 3
Due date TBA
1. Class notes exercise 1.
~
We proved in class that if R(t) is a stochastic variable defined by
Z t
~
R(t) = ~ 0)
dt0 ψ(t0 )A(t
0
where Z t+∆t
~
B(∆t) = ~ 0)
dt0 A(t
t
obeys the Gaussian probability distribution
~ 1 ~ 2
P (B(∆t)) = 3/2
e−|B(∆t)| /(2Γ∆t) ,
(2πΓ∆t)
~
then the probability density function (PDF) of R(t) is
~ t) = 1 ~ 2 /(2Γ t dt0 ψ 2 (t0 )).
−|R|
R
P (R, Rt e 0
(2πΓ 0 dt0 ψ 2 (t0 ))3/2
Use this lemma to show that the PDF for the velocity of a free Brownian
particle with initial velocity ~v (0, t) = ~v0 is given by
!3/2
β β −βt |2 /(1−e−2βt )
P (~v , t; ~v0 ) = e− Γ |~v−~v0 e .
πΓ(1 − e−2βt )
2. Class notes exercise 2.
Use the lemma of exercise 1 to derive the PDF P (~r, t) for the position of a free
Brownian particle with initial velocity ~v (0, t) = ~v0 and position ~r(0, t) = ~r0 .
Use the PDF to calculate h|~r(t) − ~r0 |2 i and compare your answer to the one
obtained in class by the direct method (page 10 of Brownian motion notes) .
3. Class notes exercise 3.
~ t) given in exercise 1 reduces in one dimension to
The expression for P (R,
1 −|R|2 /(2Γ
Rt 0 2 0
dt ψ (t )).
P (R, t) = Rt e 0
(2πΓ 0 dt0 ψ 2 (t0 ))1/2
Derive the above expression starting from the last intermediate expression
Z +∞
dq iqR − q22Γ R t dt0 ψ2 (t0 )
P (R, t) = e e 0
−∞ 2π
derived in class.
4. Class notes exercise 4.
Show that the Fokker-Planck equation for the PDF of velocity can be solved
by the separation of variables P (~v , t) = Px (vx , t)Py (vy , t)Pz (vz , t).
5. Harmonically bound Brownian particle. Consider a harmonically bound
Brownian particle in a fluid that is constrained to move in one dimension.
The Langevin equation of motion is Newton’s second law for a damped 1D
harmonic oscillator subject to a random external force
d2 x dx
2
+β + ω 2 x = A(t)
dt dt
where
hA(t)A(t0 )i = Γδ(t − t0 )
and Z t+∆t
B(∆t) = dt0 A(t0 )
t
obeys the Gaussian probability distribution
~ 1 2
P (B(∆t)) = 1/2
e−|B(∆t)| /(2Γ∆t) .
(2πΓ∆t)
Many of the answers to the questions below can be found in Chandrasekhar’s
Review of Modern Physics, 1943, but I would like you to work through the
problems and check your answers with the article.
(a) Use the method of variation of parameters to show that the solution of
the Langevin equation is given by
x(t) = a1 (t)eµ1 t + a2 (t)eµ2 t
q q
where µ1 = −β/2 + β 2 /4 − ω 2 and µ2 = −β/2 − β 2 /4 − ω 2 are the
two roots of the quadratic equation µ2 + βµ + ω 2 = 0 and
Z t
1 0
a1 (t) = a10 + dt0 e−µ1 t A(t0 ),
µ1 − µ2 0
Z t
1 0
a2 (t) = a20 − dt0 e−µ2 t A(t0 ).
µ1 − µ2 0
(b) Determine the values of the constants a10 and a20 for initial conditions
that correspond to the initial position x(0) = x0 and the initial velocity
v(0) = v0 .
(c) Use your result for the instantaneous position x(t) to derive an expres-
sion for the instantaneous velocity v(t) = dx(t)/dt.
(d) Use the 1D lemma derived in class (also exercise 3 for the last step) to ob-
tain analytical expressions for the PDFs P (x, t; x0 , v0 ) and P (v, t; x0 , v0 )
of finding the Brownian particle at position between x and x + dx and
velocity between v and v + dv, respectively, at time between t and t + dt
given that it had position x0 and velocity v0 at time t = 0. Distinguish
the solutions for the overdamped and the underdamped cases defined
by β > 2ω (µ1,2 real) and β < 2ω (µ1,2 imaginary), respectively. Those
PDFs can be used to compute statistical averages such as hx(t)i, hv(t)i,
hx2 (t)i, and hv 2 (t)i. You do not need to compute those averages but
you should check the results in Chandrasekhar’s article to make sure
that you understand how those calculations are done.
6. Velocity auto-correlation function.
Show using the result from part (a) of the previous problem that, when the
particle has relaxed to equilibrium at sufficiently large time t
Z t
v(t) = dτ φ(τ )A(τ )
−∞
where
1
φ(τ ) = µ1 eµ1 (t−τ ) − µ2 eµ2 (t−τ )
µ1 − µ2
Starting from the above expression, derive the expression for the equilibrium
velocity autocorrelation function
!
0 Γ − β (t0 −t) β
hv(t)v(t )i = e 2 − sin ω1 (t0 − t) + cos ω1 (t0 − t) , (t0 > t).
2β 2ω1
With the substitutions t = 0 and t0 = t > 0, compare your answer for
hv(0)v(t)i to equation (2) of the Science paper Measurement of the In-
stantaneous Velocity of a Brownian Particle, Vol. 328, p. 1673 (2010).
Hint: to calculate hv(t)v(t0 )i, combine the above expression with v(t) with
Z t0
0
v(t ) = dτ 0 φ(τ 0 )A(τ 0 )
−∞
and use the fluctuating accelerating correlation function
hA(τ )A(τ 0 )i = Γδ(τ − τ 0 ),
using t0 > t to evaluate the double integrals properly. Extra credit: Write a
Python code to simulate the Langevin equation for the harmonically bound
particle and use it to numerically calculate hv(t)v(t0 )i and include a compar-
ison of the numerical and analytical results to make sure that they agree.