In hopping model, we can get the Hamitonian as $H_0=-t\sum a^\dagger_ia_{i'}$. Then we take the fourier transform and put the operator which are in momentum space in the Hamitonian above. However, I found that there will be terms like $blabla~a^\dagger_ka_{k'}$, which is not remained at last when you finally got the energy spectrum like $E_ka^\dagger_ka_k$. So my question is how do these non-diagonal terms vanish.

  • $\begingroup$ Is that the tight-binding model? If so you're missing the h.c. Your $H_0$ is not Hermitian. $\endgroup$ – Bubble Oct 4 '14 at 15:23
  • $\begingroup$ Yeah, it's the tight-binding model. It seems that if $i$ and $i'$ can run all of the lattice, there is no need to add the hc term, and it is Hermitian. $\endgroup$ – Simon Oct 4 '14 at 15:26
  • $\begingroup$ Aha, so the summation takes care of h.c. and it is Hermitian hence. But its not nearest neighbor if $i$ and $i'$ run over the entire lattice? If so in that case I don't think you can diagonalize the model with Fourier transform, though I'm unsure. $\endgroup$ – Bubble Oct 4 '14 at 15:32
  • $\begingroup$ Well, I think $i$ can run all of the lattice, and keep $i$ and $i'$ to be the nearest ones at the same time. That's what the sum means. $\endgroup$ – Simon Oct 4 '14 at 15:36
  • $\begingroup$ Then you're making a mistake in your calculation. Without seeing the details we can't know what it is. $\endgroup$ – Bubble Oct 4 '14 at 15:37

When you Fourier transform the tight-binding Hamiltonian, $$c_{j,s}=\sum_{k,s} a_{k,s} e^{i R_j k},$$ with periodic boundary conditions, you will be left with a diagonalized Hamiltonian in the desired form. For details see section 2.3 of these lectures notes, starting on p.18: http://manybody.skku.edu/Lecture%20notes/Solid%20State%20Physics.pdf

  • $\begingroup$ Yeah, I think I find where I made the mistake. I forget to use the anti-commutation relations with $a^\dagger_k$ and $a_{k'}$, this will help me cancel those crossing terms. Thank you very much. $\endgroup$ – Simon Oct 4 '14 at 15:52
  • 1
    $\begingroup$ @Simon.Z If you found Bubble's answer helpful, perhaps you might consider upvoting it. $\endgroup$ – WetSavannaAnimal Nov 5 '14 at 13:29

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.