I was wondering about how could one define the exchange interaction in the most general way. I am aware of the definition of exchange for the Coulomb interaction, which looks basically like $\int \phi^\ast_1(r)\phi^\ast_2(r^\prime) \dfrac{1}{\vert r-r^\prime \vert} \phi_2(r^\prime) \phi_1(r),$ and I have also seen, for example, how one defines the exchange energy for Hund-like terms in second-quantized hamiltonians, $\hat{n}_{\alpha,s}\hat{n}_{\beta,s^{\prime}}$ (here $\hat{n}_{\alpha,s}=\hat{c}^\dagger_{\alpha,s}\hat{c}_{\alpha,s}$ is the number operator for state $\alpha$ with spin $s$) as $\langle\hat{c}^\dagger_{\alpha,s} \hat{c}_{\beta,s}\rangle \langle \hat{c}^\dagger_{\beta,s^\prime}\hat{c}_{\alpha,s^\prime} \rangle. $

My first doubt is that I don't understand at all how these two definitions relate to each other, although it is clear that they share the logic of "twisting the indeces".

Secondly, I was wondering whether it would make sense to define a Hartree and an exchange contribution for very general terms like $\hat{c}^\dagger_i\hat{c}^\dagger_j\hat{c}_k\hat{c}_l.$

I believe that, intuitively at least, the exchange are just those non-classical terms which arise due to the antisymmetry of the wave-function when computing expected values over the lowest-energy Slater determinant. Right?

Thus, given a Hamiltonian, I thought of trying to define "the exchange of $\hat{c}^\dagger_i\hat{c}^\dagger_j\hat{c}_k\hat{c}_l$" as the diference: $$\langle \Phi \vert \hat{c}^\dagger_i\hat{c}^\dagger_j\hat{c}_k\hat{c}_l \vert \Phi \rangle -\langle \bigotimes_{ n=1}^{N}\phi_{n} \vert \hat{c}^\dagger_i\hat{c}^\dagger_j\hat{c}_k\hat{c}_l \vert \bigotimes_{n=1}^{N} \phi_n \rangle,$$ where $\text{Alt}\left( \bigotimes_{ n=1}^{N}\vert \phi_{n} \rangle \right),$ where the $\vert \phi_n \rangle$ are the Hartree-Fock spin-orbitals (or they could be the Kohn-Sham spin-orbitals, if you want. The point is just we define the exchange in a precise way in the context of a one-electron approximation).

This seemed to me like a fine candidate for a pretty general definition of exchange, but the problem is, of course, that the action of operators $\hat{c}^{\dagger}_{i}, \hat{c}_j$ is not well defined for states which are not antisymmetrized. Is there any way around this difficulty so that we can indeed define an exchange term in a meaningful way for the expected values of so general operators? Perhaps something in the line of Wick's theorem?


A 2-body operator in second quantization $\hat{O}$ has the form (assuming spin-independent interactions) $$\dfrac{1}{2}\sum_{a,b,c,d,s,s^\prime}O_{a,b}^{c,d}\hat{c}^\dagger_{a,s}\hat{c}^\dagger_{b,s^\prime}\hat{c}_{d,s^\prime}\hat{c}_{c,s},$$ where $$O_{a,b}^{c,d}=(a,b\vert \hat{O} \vert c,d)=\int dr dr^\prime O(r,r^\prime) \phi_{a,s}^\ast(r)\phi_{b,s^\prime}^\ast(r^\prime)\phi_{c,s}(r)\phi_{d,s^\prime}(r^\prime ).$$

The Hartree part has to do with pairing up the terms with the same variable and spin, $\phi_{a,s}(r)$ with $\phi_{c,s}(r)$ and $\phi_{b,s^\prime}(r^\prime)$ with $\phi_{d,s^\prime}(r^\prime).$ The exchange part correspond to the "twisted" pairing in which we associate the $a$ with the $d$ and the $b$ with the $c.$

Using Wick/Isserlis theorem, valid for expected values over Slater determinants, one has: $$\langle \hat{c}^\dagger_{a,s}\hat{c}^\dagger_{b,s^\prime}\hat{c}_{d,s^\prime}\hat{c}_{c,s} \rangle= \langle \hat{c}^\dagger_{a,s}\hat{c}_{c,s} \rangle \langle \hat{c}^\dagger_{b,s^\prime}\hat{c}_{d,s^\prime} \rangle-\langle \hat{c}^\dagger_{a,s}\hat{c}_{d,s} \rangle \langle \hat{c}^\dagger_{b,s^\prime}\hat{c}_{c,s^\prime} \rangle.$$

From the considerations above, is clear that the first terms should be defined as the Hartree contribution and the second one as the exchange part.


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.