# 8.

333: Statistical Mechanics I

Fall 2005 Review Problems

Mid-term Quiz�

The Mid-term quiz will take place on Monday 10/24/05 in room 1-190 from 2:30 to 4:00 pm. There will be a recitation with quiz review on Friday 10/21/05. All topics up to (but not including) the micro-canonical ensemble will be covered. The exam is ‘closed book,’ but if you wish you may bring a two-sided sheet of formulas. The enclosed exams (and solutions) from the previous years are intended to help you review the material. Solutions to the midterm are available in the exams section. ******** Answer all three problems, but note that the ﬁrst parts of each problem are easier than its last parts. Therefore, make sure to proceed to the next problem when you get stuck. You may ﬁnd the following information helpful: Physical Constants Electron mass Electron Charge Speed of light Boltzmann’s const. me ∝ 9.1 × 10−31 Kg e ∝ 1.6 × 10−19 C c ∝ 3.0 × 108 ms−1 kB ∝ 1.4 × 10−23 J K −1 Proton mass Planck’s const./2α Stefan’s const. Avogadro’s number mp ∝ 1.7 × 10−27 Kg h ∝ 1.1 × 10−34 J s−1 ¯ δ ∝ 5.7 × 10−8 W m−2 K −4 N0 ∝ 6.0 × 1023 mol−1 1eV ∞ 1.1 × 104 K

**Conversion Factors 1atm ∞ 1.0 × 105 N m−2 ˚ 1A ∞ 10−10 m Thermodynamics dE = T dS +dW ¯ For a gas: dW = −P dV ¯ Mathematical Formulas �1� != 2
**

� λ 2

For a wire: dW = J dx ¯

⎪�

0

dx xn e−�x =

�n+1 x2 2β 2

n!

⎪� �

� dx exp −ikx − −� � �� e−ikx = n=0

(−ik)n n!

�

=

⇒

� 2 2� 2αδ 2 exp − β 2k

limN �� ln N ! = N ln N − N

(−ik)n n! x5 5!

�xn ≡ + ···

cosh(x) = 1 +

x2 2!

+

x4 4!

sinh(x) = x + Sd =

2λ d/2 (d/2−1)!

� � �� ln e−ikx = n=1

x3 3!

�xn ≡c

+

+···

Surface area of a unit sphere in d dimensions

1

8.333: Statistical Mechanics I�

Fall 1999

Mid-term Quiz�

1. Photon gas Carnot cycle: The aim of this problem is to obtain the blackbody ra diation relation, E(T, V ) → V T 4 , starting from the equation of state, by performing an inﬁnitesimal Carnot cycle on the photon gas.

P

P+dP T+dT V V+dV T

P

V

(b) Express the heat absorbed, Q, in expanding the gas along an isotherm, in terms of P , dV , and an appropriate derivative of E(T, V ). • Applying the ﬁrst law, the heat absorbed is Q = dE + P dV = �� � � � � �� � � γE� γE γE dT + dV + P dV = + P dV. γT V� γV T γV T isotherm (a) Express the work done, W , in the above cycle, in terms of dV and dP . • Ignoring higher order terms, net work is the area of the cycle, given by W = dP dV .

(c) Using the eﬃciency of the Carnot cycle, relate the above expressions for W and Q to T and dT . • The eﬃciency of the Carnot cycle (σ = dT /T ) is here calculated as σ= W dP dT =� = . Q [(γE/γV )T + P ] T

(d) Observations indicate that the pressure of the photon gas is given� P = AT 4 , by 3 2 4 hc) is a constant. Use this information to obtain E(T, V ), assuming where A = α kB /45 (¯ E(0, V ) = 0. • From the result of part (c) and the relation P = AT 4 , 4AT = so that E = 3AV T 4 . 2

4

�

�� γE + AT 4 , γV� T�

or

�

γE γV

�

= 3AT 4 ,

T

for coordinate q . i. as � �m � �n � �σ � � �� � σ m n� γ γ γ � px py pz c = ln p τ � ˜ k �γ γ (−ikx ) γ (−iky ) γ (−ikz ) = mkB T (βσ2 βm0 βn0 + βσ0 βm2 βn0 + βσ0 βm0 βn2 ) . (τ τ ).e. (a) Write down the equilibrium one particle density feq.e. P V 4/3 = constant. exp −iτ · p . all other cumulants are zero. Moments of momentum: Consider a gas of N classical particles of mass m in thermal equilibrium at a temperature T . q 3/2 2mkB T (2αmkB T ) � � �� (b) Calculate the joint characteristic function. τ • The equilibrium Maxwell-Boltzmann distribution reads � � n p2 exp − . in a box of volume V . there are only second cumulants. or � � � � γE γE 0= dT + dV + P dV = 3V dP + 4P dV. q τ mentum p. ******** 2. k τ • Performing the Gaussian average yields � � � � � � k γ τ = e−iγ ·p = exp − mkB T k 2 . f (τ τ ) = p. p τ • Using Wick’s theorem �p� p� (τ · p )≡ = �p� p� pπ pπ ≡ p τ
2
k=0
= �p� p� ≡ �pπ pπ ≡ + 2 �p� pπ ≡ �p� pπ ≡
2 2
= (mkB T ) β�� βππ + 2 (mkB T ) β�π β�π = 5 (mkB T ) β�� . γT V γV T i. x y • The cumulants are calculated from the characteristic function. and mo p. • Adiabatic curves are given by dQ = 0.(e) Find the relation describing the adiabatic paths in the above cycle. for momentum. p k ˜ 2 � � n (c) Find all the joint cumulants pσ pm pz c .. (d) Calculate the joint moment �p� p� (τ · p )≡. 3
.

