• Home
  • Textbooks
  • A Kinetic View of Statistical Physics
  • Collisions

A Kinetic View of Statistical Physics

Pavel L. Krapivsky, Sidney Redner, Eli Ben-Naim

Chapter 3

Collisions - all with Video Answers

Educators


Chapter Questions

01:02

Problem 1

The Boltzmann equation can be applied when a gas is dilute, namely when $n d^{3} \ll 1$, where $n$ is the density and $d$ is the range of the intermolecular force. Under standard conditions (room temperature and atmospheric pressure), estimate the density of the atmosphere and also the range of intermolecular forces. Show that under these conditions the atmosphere is indeed dilute.

Narayan Hari
Narayan Hari
Numerade Educator
09:35

Problem 2

Justify that quantum effects are negligible for the atmosphere under standard conditions. In particular, show that the de Broglie wavelength is much smaller than the average distance between adjacent particles:
$$
\frac{\hbar}{\sqrt{2 m T}} \ll n^{-1 / 3}
$$
For a gas at atmospheric pressure, estimate the temperature at which quantum effects become important. For a gas at room temperature, estimate the density at which quantum effects become important.

Sarah Mccrumb
Sarah Mccrumb
Numerade Educator
03:59

Problem 3

Consider the scattering of a uniform parallel beam of point particles that moves in the $z$ direction on an immobile spherical obstacle in $d$ dimensions. Show that the ratio of the total backward scattering flux $\Phi_{\mathrm{b}}$ to forward scattering flux $\Phi_{\mathrm{f}}$ is
$$
\frac{\Phi_{\mathrm{b}}}{\Phi_{\mathrm{f}}}=\frac{(1 / \sqrt{2})^{d-1}}{1-(1 / \sqrt{2})^{d-1}} \equiv \lambda_{d}
$$
Here forward flux refers to particles that have a positive $z$ component (and vice versa for the backward flux). In particular $\lambda_{1}=\infty$ (only backward scattering in one dimension), $\lambda_{2}=(\sqrt{2}-1)^{-1}>1$ (predominantly backward scattering in two dimensions), and $\lambda_{3}=1$ (a manifestation of isotropic scattering in three dimensions).

Supratim Pal
Supratim Pal
Numerade Educator
05:57

Problem 4

This problem is concerned with the derivation of the Einstein-Green-Kubo formula.
(a) Show that the diffusion equation implies that the mean-square displacement $\Delta(t)=\left\langle[\mathbf{r}(t)-\mathbf{r}(0)]^{2}\right\rangle$ grows as $\Delta \simeq 2 d D t$
(b) Derive from the above result
$$
\frac{d \Delta}{d t}=2\langle\mathbf{v}(t) \cdot[\mathbf{r}(t)-\mathbf{r}(0)]\rangle=2 \int_{0}^{t} d t^{\prime}\left\langle\mathbf{v}(t) \cdot \mathbf{v}\left(t^{\prime}\right)\right\rangle
$$
(c) Argue that the equilibrium average depends only on the difference of the time arguments $t^{\prime \prime}=t-t^{\prime}$ and therefore
$$
\frac{d \Delta}{d t}=2 \int_{0}^{t} d t^{\prime \prime}\left\langle\mathbf{v}(0) \cdot \mathbf{v}\left(t^{\prime \prime}\right)\right\rangle
$$
(d) Using $\Delta \simeq 2 d D t$ and taking the $t \rightarrow \infty$ limit show that the previous equation yields
$$
D=\frac{1}{d} \int_{0}^{\infty} d t(\mathbf{v}(0) \cdot \mathbf{v}(t)\rangle
$$.

Yiyang Wang
Yiyang Wang
Numerade Educator
01:32

Problem 5

Any isotropic velocity distribution is an equilibrium distribution for the Lorentz model. If, however, there is a weak interaction between moving particles, they will eventually equilibrate at some temperature $T$ so that the velocity distribution is Maxwellian. Show that in this situation the average diffusion coefficient (3.9) becomes
$$
D=\frac{4 \ell}{3} \sqrt{\frac{T}{2 \pi m}}=\frac{16 a}{9} \sqrt{\frac{T}{2 \pi m}} \frac{1}{v}
$$
where $v=n\left(4 \pi a^{3} / 3\right) \ll 1$ is the volume fraction occupied by the scatterers.

