PH503 Homework


































1.

(a)

Liouville's equation $\frac{\partial \rho}{\partial t} = -\{\rho, \mathcal{H}\}$, or

    $$\frac{d\rho}{dt} = \frac{\partial \rho}{\partial t} + \sum_i \left( \frac{\partial \rho}{\partial q_i}\dot{q}_i + \frac{\partial \rho}{\partial p_i}\dot{p}_i \right) = \frac{\partial \rho}{\partial t} + \{\rho, \mathcal{H}\}$$

Which is $\frac{d\rho}{dt} = 0$, volume element $d\Gamma = dpdq$ is time invariant.

Entropy is  $S = - \int \rho \ln \rho \, d\Gamma$.

Then $\rho$ invariant, $\rho \ln \rho$ invariant, $(\rho \ln \rho) \times d\Gamma$ invariant.

$\frac{dS}{dt} = 0$


(b)

Normalization: $C_1 = \int dpdq \, \rho(p, q) = 1$

Fixed average energy: $C_2 = \int dpdq \, \rho(p, q) \mathcal{H}(p, q) = E$

Lagrange multipliers:

$$L[\rho] = S - \alpha C_1 - \beta C_2$$

$$L[\rho] = - \int \rho \ln \rho \, dpdq - \alpha \left(\int \rho \, dpdq - 1\right) - \beta \left(\int \rho \mathcal{H} \, dpdq - E\right)$$

$$L[\rho] = \int (-\rho \ln \rho - \alpha \rho - \beta \rho \mathcal{H}) \, dpdq + \alpha + \beta E$$

To find the function $\rho$ that maximizes $L$, 

$$\frac{\partial}{\partial \rho} (-\rho \ln \rho - \alpha \rho - \beta \rho \mathcal{H}) = 0$$

$$\ln \rho = -1 - \alpha - \beta \mathcal{H}$$

Then

$$\rho_{\text{max}} = e^{-1-\alpha} e^{-\beta \mathcal{H}}$$

$$\int \rho_{\text{max}} \, dpdq = \int e^{-1-\alpha} e^{-\beta \mathcal{H}} \, dpdq = e^{-1-\alpha} \int e^{-\beta \mathcal{H}} \, dpdq = 1$$

Partition function $Z = \int e^{-\beta \mathcal{H}} \, dpdq$.

$$e^{-1-\alpha} Z = 1 \quad \implies \quad e^{-1-\alpha} = \frac{1}{Z}$$

Finally,

$$\rho_{\text{max}}(p, q) = \frac{1}{Z} e^{-\beta \mathcal{H}(p, q)}$$

This is the probability distribution of the canonical ensemble.


(c)

A density function is stationary if its partial time derivative is zero $\frac{\partial \rho}{\partial t} = 0$.

Use Liouville's equation($\frac{\partial \rho}{\partial t} = -\{\rho, \mathcal{H}\}$). 

Goal is $\{\rho_{\text{max}}, \mathcal{H}\}=0$.


$$\rho_{\text{max}} = f(\mathcal{H}) = \frac{1}{Z} e^{-\beta \mathcal{H}}$$

$$\{f(\mathcal{H}), \mathcal{H}\} = \sum_i \left( \frac{\partial f(\mathcal{H})}{\partial q_i} \frac{\partial \mathcal{H}}{\partial p_i} - \frac{\partial f(\mathcal{H})}{\partial p_i} \frac{\partial \mathcal{H}}{\partial q_i} \right)$$

Using the chain rule:

$$\frac{\partial f(\mathcal{H})}{\partial q_i} = \frac{df}{d\mathcal{H}} \frac{\partial \mathcal{H}}{\partial q_i} \quad \text{and} \quad \frac{\partial f(\mathcal{H})}{\partial p_i} = \frac{df}{d\mathcal{H}} \frac{\partial \mathcal{H}}{\partial p_i}$$

$$\{f(\mathcal{H}), \mathcal{H}\} = \sum_i \left( \left(\frac{df}{d\mathcal{H}} \frac{\partial \mathcal{H}}{\partial q_i}\right) \frac{\partial \mathcal{H}}{\partial p_i} - \left(\frac{df}{d\mathcal{H}} \frac{\partial \mathcal{H}}{\partial p_i}\right) \frac{\partial \mathcal{H}}{\partial q_i} \right)$$

$$\{f(\mathcal{H}), \mathcal{H}\} = \frac{df}{d\mathcal{H}} \sum_i \left( \frac{\partial \mathcal{H}}{\partial q_i} \frac{\partial \mathcal{H}}{\partial p_i} - \frac{\partial \mathcal{H}}{\partial p_i} \frac{\partial \mathcal{H}}{\partial q_i} \right)$$

$$\{\rho_{\text{max}}, \mathcal{H}\} = 0$$

Since $\frac{\partial \rho_{\text{max}}}{\partial t} = -\{\rho_{\text{max}}, \mathcal{H}\}$:

$$\frac{\partial \rho_{\text{max}}}{\partial t} = 0$$



2. 

(a)

Define the Systems:

- Subsystem 'a' has energy $E_a$, particle number $N_a$, and its number of accessible microstates is given by the phase space volume $\Gamma_a(E_a)$.

- Subsystem 'b' has energy $E_b$, particle number $N_b$, and its number of accessible microstates is $\Gamma_b(E_b)$.

- The total system has energy $E = E_a + E_b$ and particle number $N = N_a + N_b$.


Calculate the Total Entropy:

    $$S_{total} = k_B \log \Gamma(E)$$

If system is statistically independent,

    $$\Gamma(E) = \Gamma_a(E_a) \cdot \Gamma_b(E_b)$$

    $$S_{total} = k_B \log(\Gamma_a(E_a) \cdot \Gamma_b(E_b))$$

    Using $\log(xy) = \log(x) + \log(y)$:

    $$S_{total} = k_B (\log \Gamma_a(E_a) + \log \Gamma_b(E_b))$$

    $$S_{total} = k_B \log \Gamma_a(E_a) + k_B \log \Gamma_b(E_b)$$

    $$S_{total} = S_a + S_b$$


(b)

$$\Gamma(E) = \Sigma(E+\Delta) - \Sigma(E)$$

    Since $\Delta \ll E$, first-order Taylor expansion:

    $$\Sigma(E+\Delta) \approx \Sigma(E) + \frac{\partial \Sigma(E)}{\partial E} \Delta$$

    $$\Gamma(E) \approx \left(\Sigma(E) + \frac{\partial \Sigma(E)}{\partial E} \Delta\right) - \Sigma(E) = \frac{\partial \Sigma(E)}{\partial E} \Delta$$

Also,

    $$\Gamma(E) \approx g(E) \Delta$$


Compare $S_1$ and $S_3$:

    $$S_1 = k_B \log \Gamma(E) \approx k_B \log(g(E) \Delta)$$

    $$S_1 \approx k_B (\log g(E) + \log \Delta) = S_3 + k_B \log \Delta$$

Since the entropy $S$ is extensive ($S \propto N$), this constant difference is negligible for large $N$. Then, $S_1 \approx S_3$.


Compare $S_2$ and $S_3$:

For large DoF $f$ system, let

    $$\Sigma(E) \propto E^f$$

    $$g(E) = \frac{\partial \Sigma(E)}{\partial E} \propto \frac{\partial}{\partial E}(E^f) = f E^{f-1} = \frac{f}{E} E^f \propto \frac{f}{E} \Sigma(E)$$

    $$S_3 = k_B \log g(E) \approx k_B \log \left(C \frac{f}{E} \Sigma(E)\right) $$

    $$S_3 \approx k_B \log \Sigma(E) + k_B \log\left(C\frac{f}{E}\right) = S_2 + k_B \log\left(C\frac{f}{E}\right)$$

Both the number of degrees of freedom ($f \propto N$) and the energy ($E \propto N$) are extensive. Then second term $O(\log N)$ at most. Since $S_2$ is of order $O(N)$, the difference is negligible for large $N$. Then, $S_2 \approx S_3$.



3.

The entropy of a monatomic ideal gas is:

$$S = N k_B \ln \left[ V \left( \frac{4\pi m E}{3N h^2} \right)^{3/2} \right] + \frac{3}{2} N k_B$$

The entropy with the Gibbs correction factor ($S = S_{dist} - k_B \ln(N!)$) is given by the Sackur-Tetrode equation:

$$S(E, V, N) = N k_B \ln \left[ \frac{V}{N} \left( \frac{4\pi m E}{3N h^2} \right)^{3/2} \right] + \frac{5}{2} N k_B$$


(a)

Let's scale the variables $E \to \lambda E$, $V \to \lambda V$, $N \to \lambda N$:

$S(\lambda E, \lambda V, \lambda N) = (\lambda N) k_B \ln \left[ \frac{(\lambda V)}{(\lambda N)} \left( \frac{4\pi m (\lambda E)}{3(\lambda N) h^2} \right)^{3/2} \right] + \frac{5}{2} (\lambda N) k_B$

$S(\lambda E, \lambda V, \lambda N) = \lambda N k_B \ln \left[ \frac{V}{N} \left( \frac{4\pi m E}{3N h^2} \right)^{3/2} \right] + \frac{5}{2} \lambda N k_B$

$S(\lambda E, \lambda V, \lambda N) = \lambda \left( N k_B \ln \left[ \frac{V}{N} \left( \frac{4\pi m E}{3N h^2} \right)^{3/2} \right] + \frac{5}{2} N k_B \right)$

$$S(\lambda E, \lambda V, \lambda N) = \lambda S(E, V, N)$$


(b)

The Gibbs correction is a term that depends only on $N$ ($S_{corr} = -k_B \ln(N!)$). 


Pressure is defined as $p = T \left( \frac{\partial S}{\partial V} \right)_{E,N}$.

Since the correction term, $-k_B \ln(N!)$, does not depend on volume $V$, its derivative with respect to $V$ is zero, pressure is unchanged. 


Heat capacity is defined as $C_V = \left( \frac{\partial E}{\partial T} \right)_{V,N}$. Monatomic ideal gas, $E = \frac{3}{2} N k_B T$.

Since the relationship between $E$ and $T$ is unchanged, the heat capacity is also unchanged.


The exponent $\gamma$ is the ratio of specific heats, $\gamma = C_p/C_V$. Since $C_V$ is unchanged and Mayer's relation for an ideal gas ($C_p = C_V + N k_B$) still holds, $C_p$ is also unchanged. Consequently, $\gamma$ is unchanged, and the adiabatic relation remains the same.


(c)

Let's use the final Sackur-Tetrode equation:

$S = N k_B \left( \ln\left[\frac{V}{N}\left(\frac{4\pi mE}{3Nh^2}\right)^{3/2}\right] + \frac{5}{2} \right) = N k_B \left( \ln(V E^{3/2} C) - \frac{5}{2}\ln N + \frac{5}{2} \right)$ where C is a constant.

$\frac{\partial S}{\partial N} = k_B \left( \ln(V E^{3/2} C) - \frac{5}{2}\ln N + \frac{5}{2} \right) + N k_B \left(-\frac{5}{2N}\right)$

$\frac{\partial S}{\partial N} = k_B \ln\left[\frac{V}{N^{5/2}} (\dots)\right] + \frac{5}{2}k_B - \frac{5}{2}k_B = k_B \ln\left[\frac{V}{N}\left(\frac{4\pi mE}{3Nh^2}\right)^{3/2}\right]$

$\mu = -T \frac{\partial S}{\partial N} = -k_B T \ln\left[\frac{V}{N}\left(\frac{4\pi mE}{3Nh^2}\right)^{3/2}\right]$


Since the entropy $S$ changes, the Helmholtz free energy $F$ must also change.

$F_{new} = E - T S_{new} = E - T(S_{old} - k_B \ln N!) = (E - T S_{old}) + k_B T \ln N!$

$$F_{new} = F_{old} + k_B T \ln(N!)$$



4.

(a)

Use the entropy formula for an ideal gas, $S(N,V) = N k_B \ln(V/N) + \text{constant}$.