1 + exp (η/kB T )
� � n1 η = exp − . (ii) Adsorption: a0 + θ � a1 . treating the atoms as ﬁxed scatterers of densities n0 and n1 . which has a higher energy π. ignoring their coordinates and momenta). n0 kB T
Consider photons θ of frequency � = π/¯ and momentum |τ | = h�/c. which can h p ¯ interact with the atoms through the following processes: (i) Spontaneous emission: a1 � a0 + θ. calculate the ratio n1 /n0 of densities of atoms in the two states as a function of temperature T .Alternatively. (b) Write down the Boltzmann equation governing the density f of the photon gas. or in an excited state a1 . Light and matter: In this problem we use kinetic theory to explore the equilibrium between atoms and radiation. = V V 1 + exp (η/kB T ) and n1 = E/η N 1 = . γt γτ q 4
. (iii) Stimulated emission: a1 + θ � a0 + θ + θ. respectively. Assume that spontaneous emission occurs with a probability δsp . By considering the atoms as a collection of N ﬁxed two-state systems of energy E (i.e. and that adsorption and stimulated emission have constant (angle-independent) diﬀerential cross-sections of δad /4α and δst /4α. directly from the characteristic function. � �� � γ γ γ γ �p� p� (τ · p )≡ = p τ p τ � ˜ k � γ (−ik� ) γ (−ik� ) γ (−ikπ ) γ (−ikπ ) γ k=0 � � mkB T 2 � γ γ γ � 2 = 3mkB − (mkB T ) τ 2 e− 2 k � k �γ γ (−ik� ) γ (−ik� ) k=0 = 5 (mkB T ) β�� . • The energy and temperature of a two-state system are related by E= leading to n0 = so that N exp (η/kB T ) N − E/η . V V 1 + exp (η/kB T ) Nη . (a) The atoms are assumed to be either in their ground state a0 .
2
******** 3. • The Boltzmann equation for photons in the presence of ﬁxed scatterers reads γf γf +p· τ = −δad n0 cf + δst n1 cf + δsp n1 .

5
. • In uniform equilibrium. n1 )
Clearly. one can add an exponentially decaying term for x > 0. f (px . and δsp = c δst . What does this imply about the above h cross sections? • The result of part (c) agrees with Planck’s law if δad = δst . for the photons of the above frequency. f will depend on x (and px ). px γx h
To the uniform solution obtained before. Adapt the Boltzmann equation you propose in part (b) to the case of a uniform incoming ﬂux of photons with momentum p = h� x/c. leaving −δad n0 cfeq. c δad n0 /n1 − δst c δad exp (η/kB T ) − δst
(d) According to Planck’s law. + δst n1 cfeq.
�
x
vacuum
matter (n0 . (px ) . What is the penetration length across which the τ ¯ ˆ incoming ﬂux decays? • In this situation. h3
a conclusion ﬁrst reached by Einstein.(c) Find the equilibrium density feq. the Boltzmann equation reduces to � γf n1 � = δst c (n1 − n0 ) f + 3 χ (x) . x > 0) = A (px ) e−ax/px + feq. i. the density of photons at a temperature T depends on their −1 frequency � as feq. but will be independent of y and z. i. feq. + δsp n1 = 0.e. = [exp (¯ �/kB T ) − 1] /h3 . and veriﬁed later with explicit quantum mechanical calculations of cross-sections. = 1 δsp 1 δsp = . as illustrated in the ﬁgure. (e) Consider a situation in which light shines along the x axis on a collection of atoms whose boundary coincides with the x = 0 plane. the left-hand side vanishes.e.

and given by px d= . and is related to the incoming ﬂux. The penetration depth d is the inverse of the decay parameter. a ********
6�
.The constant A (px ) can be determined by matching to solution for x < 0 at x = 0. with a = δst c (n0 − n1 ) > 0.

