• Home
  • Textbooks
  • Statistical mechanics: Entropy, Order Parameters and Complexity
  • Temperature and Equilibrium

Statistical mechanics: Entropy, Order Parameters and Complexity

James P. Sethna

Chapter 3

Temperature and Equilibrium - all with Video Answers

Educators


Chapter Questions

01:52

Problem 1

The Damped Pendulum vs. Liouville's Theorem. (Basic, Mathematics)
The damped pendulum has a force $-\gamma p$ proportional to the momentum slowing down the pendulum. It satisfies the equations
$$
\begin{aligned}
&\dot{x}=p / M \\
&\dot{p}=-\gamma p-K \sin (x)
\end{aligned}
$$
At long times, the pendulum will tend to an equilibrium stationary state, zero velocity at $x=0$ (or more generally at the equivalent positions $x=2 m \pi$, for $m$ an integer $)$ : $(p, x)=(0,0)$ is an attractor for the damped pendulum. An ensemble of damped pendulums is started with initial conditions distributed with probability $\rho\left(p_{0}, x_{0}\right) .$ At late times, these initial conditions are gathered together near the equilibrium stationary state: Liouville's theorem clearly is not satisfied.
(a) In the steps leading from equation $4.5$ to equation 4.7, why does Liouville's theorem not apply to the damped pendulum? More specifically, what are $\partial \dot{p} / \partial p$ and $\partial \dot{q} / \partial q$ ?
(b) Find an expression for the total derivative $d \rho / d t$ in terms of $\rho$ for the damped pendulum. If we evolve a region of phase space of initial volume $A=\Delta p \Delta x$ how will its volume depend upon time?

Narayan Hari
Narayan Hari
Numerade Educator
01:44

Problem 1

is the classic problem of planetary atmo- functions; parts (b) and (c) calculate the energy flucspheres. Exercise $3.3$ is a nice generalization of the ideal tuations for a mixture of two ideal gases, and could be gas law. Part (a) of exercise $3.4$ is a workout in $\delta$ - assigned separately. Exercise $3.5$ extends the calculation
Ta be pub. Oxford UP, $\sim$ Fallos
of the density fluctuations from two subvolumes to $K$ subvolumes, and introduces the Poisson distribution. Finally, exercise $3.6$ introduces some of the tricky partial derivative relations in thermodynamics (the triple product of equation $3.37$ and the Maxwell relations) and applys them to the ideal gas.

Dominador Tan
Dominador Tan
Numerade Educator
13:50

Problem 2

(4.2) Jupiter! and the KAM Theorem (Astrophysics, Mathematics)
See also the Jupiter Web pages [99].
The foundation of statistical mechanics is the ergodic hypothesis: any large system will explore the entire energy surface. We focus on large systems because it is well known that many systems with a few interacting particles are definitely not ergodic.
The classic example of a non-ergodic system is the Solar system. Jupiter has plenty of energy to send the other planets out of the Solar system. Most of the phase-space volume of the energy surface has eight planets evaporated and Jupiter orbiting the Sun alone: the ergodic hypothesis would doom us to one long harsh winter. So, the big question is: Why hasn't the Earth been kicked out into interstellar space?
Mathematical physicists have studied this problem for hundreds of years. For simplicity, they focused on the three-body problem: for example, the Sun, Jupiter, and the Earth. The early (failed) attempts tried to do perturbation theory in the strength of the interaction between planets. Jupiter's gravitational force on the Earth is not tiny, though: if it acted as a constant brake or accelerator, our orbit would be way out of whack in a few thousand years. Jupiter's effects must cancel out over time rather perfectly...

Emily Coulter
Emily Coulter
Numerade Educator
02:47

Problem 2

Escape Velocity. (Basic)
Assuming the probability distribution for the $z$ component of momentum given in equation 3.21, $\rho\left(p_{z}\right)=$ $1 / \sqrt{2 \pi m k_{B} T} \exp \left(-p_{z}^{2} / 2 m k_{B} T\right)$, give the probability density that an $N_{2}$ molecule will have a vertical component of the velocity equal to the escape velocity from the Earth (about $10 \mathrm{~km} / \mathrm{sec}$, if I remember right). Do we need to worry about losing our atmosphere? Optional: Try the same calculation for $\mathrm{H}_{2}$, where you'll find a substantial leakage.
(Hint:You'll want to know that there are about $\pi \times 10^{7}$ seconds in a year, and molecules collide (and scramble their velocities) many times per second. That's why Jupiter has hydrogen gas in its atmosphere, and Earth does not.)

