3
$\begingroup$

I have been reading David Tong's notes on Phonons: http://www.damtp.cam.ac.uk/user/tong/aqm/aqmfour.pdf I am quite interested in Section 4.1.4, where he quantises the vibrations. First, he defines the most general (classical) solution $u_{n}(t)$, for the displacement of the $n^{th}$ atom in the chain:

(A) \begin{equation} u_{n}(t) = X_0(t) + \sum_{l\neq0}\bigg[\alpha_l\ e^{-i(\omega_lt-k_lna)} + \alpha_l^{\dagger}\ e^{i(\omega_lt-k_lna)}\bigg] \end{equation} and the corresponding momentum:

(B) \begin{equation} p_{n}(t)=P_{0}(t)+\sum_{l \neq 0}\left[-i m \omega_{l} \alpha_{l} e^{-i\left(\omega_{l} t-k_{l} n a\right)}+i m \omega_{l} \alpha_{l}^{\dagger} e^{i\left(\omega_{l} t-k_{l} n a\right)}\right] \end{equation} where $l$ indexes the wave-mode ($l = -N/2 , ... , N/2$) and wavenumber: $k_l = 2\pi\ l/Na$, with $N$ the number of unit cells and $a$ the lattice constant. These are treated as operators in the Heisenberg picture, which can be inverted to find the operators $\alpha_{l}$ and $\alpha_{l}^{\dagger}$. My confusion arises in the following step:

We can invert the equations above by setting t = 0 and looking at \begin{equation} \sum_{n=1}^{N} u_{n} e^{-i k_{l} n a}=\sum_{n} \sum_{l^{\prime}}\left[\alpha_{l} e^{-i\left(k_{l}-k_{l^{\prime}}\right) n a}+\alpha_{l}^{\dagger} e^{-i\left(k_{l}+k_{l^{\prime}}\right) n a}\right]=N\left(\alpha_{l}+\alpha_{-l}^{\dagger}\right) \end{equation}

  1. How did the last equality come about and where did $\alpha_{-l}^{\dagger}$ come from?
  2. Is it mathematically wrong to just add $u_{n}(t = 0)$ and $p_n(t = 0)$ and rearrange to find $\alpha_{l}$, rather than looking at $\sum_{n=1}^{N} u_{n} e^{-i k_{l} n a}$? If so, why?

If someone can help me parse this step, I'd be very grateful :)

$\endgroup$

1 Answer 1

3
$\begingroup$
  1. Using the representation of the Kronecker delta $$\delta_{l,l'} = \frac{1}{N} \sum_{n=1}^N e^{2\pi i (l'-l) n/N}$$ we have $$\sum_n \sum_{l'} \alpha_{l'} e^{-i(k_l-k_{l'})na} = \sum_{l'} \alpha_{l'} N \delta_{l,l'} = N \alpha_l$$ and $$\sum_n \sum_{l'} \alpha^\dagger_{l'} e^{-i(k_l+k_{l'})na} = \sum_{l'} \alpha^\dagger_{l'} N \delta_{-l,l'} = N \alpha^\dagger_{-l}$$ as desired.

  2. To find $\alpha_l$ we need to invert the given relations. Since the these relations take the form of a Fourier series, what Tong does is the most natural. Other methods will amount to the same thing.

$\endgroup$
2
  • 1
    $\begingroup$ Ah, so there was a typo, $a_l$ should probably have been $a_{l'}$ in the original test, right? Otherwise the wrong index is contracted. $\endgroup$
    – qwyxivi
    Commented Jun 28, 2020 at 23:42
  • $\begingroup$ Yes, there must be a typo either in the text or in your transcription of it. $\endgroup$
    – eric
    Commented Jun 29, 2020 at 1:46

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.