we obtain � 3 3 ωTc 3θ �Tc = + θTc . =≥ S = V ωT + θT ⎭ n n dT 3 (b) Experiments indicate that there is no latent heat (L = 0) for the transition between the normal and superconducting phases at zero ﬁeld. using the third law of thermodynamics. The heat capacities of the two phases at zero magnetic ﬁeld are approximately given by � Cs (T ) = V �T 3 in the superconducting phase . and {�. ω. and θ. As a result. Cs = V �T = T � dT �3 3 � . respectively. • Finite temperature entropies are obtained by integrating dS = dQ/T . • The Latent hear for the transition is related to the diﬀerence in entropies.) (a) Calculate the entropies Ss (T ) and Sn (T ) of the two phases at zero ﬁeld.e. the electrons in the superconductor form bound Cooper pairs. Calculate the internal energies of both phases at ﬁnite temperatures. yields � � � � ⎧ Es (T ) = E0 + V −� + T 4 � 4 � � . and thus L = Tc (Sn (Tc ) − Ss (Tc )) = 0. Using the entropies calculated in the previous part. and magnetic ﬁelds B. and mechanical work can be ignored throughout this problem. ⎧ ⎧ C = V �ωT 3 + θT ⎬ = T dSn . the internal energy of the superconductor is reduced by an amount V �. θ} are constants. we ﬁnd �� �T 3 dSs ⎧� 3 ⎧ . Integrating the given expressions for heat capacity. =≥ Ss = V . for dN = 0. =≥ Tc = . ω. we have dE = T dS = CdT . i. Using the heat capacities to obtain the heat inputs. and B = 0. as a function of �. (There is no appreciable change in volume at this transition. Superconducting transition: Many metals become superconductors at low temperatures T .333: Statistical Mechanics I
Fall 2000
Mid-term Quiz�
where V is the volume. • Since dE = T dS + BdM + µdN . ⎧ E (T ) = E + V ω T 4 + θ T 2 ⎭ n 0 4 2 7
1. Use this information to obtain the transition temperature Tc .8. � 3 ⎬ Cn (T ) = V ωT + θT in the normal phase
. starting from ¯ S(T = 0) = 0. En (T = 0) = E0 and Es (T = 0) = E0 −V � for the metal and superconductor. 3 3 �−ω (c) At zero temperature. and starting with the internal energies E0 and E0 − V � at T = 0.

B) = E0 − V ω T 4 + θ T 2 ⎭ 12 2 8�
. expelling the magnetic ﬁeld from its interior. using the results obtained before. in conjunction with previous results. as � � � � � � � � ⎧ Gs (T ) = E0 + V −� + T 4 − T V T 3 = E0 − V � + T 4 � 4 3� 12 � � � � � ω 4 θ 2 ω 3 ω 4 θ 2 . =≥ 12 12 c 2 � Using the value of Tc = 3θ/(� − ω). inclusion of magnetic work results in dE = T dS +BdM +µdN . 4�−ω
(e) In the presence of a magnetic ﬁeld B. B) = E0 − V � + � T 4 + V B � 12 � � 8α . Use this information. we have to add the integral of −M dB to the Gibbs free energies calculated in the previous section for B = 0. leading to � 4 ω 4 θ 2 Tc = T + Tc . while in the superconducting phase there is an additional ⎪ ⎪ contribution of − Ms dB = (V /4α) BdB = (V /8α)B 2 . to show that the superconducting phase becomes normal for magnetic ﬁelds larger than Bc (T ) = B0 � T2 1− 2 Tc � . The superconducting phase is a perfect diamagnet. • The Gibbs free energy G = E − T S − BM = µN can be calculated for B = 0 in each phase. ⎧ ⎧ Gn (T. ⎧ G (T ) = E + V ⎭ n T + T − TV T + θT = E0 − V T + T 0 4 2 3 12 2
At the transition point. ω. • Since dG = −SdT − M dB + µdN . and θ.
giving an expression for B0 . obtain an expression for the energy gap � in terms of �. such that Ms = −V B/(4α) in appropriate units. 2 c 12 c
3 θ2 . where M is the magnetization. Hence the Gibbs free energies at ﬁnite ﬁeld are � � � 2 ⎧ ⎧ Gs (T.(d) By comparing the Gibbs free energies (or chemical potentials) in the two phases. There is no change in the metallic phase since Mn = 0. the chemical potentials (and hence the Gibbs free energies) must be equal. we obtain �+ �= �= θ 2 �−ω 4 T − T . with Mn = 0. The normal metal can be regarded as approximately non-magnetic.

