A Unified Formulation for ⟨ S ^ 2 ⟩ \langle\hat{S}^{2}\rangle in Two-Component TDDFT
Xiaoyu Zhang · 2026
Read A Unified Formulation for ⟨ S ^ 2 ⟩ \langle\hat{S}^{2}\rangle in Two-Component TDDFT by Xiaoyu Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-16325.txt.
Introduction
Xiaoyu Zhang, writing alone, develops a unified way to evaluate the squared total spin of excited states in two-component time-dependent density functional theory. That theory already covers noncollinear excitations in noncollinear starting states, plus spin-conserving and spin-flip excitations in collinear ones. The paper sets out a general formalism for the expectation value of the squared total spin within that setting.
Zhang then derives specialized forms for collinear cases, where the theory splits into conventional spin-conserving and spin-flip theory. The working forms are compared in detail with earlier theoretical approaches. The analysis shows the excited-state value comes from two distinct sources: the starting state's own value, plus an extra part added when the state is excited. Calculations based on the same theory, unrestricted Kohn-Sham, and restricted open-shell Kohn-Sham starting states put the method to work.
Catalog record
| CREATOR | Xiaoyu Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-16325
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Unified Formulation for ⟨ S ^ 2 ⟩ \langle\hat{S}^{2}\rangle in Two-Component TDDFT Xiaoyu Zhang Abstract Two-component linear-response time-dependent density functional theory (TDDFT) provides a unified framework that encompasses noncollinear excitations in noncollinear reference states, as well as both spin-conserving and spin-flip excitations in collinear reference states. In this work, we present a general formalism for evaluating the expectation value ⟨ S ^ 2 ⟩ \langle\hat{S}^{2}\rangle of electronically excited states obtained within two-component TDDFT. We then derive and analyze specialized forms of the resulting equations for collinear reference determinants, for which the two-component formalism decomposes into conventional spin-conserving and spin-flip TDDFT. The resulting working equations are systematically compared with previously proposed theoretical approaches. On the basis of our analysis, ⟨ S ^ 2 ⟩ \langle\hat{S}^{2}\rangle in the excited states is shown to arise from two distinct sources: (i) ⟨ S ^ 2 ⟩ 0 \langle\hat{S}^{2}\rangle_{0} in the reference state and (ii) additional Δ ⟨ S ^ 2 ⟩ \Delta\langle\hat{S}^{2}\rangle introduced by the excitation process itself. Finally, we evaluate the expectation value ⟨ S ^ 2 ⟩ \langle\hat{S}^{2}\rangle by performing two-component TDDFT calculations based on two-component DFT, unrestricted Kohn–Sham (UKS), and restricted open-shell Kohn–Sham (ROKS) reference states, respectively. † † email: [email protected] † † affiliation: College of Chemistry and Molecular Engineering, Peking University, Beijing 100871, the People’s Republic of China 1 Introduction Linear-response TDDFT is prevalent due to its balance between efficiency and accuracy. 