Initial Entropy: $S_i = S(N_1, V_1) + S(N_2, V_2) = N_1 k_B \ln(V_1/N_1) + N_2 k_B \ln(V_2/N_2)$.

Final Entropy: $S_f = S(N_1+N_2, V_1+V_2) = (N_1+N_2) k_B \ln\left(\frac{V_1+V_2}{N_1+N_2}\right)$.


        $$(\Delta S)_{1=2} = S_f - S_i = k_B \left[ (N_1+N_2) \ln\left(\frac{V_1+V_2}{N_1+N_2}\right) - N_1 \ln\left(\frac{V_1}{N_1}\right) - N_2 \ln\left(\frac{V_2}{N_2}\right) \right]$$


Set volume $v_1 = V_1/N_1$ and $v_2 = V_2/N_2$:

    $$v_f = \frac{V_1+V_2}{N_1+N_2} = \frac{N_1 v_1 + N_2 v_2}{N_1+N_2}$$

Then it changes with $N = N_1+N_2$, $x_1 = N_1/N$, and $x_2 = N_2/N$:

    $$\ln(v_f) - x_1 \ln(v_1) - x_2 \ln(v_2) \ge 0$$

    $$\ln(x_1 v_1 + x_2 v_2) \ge x_1 \ln(v_1) + x_2 \ln(v_2)$$

Use Jensen's inequality ($f(x) = \ln(x)$, is strictly concave), it's true.

Equal at 

    $$v_1 = v_2 \implies \frac{V_1}{N_1} = \frac{V_2}{N_2}$$


(b)

$$(\Delta S)^* = -k_B (N_1 \ln x_1 + N_2 \ln x_2)$$

where $N = N_1 + N_2$ and $x_1 = N_1/N$ and $x_2 = N_2/N$.

Let $N = N_1+N_2$. then $x = x_1 = N_1/N$, so that $x_2 = 1-x$,

    $$(\Delta S)^*(x) = -N k_B [x \ln x + (1-x) \ln(1-x)]$$

To find the maximum,

    $$\frac{d(\Delta S)^*}{dx} = -N k_B \frac{d}{dx}[x \ln x + (1-x) \ln(1-x)]$$   $$\frac{d(\Delta S)^*}{dx} = -N k_B \left[ (\ln x + 1) + (-\ln(1-x) - 1) \right] = -N k_B [\ln x - \ln(1-x)]$$

    $$\ln x - \ln(1-x) = 0 \implies \ln x = \ln(1-x) \implies x = 1-x \implies 2x=1$$

at $x = 1/2$.

    The second derivative is $\frac{d^2(\Delta S)^*}{dx^2} = -N k_B (\frac{1}{x} + \frac{1}{1-x})$, which is negative for $x \in (0,1)$, then $x=1/2$ is a maximum.


Maximum  when $x_1 = x_2 = 1/2$, which means $N_1 = N_2$.

    $$(\Delta S)^*_{max} = -N k_B \left[ \frac{1}{2} \ln\left(\frac{1}{2}\right) + \frac{1}{2} \ln\left(\frac{1}{2}\right) \right]$$   $$(\Delta S)^*_{max} = -N k_B \left[ \ln\left(\frac{1}{2}\right) \right] = -N k_B (-\ln 2) = N k_B \ln 2$$

    $$(\Delta S)^* \le (N_1+N_2)k_B \ln 2$$



5.

For ideal gas, the adiabatic exponent $\gamma$ and the molar heat capacities $C_{P,m}$ and $C_{V,m}$ are related by:

$$\gamma = \frac{C_{P,m}}{C_{V,m}}$$

$$C_{P,m} - C_{V,m} = R$$

$$\gamma = \frac{C_{V,m} + R}{C_{V,m}} = 1 + \frac{R}{C_{V,m}}$$

$$\gamma - 1 = \frac{R}{C_{V,m}} \quad \implies \quad C_{V,m} = \frac{R}{\gamma - 1}$$


For a mixture of ideal gases,

$$C_{V,mix} = f_1 C_{V,m1} + f_2 C_{V,m2}$$

where $f_1$ and $f_2$ are the mole fractions, and $C_{V,m1}$ and $C_{V,m2}$ are the molar heat capacities of gas 1 and gas 2.


Use $C_{V,mix} = \frac{R}{\gamma - 1}$, $C_{V,m1} = \frac{R}{\gamma_1 - 1}$, $C_{V,m2} = \frac{R}{\gamma_2 - 1}$

$$\frac{R}{\gamma - 1} = f_1 \left(\frac{R}{\gamma_1 - 1}\right) + f_2 \left(\frac{R}{\gamma_2 - 1}\right)$$

$$\frac{1}{\gamma - 1} = \frac{f_1}{\gamma_1 - 1} + \frac{f_2}{\gamma_2 - 1}$$













1

(a) 

The classical partition function ($Z_{cl}$) is defined as the integral over all of phase space. Let $\beta = \frac{1}{k_BT}$.

$$Z_{cl} = \int_{-\infty}^{\infty} \int_{-\infty}^{\infty} e^{-\beta\mathcal{H}} \,dp\,dx$$

$$Z_{cl} = \int_{-\infty}^{\infty} e^{-\frac{\beta p^2}{2m}} \,dp \int_{-\infty}^{\infty} e^{-\frac{\beta m\omega^2x^2}{2}} \,dx$$


Both integrals are Gaussian integrals of the form $\int_{-\infty}^{\infty} e^{-ax^2} dx = \sqrt{\frac{\pi}{a}}$.


Integral over momentum ($p$): Here, $a = \frac{\beta}{2m}$.

    $$\int_{-\infty}^{\infty} e^{-\frac{\beta p^2}{2m}} \,dp = \sqrt{\frac{\pi}{\beta/2m}} = \sqrt{\frac{2\pi m}{\beta}}$$

Integral over position ($x$): Here, $a = \frac{\beta m\omega^2}{2}$.

    $$\int_{-\infty}^{\infty} e^{-\frac{\beta m\omega^2x^2}{2}} \,dx = \sqrt{\frac{\pi}{\beta m\omega^2/2}} = \sqrt{\frac{2\pi}{\beta m\omega^2}}$$


Multiplying the two results gives the partition function:

$$Z_{cl} = \left(\sqrt{\frac{2\pi m}{\beta}}\right) \left(\sqrt{\frac{2\pi}{\beta m\omega^2}}\right) = \frac{2\pi}{\beta\omega}= \frac{2\pi k_BT}{\omega}$$


(b)

$$\langle E \rangle = -\frac{\partial}{\partial\beta} \ln Z_{cl} = -\frac{\partial}{\partial\beta} \ln\left(\frac{2\pi}{\beta\omega}\right) = -\left(-\frac{1}{\beta}\right) = \frac{1}{\beta} = k_BT$$


The Hamiltonian is separable: $\mathcal{H} = \frac{p^2}{2m} + \frac{1}{2}m\omega^2x^2$. The average energy is the sum of the averages of its parts:

$$\langle E \rangle = \left\langle \frac{p^2}{2m} \right\rangle + \left\langle \frac{1}{2}m\omega^2x^2 \right\rangle$$

The partition function is also separable into a product $Z_{cl} = Z_p \cdot Z_x$, where $Z_p = \sqrt{\frac{2\pi m}{\beta}}$ and $Z_x = \sqrt{\frac{2\pi}{\beta m\omega^2}}$. We can calculate the average of each energy term from its corresponding part of the partition function.


Average kinetic energy:

$$\left\langle \frac{p^2}{2m} \right\rangle = -\frac{\partial}{\partial\beta} \ln Z_p = -\frac{\partial}{\partial\beta} \ln\left(\sqrt{\frac{2\pi m}{\beta}}\right) = -\frac{\partial}{\partial\beta}\left[\frac{1}{2}\ln(2\pi m) - \frac{1}{2}\ln\beta\right]$$

$$= -\left(-\frac{1}{2\beta}\right) = \frac{1}{2\beta} = \frac{1}{2}k_BT$$


Average potential energy:

$$\left\langle \frac{1}{2}m\omega^2x^2 \right\rangle = -\frac{\partial}{\partial\beta} \ln Z_x = -\frac{\partial}{\partial\beta} \ln\left(\sqrt{\frac{2\pi}{\beta m\omega^2}}\right) = -\frac{\partial}{\partial\beta}\left[\frac{1}{2}\ln\left(\frac{2\pi}{m\omega^2}\right) - \frac{1}{2}\ln\beta\right]$$

$$= -\left(-\frac{1}{2\beta}\right) = \frac{1}{2\beta} = \frac{1}{2}k_BT$$

Calculation from the partition function shows that the kinetic energy degree of freedom and the potential energy degree of freedom each have an average energy of $\frac{1}{2}k_BT$. 


(c)

The quantum mechanical partition function ($Z_{QM}$) is the sum of the Boltzmann factors over all quantum states:

$$Z_{QM} = \sum_{n=0}^{\infty} e^{-\beta E_n} = \sum_{n=0}^{\infty} e^{-\beta(n+\frac{1}{2})\hbar\omega}$$

$$Z_{QM} = e^{-\frac{\beta\hbar\omega}{2}} \sum_{n=0}^{\infty} e^{-n\beta\hbar\omega} = e^{-\frac{\beta\hbar\omega}{2}} \sum_{n=0}^{\infty} (e^{-\beta\hbar\omega})^n$$


The summation is a geometric series with the first term $a=1$ and the common ratio $r = e^{-\beta\hbar\omega}$. Since $|r|<1$, the sum converges to $\frac{1}{1-r}$.

$$\sum_{n=0}^{\infty} (e^{-\beta\hbar\omega})^n = \frac{1}{1-e^{-\beta\hbar\omega}}$$


Therefore, the quantum mechanical partition function is:

$$Z_{QM} = \frac{e^{-\frac{\beta\hbar\omega}{2}}}{1-e^{-\beta\hbar\omega}}$$

This can also be expressed compactly using the hyperbolic sine function:

$$\Large Z_{QM} = \frac{1}{e^{\frac{\beta\hbar\omega}{2}} - e^{-\frac{\beta\hbar\omega}{2}}} = \frac{1}{2\sinh\left(\frac{\beta\hbar\omega}{2}\right)}$$


(d)

Average Energy $\langle E \rangle$

The quantum average energy is calculated from $Z_{QM}$:

$$\langle E \rangle = -\frac{\partial}{\partial\beta} \ln Z_{QM}$$

$$\ln Z_{QM} = \ln\left(\frac{e^{-\frac{\beta\hbar\omega}{2}}}{1-e^{-\beta\hbar\omega}}\right) = -\frac{\beta\hbar\omega}{2} - \ln(1-e^{-\beta\hbar\omega})$$

Differentiating with respect to $\beta$:

$$\frac{\partial}{\partial\beta} \ln Z_{QM} = -\frac{\hbar\omega}{2} - \frac{1}{1-e^{-\beta\hbar\omega}}(-e^{-\beta\hbar\omega})(-\hbar\omega) = -\frac{\hbar\omega}{2} - \frac{\hbar\omega e^{-\beta\hbar\omega}}{1-e^{-\beta\hbar\omega}}$$

Thus, the average energy is:

$$\langle E \rangle = - \left(-\frac{\hbar\omega}{2} - \frac{\hbar\omega e^{-\beta\hbar\omega}}{1-e^{-\beta\hbar\omega}}\right) = \frac{\hbar\omega}{2} + \frac{\hbar\omega e^{-\beta\hbar\omega}}{1-e^{-\beta\hbar\omega}}$$

By multiplying the numerator and denominator of the second term by $e^{\beta\hbar\omega}$, we get the standard form:

$$\langle E \rangle = \frac{\hbar\omega}{2} + \frac{\hbar\omega}{e^{\beta\hbar\omega}-1}$$


High-Temperature Limit ($T \to \infty$)

The high-temperature limit corresponds to $T \to \infty$, which means $\beta = \frac{1}{k_BT} \to 0$. In this limit, $\beta\hbar\omega \ll 1$, so we can use the Taylor expansion for the exponential, 

$$e^{\beta\hbar\omega} \approx 1+\beta\hbar\omega$$

Average energy:

$$\langle E \rangle \approx \frac{\hbar\omega}{2} + \frac{\hbar\omega}{(1+\beta\hbar\omega)-1} = \frac{\hbar\omega}{2} + \frac{\hbar\omega}{\beta\hbar\omega} = \frac{\hbar\omega}{2} + \frac{1}{\beta}$$


As $T \to \infty$ (or $\beta \to 0$), the first term, the zero-point energy $\frac{\hbar\omega}{2}$, is a finite constant, while the second term $\frac{1}{\beta}$ becomes dominant.

$$\lim_{T \to \infty} \langle E \rangle = \lim_{\beta \to 0} \left( \frac{\hbar\omega}{2} + \frac{1}{\beta} \right) = \frac{1}{\beta} = k_BT$$


This result, $\langle E \rangle = k_BT$, exactly matches the average energy of the classical harmonic oscillator found in part (b).