Similarly. since the {ti } are independent variables. 12 �−ω �−ω 12
where we have used the values of � and Tc obtained before.Equating the Gibbs free energies gives a critical magnetic ﬁeld
2 Bc θ �−ω 4 3 θ2 θ �−ω 4 = � − T2 + T = − T2 + T 8α 2 12 4�−ω 2 12 �� � �2 �2 3θ 6θT 2 �−ω � 2 �−ω − = + T4 = Tc − T 2 . N ) for large N .
i=1
where ti = 1 if the i-th particle is A. N ) = N! pNA pN −NB . 9
. �
2 NA
�
c
=
N N � �� � �� � � 2 2 t2 − �ti ≡ = pA − p A = N p A pB . (a) What is the probability p(NA . The mean value is then equal to �NA ≡ =
N � i=1
�ti ≡ =
N � i=1
(pA × 1 + pB × 0) = N pA . N ) that NA out of the N particles are of type A? • The answer is the binomial probability distribution p(NA . • We can write N � nA = ti . Taking the square root of the above expression gives Bc = B 0 � T2 1− 2 Tc � . NA !(N − NA )! A B
(b) Calculate the mean and the variance of NA . �−ω
******** 2. where B0 = � 2α(� − ω) 2 Tc = 3 � � 6αθ 2 = Tc 2αθ. Probabilities: Particles of type A or B are chosen independently with probabilities p A and pB . i i=1 i=1
(c) Use the central limit theorem to obtain the probability p(NA . and 0 if it is B.

One plate at y = 0 is maintained at a temperature T1 . − 2 = − dNA N p A pB NA N − NA NA N − NA (NA − pA N ) . and check for consistency with the result from the central limit theorem. N ) = ln N ! − ln NA ! − ln(N − NA )! + NA ln pA + (N − NA ) ln pB � � � � NA NA − (N − NA ) ln 1 − + NA ln pA + (N − NA ) ln pB . The correct normaliza tion is also obtained if the next term in the Stirling approximation is included. for N √ 1. ∝ −NA ln N N The most likely value. N ) [using the prob ability calculated in part (a). N ) calculated in (d) around its maximum to second order in � � NA − NA . • Taking a second derivative of ln p gives d2 ln p 1 1 N 1 � � =− =− . • Applying Stirling’s approximation to the logarithm of the binomial distribution gives ln p(NA . ******** 3. NA .
Thus the most likely value is the same as the mean in this limit. NA . is obtained by setting the derivative of the above expression with respect to NA to zero. not part (c)] to ﬁnd the most likely value. Thermal Conductivity: Consider a classical gas between two plates separated by a distance w. i.• According to the central limit theorem the PDF of the sum of independent variables for large N approaches a Gaussian of the right mean and variance. N ) ∝ exp − N ∞1 2N pA pB �
2
�
⇒
1 . dNA N N − NA pB =≥ NA = pA N. we get (NA − N pA ) lim p(NA . � � d ln p NA N pA = − ln + ln = 0. 2αN pA pB
(d) Apply Stirling’s approximation (ln N ! ∝ N ln N − N ) to ln p(NA . while the other plate at 10�
.e. 2N pA pB
2
The expansion of ln p around its maximum thus gives ln p ∝ −
which is consistent with the result from the central limit theorem. (e) Expand ln p(NA . Using the mean and variance calculated in the previous part.

γt m γy εK 11
.) • Since there is no external force acting on the gas between plates. Use the result p6 = 105(mkB T )3 (you don’t have to derive this) in conjunction with symmetry arguments to conclude � 2 4 �0 3 py p = 35 (mkB T ) . so that the initial zeroth order approximation to the one particle density is. to � � 0 � 1⎬ γ py γ f 1 − f1 0 L f1 ∝ + f1 ∝ − 1 .y = w is at a diﬀerent temperature T2 . � �0 0 0 where �O≡ indicates local averages with the Gaussian weight f1 . show that � 2 �0 � 4 �0 0 0 2 p ∞ �p� p� ≡ = 3 (mkB T ) . � � n(y) p·p τ τ 0 f1 (τ x. and p ∞ �p� p� p� p� ≡ = 15 (mkB T ) . y). Since the local pressure is P (y) = n(y)k B T (y). exp − . y. The gas velocity is zero. the condition for the ﬂuid to be stationary is n(y)T (y) = constant. � �0 � �0 1 � 2 4 �0 1 3 3 = p2 p4 = p2 p 4 = p p = × 105 (mkB T ) = 35 (mkB T ) .
(b) Using Wick’s theorem. z) = p. p by linearizing the Boltzmann equation in the single collision time approximation. Similarly � 4 �0 0 2 2 p = �p� p� p� p� ≡ = (mkB T ) (β�� + 2β�� β�� ) = 15 (mkB T ) . to ensure that the gas velocity τ u remains zero? (Use this relation between n(y) and T (y) in the remainder of this problem. y z 3 3
0
The symmetry along the three directions implies � p2 p4 x �0
(c) The zeroth order approximation does not lead to relaxation of temperature/density 1 variations related as in part (a). 3/2 2mkB T (y) [2αmkB T (y)] (a) What is the necessary relation between n(y) and T (y). • The Gaussian weight has a covariance �p� p� ≡ = β�� (mkB T ). or otherwise. Find a better (time independent) approximation f1 (τ. the gas can only ﬂow locally if there are variations in pressure. Using Wick’s theorem gives � 2 �0 0 p = �p� p� ≡ = (mkB T ) β�� = 3 (mkB T ) .