1 In the non-relativistic limit, two-component TDDFT is an adequate framework even in a noncollinear magnetic system 6 . On a collinear reference state, two-component TDDFT is divided into spin-conserving and spin-flip TDDFT. Spin-flip TDDFT, formulated on an open-shell reference state, has been demonstrated to be particularly useful for the description of conical intersections, charge-transfer excited states, nonadiabatic derivative couplings, spin–orbit couplings, and bond dissociation processes. 16 In order to further enable applications in spectroscopy, it is vital to know the spin multiplicity of excited states; that is, to calculate \langle\hat{S}^{2}\rangle . However, calculating \langle\hat{S}^{2}\rangle is not trivial and hasn’t been systematically explored in two-component TDDFT. Maurice and Head-Gordon 9 derived analytical expressions for the expectation value of the total spin-squared operator, \langle\hat{S}^{2}\rangle , within the framework of spin-conserving configuration interaction singles (CIS). Ipatov et al. 5 subsequently obtained expressions for \langle\hat{S}^{2}\rangle in spin-conserving TDDFT. Myneni and Casida 10 addressed and resolved several ambiguities associated with the evaluation of \langle\hat{S}^{2}\rangle in spin-conserving TDDFT and proposed an ab initio formulation. Building on the conceptual framework introduced by Ipatov et al. , Li et al. 7 derived the corresponding working equations for spin-flip TDDFT, which they reported in the appendix of their work. To enrich and systematize the theoretical framework in this area, we introduce a unified formulation for the spin-squared expectation value, \langle\hat{S}^{2}\rangle , within two-component time-dependent density functional theory (TDDFT). We begin from the second-quantized representation of \hat{S}^{2} and demonstrate that its algebraic structure closely parallels that of the second-quantized electronic Hamiltonian \hat{H} . On this basis, we derive explicit expressions for \langle\hat{S}^{2}\rangle evaluated over general Slater determinants, which are found to be analogous to the corresponding energy expectation values derived from the Slater–Condon rules. Subsequently, we formulate two-component TDDFT in the second-quantized equation-of-motion (EOM) framework, which is formally equivalent to the Casida equation in the first-quantized matrix representation. By substituting \hat{H} with \hat{S}^{2} in the EOM-like TDDFT expressions, we obtain a Casida-type eigenvalue equation in the first-quantized representation for the evaluation of spin-squared expectation values. In close analogy to the electronic excitation energies \omega ( \Delta\langle\hat{H}\rangle ), the changes in spin-squared expectation values, \Delta\langle\hat{S}^{2}\rangle , can be computed within this Casida-like formalism. We then specialize the treatment to collinear reference states and derive further simplified working equations, which enable direct comparison with previously proposed spin-analysis schemes. Finally, we present numerical results for a range of molecular systems to illustrate and benchmark the proposed methodology. 