2

(a)

The Helmholtz free energy $F$ is defined as $F = U - TS$, where $U$ is the internal energy, $T$ is the temperature, and $S$ is the entropy. 

The internal energy $U$ is the average energy of the system: $U = \langle E \rangle = \sum_s P_s E_s$.

The Gibbs entropy is given by $S = -k_B \sum_s P_s \ln P_s$.

The probability $P_s$ of the system being in a microstate $s$ with energy $E_s$ is $P_s = \frac{e^{-\beta E_s}}{Z}$, where $\beta = \frac{1}{k_B T}$ and $Z = \sum_s e^{-\beta E_s}$ is the partition function.


    $$S = -k_B \sum_s \left( \frac{e^{-\beta E_s}}{Z} \right) \ln \left( \frac{e^{-\beta E_s}}{Z} \right)$$

    $$S = -k_B \sum_s \frac{e^{-\beta E_s}}{Z} (-\beta E_s - \ln Z)$$

    $$S = k_B \beta \sum_s \frac{E_s e^{-\beta E_s}}{Z} + k_B \ln Z \sum_s \frac{e^{-\beta E_s}}{Z}$$


Note $\sum_s \frac{E_s e^{-\beta E_s}}{Z} = U$, $\sum_s \frac{e^{-\beta E_s}}{Z} = \frac{1}{Z} \sum_s e^{-\beta E_s} = \frac{Z}{Z} = 1$.


    $$S = k_B \beta U + k_B \ln Z$$

    $$S = \frac{U}{T} + k_B \ln Z$$

    $$TS = U + k_B T \ln Z$$

    $$U - TS = -k_B T \ln Z$$

    $$\boldsymbol{F = -k_B T \ln Z}$$


(b)

    $$Z_1 = \frac{1}{h^2} \int d^2q \int d^2p \ e^{-\beta H_1(\vec{p})}$$

    Here, $H_1(\vec{p}) = c|\vec{p}|$ is the Hamiltonian for a single particle, and the integral over the position coordinates $\int d^2q$ is simply the area $A$.

Use polar coordinates, where $d^2p = p \, dp \, d\theta$ and $|\vec{p}| = p$.

    $$Z_1 = \frac{A}{h^2} \int_0^{2\pi} d\theta \int_0^\infty dp \ p e^{-\beta c p}$$

    The integral over $\theta$ gives $2\pi$. The integral over $p$ is a standard form $\int_0^\infty x e^{-ax} dx = 1/a^2$. Here, $x=p$ and $a = \beta c$.

    $$\int_0^\infty p e^{-\beta c p} dp = \frac{1}{(\beta c)^2}$$

    Combining the parts, the single-particle partition function is:

    $$Z_1 = \frac{A}{h^2} \cdot 2\pi \cdot \frac{1}{(\beta c)^2} = \frac{2\pi A}{(h\beta c)^2} = \frac{2\pi A(k_B T)^2}{(hc)^2}$$


    For $N$ non-interacting, indistinguishable particles, the partition function is $Z = \frac{Z_1^N}{N!}$.

    $$Z = \frac{1}{N!} \left( \frac{2\pi A(k_B T)^2}{(hc)^2} \right)^N$$


    Now, we use the result from part (a), $F = -k_B T \ln Z$.

    $$F = -k_B T \ln \left[ \frac{1}{N!} \left( \frac{2\pi A(k_B T)^2}{(hc)^2} \right)^N \right]$$

    $$F = -k_B T \left[ N \ln\left( \frac{2\pi A(k_B T)^2}{(hc)^2} \right) - \ln(N!) \right]$$

    Using the Stirling approximation, $\ln(N!) \approx N \ln N - N$:

    $$F \approx -k_B T \left[ N \ln\left( \frac{2\pi A(k_B T)^2}{(hc)^2} \right) - (N \ln N - N) \right]$$

    $$F = -N k_B T \left[ \ln\left( \frac{2\pi A(k_B T)^2}{(hc)^2} \right) - \ln N + 1 \right]$$

    Combining the logarithmic terms, we get the final expression for the free energy:

    $$\boldsymbol{F = -N k_B T \left[ \ln \left( \frac{2\pi A (k_B T)^2}{N(hc)^2} \right) + 1 \right]}$$


(c)

Entropy (S)

$S = -\left(\frac{\partial F}{\partial T}\right)_{A,N}$

Let's first write $F$ as $F = -N k_B T \left[ \ln \left( \frac{2\pi A k_B^2}{N(hc)^2} \right) + 2\ln T + 1 \right]$.

$S = - \left( -N k_B \left[ \ln \left( \dots \right) + 2\ln T + 1 \right] - N k_B T \left[ \frac{2}{T} \right] \right)$

$S = N k_B \left[ \ln \left( \frac{2\pi A (k_B T)^2}{N(hc)^2} \right) + 1 \right] + 2N k_B$

$$\boldsymbol{S = N k_B \left[ \ln \left( \frac{2\pi A (k_B T)^2}{N(hc)^2} \right) + 3 \right]}$$


Internal Energy ($U = F + TS$)

$U = -N k_B T \left[ \ln \left( \dots \right) + 1 \right] + T \left( N k_B \left[ \ln \left( \dots \right) + 3 \right] \right)$

$U = -N k_B T \ln(\dots) - N k_B T + N k_B T \ln(\dots) + 3N k_B T$

$$\boldsymbol{U = 2N k_B T}$$


Pressure (P)

$P = -\left(\frac{\partial F}{\partial A}\right)_{T,N}$

$F = -N k_B T \left[ \ln A + \ln \left( \frac{2\pi (k_B T)^2}{N(hc)^2} \right) + 1 \right]$

$P = - \left( -N k_B T \cdot \frac{1}{A} \right)$

$$\boldsymbol{P = \frac{N k_B T}{A}}$$


Chemical Potential ($\mu$)

$\mu = \left(\frac{\partial F}{\partial N}\right)_{T,A}$

$F = -N k_B T \left[ \ln \left( \frac{2\pi A (k_B T)^2}{(hc)^2} \right) - \ln N + 1 \right]$

Let $C = k_B T \left[ \ln \left( \frac{2\pi A (k_B T)^2}{(hc)^2} \right) + 1 \right]$. Then $F = -NC + N k_B T \ln N$.

$\mu = \frac{\partial}{\partial N} (-NC + N k_B T \ln N)$

$\mu = -C + \left( k_B T \ln N + N k_B T \frac{1}{N} \right)$

$\mu = -C + k_B T \ln N + k_B T$

$\mu = -k_B T \left[ \ln \left( \frac{2\pi A (k_B T)^2}{(hc)^2} \right) + 1 \right] + k_B T \ln N + k_B T$

$\mu = k_B T \left[ -\ln \left( \frac{2\pi A (k_B T)^2}{(hc)^2} \right) -1 + \ln N + 1 \right]$

$\mu = k_B T \left[ \ln N - \ln \left( \frac{2\pi A (k_B T)^2}{(hc)^2} \right) \right]$

$$\boldsymbol{\mu = k_B T \ln \left( \frac{N(hc)^2}{2\pi A (k_B T)^2} \right)}$$


3

(a)

The canonical partition function for $N_i$ indistinguishable particles of mass $m_i$ in a volume $V$ at temperature $T$ is given by:

$$Q_i = \frac{q_i^{N_i}}{N_i!}$$

where $q_i$ is the single-particle translational partition function:

$$q_i = V \left( \frac{2\pi m_i k_B T}{h^2} \right)^{3/2} = \frac{V}{\Lambda_i^3}$$

Here, $\Lambda_i = h/\sqrt{2\pi m_i k_B T}$ is the thermal de Broglie wavelength for species $i$, $k_B$ is the Boltzmann constant, and $h$ is Planck's constant.


Since the two types of molecules are distinguishable from each other, the total partition function for the system is the product of the individual partition functions:

$$Q(N_1, N_2, V, T) = Q_1 Q_2 = \frac{q_1^{N_1}}{N_1!} \frac{q_2^{N_2}}{N_2!}$$

Substituting the expressions for $q_1$ and $q_2$:

$$Q = \frac{1}{N_1! N_2!} \left[ V \left( \frac{2\pi m_1 k_B T}{h^2} \right)^{3/2} \right]^{N_1} \left[ V \left( \frac{2\pi m_2 k_B T}{h^2} \right)^{3/2} \right]^{N_2}$$


To find the thermodynamic properties, we first compute the natural logarithm of $Q$, using Stirling's approximation ($\ln N! \approx N \ln N - N$):

$$\ln Q = N_1 \ln q_1 + N_2 \ln q_2 - (N_1 \ln N_1 - N_1) - (N_2 \ln N_2 - N_2)$$

$$\ln Q = N_1 \ln\left(\frac{q_1}{N_1}\right) + N_2 \ln\left(\frac{q_2}{N_2}\right) + (N_1 + N_2)$$


1.  Helmholtz Free Energy ($A$)

    $A = -k_B T \ln Q$

    $$A = -k_B T \left[ N_1 \ln\left(\frac{q_1}{N_1}\right) + N_2 \ln\left(\frac{q_2}{N_2}\right) + N_1 + N_2 \right]$$


2.  Pressure ($P$)

    $P = k_B T \left( \frac{\partial \ln Q}{\partial V} \right)_{T, N_1, N_2}$. Since $\ln q_i = \ln V + \text{terms independent of } V$, we have $\frac{\partial \ln q_i}{\partial V} = \frac{1}{V}$.

    $$P = k_B T \left( N_1 \frac{\partial \ln q_1}{\partial V} + N_2 \frac{\partial \ln q_2}{\partial V} \right) = k_B T \left( \frac{N_1}{V} + \frac{N_2}{V} \right)$$

    $$P = \frac{(N_1 + N_2)k_B T}{V}$$


3.  Internal Energy ($U$)

    $U = k_B T^2 \left( \frac{\partial \ln Q}{\partial T} \right)_{V, N_1, N_2}$. Since $\ln q_i = \frac{3}{2}\ln T + \text{terms independent of } T$, we have $\frac{\partial \ln q_i}{\partial T} = \frac{3}{2T}$.

    $$U = k_B T^2 \left( N_1 \frac{\partial \ln q_1}{\partial T} + N_2 \frac{\partial \ln q_2}{\partial T} \right) = k_B T^2 \left( N_1 \frac{3}{2T} + N_2 \frac{3}{2T} \right)$$

    $$U = \frac{3}{2}(N_1 + N_2)k_B T$$


4.  Entropy ($S$)

    $S = \frac{U - A}{T}$

    $$S = \frac{3}{2}(N_1 + N_2)k_B + k_B \left[ N_1 \ln\left(\frac{q_1}{N_1}\right) + N_2 \ln\left(\frac{q_2}{N_2}\right) + N_1 + N_2 \right]$$

    $$S = k_B \left[ N_1 \ln\left(\frac{V}{N_1\Lambda_1^3}\right) + N_2 \ln\left(\frac{V}{N_2\Lambda_2^3}\right) + \frac{5}{2}(N_1 + N_2) \right]$$



(b)

Properties of the Single-Component Gas


The partition function $Q'$ for this gas is:

$$Q' = \frac{q'^N}{N!}$$

where $q' = V \left( \frac{2\pi m k_B T}{h^2} \right)^{3/2} = \frac{V}{\Lambda'^3}$.