and T (y) varies linearly between the two plates. hy = −εK 2m3 T 2 2 2 m The coeﬃcient of thermal conductivity relates the heat transferred to the temperature gradient by τ = −K≈T . we have � � � � γ py γ 3 p2 3 0 0 py 0 0 py − ln (2αmkB ) + f = f1 γy ln f1 = f1 γy ln n − ln T − γt m γy 1 m m 2 2mkB T 2 � � � � 2 3 γy T p γT 5 p2 γy T 0 py γy n 0 py − − + + . this gives T2 − T 1 y. But hy = −Kγy T . respectively. = f1 = f1 m n 2 T 2mkB T T m 2 2mkB T T where in the last equality we have used nT = constant to get γy n/n = −γy T /T . and hence we can identify h 2 5 nεK kB T . and T (w) = T2 . p. The correc 1 tions in f1 . the coeﬃcient of thermal conductivity. along with the averages obtained in part (b). 2 2m2 In the zeroth order Gaussian weight all odd moments of p have zero average. where K. which is proportional to the product nT . hy = −εK 2m2 T m 2mkB T 2 �0 � � 2 �0 Note that we need the Gaussian averages of p2 p4 and py p2 . Hence the ﬁrst order result is � � � � py p2 5 γy T 1 0 f1 (τ y) = f1 (τ y) 1 − εK p. of the gas in steady state? • Since γt T is proportional to −γy hy . From the results of part y (b). Hence γy T must be constant. the y component of the heat transfer vector. m 2mkB T 2 T
1 (d) Use f1 . • Since the velocity τ is zero.where εK is of the order of the mean time between collisions. however. and hence ﬁnd K. T (y) = T1 + w ********� 12�
. Subject to the boundary conditions of T (0) = T1 . give a non-zero heat transfer � � � �0 p2 5 n γ y T py 2 − py p . • Since there are only variations in y. there will be no time variation if hy is a constant. K= 2 m
(e) What is the temperature proﬁle. − . is a constant in the situation under investigation. T (y). these averages are equal to 35(mkB T )3 and 5(mkB T )2 . the heat transfer vector is u �1 � mc2 n � 2 �1 = py p hy = n cy . Hence � � 2 n γy T 35 5 × 5 5 nεK kB T 2 (mkB T ) − =− γy T. to calculate hy .

d(E − T S) = −SdT − P dV.8. Hard core gas: A gas obeys the equation of state P (V − N b) = N kB T .) (a) Find the Maxwell relation involving γS/γV |T . � γT �P γT �P P
But. • The heat capacity is � γQ � γE + P γV � CP = � = γT P γT � � � � � � � = γE � + P γV � . V ) = T γT �V. and has a heat capacity CV independent of temperature. γT �V.
and from the equation of state we get � γV � � = N kB .e. show that E is a function of T (and N ) only.N
(b) By calculating dE(T.333: Statistical Mechanics I
Fall 2003
Mid-term Quiz�
1. V ) = E(T.N
Using the Maxwell relation from part (a). V ) = T dT. dE = T dS − P dV = T � � � � � γS � γS � � dT + dV − P dV.
(c) Show that θ ∞ CP /CV = 1 + N kB /CV (independent of T and V ). we ﬁnd
But from the equation of state. γT �P γT �V C P = CV + N k B . γT �V.N γT �V.N γV �T .N T =≥ � � γS � dE(T.N γT �V. CV
. V ). N. dE(T. 13 =≥ θ =1+ N kB . we get N kB T P = . γT �P P =≥
� � � � γE � γE � = = CV . γV �T . since E = E(T ) only.N
i.N � γP � P � = .N . (V − N b) =≥
� � � � γS � γP � � � dT + T − P dV. N ) does not depend on V . • For dN = 0. • Writing dS in terms of dT and dV . =≥ � � γS � γP � � � = . � γT V. E(T. (N is kept ﬁxed in the following.

P (V − N b) =≥ ln [P (V − N b)π ] = constant. • From the Gaussian form we obtain �p� p� ≡ = mkB T β�� . py . calculate the averages p2 and �(p · p)(p · p)≡. which is easily calculated by Gaussian integration as � 2� �� � � ikα � � ikp2 /2m � p 1 d3 p −3/2 = (1 − ikkB T ) . or otherwise.
2 2
and using Wick’s theorem
�(p · p)(p · p)≡ = �p� p� p� p� ≡ = (mkB T ) [β�� β�� + 2β�� β�� ] = 15 (mkB T ) . � � • The characteristic function π is the average eikα . (b) Calculate the characteristic function for the energy π = p2 /2m of a gas particle.
******** 2. exp ik − e = e = 3/2 kB T 2m (2αmkB T ) 14�
. • Using the equation of state. V ). we have � � P (V − N b) CV dE = CV dT = CV d = (P dV + (V − N b)dP ) . N kB N kB Dividing by CV P (V − N b)/(N kB ) yields d(V − N b) dP +θ = 0. N kB N kB The adiabatic condition. show that an adiabatic change satisﬁes the equation P (V − N b)π =constant.which is independent of T . can now be written as � � CV CV 0 = dQ = 1 + P d(V − N b) + (V − N b)dP. since CV is independent of temperature. dQ = dE + P dV = 0. where � and ω label any of the three components of the momentum. where p2 = p · p . Therefore: � 2� p = �p� p� ≡ = mkB T β�� = 3mkB T. pz ) in a gas at temperature T is given by � � 1 p2 . (d) By writing an expression for E(P. Energy of a gas: The probability density to ﬁnd a particle of momentum p ∞ (px . The independence of CV from V also follows from part (a). p(p) = exp − 3/2 2mkB T (2αmkB T ) � � (a) Using Wick’s theorem. or otherwise.