2 Theory 2.1 Second-Quantized \hat{S}^{2} We use \Gamma,\Lambda,\Theta... for two-component AO basis functions, P,Q,R... for unspecified two-component molecular orbitals, and I,J,K... / A,B,C... for occupied / virtual orbitals. We use g_{\mu},g_{\nu},... to denote the spatial part of AO basis functions. The corresponding spin-up basis is denoted as: |\mu\rangle=\begin{pmatrix}g_{\mu}\\ 0\end{pmatrix} (1) And the spin-down basis is denoted as: |\bar{\mu}\rangle=\begin{pmatrix}0\\ g_{\mu}\end{pmatrix} (2) Also, we use the corresponding lowercase letters to denote the spin-up or spin-down parts of two-component orbitals: |P\rangle=\begin{pmatrix}p\\ \bar{p}\end{pmatrix}=\sum_{\mu}\begin{pmatrix}c_{\mu p}g_{\mu}\\ c_{\bar{\mu}\bar{p}}g_{\mu}\end{pmatrix} (3) We denote the coefficient matrix as follows: (C^{\alpha})_{\mu p}=c_{\mu p} (4) (C^{\beta})_{\mu p}=c_{\bar{\mu}\bar{p}} (5) We denote the column p of C^{\alpha} as C^{\alpha}_{p} , and that of C^{\beta} as C^{\beta}_{p} . We use short notations for double-electron integrals: (PQ|RS)=\int d\mathbf{r}_{1}d\mathbf{r}_{2}\frac{P^{\dagger}(\mathbf{r}_{1})Q(\mathbf{r}_{1})R^{\dagger}(\mathbf{r}_{2})S(\mathbf{r}_{2})}{|\mathbf{r}_{1}-\mathbf{r}_{2}|} (6) \langle PQ|RS\rangle=(PR|QS) (7) \langle PQ||RS\rangle=\langle PQ|RS\rangle-\langle PQ|SR\rangle (8) We denote the matrix representation of a one-electron spin operator \hat{s}_{u} as: (s_{u})_{PQ}=\langle P|\frac{1}{2}\hat{\sigma}_{u}|Q\rangle (9) Here, u=x,y,z . \hat{\sigma}_{u} is the Pauli matrix along the u -axis. Because \hat{s}_{u} is a one-body operator, its second-quantized form is: \hat{s}_{u}=\sum_{PQ}(s_{u})_{PQ}\hat{a}_{P}^{\dagger}\hat{a}_{Q} (10) The total spin of the electrons is defined as: \hat{S}^{2}=\sum_{u}\hat{s}_{u}\hat{s}_{u} (11) By substituting the \hat{s}_{u} , we obtain: \hat{S}^{2}=\sum_{u}\sum_{PQRS}(s_{u})_{PQ}(s_{u})_{RS}\hat{a}^{\dagger}_{P}\hat{a}_{Q}\hat{a}_{R}^{\dagger}\hat{a}_{S} (12) There are two basic anti-commutation relations: \{\hat{a}_{P},\hat{a}_{Q}\}=0 (13) \{\hat{a}_{P},\hat{a}_{Q}^{\dagger}\}=\delta_{PQ} (14) Using the two relations, we obtain: \hat{S}^{2}=\sum_{u}\sum_{PQRS}(s_{u})_{PQ}(s_{u})_{RS}(\delta_{QR}\hat{a}_{P}^{\dagger}\hat{a}_{S}+\hat{a}_{P}^{\dagger}\hat{a}_{R}^{\dagger}\hat{a}_{S}\hat{a}_{Q}) (15) We denote: W_{PQRS}=2\sum_{u}(s_{u})_{PR}(s_{u})_{QS} (16) Based on the definition, we have: W_{PQRS}=W_{QPSR}=W^{*}_{RSPQ}=W^{*}_{SRQP} (17) Finally, we obtain: \hat{S}^{2}=\frac{3}{4}\sum_{P}\hat{a}_{P}^{\dagger}\hat{a}_{P}+\frac{1}{2}\sum_{PQRS}W_{PQRS}\hat{a}_{P}^{\dagger}\hat{a}_{Q}^{\dagger}\hat{a}_{S}\hat{a}_{R} (18) 