Seniority

The seniority operator counts the number of orbitals that are singly occupied in a Slater determinant. It is a Hermitian operator and can be defined (Bytautas et al. 2011) as \[\begin{equation} \label{eq:seniority_operator} \require{physics} \hat{\Omega} % = \sum_p^K \hat{\Omega}_p % \thinspace , \end{equation}\] where \(\hat{\Omega}_p\) can be written in many different Van Raemdonck et al. (2015) ways: \[\begin{align} \hat{\Omega}_p % &= \sum_\sigma % \hat{a}^\dagger_{p \sigma} \hat{a}_{p \sigma} % - \sum_{\sigma \tau} % \hat{a}^\dagger_{p \sigma} \hat{a}^\dagger_{p \tau} % \hat{a}_{p \tau} \hat{a}_{p \sigma} \\ &= \hat{E}_{pp} - \hat{e}_{pppp} \\ &= \hat{a}^\dagger_{p \alpha} \hat{a}_{p \alpha} % + \hat{a}^\dagger_{p \beta} \hat{a}_{p \beta} % - 2 \hat{a}^\dagger_{p \alpha} \hat{a}^\dagger_{p \beta} % \hat{a}_{p \beta} \hat{a}_{p \alpha} \\ &= \hat{N}_{p \alpha} + \hat{N}_{p \beta} % - 2 \hat{N}_{p \alpha} \hat{N}_{p \beta} \\ &= (\hat{N}_{p \alpha} - \hat{N}_{p \beta})^2 % \label{eq:seniority_number_operators_squared} \\ &= \frac{4}{3} \hat{S}^2_p % \label{eq:seniority_casimir_S} \\ &= -\frac{4}{3} \hat{C}_p + 1 % \label{eq:seniority_casimir_P} \thinspace . \end{align}\]

The vacuum state has seniority zero: \[\begin{equation} \hat{\Omega} \ket{\text{vac}} % = 0 % \thinspace . \end{equation}\] The seniority of a general wave function \(\ket{\Psi}\) can be calculated as \[\begin{equation} \ev{ \hat{\Omega} }{\Psi} % = N - \sum_p^K d_{pppp} % \thinspace , \end{equation}\] in which \(\vb{d}\) is the 2-DM.

As the \(p\)-th orbital seniority operator can be written in linear terms of the \(\hat{S}\) and \(\hat{P}\) Casimir operators, i.e. \(\hat{S}^2_p\) and \(\hat{C}_p\), the following commutators are also zero: \[\begin{equation} \label{eq:commutator_seniority_P} \comm{ % \hat{\Omega}_p % }{ % \hat{P}^+_q % } % = \comm{ % \hat{\Omega}_p % }{ % \hat{P}^-_q % } % = \comm{ % \hat{\Omega}_p % }{ % \hat{P}^\circ_q % } = 0 % \thinspace , \end{equation}\] and \[\begin{equation} \label{eq:commutator_seniority_S} \comm{ % \hat{\Omega}_p % }{ % \hat{S}^+_q % } % = \comm{ % \hat{\Omega}_p % }{ % \hat{S}^-_q % } % = \comm{ % \hat{\Omega}_p % }{ % \hat{S}^z_q % } = 0 % \thinspace . \end{equation}\] Furthermore, since orbital occupation number operators commute among themselves, we have \[\begin{equation} \label{eq:commutator_seniority_on_spin} \comm{ % \hat{N}_{p \sigma} % }{ % \hat{\Omega}_q % } = 0 \end{equation}\] and therefore \[\begin{equation} \label{eq:commutator_seniority_on} \comm{ % \hat{N}_p % }{ % \hat{\Omega}_q % } = 0 % \thinspace . \end{equation}\] A single Slater determinant (an occupation number vector \(\ket{\vb{k}}\)) is an eigenvector of the seniority operator: \[\begin{equation} \hat{\Omega} \ket{\vb{k}} % = \Omega_{\vb{k}} \ket{\vb{k}} % \thinspace , \end{equation}\] with \[\begin{equation} \Omega_{\vb{k}} % = \sum_p^K \qty( % k_{p \alpha} - k_{p \beta} % )^2 % \thinspace , \end{equation}\] which means that an arbitrary linear combination of occupation number vectors, \[\begin{equation} \ket{\vb{c}} % = \sum_{\vb{k}} % c_{\vb{k}} \ket{\vb{k}} % \thinspace , \end{equation}\] can be divided in so called seniority sectors (Bytautas et al. 2011), i.e. sets of occupation number vectors (Slater determinants) that belong to a certain seniority number. Seniority being a good quantum number, or preserving the seniority number means that if an operator \(\hat{A}\), acts on for example a seniority zero wave function, the corresponding seniority number \(\Omega=0\) will not change: \[\begin{equation} \hat{A} \ket{\Omega = 0} % \propto % \ket{\Omega = 0} % \qquad \iff \qquad % \comm{ % \hat{A} % }{ % \hat{\Omega} % } = 0 % \thinspace . \end{equation}\] In other words, \(\hat{A} \ket{\Omega = 0}\) remains in the seniority zero sector, or equivalently: the amount of broken pairs (for singlet wave functions) remains the same. Another way to say this is that \(\hat{\Omega}\) is a symmetry or constant of motion for \(\hat{A}\). We have already seen examples of seniority-preserving operators: \(\hat{P}^+_p, \hat{P}^-_p\) and \(\hat{P}^\circ_p\) (see above: the commutators with \(\hat{\Omega}\) vanish).