Using the same methods as in part (a):

Pressure ($P'$):

    $$P' = \frac{N k_B T}{V} = \frac{(N_1+N_2)k_B T}{V}$$

Internal Energy ($U'$):

    $$U' = \frac{3}{2}N k_B T = \frac{3}{2}(N_1+N_2)k_B T$$

Entropy ($S'$):

    $$S' = k_B \left[ N \ln\left(\frac{q'}{N}\right) + \frac{5}{2}N \right] = k_B \left[ (N_1+N_2) \ln\left(\frac{V}{(N_1+N_2)\Lambda'^3}\right) + \frac{5}{2}(N_1+N_2) \right]$$


The pressure and internal energy are identical in both cases ($P=P'$ and $U=U'$). 

The entropy is different. Let's find the difference $\Delta S = S_{mixture} - S_{single}$.

    $$\frac{\Delta S}{k_B} = \left[ N_1 \ln\left(\frac{q_1}{N_1}\right) + N_2 \ln\left(\frac{q_2}{N_2}\right) \right] - (N_1+N_2) \ln\left(\frac{q'}{N_1+N_2}\right)$$



4

The virial $\mathcal{V}$ of a system of $N$ particles is defined as the ensemble average of the sum of the dot products of the force $\mathbf{F}_i$ on each particle with its position vector $\mathbf{r}_i$.

$$\mathcal{V} = \left\langle \sum_{i=1}^N \mathbf{F}_i \cdot \mathbf{r}_i \right\rangle$$

The total force $\mathbf{F}_i$ on any particle can be split into two components: the external force exerted by the walls of the container, $\mathbf{F}_i^{\text{ext}}$, and the internal force from other particles in the system, $\mathbf{F}_i^{\text{int}}$.

$$\mathcal{V} = \left\langle \sum_{i=1}^N \mathbf{F}_i^{\text{ext}} \cdot \mathbf{r}_i \right\rangle + \left\langle \sum_{i=1}^N \mathbf{F}_i^{\text{int}} \cdot \mathbf{r}_i \right\rangle$$


The external forces are due to the pressure $P$ exerted by the container walls. The force on a surface element $d\mathbf{S}$ of the container wall is $-P d\mathbf{S}$ (pointing into the volume). The sum over the particles feeling this force can be replaced by an integral over the container's surface area $S$.

$$\left\langle \sum_{i=1}^N \mathbf{F}_i^{\text{ext}} \cdot \mathbf{r}_i \right\rangle = \oint_S (-P d\mathbf{S}) \cdot \mathbf{r} = -P \oint_S \mathbf{r} \cdot d\mathbf{S}$$

Using the divergence theorem, which states $\oint_S \mathbf{A} \cdot d\mathbf{S} = \int_V (\nabla \cdot \mathbf{A}) dV$, we can convert the surface integral into a volume integral. Here, our vector field is the position vector $\mathbf{A} = \mathbf{r} = (x, y, z)$.


The divergence of $\mathbf{r}$ is:

$$\nabla \cdot \mathbf{r} = \frac{\partial x}{\partial x} + \frac{\partial y}{\partial y} + \frac{\partial z}{\partial z} = 1 + 1 + 1 = 3$$

Substituting this back into the integral:

$$-P \oint_S \mathbf{r} \cdot d\mathbf{S} = -P \int_V (\nabla \cdot \mathbf{r}) dV = -P \int_V 3 dV = -3PV$$

So, the contribution to the virial from external forces is -3PV.


The internal force on particle $i$ is the negative gradient of the total potential energy $U_{\text{total}}$ with respect to its coordinates: $\mathbf{F}_i^{\text{int}} = -\nabla_i U_{\text{total}}$.

The contribution to the virial is:

$$\left\langle \sum_{i=1}^N \mathbf{F}_i^{\text{int}} \cdot \mathbf{r}_i \right\rangle = \left\langle \sum_{i=1}^N (-\nabla_i U_{\text{total}}) \cdot \mathbf{r}_i \right\rangle = -\left\langle \sum_{i=1}^N \mathbf{r}_i \cdot \nabla_i U_{\text{total}} \right\rangle$$


The problem states that the interparticle potential energy $u(\mathbf{r})$ is a homogeneous function of degree $n$. This implies that the total potential energy of the system, $U_{\text{total}}(\mathbf{r}_1, \dots, \mathbf{r}_N)$, is also a homogeneous function of degree $n$ with respect to all particle coordinates. That is, if we scale all coordinates by a factor $\lambda$, the potential energy scales as:

$$U_{\text{total}}(\lambda\mathbf{r}_1, \dots, \lambda\mathbf{r}_N) = \lambda^n U_{\text{total}}(\mathbf{r}_1, \dots, \mathbf{r}_N)$$


According to Euler's theorem for homogeneous functions, if a function $f(x_1, \dots, x_m)$ is homogeneous of degree $k$, then $\sum_{j=1}^m x_j \frac{\partial f}{\partial x_j} = kf$. Applying this theorem to our potential energy $U_{\text{total}}$:

$$\sum_{i=1}^N \mathbf{r}_i \cdot \nabla_i U_{\text{total}} = n U_{\text{total}}$$


Taking the ensemble average gives $\langle n U_{\text{total}} \rangle = nU$, where $U$ is the mean potential energy.

Thus, the contribution to the virial from internal forces is:

$$-\left\langle \sum_{i=1}^N \mathbf{r}_i \cdot \nabla_i U_{\text{total}} \right\rangle = -nU$$


Combining the contributions from external and internal forces, we get the total virial:

$$\mathcal{V} = -3PV - nU$$



The virial theorem provides a general relationship between the mean kinetic energy $K$ and the virial $\mathcal{V}$:

$$K = \left\langle \sum_{i=1}^N \frac{p_i^2}{2m_i} \right\rangle = -\frac{1}{2} \mathcal{V}$$

Substituting the expression for $\mathcal{V}$ we just derived:

$$K = \frac{1}{2}(3PV + nU)$$


The total mean energy of the system is the sum of the mean kinetic and mean potential energies: $E = K + U$. We can rearrange this to write the potential energy as $U = E - K$.

Now, substitute this into our expression for $K$:

$$K = \frac{1}{2}(3PV + n(E - K))$$

$$K = \frac{1}{n+2}(3PV + nE)$$




5

(a)

The total energy of a single molecule is given as $\epsilon = \epsilon_{trans} + \epsilon_{rot} + \epsilon_{pot}$. Since these energy terms are independent, the single-particle partition function $Z_1$ is a product of the partition functions for translation, rotation, and potential energy.

$Z_1 = Z_{trans} \cdot Z_{rot,pot}$


Translational Part: The standard result for a classical gas in a volume $V$ is:

    $$Z_{trans} = \frac{V}{h^3} \int e^{-\beta p^2/2m} d^3p = V \left(\frac{2\pi m k_B T}{h^2}\right)^{3/2}$$

Rotational and Potential Part: This part requires integrating over the rotational phase space $(\theta, \phi, p_\theta, p_\phi)$.

    $$Z_{rot,pot} = \frac{1}{h^2} \int e^{-\beta \left( \frac{p_\theta^2}{2I} + \frac{p_\phi^2}{2I \sin^2\theta} - \mu E \cos\theta \right)} dp_\theta dp_\phi d\theta d\phi$$

    Integrating over the momenta $p_\theta$ and $p_\phi$ (which are Gaussian integrals) yields:

    $$Z_{rot,pot} = \frac{2\pi I k_B T}{h^2} \int_0^{2\pi} d\phi \int_0^\pi e^{\beta \mu E \cos\theta} \sin\theta \, d\theta$$

    Let $x = \beta \mu E = \frac{\mu E}{k_B T}$. The integral over the angles becomes:

    $$2\pi \int_{-1}^{1} e^{xu} du = 4\pi \frac{\sinh(x)}{x}$$

    where $u = \cos\theta$. Combining these gives the rotational partition function:

    $$Z_{rot,pot} = \frac{8\pi^2 I k_B T}{h^2} \left( \frac{\sinh(\beta \mu E)}{\beta \mu E} \right)$$

    The total partition function for $N$ non-interacting molecules is $Z_N = \frac{Z_1^N}{N!}$.


The average alignment of the dipoles with the electric field gives the electric polarization $P$, defined as the average dipole moment per unit volume.

$$P = \frac{N}{V} \langle \mu \cos\theta \rangle$$

The average value $\langle \cos\theta \rangle$ is given by the Langevin function, $L(x) = \coth(x) - 1/x$, where $x = \beta \mu E$.

$$P = \frac{N\mu}{V} \left( \coth(\beta \mu E) - \frac{1}{\beta \mu E} \right)$$

Using the assumption that $|\mu E| \ll k_B T$ (i.e., $x \ll 1$), we can approximate the Langevin function by expanding it: $L(x) \approx x/3$.

$$P \approx \frac{N\mu}{V} \left( \frac{\beta \mu E}{3} \right) = \frac{N \mu^2}{3V k_B T} E$$


The dielectric constant $\epsilon_r$ relates the polarization to the electric field. In cgs units, the relation is:

$$\epsilon_r = 1 + 4\pi \frac{P}{E}$$

Substituting our approximate expression for $P$:

$$\epsilon_r = 1 + 4\pi \frac{N \mu^2}{3V k_B T}$$

Letting $n = N/V$ be the number density of the molecules, we get the final expression:

$$\boldsymbol{\epsilon_r \approx 1 + \frac{4\pi n \mu^2}{3 k_B T}}$$

This is the Debye equation for the dielectric constant of a polar gas.



(b)

We can treat steam as an ideal gas. The ideal gas law is $PV = N k_B T$, so the number density $n=N/V$ is:

$$n = \frac{P}{k_B T}$$

We will use cgs units for all quantities:

Pressure $P$: $1 \text{ atm} = 1.01325 \times 10^6 \text{ dyne/cm}^2$.

Temperature $T$: $100^\circ C = 373.15 \text{ K}$.

Boltzmann constant $k_B$: $1.3806 \times 10^{-16} \text{ erg/K}$.


First, calculate the thermal energy $k_B T$:

$k_B T = (1.3806 \times 10^{-16} \text{ erg/K}) \times (373.15 \text{ K}) = 5.152 \times 10^{-14} \text{ erg}$.

Now, calculate the number density:

$$n = \frac{1.01325 \times 10^6 \text{ dyne/cm}^2}{5.152 \times 10^{-14} \text{ erg}} = 1.9667 \times 10^{19} \text{ cm}^{-3}$$



Now we plug our values into the Debye equation:

$$\epsilon_r - 1 = \frac{4\pi n \mu^2}{3 k_B T}$$

We are given:

Dipole moment $\mu = 1.85 \times 10^{-18}$ e.s.u. (statcoulomb-cm).


Let's calculate the terms:

$\mu^2 = (1.85 \times 10^{-18})^2 = 3.4225 \times 10^{-36} \text{ (statC}\cdot\text{cm)}^2$.

$3 k_B T = 3 \times (5.152 \times 10^{-14} \text{ erg}) = 1.5456 \times 10^{-13} \text{ erg}$.


Now, combine everything:

$$\epsilon_r - 1 = \frac{4\pi (1.9667 \times 10^{19} \text{ cm}^{-3}) (3.4225 \times 10^{-36} \text{ (statC}\cdot\text{cm)}^2)}{1.5456 \times 10^{-13} \text{ erg}}$$

Recalling that in cgs units, the units combine to be dimensionless (since $\text{erg} = \text{dyne}\cdot\text{cm}$ and $\text{dyne} = \text{statC}^2/\text{cm}^2$), the calculation gives:

$$\epsilon_r - 1 \approx 0.00547$$

$$\boldsymbol{\epsilon_r \approx 1.00547}$$


Thus, the dielectric constant of steam at $100^\circ C$ and atmospheric pressure is approximately 1.0055.























1

(a)

The grand canonical average of any quantity $X$, denoted by $\langle\langle X \rangle\rangle$. First, we average over the microstates for a fixed number of particles $N$ (this is the canonical average, $\langle X \rangle_{can, N}$). Second, we average over all possible values of $N$.


