If I look at a gaussian wave package, and then interpret (in the usual Quantum-Mechanics Way) its square value as the propability density, then I can calculate a mean value for the position: $$ x_{mean}(t) = \int \Psi^*(x,t) x \Psi(x,t) dx $$

Is the time derivative of this mean position the same as the group velocity of the wave package, that means, is $ \dot{x}_{mean} = \frac{d \omega}{ d k}$?

I tried to derive it (because this is what I imagine the group velocity to be), but I failed.

  • $\begingroup$ Possible duplicates: physics.stackexchange.com/q/143696, physics.stackexchange.com/q/52319, physics.stackexchange.com/q/73328 $\endgroup$ – TotallyRhombus Jul 1 '16 at 14:02
  • $\begingroup$ None of these questions have the same content as mine, and none of these answers answer my question. To me it seams that the equivalence I want to show is only an approximation. Can you verify that? $\endgroup$ – Quantumwhisp Jul 1 '16 at 16:41
  • $\begingroup$ It is true that $\dot x_{mean}=\langle \frac{d\omega}{dk}\rangle$, and usually $\dot x_{mean}=\frac{d\omega}{dk}\big|_{k_0}$ for some value of $k_0$ in the support of the original Fourier transform $\hat\psi_0(k)$. However, group velocity itself is defined as the limiting average velocity as the wave packet width approaches zero, so $v_{group}(k_0)=\frac{d\omega}{dk}|_{k_0}$ holds exactly. Does this help? $\endgroup$ – TotallyRhombus Jul 1 '16 at 16:55


Group velocity is a property of a linear superposition of plane wave solutions of the form $e^{iq\cdot x-i\omega t}$, with dispersion relation $\omega(q)$ that relates frequency to wave number. From this, you can assume $\psi(x,t)=\int\frac{d^3q}{(2\pi)^3}g(q)e^{iq\cdot x-i\omega(q)t}$ for some initial (normalized) wave function $g(q)$. For your question, you can assume that $g(q)$ is a Gaussian sharply peaked near some frequency $q_0$.

Also, $xe^{iq\cdot x}=-i\frac{d}{dq}e^{iq\cdot x}$.

ANSWER \begin{align*} \langle x \rangle(t) &= \int d^3x' \bigg[x'\int \frac{d^3qd^3q'}{(2\pi)^6}g(q)g^*(q')e^{i(q-q')\cdot x'-i(\omega(q)-\omega(q'))t} \bigg]\\ &=\int d^3x'\bigg[\int\frac{d^3 qd^3 q'}{(2\pi)^6}g(q)g^*(q')(-i\nabla_q e^{i(q-q')\cdot x'})e^{-i(\omega(q)-\omega(q'))t}\bigg]\\ &=\int d^3x'\bigg[\int\frac{d^3qd^3q'}{(2\pi)^6}\Big(i\nabla_qg(q)g^*(q')+\nabla_q\omega(q)tg(q)g^*(q')\Big)e^{i(q-q')\cdot x-i(\omega(q)-\omega(q'))t}\bigg]\\ &\text{(from integration by parts)}\\ &= \int\frac{d^3qd^3q'}{(2\pi)^6}(2\pi)^3\delta^3(q-q')\Big(i\nabla_q g(q)g^*(q')+\nabla_q\omega(q)tg(q)g^*(-q)\Big)e^{-i(\omega(q)-\omega(q'))t}\\ &\text{(using the momentum-space delta function with Phys. convention)}\\ &=\int \frac{d^3 q}{(2\pi)^3}\Big(i\nabla_q g(q)g^*(q)+\nabla_q\omega(q)|g(q)|^2t\Big)\\ &\approx\nabla_q\omega(q_0)t + \mathcal O(1=t^0) \\ &\text{(assuming $g(q)$ is sharply peaked about $q_0$)} \end{align*} Hence, $\lim_{t\rightarrow\infty}\frac{\langle x\rangle(t)}{t}=\nabla_q\omega(q_0)$

  • $\begingroup$ In this case the group velocity is just approximately the time derivative of the mean position. What is the exact value of the group velocity? It's strange to me that a physical quantitiy so often used has no "exact meaning". $\endgroup$ – Quantumwhisp Jul 1 '16 at 12:06
  • $\begingroup$ The group velocity is the unique limiting average velocity as the distribution $g(k)$ approaches a delta measure concentrated at $k_0$. $\endgroup$ – TotallyRhombus Jul 1 '16 at 13:48
  • $\begingroup$ Do you have any recomendations on further more to read about this? $\endgroup$ – Quantumwhisp Jul 1 '16 at 13:50
  • $\begingroup$ The wikipedia page on group velocity is pretty informative: en.wikipedia.org/wiki/Group_velocity $\endgroup$ – TotallyRhombus Jul 1 '16 at 13:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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