3 m N (m − 1)! (kB T ) . where |p| = p2 + p2 + pz is the magnitude of the momentum. • Since the energy E is the sum of N identically distributed independent variables. 2 According to the central limit theorem. i. t).) Throughout this problem treat the two body interactions and collisions precisely as in the case of classical particles considered in lectures. f1 ]. expect at the edges of the box conﬁning the gas particles. its cumulants are simply N times those for a signle variable. ‘Relativistic’ gas: Consider� gas of particles with a ‘relativistic’ one particle Hamil a 2 tonian H1 = c|p|. as given above. p(E). for N √ 1. (a) Write down the Boltzmann equation for the one-particle density f1 (p.e. • The Boltzmann equation has the general form Lf1 = C[f1 . (The x y external potential is assumed to be zero. calculate the mth cumulant of the particle energy �πm ≡c . which thus assumes the Gaussian form � � 2 1 (E − 3N kB T /2) p(E) = ⇒ . 1 2 15
. where πi is the kinetic energy of the ith particle.(c) Using the characteristic function. ln eikα = m! 2 2 m=1 m m=1
as
�πm ≡c =
3 m (m − 1)! (kB T ) . • The cumulants are obtained from the expansion
� � m � � � (ik)m m 3 3 � (kB T ) �π ≡c = − ln (1 − ikkB T ) = (ik)m . and thus given by � C[f1 . f1 ] = − d3 p2 d2 b|v2 − v1 | [f1 (p1 )f1 (p2 ) − f1 (p→ )f1 (p→ )] . or otherwise. exp − 3N kB T 3αN kB T �E m ≡c = N �πm ≡c = ******** 3. Use the central limit theorem to compute the probability density for energy. using the same collision form as employed in lectures (without derivation). 2
�N (d) The total energy of a gas of N (independent) particles is given by E = i=1 πi . in the large N limit the mean and variance are suﬃcient to describe the probability density. q. The collision term is assumed to be the same as in the classical case derived in lectures.

and aµ are functions independent of p. γt (n �λ≡) + γ� n c λ |p| |p| ⎪ where n(q. such as the meaning of |v2 − v1 |. we can obtain a hydrodynamic equation by putting λ� = p� − α� in the general conservation form. Since �λ� ≡ = 0.
For any function λ(p) which is conserved in the collisions. • Substituting λ = 1 in the conservation equation gives γt n + γ� (nu� ) = 0. n (c) Obtain the equation governing the density n(q.) The streaming terms have the form � � � � p� γH1 Lf1 (p. q. leading to
0 f1 (p. the conserved quantities are 1 (particle number). t) that sets the collision integrand in the Boltzmann equation to zero. in terms of the average local velocity u� = �cp� /|p|≡. • Since momentum is conserved in the collisions. there is a hydrodynamic equation of the form � � � � �� p� p� − n �γt λ≡ − n c γ� λ = 0.(There are various subtleties in treatment of relativistic collisions. which shall be ignored here. γ� n c λ� |p| 16�
. and c|p| (energy). q. t) − a1 (q. t) is the local density. and the 0 particle energies as given by H1 . t) = exp [−a0 (q.) • The integrand in C[f1 . (You do not need to normalize this solution. t). q. γp� |p| (b) The two body collisions conserve the number of particles. ln f1 (p1 )+ln f1 (p2 ) = ln f1 (p→ )+ln f1 (p→ ). this leads to � � �� λ� + α � + nγt α� + nu� γ� α� = 0. f1 ] is zero if at each q. t) ∞ �p� ≡. t) = γt f1 + {H1 . In our case. the momentum. and � 1 �O≡ = d3 pf1 (p.
(d) Find the hydrodynamic equation for the local momentum density α� (q. t) = d3 pf1 (p. t)λµ (p). f1 } = γt + γ� f1 = γt + c γ� f1 . in terms of the pressure tensor P�� = nc �(p� − α� ) (p� − α� ) /|p|≡. Write down the most general form f1 (p. with u� = �cp� /|p|≡ . p (momentum). where λµ (p) are quantities conserved in a two body collision. t)c|p|] . q. t)·p − a2 (q. 1 2 � This can be achieved if ln f1 = µ aµ (q. q. t)O.