Victor Salazar
Victor Salazar
Numerade Educator
02:37

Problem 6

Verify that Eq. (3.34) is the solution to the telegraph equation.

Amit Srivastava
Amit Srivastava
Numerade Educator
01:16

Problem 7

Using (3.47a) compute the root mean-square velocity of a Lorentz gas particle in an electric field. Show that this velocity coincides with the estimate $\left(\mathcal{E}^{2} \ell t\right)^{1 / 3}$ up to a numerical factor $3^{2 / 3}[\Gamma(1 / 3)]^{-1 / 2}$.

Sean Dougherty
Sean Dougherty
Numerade Educator
03:44

Problem 8

Using the Green-Kubo formula as in the case of zero magnetic field calculate the non-trivial components of the diffusion tensor in Eq. (3.52) and show that they are given by
$$
D_{1}=\frac{1}{3} v_{0}^{2} \tau, \quad D_{\perp}=\frac{1}{3} v_{0}^{2} \tau \frac{1}{1+(\omega \tau)^{2}}, \quad D_{x y}=-D_{y x}=-\frac{1}{3} v_{0}^{2} \tau \frac{\omega \tau}{1+(\omega \tau)^{2}}
$$.

Laszlo Zalavari
Laszlo Zalavari
Numerade Educator
09:10

Problem 9

Use a kinetic theory argument, similar to that developed for the number of moving particles, Eq. (3.55), to determine the number of collisions $C(t)$ up to time $t$ in the collision cascade discussed in Section 3.4.

Naresh Bagrecha
Naresh Bagrecha
Numerade Educator
25:40

Problem 10

Consider a zero-temperature gas that occupies the half-space $x \geq 0$. When a particle that moves along the $x$ axis hits this system, it generates a collisional cascade that propagates into the medium. Particles are also ejected backward to create a splashlike pattern (Fig. 3.3). The goal of this problem is to argue that the energy $E(t)$ of the particles in the initially occupied half-space ultimately approaches zero algebraically in time.
(a) Using dimensional analysis show that the size $R$ of the cascade is given by Eq. (3.56), although we should now think about the energy $E(t)$ as decreasing with time.
(b) Show that the total number of particles in the cascade scales as
$$
N \sim n R^{d} \sim(E / m)^{d /(d+2)}\left(n t^{d}\right)^{2 /(d+2)}
$$
(c) Show that the typical velocity of the ejected particles $v$ is related to $N$ and $E$ via $N m v^{2} \sim E$, from which
$$
v \sim\left(\frac{E}{n m t^{d}}\right)^{1 /(d+2)}
$$
(d) Argue that the energy $E(t)$ decreases according to
$$
\frac{d E}{d t} \sim-\frac{m v^{2}}{2} \times n v R^{d-1}
$$
(e) Using results from (a)-(c) show that (d) reduces to $d E / d t \sim-E / t$, which leads to an algebraic decay of $E$ with time.

