1
$\begingroup$

I am stuck calculating a simple expectation value for an operator, which is expressed in second quantization. I know the result, but I fail to proof it.

Lets say I have one-particle wave function $|\phi_n\rangle$ given by $|\phi_n\rangle=\sum_{j=1}^K |\alpha_j\rangle A_{j,n}$, where $K$ is the number of orbitals/sites in the system and the $A_{j,n}$ are the probability amplitudes of the oribitals $|\alpha_j\rangle$. The index $n$ labels the particles in the system, of which we have $N$.

The orbitals are orthonormal, i.e. $$\sum_j A_{j,m}^* A_{j,n} = \delta_{m,n}.$$

Let's ignore any spin degrees of freedom. The many particle wave function is now given by $$|\Psi\rangle = \left(\prod_{n=1}^N \sum_{j=1}^K \hat{c}^\dagger_j A_{j,n}\right) |\text{vac}\rangle,$$ where the $\hat{c}^\dagger_j$ is the usual creation operator on site $j$.

What I now want to calculate, is the expectation value $$\langle \Psi|H_\text{hop}|\Psi\rangle$$ with $$H_\text{hop}=-t \sum_{j=1}^K \hat{c}^\dagger_{j+1}\hat{c}_j + h.c.$$ the usual hopping Hamiltonian. I have the strong feeling (and one example calculation supported this), that the result is just $$\langle \Psi|H_\text{hop}|\Psi\rangle = -t \sum_{n=1}^N \sum_{j=1}^K ( A_{j,n}^* A_{j+1,n} + A_{j+1,n}^* A_{j,n} )$$

I think this result is trivially related to the Slater-Condon rules, but I fail to see the connection. Additionally, I fail to explicitly calculate the expectation value, which contains the sum and products of the creation/annihilation operators.

What is a good way to prove my result?

$\endgroup$
  • $\begingroup$ Hint: use the CCR-s associated with the ${\hat c}_j$-s. Doesn't your sample calculation already suggest something like this? $\endgroup$ – udrv May 31 '16 at 23:39
  • $\begingroup$ Well, I bet I have to use the commutation relations at some point. But I really don't see, how I can arrive at the general result. As the scalar product is linear, it is surely enough to just consider one term from $H_\text{hop}$, so maybe one can set w.l.o.g. $j=1$. But then I have still lots of $\hat{c}^\dagger$ from the $|\Psi\rangle$ (Slater determinants). How do I proceed? $\endgroup$ – Merlin1896 Jun 1 '16 at 7:47
2
$\begingroup$