The dimension of the Fock space spanned by the determinants having \(N_\alpha\) \(\alpha\)-electrons and \(N_\beta\) \(\beta\)-electrons and is equal to \[\begin{equation} \dim \mathcal{F}(2K, N_\alpha, N_\beta) % = \binom{K}{N_\alpha} \binom{K}{N_\beta} % \thinspace , \end{equation}\] in which \(K\) is the number of spatial orbitals. We can arrivate at since every electron by realizing that every \(\alpha\)-electron can only occupy one of the \(K\) spin orbitals, and equivalently for the \(\beta\)-electrons. Note that this is not equal to \[\begin{equation*} \binom{2K}{N_\alpha + N_\beta} % \thinspace , \end{equation*}\] an expression that could have been derived from equation \(\eqref{eq:dim_fock_M_N}\). The reason for inequality is because the latter way of looking at it is by looking at \(N = N_\alpha + N_\beta\) electrons as a whole, while the former looks at \(N_\alpha\) \(\alpha\)-electrons and \(N_\beta\) \(\beta\)-electrons separately (i.e. with a defined spin projection \(S_z\)). Given an equal amount of \(\alpha\)- and \(\beta\)-electrons (in total summed up to \(N\)), Bytautas (Bytautas et al. 2011) writes the number of seniority \(\Omega=2k\) determinants as \[\begin{equation} D_{\Omega = 2k} % = \binom{K}{N/2} \binom{N/2}{k} \binom{K - N/2}{k} % \thinspace . \end{equation}\] Alcoba (Alcoba et al. 2014) extended this formula to \[\begin{equation} D_{\Omega = N - 2r} % = \binom{K}{N/2 + S_z} % \binom{N/2 + S_z}{r} % \binom{K - (N/2 + S_z)}{N/2 - S_z - r} % \thinspace , \end{equation}\] which in terms of \(N_\alpha\), \(N_\beta\), \(K\) and the seniority \(\Omega\) is \[\begin{equation} D(\Omega, K, N_\alpha, N_\beta) % = \binom{K}{N_\alpha} % \binom{N_\alpha}{ % \frac{ N_\alpha + N_\beta - \Omega }{2} % } % \binom{K-N_\alpha}{ % \frac{ \Omega - N_\alpha + N_\beta }{2} % } % \thinspace . \end{equation}\]

References

Alcoba, Diego R., Alicia Torre, Luis Lain, Gustavo E. Massaccesi, and Ofelia B. Oña. 2014. Configuration interaction wave functions: A seniority number approach.” The Journal of Chemical Physics 140 (23): 234103. https://doi.org/10.1063/1.4882881.
Bytautas, Laimutis, Thomas M. Henderson, Carlos A. Jiménez-Hoyos, Jason K. Ellis, and Gustavo E. Scuseria. 2011. Seniority and orbital symmetry as tools for establishing a full configuration interaction hierarchy.” The Journal of Chemical Physics 135: 044199. https://doi.org/10.1063/1.3613706.
Van Raemdonck, Mario, Diego R. Alcoba, Ward Poelmans, Stijn De Baerdemacker, Alicia Torre, Luis Lain, Gustavo E. Massaccesi, Dimitri Van Neck, and Patrick Bultinck. 2015. Polynomial scaling approximations and dynamic correlation corrections to doubly occupied configuration interaction wave functions.” Journal of Chemical Physics 143 (10). https://doi.org/10.1063/1.4930260.