![]() |
(git:27c966b)
|
Fully periodic small-cell GW using RI-RS and explicit full k-point sampling. More...
Functions/Subroutines | |
| subroutine, public | gw_calc_ri_rs_small_cell_full_kp (qs_env, bs_env) |
| Periodic RI-RS GW from the converged SCF orbitals and eigenvalues. | |
Fully periodic small-cell GW using RI-RS and explicit full k-point sampling.
| subroutine, public gw_ri_rs_small_cell_full_kp::gw_calc_ri_rs_small_cell_full_kp | ( | type(qs_environment_type), pointer | qs_env, |
| type(post_scf_bandstructure_type), pointer | bs_env ) |
Periodic RI-RS GW from the converged SCF orbitals and eigenvalues.
l,l′ label real-space points r_l,r_l′; μ labels AOs, n bands, σ spins, P,Q RI functions. k samples the SCF mesh, q transfers momentum, and R,T are lattice vectors. All phases use Cartesian vectors. ψ_nkσ(r_l) is a sampled Bloch orbital; C_μn(k,σ) is its AO coefficient. σ sums over the two physical spin channels.
(1) ϕ_μ(k,r_l) = Σ_R exp(ik·R) ϕ_μ(r_l-R), ψ_nkσ(r_l) = Σ_μ ϕ_μ(k,r_l) C_μn(k,σ).
(2) ρ_l(r) = Σ_μR ϕ_μ(r_l-R) ϕ_μ(r-R), D_ll′(T) = ρ_l(r_l′+T)², d_lP(T) = ∫dr ∫dr′ ρ_l(r)² m(r-r′) P(r′-T), D_ll′(q) = Σ_T exp(iq·T) D_ll′(T), d_lP(q) = Σ_T exp(iq·T) d_lP(T), S_l(q) = D_ll(q)^(-1/2), D̂_ll′(q) = S_l(q) D_ll′(q) S_l′(q), d̂_lP(q) = S_l(q) d_lP(q), [D̂(q) + ηI] z_P(q) = d̂_P(q), Z_lP(q) = S_l(q) z_lP(q). P(r) is an RI basis function, m(r-r′) the selected RI fitting kernel, and η = RI_RS/TIKHONOV.
(3) G^occ/vir_ll′(k,iτ;σ) = Σ_n∈occ/vir ψ_nkσ(r_l) ψ_nkσ(r_l′)* × exp(-|ε_nkσ-ε_Fσ| τ), τ≥0.
(4) G^occ/vir_ll′(R,iτ;σ) = Re[Σ_k w_k exp(-ik·R) G^occ/vir_ll′(k,iτ;σ)].
(5) χ_ll′(R,iτ) = Σ_σ G^occ_ll′(R,iτ;σ) G^vir_ll′(R,iτ;σ), χ_ll′(q,iτ) = Σ_R exp(iq·R) χ_ll′(R,iτ).
(6) χ_ll′(q,iω_j) = Σ_t C_jt χ_ll′(q,iτ_t), χ_PQ(q,iω_j) = Σ_ll′ Z_lP(q)* χ_ll′(q,iω_j) Z_l′Q(q). C_jt supplies the imaginary-time Fourier transformation by cosine quadrature.
(7) K(q) = M(q)⁻¹ V(q)½, ε_PQ(q,iω) = δ_PQ + Σ_P′Q′ K_P′P(q)* χ_P′Q′(q,iω) K_Q′Q(q), Wᶜ_PQ(q,iω) = Σ_P′Q′ K_PP′(q) [ε⁻¹_P′Q′(q,iω)-δ_P′Q′] K_QQ′(q)*. M is the RI fitting metric and V the Coulomb matrix.
(8) Wᶜ_ll′(q,iτ) = Σ_PQ Z_lP(q) Wᶜ_PQ(q,iτ) Z_l′Q(q)*, Wᶜ_ll′(R,iτ) = Re[Σ_q w_q exp(-iq·R) Wᶜ_ll′(q,iτ)] + ΔWᶜ_Γ,ll′(R,iτ).
(9) V_ll′(q) = Σ_PQ Z_lP(q) [K(q)K(q)†]_PQ Z_l′Q(q)*, V_ll′(R) = Re[Σ_q w_q exp(-iq·R) V_ll′(q)] + ΔV_Γ,ll′(R), Σˣ_ll′(R;σ) = -G^occ_ll′(R,0;σ) V_ll′(R).
(10) Σˣ_ll′(k;σ) = Σ_R exp(ik·R) Σˣ_ll′(R;σ), Σˣ_nkσ = Σ_ll′ ψ_nkσ(r_l)* Σˣ_ll′(k;σ) ψ_nkσ(r_l′).
(11) Σᶜ_ll′(R,-iτ;σ) = -G^occ_ll′(R,iτ;σ) Wᶜ_ll′(R,iτ), Σᶜ_ll′(R,+iτ;σ) = +G^vir_ll′(R,iτ;σ) Wᶜ_ll′(R,iτ).
(12) Σᶜ_ll′(k,±iτ;σ) = Σ_R exp(ik·R) Σᶜ_ll′(R,±iτ;σ), Σᶜ_nkσ(±iτ) = Σ_ll′ ψ_nkσ(r_l)* Σᶜ_ll′(k,±iτ;σ) ψ_nkσ(r_l′), Σᶜ_nkσ(iω) = ∫_0^∞ dτ [exp(iωτ)Σᶜ_nkσ(+iτ)+exp(-iωτ)Σᶜ_nkσ(-iτ)]. Analytic continuation gives Σᶜ at real energies for the quasiparticle equation.
(13) ε^GW_nkσ = ε^DFT_nkσ + Σˣ_nkσ + Re Σᶜ_nkσ(ε^GW_nkσ) - vˣᶜ_nkσ. evGW₀ updates the energies entering G while retaining Wᶜ from the first cycle.
Γ-point corrections are included as ΔWᶜ_Γ and ΔV_Γ; their construction is not shown here: compute_G_and_chi prepares ΔWᶜ_Γ through gamma_point_correction. compute_Sigma_c_from_W_PQ adds it to Wᶜ before forming Σᶜ. compute_Sigma_x_from_V computes and applies ΔV_Γ via singular_head_cell_factors.
| qs_env | Quickstep environment containing the converged SCF state. |
| bs_env | Band-structure settings, orbitals, and quasiparticle results. |
Definition at line 110 of file gw_ri_rs_small_cell_full_kp.F.