2.2 Expectation Values of \hat{S}^{2} We denote |X\rangle as the reference single slater determinant wavefunction |\Psi\rangle_{0} , constructed by a set of orthonormal molecular orbitals. We usually obtain |\Psi\rangle_{0} using the two-component Hartree-Fock method. A singly excited wavefunction is denoted as |Y\rangle=\hat{a}^{\dagger}_{A}\hat{a}_{I}|X\rangle . A doubly excited wavefunction is denoted as |Z\rangle=\hat{a}^{\dagger}_{A}\hat{a}_{B}^{\dagger}\hat{a}_{J}\hat{a}_{I}|X\rangle . Based on the Slater-Condon rule 13 ; 3 , we can calculate the expectation value of \hat{S}^{2} for different determinants. \langle X|\hat{S}^{2}|X\rangle=\frac{3}{4}N+\frac{1}{2}\sum_{IJ}(W_{IJIJ}-W_{IJJI}) (19) \langle X|\hat{S}^{2}|Y\rangle=\sum_{J}(W_{IJAJ}-W_{IJJA})=\sum_{J}(W_{AJIJ}-W_{AJJI})^{*} (20) \langle X|\hat{S}^{2}|Z\rangle=W_{IJAB}-W_{IJBA}=(W_{ABIJ}-W_{ABJI})^{*} (21) Here, N is the number of electrons. We now need to further rewrite W_{PQRS} with two-component orbital indices to W_{pqrs} with one-component orbital indices. We denote the overlap matrix as: S_{pq}=\langle p|q\rangle (22a) S_{\bar{p}q}=\langle\bar{p}|q\rangle (22b) S_{p\bar{q}}=\langle p|\bar{q}\rangle (22c) S_{\bar{p}\bar{q}}=\langle\bar{p}|\bar{q}\rangle (22d) By expanding a molecular orbital into a set of basis functions, we have: S_{pq}=\sum_{\mu\nu}c^{*}_{\mu p}c_{\nu q}\langle g_{\mu}|g_{\nu}\rangle=\sum_{\mu\nu}c^{*}_{\mu p}c_{\nu q}S_{\mu\nu} (23a) S_{\bar{p}q}=\sum_{\mu\nu}c^{*}_{\bar{\mu}\bar{p}}c_{\nu q}\langle g_{\mu}|g_{\nu}\rangle=\sum_{\mu\nu}c^{*}_{\bar{\mu}\bar{p}}c_{\nu q}S_{\mu\nu} (23b) S_{p\bar{q}}=\sum_{\mu\nu}c^{*}_{\mu p}c_{\bar{\nu}\bar{q}}\langle g_{\mu}|g_{\nu}\rangle=\sum_{\mu\nu}c^{*}_{\mu p}c_{\bar{\nu}\bar{q}}S_{\mu\nu} (23c) S_{\bar{p}\bar{q}}=\sum_{\mu\nu}c^{*}_{\bar{\mu}\bar{p}}c_{\bar{\nu}\bar{q}}\langle g_{\mu}|g_{\nu}\rangle=\sum_{\mu\nu}c^{*}_{\bar{\mu}\bar{p}}c_{\bar{\nu}\bar{q}}S_{\mu\nu} (23d) Then, W_{pqrs}=\frac{1}{2}(S_{pr}-S_{\bar{p}\bar{r}})(S_{qs}-S_{\bar{q}\bar{s}})+S_{p\bar{r}}S_{\bar{q}s}+S_{\bar{p}r}S_{q\bar{s}} (24) By introducing eq. 24 into eq. 19 , we produce the same result as eq. (19) in Ref. 2 . But we need to derive eq. 19 under a common AO representation for more convenient implementation. We define the first-order reduced density matrix as: D_{\Gamma\Lambda}=\sum_{I}c^{*}_{\Gamma I}c_{\Lambda I} (25) We denote four blocks of the density matrix as: D_{\mu\nu}=\sum_{i}c^{*}_{\mu i}c_{\nu i}=(D^{\alpha\alpha})_{\mu\nu} (26a) D_{\bar{\mu}\nu}=\sum_{i}c^{*}_{\bar{\mu}\bar{i}}c_{\nu i}=(D^{\beta\alpha})_{\mu\nu} (26b) D_{\mu\bar{\nu}}=\sum_{i}c^{*}_{\mu i}c_{\bar{\nu}\bar{i}}=(D^{\alpha\beta})_{\mu\nu} (26c) D_{\bar{\mu}\bar{\nu}}=\sum_{i}c^{*}_{\bar{\mu}\bar{i}}c_{\bar{\nu}\bar{i}}=(D^{\beta\beta})_{\mu\nu} (26d) After the mathematical derivation shown in Appendix A, we have the expectation value of the total electronic spin in the reference state: \begin{split}\langle\hat{S}^{2}\rangle_{0}&=\frac{3}{4}N+\frac{1}{4}[\mathrm{Tr}(S(D^{\alpha\alpha}-D^{\beta\beta}))]^{2}+\mathrm{Tr}(SD^{\alpha\beta})\times\mathrm{Tr}(SD^{\beta\alpha})\\ &-\frac{1}{4}[\mathrm{Tr}(D^{\alpha\alpha}SD^{\alpha\alpha}S)+\mathrm{Tr}(D^{\beta\beta}SD^{\beta\beta}S)-2\mathrm{Tr}(D^{\alpha\beta}SD^{\beta\alpha}S)]-\mathrm{Tr}(D^{\alpha\alpha}SD^{\beta\beta}S)\end{split} (27) If there are no fractional occupations, we have: \langle\hat{S}^{2}\rangle_{0}=\frac{1}{2}N+\frac{1}{4}[\mathrm{Tr}(S(D^{\alpha\alpha}-D^{\beta\beta}))]^{2}+\mathrm{Tr}(SD^{\alpha\beta})\times\mathrm{Tr}(SD^{\beta\alpha})+\mathrm{Tr}(D^{\alpha\beta}SD^{\beta\alpha}S)-\mathrm{Tr}(D^{\alpha\alpha}SD^{\beta\beta}S) (28) Another way to calculate \langle\hat{S}^{2}\rangle_{0} is to decompose \hat{S}^{2} into one-body operators \hat{S}_{+} , \hat{S}_{-} , and \hat{S}_{z} , and we have: \hat{S}^{2}=\frac{1}{2}(\hat{S}_{+}\hat{S}_{-}+\hat{S}_{-}\hat{S}_{+})+\hat{S}_{z}^{2} (29) \begin{split}\langle\hat{S}_{+}\hat{S}_{-}\rangle_{0}=\mathrm{Tr}(D^{\alpha\alpha}S)+\mathrm{Tr}(D^{\beta\alpha}S)\mathrm{Tr}(D^{\alpha\beta}S)-\mathrm{Tr}[(D^{\alpha\alpha}S)(D^{\beta\beta}S)]\end{split} (30) \langle\hat{S}_{-}\hat{S}_{+}\rangle_{0}=\mathrm{\mathrm{Tr}}(D^{\beta\beta}S)+\mathrm{\mathrm{Tr}}(D^{\beta\alpha}S)\mathrm{\mathrm{Tr}}(D^{\alpha\beta}S)-\mathrm{\mathrm{Tr}}[(D^{\alpha\alpha}S)(D^{\beta\beta}S)] (31) \begin{split}\langle\hat{S}_{z}\hat{S}_{z}\rangle_{0}=\frac{1}{4}\mathrm{Tr}(D^{\alpha\alpha}S)+\frac{1}{4}\mathrm{Tr}(D^{\beta\beta}S)+\frac{1}{4}[\mathrm{Tr}(D^{\alpha\alpha}S)-\mathrm{Tr}(D^{\beta\beta}S)]^{2}\\ -\frac{1}{4}\mathrm{Tr}(D^{\alpha\alpha}SD^{\alpha\alpha}S)-\frac{1}{4}\mathrm{Tr}(D^{\beta\beta}SD^{\beta\beta}S)+\frac{1}{2}\mathrm{Tr}(D^{\alpha\beta}SD^{\beta\alpha}S)\end{split} (32) This method produces the same result as eq. 27 . Here, D and S are both under a common AO representation. After the mathematical derivation shown in Appendix B, we have: \begin{split}\langle X|\hat{S}^{2}|Y\rangle&=[\left(C^{\alpha}_{a}\right)^{\dagger}\left(\frac{1}{2}S\mathrm{Tr}\left(S\left(D^{\alpha\alpha}-D^{\beta\beta}\right)\right)-\frac{1}{2}\left(S\left(D^{\alpha\alpha}\right)^{T}S\right)-\left(S\left(D^{\beta\beta}\right)^{T}S\right)\right)\left(C^{\alpha}_{i}\right)\\ &+\left(C_{a}^{\alpha}\right)^{\dagger}\left(S\mathrm{Tr}\left(SD^{\beta\alpha}\right)+\frac{1}{2}S\left(D^{\beta\alpha}\right)^{T}S\right)\left(C_{i}^{\beta}\right)\\ &+\left(C_{a}^{\beta}\right)^{\dagger}\left(S\mathrm{Tr}\left(SD^{\alpha\beta}\right)+\frac{1}{2}S\left(D^{\alpha\beta}\right)^{T}S\right)\left(C_{i}^{\alpha}\right)\\ &+\left(C^{\beta}_{a}\right)^{\dagger}\left(-\frac{1}{2}S\mathrm{Tr}\left(S\left(D^{\alpha\alpha}-D^{\beta\beta}\right)\right)-\frac{1}{2}\left(S\left(D^{\beta\beta}\right)^{T}S\right)-\left(S\left(D^{\alpha\alpha}\right)^{T}S\right)\right)\left(C^{\beta}_{i}\right)]^{*}\end{split} (33) For eq. 21 , we directly use