The total energy variance in the grand canonical ensemble is:

$$\langle\langle(\Delta E)^2\rangle\rangle = \langle\langle E^2 \rangle\rangle - \langle\langle E \rangle\rangle^2$$


We can express $\langle\langle E^2 \rangle\rangle$ as the average of the canonical expectation of $E^2$:

$$\langle\langle E^2 \rangle\rangle = \langle\langle \langle E^2 \rangle_{can, N} \rangle\rangle$$

The variance in the canonical ensemble for a fixed $N$ is $\langle(\Delta E)^2\rangle_{can, N} = \langle E^2 \rangle_{can, N} - (\langle E \rangle_{can, N})^2$. Let's denote the average energy in the canonical ensemble, $\langle E \rangle_{can, N}$, as $U(N)$.


Rearranging this, we get $\langle E^2 \rangle_{can, N} = \langle(\Delta E)^2\rangle_{can, N} + U(N)^2$. Substituting this back into the expression for $\langle\langle E^2 \rangle\rangle$:

$$\langle\langle E^2 \rangle\rangle = \langle\langle \langle(\Delta E)^2\rangle_{can, N} + U(N)^2 \rangle\rangle = \langle\langle \langle(\Delta E)^2\rangle_{can, N} \rangle\rangle + \langle\langle U(N)^2 \rangle\rangle$$

The term $\langle\langle \langle(\Delta E)^2\rangle_{can, N} \rangle\rangle$ is what the problem calls $\langle(\Delta E)^2\rangle_{can}$, the average canonical variance.


Now, substitute this back into the formula for the grand canonical variance:

$$\langle\langle(\Delta E)^2\rangle\rangle = \left( \langle(\Delta E)^2\rangle_{can} + \langle\langle U(N)^2 \rangle\rangle \right) - \langle\langle U(N) \rangle\rangle^2$$

$$\langle\langle(\Delta E)^2\rangle\rangle = \langle(\Delta E)^2\rangle_{can} + \left( \langle\langle U(N)^2 \rangle\rangle - \langle\langle U(N) \rangle\rangle^2 \right)$$

The second part of the expression is the variance of the average energy $U(N)$ due to fluctuations in the particle number $N$.


For a large system, the particle number $N$ fluctuates only slightly around its mean value $\langle\langle N \rangle\rangle$. We can use a Taylor expansion for $U(N)$ around $\langle\langle N \rangle\rangle$:

$$U(N) \approx U(\langle\langle N \rangle\rangle) + (N - \langle\langle N \rangle\rangle) \left(\frac{\partial U}{\partial N}\right)_{T,V}$$

The variance of $U(N)$ is $\langle\langle (\Delta U)^2 \rangle\rangle = \langle\langle (U(N) - \langle\langle U \rangle\rangle)^2 \rangle\rangle$. Using the approximation, $U(N) - \langle\langle U \rangle\rangle \approx (N - \langle\langle N \rangle\rangle) \left(\frac{\partial U}{\partial N}\right)_{T,V}$.

$$\langle\langle (\Delta U)^2 \rangle\rangle \approx \left\langle\left\langle \left[ (N - \langle\langle N \rangle\rangle) \left(\frac{\partial U}{\partial N}\right)_{T,V} \right]^2 \right\rangle\right\rangle = \left(\frac{\partial U}{\partial N}\right)_{T,V}^2 \langle\langle (N - \langle\langle N \rangle\rangle)^2 \rangle\rangle$$

$$\langle\langle (\Delta U)^2 \rangle\rangle \approx \left(\frac{\partial U}{\partial N}\right)_{T,V}^2 \langle\langle(\Delta N)^2\rangle\rangle$$


Combining the results, we get the final expression:

$$\langle\langle(\Delta E)^2\rangle\rangle = \langle(\Delta E)^2\rangle_{can} + \left(\frac{\partial U}{\partial N}\right)_{T,V}^2 \langle\langle(\Delta N)^2\rangle\rangle$$



(b)

The average energy $\langle\langle E \rangle\rangle$ can be written as:

$$\langle\langle E \rangle\rangle = \frac{1}{Q} \sum_{N,j} E_{N,j} e^{\beta(\mu N - E_{N,j})}$$

Let's take the partial derivative of $\langle\langle E \rangle\rangle$ with respect to the chemical potential $\mu$ at constant $T$ and $V$:

$$\left(\frac{\partial \langle\langle E \rangle\rangle}{\partial \mu}\right)_{T,V} = \frac{\partial}{\partial \mu} \left( \frac{1}{Q} \sum E e^{\beta(\mu N - E)} \right)$$

Using the quotient rule, we find:

$$\left(\frac{\partial \langle\langle E \rangle\rangle}{\partial \mu}\right)_{T,V} = \frac{1}{Q} \sum E (\beta N) e^{\beta(\mu N - E)} - \frac{1}{Q^2} \left( \sum E e^{\beta(\mu N - E)} \right) \left( \frac{\partial Q}{\partial \mu} \right)$$

We know that $\frac{\partial Q}{\partial \mu} = \sum (\beta N) e^{\beta(\mu N - E)} = \beta Q \langle\langle N \rangle\rangle$. Substituting this in:

$$\left(\frac{\partial \langle\langle E \rangle\rangle}{\partial \mu}\right)_{T,V} = \beta \langle\langle NE \rangle\rangle - \frac{1}{Q^2} (Q \langle\langle E \rangle\rangle) (\beta Q \langle\langle N \rangle\rangle) = \beta \left( \langle\langle NE \rangle\rangle - \langle\langle E \rangle\rangle \langle\langle N \rangle\rangle \right)$$

Therefore, the covariance is:

$$\langle\langle NE \rangle\rangle - \langle\langle E \rangle\rangle \langle\langle N \rangle\rangle = \frac{1}{\beta} \left(\frac{\partial \langle\langle E \rangle\rangle}{\partial \mu}\right)_{T,V} = k_B T \left(\frac{\partial U}{\partial \mu}\right)_{T,V}$$

where we have used $U = \langle\langle E \rangle\rangle$.


Using the chain rule, we can change the variable of differentiation from $\mu$ to the average particle number $\langle\langle N \rangle\rangle$:

$$\left(\frac{\partial U}{\partial \mu}\right)_{T,V} = \left(\frac{\partial U}{\partial \langle\langle N \rangle\rangle}\right)_{T,V} \left(\frac{\partial \langle\langle N \rangle\rangle}{\partial \mu}\right)_{T,V}$$

A fundamental result for fluctuations in the grand canonical ensemble is that the variance of the particle number is given by:

$$\langle\langle(\Delta N)^2\rangle\rangle = k_B T \left(\frac{\partial \langle\langle N \rangle\rangle}{\partial \mu}\right)_{T,V}$$


Substituting these relations back into our expression for the covariance:

$$\langle\langle NE \rangle\rangle - \langle\langle E \rangle\rangle \langle\langle N \rangle\rangle = k_B T \left[ \left(\frac{\partial U}{\partial \langle\langle N \rangle\rangle}\right)_{T,V} \left(\frac{\partial \langle\langle N \rangle\rangle}{\partial \mu}\right)_{T,V} \right]$$

$$\langle\langle NE \rangle\rangle - \langle\langle E \rangle\rangle \langle\langle N \rangle\rangle = \left(\frac{\partial U}{\partial N}\right)_{T,V} \left[ k_B T \left(\frac{\partial \langle\langle N \rangle\rangle}{\partial \mu}\right)_{T,V} \right]$$

$$\langle\langle NE \rangle\rangle - \langle\langle N \rangle\rangle \langle\langle E \rangle\rangle = \left(\frac{\partial U}{\partial N}\right)_{T,V} \langle\langle(\Delta N)^2\rangle\rangle$$



(c)

Start with the statistical definition and derive the thermodynamic quantities from it.

$$\Phi(T, V, \mu) = -\frac{1}{\beta} \ln Q(T, V, \mu)$$

This implies $\ln Q = -\beta \Phi$.


The average value of the quantity $E-\mu N$ is given by:

$$\langle\langle E - \mu N \rangle\rangle = -\left(\frac{\partial \ln Q}{\partial \beta}\right)_{\mu,V}$$

Substituting $\ln Q = -\beta \Phi$:

$$\langle\langle E \rangle\rangle - \mu \langle\langle N \rangle\rangle = -\frac{\partial (-\beta \Phi)}{\partial \beta} = \frac{\partial(\beta \Phi)}{\partial \beta} = \Phi + \beta \frac{\partial \Phi}{\partial \beta}$$


From the thermodynamic definition, we can rearrange to get:

$$\langle\langle E \rangle\rangle - \mu \langle\langle N \rangle\rangle = \Phi + TS$$

By equating the two expressions for $\langle\langle E \rangle\rangle - \mu \langle\langle N \rangle\rangle$, we find:

$$\Phi + TS = \Phi + \beta \frac{\partial \Phi}{\partial \beta} \implies TS = \beta \frac{\partial \Phi}{\partial \beta}$$


We can express the derivative with respect to $\beta$ in terms of a derivative with respect to $T$ using the chain rule. Since $\beta = 1/(k_B T)$, the derivative is $\frac{d\beta}{dT} = -\frac{1}{k_B T^2}$.

$$\frac{\partial \Phi}{\partial \beta} = \frac{\partial \Phi}{\partial T} \frac{dT}{d\beta} = \frac{\partial \Phi}{\partial T} \left( -k_B T^2 \right)$$

Substituting this into our expression for entropy:

$$TS = \beta \left[ \frac{\partial \Phi}{\partial T} (-k_B T^2) \right] = \frac{1}{k_B T} \left[ \frac{\partial \Phi}{\partial T} (-k_B T^2) \right] = -T \left(\frac{\partial \Phi}{\partial T}\right)_{\mu,V}$$

Dividing by $T$ gives the desired result:

$$S = -\left(\frac{\partial \Phi}{\partial T}\right)_{\mu,V}$$




2

(a)

The probability of a single, specific molecule being in the volume $V$ is $p = \frac{V}{V^{(0)}}$. The probability of it being outside this volume is $1-p$. Since there are $N^{(0)}$ molecules and their locations are uncorrelated, the number of molecules $N$ found in volume $V$ follows a binomial distribution.

$$P(N) = \binom{N^{(0)}}{N} p^N (1-p)^{N^{(0)}-N}$$


The mean number of successes in $N^{(0)}$ independent Bernoulli trials is the number of trials multiplied by the probability of success, $p$.

    $$\bar{N} = \sum_{N=0}^{N^{(0)}} N P(N) = N^{(0)}p$$


The variance, $\sigma^2$, of a binomial distribution is given by $\sigma^2 = N^{(0)}p(1-p)$. The root-mean-square (r.m.s.) deviation is the square root of the variance.

    $$(\Delta N)_{r.m.s.} = \sqrt{\sigma^2} = \sqrt{\langle (N - \bar{N})^2 \rangle} = \sqrt{N^{(0)}p(1-p)}$$

This can also be written as:

    $$(\Delta N)_{r.m.s.} = \{N^{(0)}p(1-p)\}^{1/2}$$


(b)

    $$\ln P(N) = \ln(N^{(0)}!) - \ln(N!) - \ln((N^{(0)}-N)!) + N\ln p + (N^{(0)}-N)\ln(1-p)$$

    Applying Stirling's approximation and simplifying gives:

    $$\ln P(N) \approx -N\ln\left(\frac{N}{N^{(0)}p}\right) - (N^{(0)}-N)\ln\left(\frac{N^{(0)}-N}{N^{(0)}(1-p)}\right)$$


The distribution is sharply peaked around its mean $\bar{N} = N^{(0)}p$. We can perform a Taylor expansion of $\ln P(N)$ for small deviations $x = N - \bar{N}$ around the mean.

    $$\ln P(N) \approx \ln P(\bar{N}) + (N-\bar{N})\left.\frac{d\ln P}{dN}\right|_{\bar{N}} + \frac{1}{2}(N-\bar{N})^2\left.\frac{d^2\ln P}{dN^2}\right|_{\bar{N}}$$

The first derivative is zero at the mean, as $\bar{N}$ is the most probable value (the maximum of the distribution).

