A particle suffers elastic collisions with scattering centres with a probability of collision per unit time $\lambda$. After a collision the particle is in a direction characterized by a solid angle $\mathrm d\Omega$ with probability $\omega(\theta)\,\mathrm d\Omega\,,$ that depends only on the angle between the intcial direction $\vec{{p}'}$ and the final direction $\vec{p}$. Assume only elastic collisions, $p={p}'$

a) Obtain the following equation of motion for the density of probability $f(\vec{p},t)$:

$$\frac{\partial f(\vec{p},t)}{\partial t}=-\lambda\cdot f(\vec{p},t)+\lambda \cdot \int\,\mathrm d{\Omega}' \cdot \omega(\theta)f(\vec{{p}'},t)$$

where the integration is over the solid angle of $\vec{{p}'}$ ($\mathrm d{\Omega}'=\sin{\theta}' \,\mathrm d{\theta}'\,\mathrm d{\phi}')$

b)Show that the equation of movement of the average momentum is:

$$\frac{\partial \langle\vec{p}\rangle}{\partial t}=-\frac{\langle\vec{p}\rangle}{\tau_\rm{tr}}$$

where $\tau_\rm{tr}\,,$ the transport time is defined by:

$$\frac{1}{\tau_\textrm{tr}}=\lambda \int \mathrm d\Omega (1-\cos \;\theta) \,\omega(\theta)$$

Attempt at a solution

a) I can arrive at the given expression, so no problems here

b) I start by writing:

$$\langle\vec{p}\rangle =\int \mathrm d^3p \; f(\vec{p},t)\cdot \vec{p}$$

And I derive this expressions with respect to $t,$ getting from a) that:

$$\frac{\partial \langle\vec{p}\rangle}{\partial t}=-\lambda\cdot \langle\vec{p}\rangle +\lambda \cdot \int d^3 p \; \vec{p} \int \,\mathrm d{\Omega}' \cdot \omega(\theta) f(\vec{{p}'},t)$$

And from here I don't know what to do, is there something that I'm missing or does it need some kind of trick?


2 Answers 2


One can perform the $\textbf{p}$ integral before the $\Omega$ integral in the last equation you wrote: $$ \int d{\Omega} \int d^3 p \; \vec{p}\, \omega(\theta) f(\vec{{p}'},t), $$ where I removed primes in angular variables because they are integrated over anyway.

Here, $\textbf{p} = R(\theta,\phi) \textbf{p}^{\prime}$, with some appropriate rotation matrix $R(\theta,\phi)$, and with this, you can express $\textbf{p}$ and $d^{3}p$ in terms of $\textbf{p}^{\prime}$ and $d^{3}p^{\prime}$ . Then, the rest will follow.

I think this may be a sufficient hint, but I'll provide more details if you want.

  • $\begingroup$ Don't understand why I didn't realize that this question is almost a year old. $\endgroup$
    – higgsss
    Commented Jan 31, 2014 at 13:12

Sorry, I don't have the answer for now either but I am at least confused by the notations. Is $\theta$ in expression a) the same as the one in the expression of $\tau_{tr}$?

Is $\omega(\theta)$ in fact $\omega(\theta|\theta')=\omega(|\theta-\theta'|)$?

Also, to be consistent, I think that is average $\langle \vec{p}\rangle$ can be written as

$\langle \vec{p} \rangle = \int d \Omega \:f(\vec{p},t)\vec{p}$

since your probability space is simply that of orientations for your vectors.

  • $\begingroup$ It's okay, just having a discussion about the problem should be pretty helpful. Thanks! Regarding your remarks: yeah, i agree about the writing of $<\vec{p}>$; The $\theta$ in $\omega(\theta)$ was also giving me some confusion but I'm pretty sure it is $\omega(\left |\theta-{\theta}' \right|)$ as you claim (it is what makes the most sense). $\endgroup$
    – benfstokes
    Commented Apr 4, 2013 at 13:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.