For future convenience denote $$ {\hat \phi}^\dagger_n = \sum_j{A_{j,n} {\hat c}^\dagger_j} $$ $$ {\hat \chi}^\dagger_n = \sum_j{A_{j,n} {\hat c}^\dagger_{j+1}} $$ The average you want to calculate reads then $$ \langle \Psi | {\hat H}_0| \Psi \rangle = - t\;\langle 0 | \prod_n{{\hat \phi}_n} \left(\sum_j{{\hat c}^\dagger_{j+1}{\hat c}_j} \right)\prod_n{{\hat \phi}^\dagger_n} |0\rangle $$ Starting with the regular CCR-s, $$ \left[ {\hat c}_j, {\hat c}^\dagger_k \right]_\mp = \delta_{jk},\;\;\;\;\; \left[ {\hat c}_j, {\hat c}_k\right]_\mp = \left[ {\hat c}^\dagger_j, {\hat c}^\dagger_k\right]_\mp = 0 $$ obtain $$ \left[ \sum_j{{\hat c}^\dagger_{j+1}{\hat c}_j}, {\hat \phi}^\dagger_n\right] = {\hat \chi}^\dagger_n $$ and $$ {\hat \phi}^\dagger_n {\hat \chi}^\dagger_m = \pm {\hat \chi}^\dagger_m {\hat \phi}^\dagger_n $$ $$ {\hat \phi}_n {\hat \chi}^\dagger_m = \pm {\hat \chi}^\dagger_m {\hat \phi}_n + \sum_j{A^*_{j+1, n}A_{j,m}} $$ Now, in the average you need to calculate, use the above to successively move $\left(\sum_j{{\hat c}^\dagger_{j+1}{\hat c}_j} \right)$ past the orbital operators to its right: $$ \left(\sum_j{{\hat c}^\dagger_{j+1}{\hat c}_j} \right)\prod_n{{\hat \phi}^\dagger_n} = {\hat \phi}^\dagger_1 \left(\sum_j{{\hat c}^\dagger_{j+1}{\hat c}_j} \right) \prod_{n>1}{{\hat \phi}^\dagger_n} + {\hat \chi}^\dagger_1 \prod_{n\neq 1} {{\hat \phi}^\dagger_n} = $$ $$ = \prod_{m=1}^{m=2} {{\hat \phi}^\dagger_m} \left(\sum_j{{\hat c}^\dagger_{j+1}{\hat c}_j} \right) \prod_{n>2}{{\hat \phi}^\dagger_n} + {\hat \chi}^\dagger_1 \prod_{n\neq 1} {{\hat \phi}^\dagger_n} + {\hat \phi}^\dagger_1 {\hat \chi}^\dagger_2 \prod_{n > 2} {{\hat \phi}^\dagger_n} = $$ $$ = \prod_{m=1}^{m=2} {\hat \phi}^\dagger_m \left(\sum_j{{\hat c}^\dagger_{j+1}{\hat c}_j} \right) \prod_{n>2}{{\hat \phi}^\dagger_n} + {\hat \chi}^\dagger_1 \prod_{n\neq 1} {{\hat \phi}^\dagger_n} \pm {\hat \chi}^\dagger_2 \prod_{n \neq 2} {{\hat \phi}^\dagger_n} = \dots = $$ $$ = \prod_m {{\hat \phi}^\dagger_m} \left(\sum_j{{\hat c}^\dagger_{j+1}{\hat c}_j} \right) + \sum_m{ (\pm 1)^{m-1} {\hat \chi}^\dagger_m \prod_{n \neq m} {{\hat \phi}^\dagger_n}} $$ The first term will annihilate the rhs vacuum, so only the sum remains. Now bring in the orbital operators on the left and flip each ${\hat \chi}^\dagger_m$ past them: $$ \left(\prod_n{{\hat \phi}_n}\right) {\hat \chi}^\dagger_m = \pm \left(\prod_{n>1}{{\hat \phi}_n}\right) {\hat \chi}^\dagger_m {\hat \phi}_1 + \left(\prod_{n\neq 1}{{\hat \phi}_n}\right) \sum_j{A^*_{j+1, 1}A_{j,m}} = $$ $$ \left(\prod_{n>1}{{\hat \phi}_n}\right) {\hat \chi}^\dagger_m \prod_{l=1}^{l=2} {{\hat \phi}_l} + \left(\prod_{n\neq 1}{{\hat \phi}_n}\right) \sum_j{A^*_{j+1, 1}A_{j,m}} \pm \left(\prod_{n\neq 2}{{\hat \phi}_n}\right)\sum_j{A^*_{j+1, 2}A_{j,m}} = \dots = $$ $$ = {\hat \chi}^\dagger_m \left(\prod_n{{\hat \phi}_n}\right) + \sum_l{(\pm 1)^{l-1}\left(\prod_{n\neq l}{{\hat \phi}_n}\right) \sum_j{A^*_{j+1, l}A_{j,m}} } $$ The first term now annihilates the lhs vacuum, and after substituting everything the desired average becomes $$ \langle \Psi | {\hat H}_0| \Psi \rangle = - t\;\sum_{j, l,m}{ (\pm 1)^{m-1} (\pm 1)^{l-1} A^*_{j+1, l}A_{j,m} \langle 0 |\left(\prod_{n\neq l}{{\hat \phi}_n}\right)\left( \prod_{n' \neq m} {{\hat \phi}^\dagger_{n'}}\right) |0\rangle } = $$ $$ \langle \Psi | {\hat H}_0| \Psi \rangle = - t\;\sum_{j, l,m}{ (\pm 1)^{m-1} (\pm 1)^{l-1} A^*_{j+1, l}A_{j,m}\delta_{l,m} } $$ and finally $$ \langle \Psi | {\hat H}_0| \Psi \rangle = - t\;\sum_{j, m}{ A^*_{j+1, m}A_{j,m} } $$ There may be some bugs I missed, but this is the general idea.

$\endgroup$

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.