I have difficulty understanding the exact concept of group velocity. Consider a wave packet as a linear combination of a number of eigenstates of a 1-D particle in box. The dispersion curve([itex]\omega[/itex]versus k) is composed of discrete points located on a parabola. Well, for each point one can calculate the phase velocity but how to calculate the group velocity? Firstly the points are discrete and it is not possible to differentiate and secondly the group velocity changes from a point to another so that we can not specify a special group velocity for that wave pocket!