t)
� � � �3 c|p| 1 c N = exp − .
In rewriting the above equation we have taken advantage of the rotational symmetry of the system. 2 e−cp/kB T c dpp 0 P�� = β�� nkB T. and there is no local velocity. q.) (e) Find the (normalized) one particle density f1 (p. in equilibrium at a temperature T . in that there is a second term in the above result that does not depend on P�� .Further simpliﬁcation and rearrangements leads to� � � �� 1 1 λ� Dt α� ∞ γt α� + u� γ� α� = − γ� P�� − γ� nα� c . the temperature T and the density n = N/V are uniform across the system. The general form obtained in part (b) now gives
0 f1 (p. • At equilibrium. t) for a gas of N such particles in a box of volume V . • For the gas at equilibrium α� = u� = 0. V kB T 8α kB T ⎪ d3 pf1 . noting that d3 p =
The normalization factor is obtained by requiring N = V ⎪� 4αp2 dp. and the pressure tensor is given by P�� = nc = ncβ�� nc = β�� 3 � p·p |p| � . the problem does not lead to a clean answer. � p� p� |p| � � px px |p| �
(f) Evaluate the pressure tensor P�� for the above gas in equilibrium at temperature T . ********
leading to
17�
. which is the usual formula for an ideal gas. as currently formulated. |p| n n (Unfortunately. The expectation value is simply ⎪� dpp2 pe−cp/kB T kB T 0 =3 �|p|≡ = ⎪ � . and using 0 dppn e−p/a = n!an+1 . q.

x→ ) dx γx→
However. dA/dx = 0. a displacement x requires a force J = ax − bT + cT x. leads to � � � � γS � γS � γS � γx � � = � + � � . =− γx �T γT �x (b) Show that A has to be independent of x. 0) + AT + bx − x2 . where a.8. now explicitly depends on x. J )). Furthermore. x) = S(0. T ) can be calculated as γS(T → . as a function of T and J . 0) + bx − cx2 /2. S(x. • By integrating the derivatives of S given above. and c are constants. b. Wire: Experiments on stretching an elastic wire indicate that. 2
0 0 T � =T
�
x� =x x� =0
γS(T. Thus γA γ γS γ γS = = =0 γx γx γT γT γx from part (a). where S = S(T. its heat capacity at constant displacement is proportional to temperature. x).e. • Writing the entropy as S(T. 0) + dT + γT → T � =0 � T � x → = S(0. (a) Use an appropriate Maxwell relation to calculate γS/γx|T . at a temperature T . (d) Calculate the heat capacity at constant tension. γT �J γT �x γx �T γT �J 18�
. x). S(T = 0.333: Statistical Mechanics I
Fall 2004
Mid-term Quiz�
1. i.e. (c) Give the expression for S(T.e. and comment on whether it is compatible with the third law of thermodynamics. x(T. CJ = T γS/γT |J . x) = S(T. in violation of the third law of thermodynamics. Cx = A(x)T . implying that A is independent of x. � εS • We have Cx = T εT �x = A(x)T . i. x = 0) → S(x. 0) + AdT + (b − cx→ )dx→ � c = S(0. i. T ) = S(0. we obtain � � � γS � γJ � � = b − cx. • From dF = −SdT + J dx.

εS �T = b − cx and εT � x = A.� � εS From parts (a) and (b). and p(Mij ) = 0 otherwise. Let us assume that each element Mij (for i � j) is an independent random variable taken from the probability density function p(Mij ) = 1 2a for − a < Mij < a . a). and thus pT (k) = ˜
N �
pii (k) = ˜
i=1
�
sin ak ak
�N
.
(a) Calculate the characteristic function for each element Mij . (a + cT )3 ********
2. γT a + cT Thus � � (b − cx)2 . εx εx εx a εT − b + cx + cT εT = 0.e. we can rewrite the heat capacity as a function of T and J . as CJ = T A + � (b − c J +bT )2 a+cT �
(a + cT ) � � (ab − cJ )2 =T A+ . the characteristic function is � a eika − e−ika sin ak 1 = = . • The trace of the matrix is the sum of the N diagonal elements which are independent random variables. Wigner considered N × N symmetric matrices whose elements are random. T ∞ tr M = i Mii . Furthermore. • Since each element is uniformly distributed in the interval (−a. CJ = T A + (a + cT )
J +bT a+cT
εx � εT J
�
is given by
Since x =
. i. γx b − cx = . Random matrices: As a model for energy levels of complex nuclei. The characteristic function for the sum of independent variables is simply the product of the corresponding characteristic functions. pij (k) = ˜ dx e−ikx 2a 2aik ak −a � (b) Calculate the characteristic function for the trace of the matrix.
(c) What does the central limit theorem imply about the probability density function of the trace at large N ? 19
.

