I will try to frame my question using the transverse field Ising model in the low spin-coupling limit as motivation. I'll begin by discussing a case I believe I understand, that of eigenstates of continuous translations.

In quantum mechanics, we identify momentum as the generator of translations, which is expressed as $\hat{T}(a) = e^{-i a\hat{p}}$. Here, $\hat{p}$ refers to the total momentum of the system. For example, if we have two free particles, 1 and 2, with position wavefunctions $\Psi(x_1, x_2)=\frac{1}{2\pi}e^{-ix_1p_1}e^{-ix_2p_2}$, then $\hat{T}(a)\Psi(x_1, x_2)=\Psi(x_1+a,x_2+a)=e^{-(p_1+p_2)a}\Psi(x_1,x_2)$. Here it is very natural to read off the total momentum from the argument of the exponential on the right hand side as $p_1+p_2$.

We also often handle periodic lattices in quantum mechanics. For a concrete example, let's consider the 1D transverse field Ising chain, with Hamiltonian $H = -J\sum\limits_{<ij>}\sigma_i^z\sigma_j^z-h\sum\limits_i\sigma_i^x$ with N spins on a ring. The Hamiltonian enjoys invariance under discrete translations of $i \rightarrow i+1$, and so we can write eigenstates of the Hamiltonian as eigenfunctions of the discrete time translation operator.

Let us consider the limit of $J=0$. Then our ground state will look like $|0> =|\rightarrow \rightarrow \rightarrow \rightarrow ...>$. This ground state is clearly an eigenstate of the discrete translation operator.

We will have a collection of N-fold degenerate first excited states that look like $|i>=|\rightarrow\rightarrow\leftarrow_i\rightarrow...> $. We often identify the flipped spin "i" with a particle. We can write these as discrete translation eigenstates by considering the N states $|k> = \frac{1}{N}\sum\limits_j e^{-ikj}|j>$ for $k=-\pi+\frac{2\pi}{N},-\pi+2\frac{2\pi}{N}, ...,\pi.$

These $|k>$ states become important when considering small, nonzero $J$, as they diagonalize the perturbation in the degenerate subspace of first excited states. We typically identify $k$ as the momentum in analogy to the continuous case at top.

Now, the next highest energy levels at $J=0$ feature a $1/2N(N-1)$ degeneracy, with states that look like $|i,j> = |\rightarrow\rightarrow\leftarrow_i\rightarrow\leftarrow_j...>$. My question is how may I take linear combinations of these degenerate states to make eigenstates of the discrete translation operator and how may I associate a total momentum with those eigenstates? Is there a similar canonical case like the continuous one at top, $\hat{T}(a)\Psi(x_1, x_2)=\Psi(x_1+a,x_2+a)=e^{-(p_1+p_2)a}\Psi(x_1,x_2)$, where it is clear that the individual momenta add to form the total momenta?


Your Answer

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

Browse other questions tagged or ask your own question.