Banhishikha Sinha
Banhishikha Sinha
Numerade Educator
01:24

Problem 3

Invariant Measures. (Math, Complexity) (With Myers. [75])
Reading: Reference [49], Roderick V. Jensen and Christopher R. Myers, "Images of the critical points of nonlinear maps" Physical Review $A \mathbf{3 2}, 1222-1224$ (1985).
Liouville's theorem tells us that all available points in phase space are equally weighted when a Hamiltonian system is averaged over all times. What happens for systems that evolve according to laws that are not Hamiltonian? Usually, the system does not continue to explore all points in its state space: at long times it is confined a subset of the original space known as the attractor.

We consider the behavior of the 'logistic' mapping from the unit interval $(0,1)$ into itself. $^{25}$
$$
f(x)=4 \mu x(1-x)
$$
We talk of the trajectory of an initial point $x_{0}$ as the sequence of points $x_{0}, f\left(x_{0}\right), f\left(f\left(x_{0}\right)\right), \ldots, f^{[n]}\left(x_{0}\right), \ldots$ Iteration can be thought of as a time step (one iteration of a Poincaré return map of exercise $4.2$ or one step $\Delta t$ in a time-step algorithm as in exercise $8.9$ ).
Attracting Fixed Point: For small $\mu$, our mapping has an attracting fixed point. A fixed point of a mapping is a value $x^{*}=f\left(x^{4}\right) ;$ a fixed point is stable if under small perturbations shrink:
$$
\left|f\left(x^{*}+\epsilon\right)-x^{*}\right| \approx\left|f^{\prime}\left(x^{*}\right)\right| \epsilon<\epsilon
$$

Sana Riaz
Sana Riaz
Numerade Educator
01:17

Problem 3

Temperature and Energy. (Basic)
What units [joules, millijoules, microjoules, nanojoules, $\ldots$, yoctojoules $\left(10^{-24}\right.$ joules)] would we use to measure temperature if we used energy units instead of introducing Boltzmann's constant $k_{B}=1.3807 \times 10^{-23} \mathrm{~J} / \mathrm{K} ?$

Eileen Sullivan
Eileen Sullivan
Numerade Educator
03:45

Problem 4

Hard Sphere Gas (Basic)
We can improve on the realism of the ideal gas by giving the atoms a small radius. If we make the potential energy infinite inside this radius ("hard spheres"), the potential energy is simple (zero unless the spheres overlap, which is forbidden). Let's do this in two dimensions.
A two dimensional $L \times L$ box with hard walls contains a gas of $N$ hard disks of radius $r \ll L$ (figure 3.5). The disks are dilute: the summed area $N \pi r^{2} \ll L^{2} .$ Let $A$ be the effective volume allowed for the disks in the box: $A=(L-2 r)^{2}$

Joanna Josey
Joanna Josey
Numerade Educator
20:37

Problem 5

Connecting Two Macroscopic Systems. arat, and Complexity]
An isolated system with energy $E$ is composed of two macroscopic subsystems, each of fixed volume $V$ and. number of particles $N$. The subsystems are weakly coupled, so the sum of their energies is $E_{1}+E_{2}=E$ (figure $3.4$ with only the energy door open). We can use the Dirac delta function $\delta(x)$ to define the volume of the energy surface of a system with Hamiltonian $\mathcal{H}$ to bel
$$
\begin{aligned}
\Omega(E)=& \int d \mathbb{P} d \mathbb{Q} \delta(E-\mathcal{H}(\mathbb{P}, \mathbb{Q})) \\
=& \int d \mathbb{P}_{1} d \mathbb{Q}_{1} d \mathbb{P}_{2} d \mathbb{Q}_{2} \\
& \delta\left(E-\left(\mathcal{H}_{1}\left(\mathbb{P}_{1}, Q_{1}\right)+\mathcal{H}_{2}\left(\mathbb{P}_{2}, \mathbb{Q}_{2}\right)\right)\right)
\end{aligned}
$$
(a) Derive formula 3.23 for the volume of the energy surface of the whole system using $\delta$-functions. (Hint: Insert $\int \delta\left(E_{1}-\mathcal{H}_{1}\left(\mathbb{P}_{1}, \mathbb{Q}_{1}\right)\right) d E_{1}=1$ into equation 3.65. $)$
Consider a monatomic ideal gas (He) mixed with a diatomic ideal gas $\left(\mathrm{H}_{2}\right)$. We showed that a monatomic ideal gas of $N$ atoms has $\Omega_{1}\left(E_{1}\right) \propto E_{1}^{3 N / 2}$. A diatomic molecule has $\Omega_{2}\left(E_{2}\right) \propto E_{2}^{5 N / 2} \cdot 44$
(b) Argue that the probability density of system 1 being at energy $E_{1}$ is the integrand of $3.23$ divided by the whole integral, equation 3.24. For these tuo gases, which energy $E_{1}^{\text {max }}$ has the maximum probability?
(c) Use the saddle-point method $[71$, sect. $3.6]$ to approximate the integral $3.65$ as the integral over a Gaussian. (That is, put the integrand into the form $\exp \left(f\left(E_{1}\right)\right)$ and Taylor expand $f\left(E_{1}\right)$ to second order in $\left.E_{1}-E_{1}^{\max } .\right)$ Use the saddle-point integrand as a Gaussian approximation for the probability density $\rho\left(E_{1}\right)$ (valid, for larye $N$, whenever $\rho\left(E_{1}\right)$ isn't absurdly small). In this approximation, what is the mean energy $\left(E_{1}\right\rangle$ ? What are the energy fluctuations per particle $\sqrt{\left\langle\left(E_{1}-E_{1}^{\max }\right)^{2}\right)} / N ?$
For subsystems with large numbers of particles $N$, temperature and energy density are well defined because $\Omega(E)$ for each subsystem grows extremely rapidly with increasing energy, in such a way that $\Omega_{1}\left(E_{1}\right) \Omega_{2}\left(E-E_{1}\right)$ is sharply peaked near its maximum.