each eigenvalue �� (� = 1. and p(�) = 0 otherwise. to get � � � � 2� �2 λ/2 �2 λ/2 �2 2�2 λ/2 0 0 0 2 2 2 � c= dχ cos χ sin χ = dχ sin 2χ = dχ (1 − cos 4χ) = 0 . and hence its variance is given by � � �0 � 2� �2 2 1 − 2. N ) of the matrix M is distributed according to a probability density function � 2 �2 p(�) = 1 − 2 for − �0 < � < �0 .• Since the trace is the sum of N √ 1 independent random variables. N �� 2αa2 2a N (d) For large N . we have �2 = 4N a2 /3. its cumulants are simply N times those of a single element.) • The mean value of � is zero by symmetry.
�=1
=≥
�
T
2
�
c
=
�=1
N ��
� � �2 + ��� �� ≡ . The leading cumulants are �T ≡c = N �Mij ≡ = 0. · · · . 2. � c= d� �2 α�0 �0 −�0
⇒
�
c
=N
�
2 Mij
�
=N
�
a
dx
−a
x2 a2 =N . higher order cumulants vanish in the limit of N � �. (Hint: Changing variables to � = �0 sin χ simpliﬁes the integrals. Find the variance of �. �2 = N � 3 3 20
. and � T
2
For the qunatity t = T / N . and thus � � � �� 3 tr M 3t2 lim p t = ⇒ = exp − 2 . α −λ/2 2α −λ/2 4α −λ/2 4 (e) If in the previous result. α�0 �0 (known as the Wigner semi-circle rule). change variables to � = �0 sin χ and d� = �0 cos χdχ. can the eigenvalues be independent of 0 each other? • The trace of a matrix is related to its eigenvalues by T =
N �
�� . 2a 3
In the integral. �
√ �=�
The cross-correlations of eigenvalues thus satisfy � ��� �� ≡ = T �
2
�=� √
�
− c
�=1
N ��
� 4N a2 a2 −N × �= 0.

q exp − (px − m�y) + py + pz .)
(a) The above approximation does not satisfy the Boltzmann equation as the collision term 0 (right hand side of the equation) vanishes. 3/2 2mkB T (2αmkB T ) obtained from the uniform Maxwell–Boltzmann distribution by substituting the average value of the gas velocity at each point. γt m γτ q mkB T whence � ⎫ � 1 0 f1 = f 1 1 − ε × py (px − m�y) . while (the left hand side) df1 /dt = 0. • The pressure tensor is P�� (y) = nm �c� c� ≡ = n �p� p� ≡ /m.e. A zeroth order approximation to the one particle density is. L f1 ∝ γt m γτ q ε× where ε× is a characteristic mean time between collisions. � � 0 � 1⎬ τ f 1 − f1 γ p γ 0 · + f1 ∝ − 1 . One plate at y = 0 is stationary. the oﬀ-diagonal element is calculated as � px py 1 Pxy (y) = d3 p f (y) m 1 � � � ε× � 0 3 px py − py (px − m�y)f1 = d p m mkB T � �⎨ � � �⎨ � p2 (px −m�y)2 ⎧� ⎧ y � ⎩� exp − 2mkB T ⎩ exp − 2mkB T −ε× �n � 2 2 ⇒ = 2 dpx (px − m�y) dpy py ⇒ �⎭ ⎧ ⎧ m kB T ⎭ 2αmkB T 2αmkB T � =− ε× �n (mkB T )2 = −�nε× kB T. the well known result that eigenvalues do not cross implies a repulsion between eigenvalues which leads to a much wider distribution than would result from independent eigenvalues. i. m2 k B T 21
.Clearly. ******** 3. • We have � � � γ p γ τ 0 0 · f1 = + py (px − m�y)f1 . In fact. Viscosity: Consider a classical gas between two plates separated by a distance w. From the ﬁrst order density. this is inconsistent with independent eigenvalues. while n and T are constants. f1 (τ ). � � � � n 1 0 2 2 2 f1 (τ τ ) = p. (� = u/w is the velocity gradient. while the other at y = w moves with a constant velocity vx = u. mkB T
(b) Calculate the oﬀ-diagonal component Pxy (y) of the pressure tensor. by considering the linearized Boltzmann equation in the p single collision time approximation. Find � 1 a better approximation.

and the force exerted on the top plate (or the bottom plate) is thus Fx = −Pxy = �nε× kB T. • The pressure tensor calculated in part (b) is in fact independent of the position y. deﬁned by σ = Fx /�. The coeﬃcient of viscosity is then simply σ= Fx = nε× kB T.(c) The gas exerts a transverse force per unit area Fx = −Pxy (y = w) on the moving plate. Calculate this force. � ********
22
. and hence obtain the coeﬃcient of viscosity.