Solid state physics: When do I use classical laws? Let's say I am given the dispersion relation for nearly-free electrons: 
$$ E(k) = \frac{\hbar^2}{2m}(k^2+c\,k^4)$$
Where $c$ is a small constant of appropiate dimension.
How do I calculate the velocity of an electron given a fixed $k_1$ ?
Applying "classical" laws results in 
$$v(k_1)=\sqrt{\frac{2E(k_1)}{m}} = \frac{\hbar k_1}{m}\sqrt{1+c k_1^2}$$
On the other hand, applying $$v(k_1) = \frac{\partial \omega(k)}{\partial k}\bigg|_{k=k_1} = \hbar^{-1} \frac{\partial E(k)}{\partial k}\bigg|_{k=k_1} = \frac{\hbar}{2m}(2k_1+4ck_1^3) = \frac{\hbar k_1}{m}(1+2c k_1^2)$$ 
Obviously both terms are not the same, so can anyone explain to me where is the difference ? I guess it has something to do with mixing up the velocity-concept of classical particles (1) and the group-velocity of electrons as waves (2).
 A: For wave mechanics there is the phase velocity and group velocity. For the energy $E~=~\hbar\omega$ the phase velocity is
$$
v_p~=~\frac{\omega}{k}~=~\frac{\hbar}{2m}(k~+~ck^3).
$$
This is the velocity of a wave front, or where the phase of the wave is constant. There is also the group velocity that is
$$
v_g~=~\frac{\partial\omega}{\partial k}~=~\frac{\hbar}{m}(k~+~2ck^3).
$$
The classical idea suggested in this question is that $k^2~+~ck^4~=~k'^2$ so that 
$$
p'~=~\hbar k'~=~\hbar k\sqrt{1~+~ck^2}
$$
This is a different definition of the momentum and thus velocity. I would say that a better approach is to write the Hamiltonian or energy according to $p~=~\hbar k$
$$
H~=~\frac{1}{2m}\left(p^2~+~\frac{c}{\hbar^2}p^4\right).
$$
This Hamiltonian is an operator for $p~\rightarrow~-i\partial_x$. Now put the wave function $\psi(x,t)~=~Ae^{-ikx~+~i\omega t}$ in the Schrodinger equation to arrive at
$$
H\psi~=~i\hbar\frac{\partial\psi}{\partial t}
$$
to find
$$
\hbar\omega~=~\frac{\hbar^2}{2m}\left(k^2~+~ck^4\right),
$$
which agrees with the phase velocity above.
If you were to insist on doing a sort of classical form with the Hamiltonian above with $H~=~E$ you will arrive at a rather complicated equation
$$
p^2~=~-\frac{\hbar^2}{2c}\left(1~-~\sqrt{1~+~\frac{8mc}{\hbar^2}E}\right).
$$
This is from the quadratic equation and the choice with $p^2~=~0$ at $E~=~0$. If you let $E~=~\hbar\omega$ it is not hard to see this recovers the above result with the Schrodinger equation.