Amit Srivastava
Amit Srivastava
Numerade Educator
05:28

Problem 6

Gauss and Poisson. (Basic)
In section 3.2.1, we calculated the probability distribution for having $n=N_{0}+m$ particles on the right-hand half of a box of volume $2 \mathrm{~V}$ with $2 \mathrm{~N}_{0}$ total particles. In section $10.3$ we will want to know the number fluctuations of a small subvolume in an infinite system. Studying this also introduces the Poisson distribution.
Let's calculate the probability of having $n$ particles in a subvolume $V$, for a box with total volume $K V$ and a total number of particles $T=K N_{0 .}$ For $K=2$ we will derive our previous result, equation 3.13, including the prefactor. As $K \rightarrow \infty$ we will derive the infinite volume result.
(a) Find the exact formula for this probability: $n$ particles in $V$, with total of $T$ particles in $K V .$ (Hint: What is the probability that the first $n$ particles fall in the subvolume $V$, and the remainder $T-n$ fall outside the subvolume $(K-1) V ?$ How many ways are there to pick $n$ particles from $T$ total particles?)
The Poisson probability distribution
$$
\rho_{n}=a^{n} e^{-a} / n !
$$
arises in many applications. It arises whenever there is a large number of possible events $T$ each with a small probability $a / T ;$ the number of cars passing a given point during an hour on a mostly empty street, the number of cosmic rays hitting in a given second, etc.
(b) Show that the Poisson distribution is normalized: $\sum_{n} \rho_{n}=1 .$ Calculate the mean of the distribution $\langle n\rangle$ in terms of a. Calculate the variance (standand deviation squared) $\left\langle(n-\langle n))^{2}\right\rangle$
(c) As $K \rightarrow \infty$, show that the probability that $n$ particles fall in the subvolume $V$ has the Poisson distribution 3.66. What is a? (Hint: You'll need to use the fact that $e^{-a}=\left(e^{-1 / K}\right)^{R \alpha} \rightarrow(1-1 / K)^{K a}$ as $K \rightarrow \infty$, and the fact that $n \ll T_{-}$Here don't assume that $n$ is large: the Poisson distribution is valid even if there are only a few events.).
From parts (b) and (c), you should be able to conclude that the variance in the number of particles found in a volume $V$ inside an infinite system should be equal to $N_{0}$. the expected number of particles in the volume:
$$
\left\langle(n-(n))^{2}\right\rangle=N_{0}
$$
This is twice the squared fluctuations we found for the case where the volume $V$ was half of the total volume, equation 3.13. That makes sense, since the particles can fluctuate more freely in an infinite volume than in a doubled volume.
If $N_{0}$ is large, the probability $P_{m}$ that $N_{0}+m$ particles lie inside our volume will be Gaussian for any $K$. (As a special case, if $a$ is large the Poisson distribution is well approximated as a Gaussian.) Let's derive this distribution for all $K$. First, as in section $3.2 .1$, let's use the weak
form of Stirling's approximation, equation $3.10$ dropping the square root: $n ! \sim(\bar{n} / e)^{n}$.
(d) Using your result from part (a), write the exact formula for $\log \left(P_{m}\right)$. Apply the weak form of Stirling's formula. Expand your result around $m=0$ to second order in $m$, and show that $\log \left(P_{m}\right) \approx-m^{2} / 2 \sigma_{K}^{2}$, giving a Gaussian form
$$
P_{m} \sim e^{-m^{2} / 2 \sigma_{K}^{2}}
$$
What is $\sigma_{K}$ ? In particular, what is $\sigma_{2}$ and $\sigma_{\infty}$ ? Your result for $\sigma_{2}$ should agree with the calculation in section 3.2.1, and your result for $\sigma_{\infty}$ should agree with equation $3.67$.

