Having a drifting Maxwellian (Maxwell-Boltzmann) distribution: $$f(\vec v) = n \left(\dfrac{m}{2\pi k_B T}\right)^\frac{3}{2}\exp\left({-\dfrac{m[(v_x-a)^2+v_y^2 + v_z^2]}{2k_BT}}\right) $$ where $a$ is the drift velocity, is it possible to derive this drift from the moment of the distribution $$ \langle \vec v \rangle = 1/n \int_{- \infty}^\infty \vec v f(\vec v) d^3v $$

I assume that both $\langle v_y \rangle=0$ and $\langle v_z \rangle=0$ (from the Maxwellian being an even function in these directions), but does $\langle v_x \rangle=a$? If so, please help me derive it, I couldn't do it myself neither could I find the anwer in textbooks.

  • $\begingroup$ Yes, the even integrals (y and z directions) will be zero... $\endgroup$ Commented Aug 9, 2023 at 20:36

1 Answer 1


According to your formulas, $$\langle v_x \rangle = \left( \frac{m}{2\pi k_B T} \right)^{3/2} \int_{-\infty}^\infty v_x e^{-\frac{m[(v_x - a)^2 + v_y^2 +v_z^2]}{2k_B T}} d v_x d v_y dv_z $$

Shifting $v_x$ by defining $v'_x = v_x - a$ we have

$$\langle v_x \rangle = \left(\frac{m}{2\pi k_B T}\right)^{3/2} \int_{-\infty}^\infty (v'_x +a) e^{-\frac{m[v_x'^2 + v_y^2 +v_z^2]}{2k_B T}} d v'_x d v_y dv_z = a \left( \left(\frac{m}{2\pi k_B T}\right)^{3/2} \int_{-\infty}^\infty e^{-\frac{m[v_x'^2 + v_y^2 +v_z^2]}{2k_B T}} d v'_x d v_y dv_z\right) = a$$ I used that the $v'_x e^{-(...)}$ integral vanishes because it is odd in $v'_x$, and the remaining $a e^{-(...)}$ integral simplifies with the prefactor to give simply $a$ as expected.

This result just follows from identifying this distribution with a gaussian distribution of mean $(a,0,0)$ and standard deviation $\sigma = \sqrt{\frac{k_B T}{m}}$


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.