The second derivative at the mean is:

        $$\left.\frac{d^2\ln P}{dN^2}\right|_{\bar{N}} = -\frac{1}{\bar{N}} - \frac{1}{N^{(0)}-\bar{N}} = -\frac{1}{N^{(0)}p} - \frac{1}{N^{(0)}(1-p)} = -\frac{1}{N^{(0)}p(1-p)} = -\frac{1}{\sigma^2}$$


Substituting this back into the expansion:

    $$\ln P(N) \approx \ln P(\bar{N}) - \frac{(N-\bar{N})^2}{2\sigma^2}$$

    Exponentiating both sides gives the Gaussian form:

    $$P(N) \approx C \cdot e^{-\frac{(N-\bar{N})^2}{2\sigma^2}}$$

    where $C$ is a normalization constant, which for a continuous distribution is $1/\sqrt{2\pi\sigma^2}$.



(c)

Since $N \ll N^{(0)}$, the term $\frac{N^{(0)}!}{(N^{(0)}-N)!} = N^{(0)}(N^{(0)}-1)\cdots(N^{(0)}-N+1)$ is a product of $N$ terms, each of which is approximately $N^{(0)}$.

        $$\frac{N^{(0)}!}{(N^{(0)}-N)!} \approx (N^{(0)})^N$$

Since $p \ll 1$, we can use the limit definition of the exponential function. As $N \ll N^{(0)}$, we can approximate $N^{(0)}-N \approx N^{(0)}$.

        $$(1-p)^{N^{(0)}-N} \approx (1-p)^{N^{(0)}} = \left(1-\frac{\bar{N}}{N^{(0)}}\right)^{N^{(0)}} \approx e^{-\bar{N}}$$


Substitute these back into the binomial expression:

    $$P(N) \approx \frac{(N^{(0)})^N}{N!} p^N e^{-\bar{N}}$$

    Now, we group the terms:

    $$P(N) \approx \frac{(N^{(0)}p)^N}{N!} e^{-\bar{N}}$$


Substituting $\bar{N} = N^{(0)}p$, we arrive at the Poisson distribution:

    $$P(N) = \frac{\bar{N}^N}{N!} e^{-\bar{N}}$$




3

The total energy of a single molecule separates into two independent parts: the translational motion of the center of mass(CM) and the relative motion(vibration).

$$H = H_{CM} + H_{rel} = \left( \frac{P^2}{2M} \right) + \left( \frac{p^2}{2\mu} + \frac{1}{2}K r^2 \right)$$

Because the parts are independent, the single-particle partition function $Z_1$ is the product of the partition functions for each part:

$$Z_1 = Z_{CM} \cdot Z_{rel}$$

The partition function for a single particle is found by integrating over all possible positions and momenta (phase space):

$$Z_1 = \frac{1}{h^6} \int d^3\boldsymbol{R} \int d^3\boldsymbol{P} \int d^3\boldsymbol{r} \int d^3\boldsymbol{p} \ e^{-\beta H(\boldsymbol{R}, \boldsymbol{P}, \boldsymbol{r}, \boldsymbol{p})}$$

where $\beta = 1/(k_B T)$.


The center of mass behaves like a free particle of mass $M=2m$ in a box of volume $V$.

$$Z_{CM} = \frac{1}{h^3} \int_V d^3\boldsymbol{R} \int d^3\boldsymbol{P} \ e^{-\beta \frac{P^2}{2M}}$$

The position integral simply gives the volume $V$. The momentum integral is a standard Gaussian integral:

$$\int_{-\infty}^{\infty} e^{-a x^2} dx = \sqrt{\frac{\pi}{a}}$$

So, for the three momentum components ($P_x, P_y, P_z$):

$$\int d^3\boldsymbol{P} \ e^{-\beta \frac{P^2}{2M}} = \left( \int_{-\infty}^{\infty} e^{-\frac{\beta}{2M} P_x^2} dP_x \right)^3 = \left( \sqrt{\frac{2M\pi}{\beta}} \right)^3$$

Combining these results:

$$Z_{CM} = \frac{V}{h^3} (2\pi M k_B T)^{3/2}$$


Relative Motion Partition Function ($Z_{rel}$)

The relative motion is described by a 3D simple harmonic oscillator with reduced mass $\mu=m/2$.

$$Z_{rel} = \frac{1}{h^3} \int d^3\boldsymbol{r} \int d^3\boldsymbol{p} \ e^{-\beta \left( \frac{p^2}{2\mu} + \frac{1}{2}Kr^2 \right)}$$

This integral itself separates into kinetic and potential parts. The kinetic part is another Gaussian integral over momentum, just like the one for $Z_{CM}$:

$$\int d^3\boldsymbol{p} \ e^{-\beta \frac{p^2}{2\mu}} = \left( \sqrt{\frac{2\mu\pi}{\beta}} \right)^3$$

The potential part is a Gaussian integral over position:

$$\int d^3\boldsymbol{r} \ e^{-\beta \frac{Kr^2}{2}} = \left( \int_{-\infty}^{\infty} e^{-\frac{\beta K}{2} x^2} dx \right)^3 = \left( \sqrt{\frac{2\pi}{\beta K}} \right)^3$$

Combining these for $Z_{rel}$:

$$Z_{rel} = \frac{1}{h^3} \left( \frac{2\pi\mu}{\beta} \right)^{3/2} \left( \frac{2\pi}{\beta K} \right)^{3/2} = \left( \frac{2\pi}{h\beta} \right)^3 \sqrt{\frac{\mu^3}{K^3}} = \left( \frac{k_B T}{h\nu} \right)^3$$

where $\nu = \frac{1}{2\pi}\sqrt{K/\mu}$ is the classical vibrational frequency.


Total Partition Function ($Z_N$)

For a system of $N$ non-interacting, indistinguishable molecules, the total partition function is $Z_N = \frac{Z_1^N}{N!}$.

$$Z_N = \frac{1}{N!} \left[ \frac{V}{h^3} (2\pi M k_B T)^{3/2} \left( \frac{k_B T}{h\nu} \right)^3 \right]^N$$


Helmholtz Free Energy (F)

$$F = -k_B T \ln Z_N = -k_B T \left[ N \ln Z_1 - \ln N! \right]$$

Using Stirling's approximation ($\ln N! \approx N\ln N - N$):

$$F \approx -N k_B T \left[ \ln\left( \frac{Z_1}{N} \right) + 1 \right]$$

Substituting $Z_1$:

$$F \approx -N k_B T \left[ \ln\left( \frac{V}{N h^3} (2\pi M k_B T)^{3/2} \left( \frac{k_B T}{h\nu} \right)^3 \right) + 1 \right]$$

Let's look at the temperature dependence inside the logarithm: $T^{3/2} \cdot T^3 = T^{9/2}$.

$$F \approx -N k_B T \left[ \ln(\text{const}) + \frac{9}{2} \ln(T) \right]$$


Internal Energy (U)

The internal energy can be found from the partition function using the relation $U = -\frac{\partial}{\partial \beta} \ln Z_N$.

$$\ln Z_N = N \ln Z_1 - \ln N! = N \left[ \ln V + \frac{3}{2}\ln(2\pi M) - \frac{3}{2}\ln\beta + 3\ln(k_B) + 3\ln T + \dots \right]$$

Since $\beta = 1/(k_B T)$, we have $\ln T = -\ln(k_B\beta)$.

$$\ln Z_N = N \left[ \ln(\dots) - \frac{3}{2}\ln\beta - 3\ln\beta \right] = N \left[ \ln(\dots) - \frac{9}{2}\ln\beta \right]$$

Now, taking the derivative:

$$U = -\frac{\partial}{\partial \beta} \left[ N \ln(\dots) - \frac{9N}{2}\ln\beta \right] = - \left(-\frac{9N}{2\beta}\right) = \frac{9N}{2\beta}$$

$$U = \frac{9}{2} N k_B T$$


Mean Squared Separation $\langle r_{12}^2 \rangle$

We want to find the thermal average $\langle r_{12}^2 \rangle = \langle r^2 \rangle$. This value is related to the potential energy part of the Hamiltonian, $U_{pot} = \frac{1}{2}Kr^2$.


The average potential energy can be calculated from the relevant part of the partition function, which is the position integral in $Z_{rel}$:

$$Z_{pot} = \int d^3\boldsymbol{r} \ e^{-\beta \frac{Kr^2}{2}} = \left( \frac{2\pi}{\beta K} \right)^{3/2}$$

The average potential energy is:

$$\langle U_{pot} \rangle = -\frac{\partial}{\partial \beta} \ln Z_{pot} = -\frac{\partial}{\partial \beta} \left[ \frac{3}{2} (\ln(2\pi) - \ln\beta - \ln K) \right]$$

$$\langle U_{pot} \rangle = - \frac{3}{2} \left( -\frac{1}{\beta} \right) = \frac{3}{2\beta} = \frac{3}{2} k_B T$$

We also know from the definition of the average that $\langle U_{pot} \rangle = \left\langle \frac{1}{2}Kr^2 \right\rangle = \frac{1}{2}K\langle r^2 \rangle$. Equating the two expressions gives:

$$\frac{1}{2}K\langle r^2 \rangle = \frac{3}{2} k_B T$$

Solving for $\langle r^2 \rangle$:

$$\langle r_{12}^2 \rangle = \frac{3 k_B T}{K}$$

This confirms the result from the equipartition theorem, showing that the mean squared separation is directly proportional to temperature.



4

The grand partition function for a system of non-interacting identical particles is given by:

$$\mathcal{Z} = e^{z Z_1}$$

where $z = e^{\beta \mu}$ is the fugacity and $Z_1$ is the single-particle partition function.


The energy of a single atom is the sum of its kinetic and magnetic potential energies, which are independent: $\epsilon = \epsilon_{kin} + \epsilon_{mag}$. Therefore, the single-particle partition function $Z_1$ is a product of the translational and magnetic partition functions:

$$Z_1 = Z_{trans} \cdot Z_{mag}$$


For a particle of mass $m$ in a volume $V$, the translational partition function is:

$$Z_{trans} = \frac{V}{\lambda_{th}^3}$$

where $\lambda_{th} = \frac{h}{\sqrt{2\pi m k_B T}}$ is the thermal de Broglie wavelength.


The problem states the atom has two magnetic energy states: $-\mu_B H$ and $+\mu_B H$. The partition function is the sum over the Boltzmann factors for these states:

$$Z_{mag} = \sum_{i} e^{-\beta \epsilon_{mag,i}} = e^{-\beta(-\mu_B H)} + e^{-\beta(+\mu_B H)}$$

$$Z_{mag} = e^{\beta \mu_B H} + e^{-\beta \mu_B H} = 2 \cosh(\beta \mu_B H)$$


The full single-particle partition function is:

$$Z_1 = \frac{V}{\lambda_{th}^3} \left[ 2 \cosh(\beta \mu_B H) \right]$$

Finally, the grand partition function is:

$$\mathcal{Z}(T, V, \mu, H) = \exp\left(z \frac{V}{\lambda_{th}^3} \left[ 2 \cosh(\beta \mu_B H) \right] \right)$$


Magnetization is found from the grand potential, $\Omega = -k_B T \ln \mathcal{Z}$, using the thermodynamic relation $M = -\left(\frac{\partial \Omega}{\partial H}\right)_{T, V, \mu}$.


First, let's find the grand potential:

$$\Omega = -k_B T \ln(\mathcal{Z}) = -k_B T \left( z \frac{V}{\lambda_{th}^3} \left[ 2 \cosh(\beta \mu_B H) \right] \right)$$

The average number of particles in the system is $\langle N \rangle = -\left(\frac{\partial \Omega}{\partial \mu}\right)_{T, V, H} = z \frac{\partial (-\Omega)}{\partial z} = z Z_1$. So we can write:

$$\Omega = -k_B T \langle N \rangle$$

Note that $\langle N \rangle$ itself depends on $H$: $\langle N \rangle(H) = z \frac{V}{\lambda_{th}^3} 2 \cosh(\beta \mu_B H)$.