Finally, we should address the normalization of the Gaussian. Notice that the ratio of the strong and weak forms of Stirling's formula, (equation 3.10) is $\sqrt{2 \pi n}$. We need to use this to produce the normalization $\frac{1}{\sqrt{2 \pi \sigma} K}$ of our Gaussian.
(e) In terms of $T$ and $n$, what factor would the squareroot term have contributed if you had kept it in Stirling's formula going from part (a) to part (d)? (It should look like a ratio involving three terms like $\sqrt{2 \pi X}$.) Show from equation $3.68$ that the fluctuations are small, $m=$ $n-N_{0} \ll N_{0}$ for large $N_{0}$. Ignoring these fluctuations, set $n=N_{0}$ in your factor, and give the prefactor multiplying the Gaussian in equation 3.68. (Hint: your answer should be normalized.)

Carolyn Behr-Jerome
Carolyn Behr-Jerome
Numerade Educator
10:01

Problem 7

Microcanonical Energy Fluctuations. (Basic)
We argued in section $3.3$ that the energy fluctuations between two weakly coupled subsystems are of order $\sqrt{N}$. Let us calculate them explicitly.
Equation $3.30$ showed that for two subsystems with energy $E_{1}$ and $E_{2}=E-E_{1}$ the probability density of $E_{1}$ is a Gaussian with variance (standard deviation squared):
$$
\sigma_{E_{1}}^{2}=-k_{B} /\left(\partial^{2} S_{1} / \partial E_{1}^{2}+\partial^{2} S_{2} / \partial E_{2}^{2}\right)
$$
(a) Show that
$$
\frac{1}{k_{B}} \frac{\partial^{2} S}{\partial E^{2}}=-\frac{1}{k_{B} T} \frac{1}{N c_{v} T}
$$
where $c_{v}$ is the inverse of the total specific heat at constant volume. (The specific heat $c_{v}$ is the energy needed per particle to change the temperature by one unit: $\left.N c_{\mathrm{t}}=\left.\frac{\partial E}{\partial T}\right|_{V, N^{-}}\right)$
The denominator of equation $3.73$ is the product of two energies. The second term $N c_{v} T$ is a system-scale energy: it is the total energy that would be needed to raise the, temperature of the system from absolute zero, if the specific heat per particle $c_{e}$ were temperature independent. However, the first energy, $k_{B} T$, is an atomic-scale energy independent of $N$. The fluctuations in energy, therefore, scale like the geometric mean of the two, summed over the two subsystems in equation $3.30$, and hence scale as $\sqrt{N}$ : the total energy fluctuations per particle thus are roughly $1 / \sqrt{N}$ times a typical energy per particle.
This formula is quite remarkable. Normally, to measure a specific heat one would add a small energy and watch the temperature change. This formula allows us to measure he specific heat of an object by watching the equilibium fluctuations in the energy. These fluctuations are iny in most experiments, but can be quite substantial in fomputer simulations.
b) If $c_{v}^{(1)}$ and $c_{v}^{(2)}$ are the specific heats per particle for hat
$$
\frac{1}{c_{E}^{(1)}}+\frac{1}{c_{v}^{(2)}}=N k_{B} T^{2} / \sigma_{E}^{2}
$$
Ne don't even need to couple two systems. The positions and momenta of a molecular dynamics simulation (atoms moving under Newton's laws of motion) can be thought of as two subsystems, since the kinetic energy does not depend on the configuration $\mathbb{Q}$, and the potential energy does not depend on the momenta $\mathbb{P} .$ Let $E_{1}$ be the kinetic energy of a microcanonical molecular dynamics simulation with total energy $E$. Assume that the simulation is hat $c_{t}^{(1)}=3 k_{B} / 2$ for the momentum degrees of freedom. fn terms of the mean and standard deviation of the kinetic mergy, solve for the total specific heat of the molecular lynamics simulation (configurational plus kinetic).

Guilherme Barros
Guilherme Barros
Numerade Educator