Problem Set 6

Due: 11/08/2021

Problem 1

a. Describe the fundamental differences between the Monte Carlo and molecular
dynamics methods.

b. Why is molecular dynamics particularly suited to simulations in the microcanonical

ensemble? What difficulties are encountered when the ensemble must be changed in
molecular dynamics?

Problem 2:

A trivial Monte Carlo simulation is performed in the canonical ensemble at 300K with
two argon atoms (atom A and atom B) interacting according to the Lennard-Jones
potential:

𝜎 )* 𝜎 ,
𝑈"# = 4𝜖 −
𝑟 𝑟

and parameters, 𝜎 = 3.405 Å and 𝜖/𝑘3 = 119.8 K (𝑘3 is Boltzmann’s constant). If the atoms are
initially beyond the cutoff distance from one another, i.e., no interaction, calculate the probability
of moving atom A:

a. 3.1 Å from atom B

b. 4.1 Å from atom B
c. 8.1 Å from atom B
d. At what separation, r, does randomness become involved?

Problem 3:

In this problem, you will derive the semi-classical canonical partition function for a pure ideal
gas of a non-linear tri-atomic molecule represented as A-B-A, where A and B are two distinct
elements. This molecule is assumed to have kinetic, configurational, rotational, electronic, and
vibrational energy modes, but no internal rotational degrees of freedom. You may assume that
each mode is independent of the others. The molecule possesses unique moments of inertia about
three axes. The first four (non-degenerate) electronic states of the isolated molecule have
energies of 0.1 eV, 20.0 eV, 21.0 eV, and 27.5 eV. The molecule has relevant (non-degenerate)
vibrational modes with energies of 0.2 eV and 0.22 eV.

a. Write an expression for the complete canonical partition function in terms of the partition
functions associated with each degree of freedom above. [Q = f(QKE,Qconf,…)].

b. In the Semi-Classical Approximation, which molecular energetic modes listed in your part A
response are described classically? And which are described according to quantum mechanical
expressions (quantally)?

c. State the appropriate (classical or quantum) canonical partition function for each of the
following energy modes symbolically, in terms of h, c, k, Eelec,a, Evib,a, m, N, T, V, I1, I2, I3, ω ,
1

ω , and the corresponding de Broglie wavelengths. If you use a de Broglie wavelength, you must
2

define it in terms of the other parameters.

i. Kinetic energy (QKE)

ii. Configurational energy (Qconf)
iii. Rotational energy (Qrot)
iv. Electronic energy (Qelec) [In this case, demonstrate quantitatively that only the lowest
energy state contributes significantly to Qelec below 10,000 K. Accordingly, include only
the electronic ground state in your symbolic expression for Qelec.]
v. Vibrational energy (Qvib)

d. State the complete canonical total partition function for this system in terms of the parameters
used in part c

e. Derive an expression for the pressure in a pure fluid of this molecule at low density, and
compare your expression with the ideal gas law.

Problem 4: Argon can be described using a Lennard-Jones potential

𝜎 )* 𝜎 ,
𝑈"# = 4𝜖 −
𝑟 𝑟

with parameters 𝜎 = 3.405 Å and 𝜖/𝑘3 = 119.8 K (𝑘3 is Boltzmann’s constant).

(a) Derive an expression for the force acting between Lennard-Jones particles a function of
separation distance r
(b) Using Python and Matplotlib, plot the Lennard-Jones potential and force for argon as a
function of separation distance r

Problem 5: Consider a system of spherical particles in three-dimensions. At equilibrium, the

distribution of particle velocities is given by the Maxwell-Boltzmann distribution:
@
𝑚 * 𝑚(𝑣7* + 𝑣9* + 𝑣:* )
𝑝(𝑣7 , 𝑣9 , 𝑣: ) = exp −
2𝜋𝑘3 𝑇 2𝑘3 𝑇

Note that the distribution of velocities in in the x-,y-, and z-directions are identical and
independent. Hence, the joint probability distribution above can be factored as follows

𝑝 𝑣7 , 𝑣9 , 𝑣: = 𝑓 𝑣7 𝑓 𝑣9 𝑓 𝑣:
where
 )
𝑚 * 𝑚(𝑣 * )
𝑓 𝑣 = exp −
2𝜋𝑘3 𝑇 2𝑘3 𝑇

is the distribution function for particle velocities for a 1 dimensional system.

In molecular dynamics simulations, it is common to generate initial velocities by randomly

drawing velocities from the distribution 𝑓 𝑣 using a Gaussian random number generator.

Modify the example python script uniform_velocities.py (this will be posted shortly), which
samples initial velocities form a uniform distribution, to draw initial velocities from 𝑓 𝑣 at a set
)
temperature 𝑇 = (i.e., 𝑘3 𝑇 = 1) . Assume all particles have unit mass 𝑚 = 1. Note that the
GH
x,y,z components are all generated from 𝑓 𝑣 .
(Hint: see numpy.random.normal doucmentation https://docs.scipy.org/doc/numpy-
1.14.0/reference/generated/numpy.random.normal.html)

(a) Generate random velocities from 𝑓 𝑣 for 𝑁 =100 particles. Plot a histogram of the
particle velocities in the x-direction using Matplotlib. Note that the x,y,z components of
the velocity should be statistically identical. On the same graph, plot the expected
analytical distribution 𝑓 𝑣 for comparison.

(b) Repeat part (b) for 𝑁 =100,000 particles.

(c) The instantaneous kinetic energy and temperature can be evaluated from the velocity
distribution via
) N * * *
𝐸GLM = L 𝑚L 𝑣L,7 + 𝑣L,9 + 𝑣L,: .
*

and
*
𝑇 GLM = 𝐸GLM .
@NO@ GH

Compute 𝐾𝐸 and 𝑘3 𝑇 GLM from the velocities generated in part (b) and (c). We expect
@
that 𝐸GLM = 𝑁𝑘𝑇 and 𝑘3 𝑇 = 1. How well do the computed values agree with these
*
expectations for 𝑁 = 100 and 𝑁 = 10000?