Now, we differentiate $\Omega$ with respect to $H$:

$$M = -\frac{\partial}{\partial H} \left[ -k_B T z \frac{V}{\lambda_{th}^3} 2 \cosh(\beta \mu_B H) \right]$$

$$M = k_B T z \frac{V}{\lambda_{th}^3} \cdot 2 \frac{\partial}{\partial H} \left[ \cosh(\beta \mu_B H) \right]$$

$$M = k_B T z \frac{V}{\lambda_{th}^3} \cdot 2 \left[ \sinh(\beta \mu_B H) \cdot (\beta \mu_B) \right]$$

Substituting $\beta = 1/(k_B T)$:

$$M = z \frac{V}{\lambda_{th}^3} \cdot 2 \mu_B \sinh(\beta \mu_B H)$$

We can express this in terms of the average particle number $\langle N \rangle$:

$$M = \left( z \frac{V}{\lambda_{th}^3} 2 \cosh(\beta \mu_B H) \right) \mu_B \frac{\sinh(\beta \mu_B H)}{\cosh(\beta \mu_B H)}$$

$$M = \langle N \rangle \mu_B \tanh(\beta \mu_B H)$$


The process occurs at constant temperature and volume. The first law of thermodynamics including magnetic work is $dU = dQ + dW$, where the work done on the system is $dW = -P dV + M dH$. Since volume is constant ($dV=0$), we have $dU = dQ + M dH$.


The heat absorbed by the system is $dQ = dU - M dH$. The total heat given off by the system when the field is reduced from $H$ to 0 is:

$$Q_{out} = - \int_{H \to 0} dQ = - \int_{H}^{0} (dU - M dH') = \int_{0}^{H} (dU - M dH')$$

Since the process is isothermal, $dU = \left(\frac{\partial U}{\partial H'}\right)_{T,V,\mu} dH'$.

$$Q_{out} = \int_{0}^{H} \left(\frac{\partial U}{\partial H'}\right)_{T,V,\mu} dH' - \int_{0}^{H} M dH' = U(H) - U(0) + \int_0^H M dH'$$

This expression is incorrect. The heat given off is $Q_{out} = -\Delta Q = - ( \Delta U - W)$, where $W = \int M dH$ is the work done by the system.

$$Q_{out} = W - \Delta U = \int_H^0 M dH' - (U(0) - U(H)) = U(H) - U(0) - \int_H^0 M dH'$$

$$Q_{out} = U(H) - U(0) + \int_0^H M(H') dH'$$

Let's find the total energy $U$. The average energy per particle is $\langle \epsilon \rangle = \langle \epsilon_{trans} \rangle + \langle \epsilon_{mag} \rangle$.

$\langle \epsilon_{trans} \rangle = \frac{3}{2} k_B T$

$\langle \epsilon_{mag} \rangle = -\frac{\partial}{\partial \beta} \ln(Z_{mag}) = -\frac{\partial}{\partial \beta} \ln(2\cosh(\beta\mu_B H)) = -\mu_B H \tanh(\beta \mu_B H)$


So the total energy is $U(H) = \langle N \rangle_H \left[ \frac{3}{2} k_B T - \mu_B H \tanh(\beta \mu_B H) \right]$.

When $H=0$, $U(0) = \langle N \rangle_0 \left[ \frac{3}{2} k_B T \right]$.


Let's define $\langle N \rangle_0 = \langle N \rangle(H=0) = z \frac{V}{\lambda_{th}^3} \cdot 2$. Then $\langle N \rangle_H = \langle N \rangle_0 \cosh(\beta \mu_B H)$.


Now we compute the integral part:

$$\int_0^H M(H') dH' = \int_0^H \langle N \rangle_{H'} \mu_B \tanh(\beta \mu_B H') dH'$$

$$= \int_0^H \langle N \rangle_0 \cosh(\beta \mu_B H') \mu_B \tanh(\beta \mu_B H') dH'$$

$$= \langle N \rangle_0 \mu_B \int_0^H \sinh(\beta \mu_B H') dH'$$

$$= \langle N \rangle_0 \mu_B \left[ \frac{\cosh(\beta \mu_B H')}{\beta \mu_B} \right]_0^H = \langle N \rangle_0 k_B T \left( \cosh(\beta \mu_B H) - 1 \right)$$

Now we assemble $Q_{out}$:

$$Q_{out} = \overbrace{\langle N \rangle_0 \cosh(\beta \mu_B H)\left[\frac{3}{2}k_B T - \mu_B H \tanh(\beta \mu_B H)\right]}^{U(H)} - \overbrace{\langle N \rangle_0 \frac{3}{2}k_B T}^{U(0)} + \overbrace{\langle N \rangle_0 k_B T(\cosh(\beta \mu_B H)-1)}^{\int M dH'}$$

$$Q_{out} = \langle N \rangle_0 \left[ \frac{3}{2}k_B T \cosh(\beta\mu_B H) - \mu_B H \sinh(\beta\mu_B H) - \frac{3}{2}k_B T + k_B T \cosh(\beta\mu_B H) - k_B T \right]$$

Grouping terms:

$$Q_{out} = \langle N \rangle_0 \left[ \left(\frac{3}{2}k_B T + k_B T\right)(\cosh(\beta\mu_B H)-1) - \mu_B H \sinh(\beta\mu_B H) \right]$$

$$Q_{out} = \langle N \rangle_0 \left[ \frac{5}{2}k_B T (\cosh(\beta\mu_B H)-1) - \mu_B H \sinh(\beta\mu_B H) \right]$$

This is the total heat given off by the system. $\langle N \rangle_0$ is the average number of atoms that would be in the system at zero magnetic field, which can be treated as a constant for the process.




5

The fundamental principle we'll use is that at equilibrium, the chemical potential of the molecules in the gaseous phase ($\mu_g$) must be equal to the chemical potential of the molecules in the adsorbed phase ($\mu_a$).


We can model the gas as a single-component ideal gas. The chemical potential for an ideal gas is given by:

$$\mu_g = k_B T \ln \left( \frac{n_g \lambda^3}{Z_{int}} \right)$$

where:

$k_B$ is the Boltzmann constant.

$T$ is the temperature.

$n_g = P_g / (k_B T)$ is the number density of the gas.

$P_g$ is the pressure of the gas.

$\lambda = h / \sqrt{2\pi m k_B T}$ is the thermal de Broglie wavelength.

$Z_{int}$ is the internal partition function of a gas molecule (for rotations, vibrations, etc.), which depends on temperature.


Substituting the expression for $n_g$, we get:

$$\mu_g = k_B T \ln \left( \frac{P_g \lambda^3}{k_B T Z_{int}(T)} \right)$$


For the adsorbed phase, we use the Langmuir model, which assumes:

- There is a fixed number of identical adsorption sites, $N_{sites}$.

- Each site can be occupied by at most one molecule.

- The adsorbed molecules do not interact with each other.


Let $N_a$ be the number of adsorbed molecules. The equilibrium fraction of occupied sites is $\theta = N_a / N_{sites}$.


The canonical partition function, $Z_a$, for the $N_a$ adsorbed molecules is the product of the combinatorial ways to arrange the molecules on the sites and the partition function of a single molecule raised to the power of $N_a$.

$$Z_a = \binom{N_{sites}}{N_a} [q(T)]^{N_a}$$

where:

$\binom{N_{sites}}{N_a} = \frac{N_{sites}!}{N_a!(N_{sites}-N_a)!}$ is the number of ways to arrange $N_a$ molecules on $N_{sites}$ sites.

$q(T)$ is the partition function for a single adsorbed molecule. This includes a term for the binding energy to the site ($-\epsilon_0$) and vibrational modes, so $q(T) = e^{\epsilon_0 / k_B T} Z_{vib}(T)$.


The Helmholtz free energy is $F_a = -k_B T \ln Z_a$. Using Stirling's approximation ($\ln N! \approx N \ln N - N$), we can find $\ln Z_a$:


$$\ln Z_a \approx N_{sites} \ln N_{sites} - N_a \ln N_a - (N_{sites}-N_a)\ln(N_{sites}-N_a) + N_a \ln q(T)$$


The chemical potential is the derivative of the free energy with respect to the number of adsorbed particles, $\mu_a = (\frac{\partial F_a}{\partial N_a})_{T, N_{sites}}$:


$$\mu_a = -k_B T \frac{\partial}{\partial N_a} \ln Z_a$$

$$\mu_a = -k_B T [-\ln N_a - 1 + \ln(N_{sites}-N_a) + 1 + \ln q(T)]$$

$$\mu_a = k_B T \ln \left( \frac{N_a}{N_{sites}-N_a} \right) - k_B T \ln q(T)$$


By dividing the numerator and denominator by $N_{sites}$ and using $\theta = N_a / N_{sites}$, we get:

$$\mu_a = k_B T \ln \left( \frac{\theta}{1-\theta} \right) - k_B T \ln q(T)$$



Now we set the chemical potentials equal: $\mu_g = \mu_a$.

$$k_B T \ln \left( \frac{P_g \lambda^3}{k_B T Z_{int}(T)} \right) = k_B T \ln \left( \frac{\theta}{1-\theta} \right) - k_B T \ln q(T)$$


We can cancel $k_B T$ from all terms and combine the logarithms on the right side:

$$\ln \left( \frac{P_g \lambda^3}{k_B T Z_{int}(T)} \right) = \ln \left( \frac{\theta}{(1-\theta)q(T)} \right)$$


Exponentiating both sides gives:

$$\frac{P_g \lambda^3}{k_B T Z_{int}(T)} = \frac{\theta}{(1-\theta)q(T)}$$


Finally, we solve for the gas pressure, $P_g$:

$$P_g = \frac{\theta}{1-\theta} \left[ \frac{k_B T Z_{int}(T)}{\lambda^3 q(T)} \right]$$

$\lambda^3 \propto T^{-3/2}$ is a function of temperature.


Therefore, we can define this entire term as "a certain function of temperature," call it $K(T)$.

$$P_g = \frac{\theta}{1-\theta} \times K(T)$$






























Problem 1

Density Matrix: The density matrix is: $$\rho = \frac{1}{Z} e^{-\beta H}$$

For a non-interacting system, $H = \sum_{i=1}^N h_i$ where $h_i$ is the single-particle Hamiltonian.

The matrix element is: $$\langle x'_1, x'_2, \ldots, x'_N | \rho | x_1, x_2, \ldots, x_N \rangle = \frac{1}{Z} \langle x'_1, x'_2, \ldots, x'_N | e^{-\beta \sum_i h_i} | x_1, x_2, \ldots, x_N \rangle$$

Since the states are unsymmetrized products: 

$$= \frac{1}{Z} \prod_{i=1}^N \langle x'i | e^{-\beta h_i} | x_i \rangle = \frac{1}{Z} \prod{i=1}^N \rho_1(x'_i, x_i)$$

where $\rho_1(x', x) = \langle x' | e^{-\beta h} | x \rangle$ is the single-particle density matrix.

Partition Function: $$Z = \text{Tr}(\rho) = \int dx_1 dx_2 \cdots dx_N \langle x_1, x_2, \ldots, x_N | e^{-\beta H} | x_1, x_2, \ldots, x_N \rangle$$

$$= \int dx_1 dx_2 \cdots dx_N \prod_{i=1}^N \langle x_i | e^{-\beta h_i} | x_i \rangle$$

$$= \prod_{i=1}^N \left(\int dx_i \langle x_i | e^{-\beta h} | x_i \rangle\right) = (Z_1)^N$$

where $Z_1 = \int dx \langle x | e^{-\beta h} | x \rangle$ is the single-particle partition function.



Problem 2

The canonical ensemble average of any operator $A$ is: $$\langle A \rangle = \frac{1}{Z} \text{Tr}(A e^{-\beta H})$$

where $Z = \text{Tr}(e^{-\beta H})$ and $\beta = 1/(k_B T)$.