Zachary Warner
Zachary Warner
Numerade Educator
02:10

Problem 11

This problem concerns heat conduction in a medium where the heat conduction coefficient depends algebraically on temperature: $\kappa=\kappa(T)=A T^{n}$. (For Maxwell molecules, hard spheres, and very hard particles, the exponent is $n=$ $1,1 / 2,0$, respectively, following the argument that led to (3.57) for the diffusion
coefficient.) The heat conduction equation
$$
\frac{\partial T}{\partial t}=A \nabla \cdot\left(T^{n} \nabla T\right)
$$
is mathematically a parabolic partial differential equation. When $n=0$, there is a general solution for an arbitrary initial condition. When $n \neq 0$, the resulting nonlinear parabolic equation is generally insoluble, but exact solutions are known for sufficiently simple initial conditions. Consider one such situation where heat is instantaneously released on the (hyper)plane $x=0$ while outside the initial temperature is zero. Then we need to solve the initial value problem
$$
\frac{\partial T}{\partial t}=A \frac{\partial}{\partial x}\left(T^{n} \frac{\partial T}{\partial x}\right), \quad T(x, 0)=Q \delta(x)
$$.
The goal is to solve this initial value problem when $n \geq 0$.
(a) On general grounds, $T=T(x, t \mid A, Q)$. Using dimensional analysis show that any three out of four quantities $x, t, A, Q$ have independent dimensions.
(b) Choosing $t, A, Q$ as the basic quantities, show that the solution must have the form
$$
T(x, t)=\left(\frac{Q^{2}}{A t}\right)^{1 /(2+n)} f(\xi), \quad \xi=\frac{x}{\left(Q^{n} A t\right)^{1 /(2+n)}}
$$
Thus dimensional analysis alone shows that the solution (which in principle depends on two variables $x$ and $t$ ) is, up to the scaling factor, a function of a single scaling variable $\xi$. The primary reason for this simplification is the absence of a quantity with dimensions of length in the formulation of the initial value problem.
(c) Substitute the above ansatz for $T(x, t)$ into the nonlinear heat equation and show that the scaling function satisfies the ordinary differential equation
$$
\left(f^{n} f^{\prime}\right)^{\prime}+\frac{1}{2+n}(\xi f)^{\prime}=0
$$
where the prime denotes differentiation with respect to $\xi$.
(d) Show that the solution to the above differential equation is
$$
f(\xi)=\left[\frac{n}{2(2+n)}\left(\xi_{0}^{2}-\xi^{2}\right)\right]^{1 / n}
$$
in the region $|\xi| \leq \xi_{0}$ and $f(\xi)=0$ for $|\xi| \geq \xi_{0}$. Argue that $\xi_{0}$ is determined by heat conservation.
$$
Q=\int T(x, t) d x=Q \int_{-\xi_{0}}^{\xi_{0}} f(\xi) d \xi
$$
and compute $\xi_{0}$ -
(e) The heated region is $\left[-x_{0}(t), x_{0}(t)\right]$ whose boundaries grow as $x_{0}(t)=$ $\xi_{0}\left(Q^{n} A t\right)^{1 /(2+n)}$; outside the heated region the medium remains at zero temperature. Try to explain the contradiction with the common wisdom that, for a parabolic partial differential equation, perturbations propagate instantaneously.

Surendra Kumar
Surendra Kumar
Numerade Educator
01:14

Problem 12

Consider the previous problem in three dimensions and assume that a finite amount of heat has been instantaneously released at the origin: $T(\mathbf{r}, 0)=Q \delta(\mathbf{r})$
(a) Using the same arguments as in the previous problem show that the solution must have the form
$$
T(\mathbf{r}, t)=\left(\frac{Q^{2}}{A^{3} t^{3}}\right)^{1 /(2+3 n)} f(\xi), \quad \xi=\frac{r}{\left(Q^{n} A t\right)^{1 /(2+3 n)}}
$$
(b) Derive an ordinary differential equation for $f(\xi)$ and solve it to give
$$
f(\xi)=\left[\frac{n}{2(2+3 n)}\left(\xi_{0}^{2}-\xi^{2}\right)\right]^{1 / n}
$$
(c) The heated region is a ball of radius $r_{0}(t)=\xi_{0}\left(Q^{n} A t\right)^{1 /(2+3 n)}$. Thus the volume of the heated region grows as $t^{3 / 5}$ and $t^{6 / 7}$ in the case of Maxwell molecules and hard spheres, respectively; for very hard spheres, the entire space $\mathbb{R}^{3}$ becomes heated at any $t>0$.

Manik Pulyani
Manik Pulyani
Numerade Educator
01:19

Problem 13

Consider heat conduction in the three-dimensional region $R \leq r<\infty .$ Assume that the inner spherical boundary is maintained at a constant temperature, $T(R, t)=$ $T_{0}$, and that the temperature vanishes at infinity, $T(\infty, t)=0 .$ Show that, if $\kappa(T)=A T^{n}$, the steady-state temperature satisfies $\nabla^{2}\left(T^{n+1}\right)=0$, and that the corresponding temperature profile is $T(r)=T_{0}(R / r)^{1 /(n+1)}$.

Manik Pulyani
Manik Pulyani
Numerade Educator
01:07

Problem 14

Consider ballistic annihilation in one dimension. Transform the governing integrodifferential Boltzmann equation (3.59) into
$$
\frac{\partial}{\partial t} \frac{\partial^{2}}{\partial v^{2}} \ln P(v, t)=-2 P(v, t)
$$
This (apparently very difficult) nonlinear partial differential equation has not been solved so far.

Hast Aggarwal
Hast Aggarwal
Numerade Educator
01:48

Problem 15

Explain the relation $\alpha+\beta=1$ between the decay exponents that appear in (3.60).