Average energy $$\langle H \rangle = \frac{1}{Z} \text{Tr}(H e^{-\beta H}) = -\frac{\partial \ln Z}{\partial \beta}$$


Second moment $$\langle H^2 \rangle = \frac{1}{Z} \text{Tr}(H^2 e^{-\beta H})$$

Note that: $$\frac{\partial}{\partial \beta}\left(\frac{1}{Z} \text{Tr}(H e^{-\beta H})\right) = -\frac{1}{Z^2}\frac{\partial Z}{\partial \beta}\text{Tr}(H e^{-\beta H}) - \frac{1}{Z}\text{Tr}(H^2 e^{-\beta H})$$

$$\frac{\partial \langle H \rangle}{\partial \beta} = -\frac{1}{Z^2}\frac{\partial Z}{\partial \beta}\text{Tr}(H e^{-\beta H}) - \langle H^2 \rangle$$

$$= -\langle H \rangle \frac{1}{Z}\frac{\partial Z}{\partial \beta} - \langle H^2 \rangle = \langle H \rangle^2 - \langle H^2 \rangle$$

Therefore: $$\langle H^2 \rangle - \langle H \rangle^2 = -\frac{\partial \langle H \rangle}{\partial \beta}$$


Relate to heat capacity

The heat capacity at constant volume is: $$C_v = \frac{\partial \langle H \rangle}{\partial T}$$

Using $\beta = 1/(k_B T)$, we have $\frac{\partial}{\partial T} = -\frac{\beta^2}{k_B}\frac{\partial}{\partial \beta}$:

$$C_v = -\frac{\beta^2}{k_B}\frac{\partial \langle H \rangle}{\partial \beta} = \frac{\beta^2}{k_B}(\langle H^2 \rangle - \langle H \rangle^2)$$

Therefore: $$\boxed{\langle H^2 \rangle - \langle H \rangle^2 = k_B T^2 C_v}$$



Problem 3

(a) Derive the average occupation number

Each single-particle state k can have occupation number $n_k \in {0, 1, 2, \ldots, \ell}$.

The grand canonical partition function for state k is: $$\Xi_k = \sum_{n_k=0}^{\ell} e^{-\beta(E_{n_k} - \mu n_k)} = \sum_{n_k=0}^{\ell} (z e^{-\beta \epsilon(k)})^{n_k}$$

where $z = e^{\beta \mu}$ is the fugacity, and the energy of $n_k$ particles in state k is $E_{n_k} = n_k \epsilon(k)$.

This is a geometric series: $$\Xi_k = \sum_{n_k=0}^{\ell} (z e^{-\beta \epsilon(k)})^{n_k} = \frac{1 - (z e^{-\beta \epsilon(k)})^{\ell+1}}{1 - z e^{-\beta \epsilon(k)}}$$

The average occupation number is: $$\langle n_k \rangle = \frac{1}{\Xi_k} \sum_{n_k=0}^{\ell} n_k (z e^{-\beta \epsilon(k)})^{n_k} = z \frac{\partial \ln \Xi_k}{\partial z}$$

$$\ln \Xi_k = \ln[1 - (z e^{-\beta \epsilon(k)})^{\ell+1}] - \ln[1 - z e^{-\beta \epsilon(k)}]$$

$$\frac{\partial \ln \Xi_k}{\partial z} = -\frac{(\ell+1)(z e^{-\beta \epsilon(k)})^{\ell} e^{-\beta \epsilon(k)}}{1 - (z e^{-\beta \epsilon(k)})^{\ell+1}} + \frac{e^{-\beta \epsilon(k)}}{1 - z e^{-\beta \epsilon(k)}}$$

Therefore: $$\langle n_k \rangle = \frac{z e^{-\beta \epsilon(k)}}{1 - z e^{-\beta \epsilon(k)}} - \frac{(\ell+1)(z e^{-\beta \epsilon(k)})^{\ell+1}}{1 - (z e^{-\beta \epsilon(k)})^{\ell+1}}$$

$$= \boxed{\frac{1}{z^{-1}e^{\beta\epsilon(k)} - 1} - \frac{\ell + 1}{(z^{-1}e^{\beta\epsilon(k)})^{\ell+1} - 1}}$$


(b) Check limiting cases

Case ℓ = 1 (Fermi-Dirac): $$\langle n_k \rangle = \frac{1}{z^{-1}e^{\beta\epsilon(k)} - 1} - \frac{2}{(z^{-1}e^{\beta\epsilon(k)})^{2} - 1}$$

$$= \frac{1}{z^{-1}e^{\beta\epsilon(k)} - 1} - \frac{2}{(z^{-1}e^{\beta\epsilon(k)} - 1)(z^{-1}e^{\beta\epsilon(k)} + 1)}$$

$$= \frac{z^{-1}e^{\beta\epsilon(k)} + 1 - 2}{(z^{-1}e^{\beta\epsilon(k)} - 1)(z^{-1}e^{\beta\epsilon(k)} + 1)} = \frac{z^{-1}e^{\beta\epsilon(k)} - 1}{(z^{-1}e^{\beta\epsilon(k)} - 1)(z^{-1}e^{\beta\epsilon(k)} + 1)}$$

$$= \frac{1}{z^{-1}e^{\beta\epsilon(k)} + 1} = \frac{1}{e^{\beta(\epsilon(k) - \mu)} + 1}$$

This is the Fermi-Dirac distribution

Case ℓ → ∞ (Bose-Einstein): As $\ell \to \infty$, if $|z e^{-\beta \epsilon(k)}| < 1$ (which must hold for convergence): $$(z^{-1}e^{\beta\epsilon(k)})^{\ell+1} \to \infty$$

So the second term vanishes: $$\langle n_k \rangle \to \frac{1}{z^{-1}e^{\beta\epsilon(k)} - 1} = \frac{1}{e^{\beta(\epsilon(k) - \mu)} - 1}$$

This is the Bose-Einstein distribution



Problem 4

Part 1: Derive the partition function

For N indistinguishable particles, the quantum partition function is: $$Q_N(V,T) = \frac{1}{N!} \sum_{\text{states}} e^{-\beta E_n}$$

In the classical limit (high temperature), we use the correspondence between quantum states and classical phase space: $$Q_N(V,T) = \frac{1}{N! h^{3N}} \int e^{-\beta H} d^{3N}p , d^{3N}r$$

For an ideal gas with weak interactions, $H = K + U$ where:

  • $K = \sum_{i=1}^N \frac{p_i^2}{2m}$ (kinetic energy)
  • $U = \sum_{i<j} v(r_{ij})$ (interaction potential)

The momentum integrals separate: $$\int e^{-\beta \sum_i p_i^2/2m} d^{3N}p = \left(\int e^{-\beta p^2/2m} d^3p\right)^N = \left(2\pi m k_B T\right)^{3N/2}$$

Using $h = 2\pi\hbar$ and defining the thermal wavelength $\lambda = h/\sqrt{2\pi m k_B T}$: $$\left(2\pi m k_B T\right)^{3N/2} = \frac{(2\pi\hbar)^{3N}}{2^{3N}\lambda^{3N}} = \frac{h^{3N}}{2^{3N}}$$

Therefore: $$Q_N(V,T) = \frac{1}{N! h^{3N}} \cdot \frac{h^{3N}}{2^{3N}} \int e^{-\beta U} d^{3N}r$$

$$\boxed{Q_N(V,T) = \frac{1}{N! 2^{3N}} Z_N(V,T)}$$

where $Z_N(V,T) = \int \exp\left\{-\beta \sum_{i<j} v(r_{ij})\right\} d^{3N}r$


Part 2: First-order correction to equation of state

For weak interactions, expand: $$e^{-\beta \sum_{i<j} v(r_{ij})} \approx 1 - \beta \sum_{i<j} v(r_{ij}) + O(\beta^2)$$

The configurational integral becomes: $$Z_N \approx \int \left[1 - \beta \sum_{i<j} v(r_{ij})\right] d^{3N}r$$

$$= V^N - \beta \sum_{i<j} \int v(r_{ij}) d^{3N}r$$

For each pair $(i,j)$, integrating over all other coordinates gives $V^{N-2}$: $$\int v(r_{ij}) d^{3N}r = V^{N-2} \int v(r) d^3r$$

The number of pairs is $\binom{N}{2} = \frac{N(N-1)}{2}$:

$$Z_N \approx V^N \left[1 - \beta \frac{N(N-1)}{2V^2} \int v(r) d^3r\right]$$

For large N: $$Z_N \approx V^N \left[1 - \beta \frac{N^2}{2V^2} \int v(r) d^3r\right]$$

Define the second virial coefficient: $$B_2(T) = -\frac{1}{2}\int \left(e^{-\beta v(r)} - 1\right) d^3r \approx -\frac{\beta}{2}\int v(r) d^3r$$

Then: $$\ln Z_N \approx N \ln V - \beta \frac{N^2}{V} B_2(T)$$

The Helmholtz free energy is: $$F = -k_B T \ln Q_N = -k_B T \left[\ln Z_N - \ln(N!) - 3N \ln 2\right]$$

Using Stirling's approximation $\ln N! \approx N \ln N - N$: $$F \approx -Nk_B T \ln\left(\frac{eV}{N \cdot 2^3}\right) + Nk_B T \frac{N}{V} B_2(T)$$

The pressure is: $$P = -\frac{\partial F}{\partial V}\bigg|_{T,N} = \frac{Nk_B T}{V} - \frac{N^2 k_B T}{V^2} B_2(T)$$

$$\boxed{PV = Nk_B T\left(1 + \frac{N}{V}B_2(T) + \ldots\right)}$$

This is the virial expansion to first order, where $B_2(T)$ is the second virial coefficient.



Problem 5

Let ${|E_m\rangle}$ be the complete orthonormal set of energy eigenstates with $H|E_m\rangle = E_m|E_m\rangle$.

The exact partition function is: $$Q(\beta) = \text{Tr}(e^{-\beta H}) = \sum_m e^{-\beta E_m}$$


Expand trial states in energy eigenbasis

Any trial state $|\phi_n\rangle$ can be expanded as: $$|\phi_n\rangle = \sum_m c_{nm} |E_m\rangle$$

where $\sum_m |c_{nm}|^2 = 1$ (normalization).


Calculate expectation value $$\langle \phi_n | H | \phi_n \rangle = \sum_m |c_{nm}|^2 E_m$$


Apply convexity of exponential

Since $e^{-\beta x}$ is a convex function for $\beta > 0$, by Jensen's inequality: $$e^{-\beta \sum_m |c_{nm}|^2 E_m} \leq \sum_m |c_{nm}|^2 e^{-\beta E_m}$$

Therefore: $$e^{-\beta \langle \phi_n | H | \phi_n \rangle} \leq \sum_m |c_{nm}|^2 e^{-\beta E_m}$$


Sum over all trial states

$$\sum_n e^{-\beta \langle \phi_n | H | \phi_n \rangle} \leq \sum_n \sum_m |c_{nm}|^2 e^{-\beta E_m} = \sum_m \left(\sum_n |c_{nm}|^2\right) e^{-\beta E_m}$$

Since ${\phi_n}$ is orthonormal (but not necessarily complete): $$\sum_n |c_{nm}|^2 = \sum_n |\langle E_m | \phi_n \rangle|^2 \leq 1$$

with equality when ${\phi_n}$ is complete (by completeness relation).

Therefore: $$\sum_n e^{-\beta \langle \phi_n | H | \phi_n \rangle} \leq \sum_m e^{-\beta E_m} = Q(\beta)$$

$$\boxed{Q(\beta) \geq \sum_n \exp{-\beta \langle \phi_n | H | \phi_n \rangle}}$$


Equality condition:

Equality holds if and only if:

  1. $\sum_n |c_{nm}|^2 = 1$ for all $m$ (completeness)
  2. $e^{-\beta \langle \phi_n | H | \phi_n \rangle} = \sum_m |c_{nm}|^2 e^{-\beta E_m}$ with equality in Jensen's inequality

Condition 2 requires $|\phi_n\rangle$ to be eigenstates (so all weight is on one eigenvalue). Combined with condition 1, we need ${\phi_n}$ to be a complete orthonormal set of eigenfunctions of $H$.

Physical significance: This inequality provides a variational bound for the partition function, useful for approximation methods like the variational principle in quantum mechanics.