Mike Gaerlan
Mike Gaerlan
Numerade Educator
02:04

Problem 16

Consider ballistic annihilation for very hard particles. Confirm the expressions (3.68) for the density and the mean speed and show that the decay exponent $\beta$ is indeed given by (3.69).

Narayan Hari
Narayan Hari
Numerade Educator
00:07

Problem 17

Verify the general behavior of the moments (quoted in Eq. (3.80)) for the onedimensional inelastic Maxwell gas by exploiting the inequality $a_{n}<a_{m}+a_{n-m}$ for all $1<m<n-1$.

John Johnson
John Johnson
Numerade Educator
00:07

Problem 18

Verify the general behavior of the moments (quoted in Eq. (3.80)) for the onedimensional inelastic Maxwell gas by exploiting the inequality $a_{n}<a_{m}+a_{n-m}$ for all $1<m<n-1$.

John Johnson
John Johnson
Numerade Educator
02:52

Problem 19

Consider the following toy ballistically controlled aggregation model in which the masses of all particles always remain identical. That is, when two particles with velocities $\mathbf{v}_{1}$ and $\mathbf{v}_{2}$ collide, they form an aggregate with velocity $\mathbf{v}=\mathbf{v}_{1}+\mathbf{v}_{2}$.
(a) Show that for the toy model the volume fraction occupied by particles decays indefinitely.
(b) Does the Boltzmann framework provide an asymptotically exact description? From (a) we see that the system becomes dilute, so that multiple collisions become asymptotically irrelevant. $^{28}$
(c) Using a (simplified) version of the mean-field argument applied to establish the conjectural behavior for ballistic agglomeration, Eqs (3.92) -(3.93), show that for the toy model the analog of (3.94) reads
$$
v \sim t, \quad n \sim t^{-2}
$$
independent of the spatial dimension $d$.
(d) Show that the Boltzmann equation describing the toy model is
$$
\begin{aligned}
\frac{\partial P(\mathbf{v}, t)}{\partial t}=& \int d \mathbf{u} d \mathbf{w} P(\mathbf{u}, t) P(\mathbf{w}, t)|\mathbf{u}-\mathbf{w}| \delta(\mathbf{u}+\mathbf{w}-\mathbf{v}) \\
&-2 P(\mathbf{v}, t) \int d \mathbf{w} P(\mathbf{w}, t)|\mathbf{v}-\mathbf{w}|
\end{aligned}
$$
(Numerical solution of this BE shows that the density decays algebraically, $n \sim t^{-\alpha}$, with exponent increasing with dimension: $\alpha \approx 1.33,1.55,1.65$ when $d=1,2,3 .)$
(e) Simulate the BE in one dimension and show that the exponent is indeed $\alpha \approx 4 / 3$, very different from $\alpha=2$ implied by the mean-field argument.
(f) Consider the toy model in the reaction-controlled limit. Show that $n \sim$ $t^{-2 d /(d+1)}$. The mean-field argument does not distinguish the ballistically controlled limit from the reaction-controlled one, so its prediction $n \sim t^{-2}$ is wrong in any dimension.

Chai Santi
Chai Santi
Numerade Educator
View

Problem 20

Assume that a cluster consists of $m$ cars with velocities independently drawn from the distribution $P_{0}(v)$. Using the asymptotic form $P_{0}(v) \sim v^{\mu}$ as $v \rightarrow 0$, show that $m \sim v^{-\mu-1}$.

Victor Salazar
Victor Salazar
Numerade Educator
03:24

Problem 21

Evaluate the cluster size distribution $P(v, t)$ for the special case $P_{0}(v)=e^{-v}$. Show that the result is consistent with the general scaling behavior,

Adriano Chikande
Adriano Chikande
Numerade Educator
04:08

Problem 22

Consider the simplest heterogeneous initial condition: a half-line $x<0$ is occupied by cars whose positions and velocities are uncorrelated, while the other half-line $x>0$ is initially empty. Show that the total number of clusters that infiltrate the initially empty half-line grows as $\alpha \ln t$. Here $\alpha=(\mu+1) /(\mu+2)$ is the exponent that characterizes the number of cars in the typical cluster (in the bulk) $m \sim t^{\alpha}$.

Raymond Matshanda
Raymond Matshanda
Numerade Educator