Line data Source code
1 : !--------------------------------------------------------------------------------------------------!
2 : ! CP2K: A general program to perform molecular dynamics simulations !
3 : ! Copyright 2000-2026 CP2K developers group <https://cp2k.org> !
4 : ! !
5 : ! SPDX-License-Identifier: GPL-2.0-or-later !
6 : !--------------------------------------------------------------------------------------------------!
7 :
8 : ! **************************************************************************************************
9 : !> \brief Routines needed for cubic-scaling RPA and SOS-Laplace-MP2 forces
10 : !> \author Augustin Bussy
11 : ! **************************************************************************************************
12 : MODULE rpa_im_time_force_methods
13 : USE admm_methods, ONLY: admm_projection_derivative
14 : USE admm_types, ONLY: admm_type,&
15 : get_admm_env
16 : USE ao_util, ONLY: exp_radius_very_extended
17 : USE atomic_kind_types, ONLY: atomic_kind_type,&
18 : get_atomic_kind_set
19 : USE basis_set_types, ONLY: gto_basis_set_p_type,&
20 : gto_basis_set_type
21 : USE bibliography, ONLY: Bussy2023,&
22 : cite_reference
23 : USE cell_types, ONLY: cell_type,&
24 : pbc
25 : USE cp_blacs_env, ONLY: cp_blacs_env_type
26 : USE cp_cfm_types, ONLY: cp_cfm_type
27 : USE cp_control_types, ONLY: dft_control_type
28 : USE cp_dbcsr_api, ONLY: &
29 : dbcsr_add, dbcsr_clear, dbcsr_complete_redistribute, dbcsr_copy, dbcsr_create, &
30 : dbcsr_distribution_new, dbcsr_distribution_release, dbcsr_distribution_type, &
31 : dbcsr_get_block_p, dbcsr_iterator_blocks_left, dbcsr_iterator_next_block, &
32 : dbcsr_iterator_start, dbcsr_iterator_stop, dbcsr_iterator_type, dbcsr_multiply, &
33 : dbcsr_p_type, dbcsr_release, dbcsr_scale, dbcsr_set, dbcsr_type, dbcsr_type_antisymmetric, &
34 : dbcsr_type_no_symmetry, dbcsr_type_symmetric
35 : USE cp_dbcsr_cholesky, ONLY: cp_dbcsr_cholesky_decompose,&
36 : cp_dbcsr_cholesky_invert
37 : USE cp_dbcsr_contrib, ONLY: dbcsr_add_on_diag,&
38 : dbcsr_frobenius_norm
39 : USE cp_dbcsr_diag, ONLY: cp_dbcsr_power
40 : USE cp_dbcsr_operations, ONLY: copy_dbcsr_to_fm,&
41 : copy_fm_to_dbcsr,&
42 : cp_dbcsr_dist2d_to_dist,&
43 : cp_dbcsr_sm_fm_multiply,&
44 : dbcsr_allocate_matrix_set,&
45 : dbcsr_deallocate_matrix_set
46 : USE cp_eri_mme_interface, ONLY: cp_eri_mme_update_local_counts
47 : USE cp_fm_struct, ONLY: cp_fm_struct_create,&
48 : cp_fm_struct_release,&
49 : cp_fm_struct_type
50 : USE cp_fm_types, ONLY: cp_fm_create,&
51 : cp_fm_release,&
52 : cp_fm_set_all,&
53 : cp_fm_to_fm,&
54 : cp_fm_type
55 : USE dbt_api, ONLY: &
56 : dbt_batched_contract_finalize, dbt_batched_contract_init, dbt_clear, dbt_contract, &
57 : dbt_copy, dbt_copy_matrix_to_tensor, dbt_copy_tensor_to_matrix, dbt_create, dbt_destroy, &
58 : dbt_filter, dbt_get_info, dbt_mp_environ_pgrid, dbt_pgrid_create, dbt_pgrid_destroy, &
59 : dbt_pgrid_type, dbt_scale, dbt_type
60 : USE distribution_2d_types, ONLY: distribution_2d_type
61 : USE gaussian_gridlevels, ONLY: gaussian_gridlevel
62 : USE hfx_admm_utils, ONLY: tddft_hfx_matrix
63 : USE hfx_derivatives, ONLY: derivatives_four_center
64 : USE hfx_exx, ONLY: add_exx_to_rhs
65 : USE hfx_ri, ONLY: get_2c_der_force,&
66 : get_force_from_3c_trace,&
67 : get_idx_to_atom,&
68 : hfx_ri_update_forces
69 : USE hfx_types, ONLY: alloc_containers,&
70 : block_ind_type,&
71 : dealloc_containers,&
72 : hfx_compression_type,&
73 : hfx_type
74 : USE input_constants, ONLY: do_admm_aux_exch_func_none,&
75 : do_eri_gpw,&
76 : do_eri_mme,&
77 : do_potential_id,&
78 : ri_rpa_method_gpw
79 : USE input_section_types, ONLY: section_vals_get,&
80 : section_vals_get_subs_vals,&
81 : section_vals_type,&
82 : section_vals_val_get
83 : USE iterate_matrix, ONLY: matrix_exponential
84 : USE kinds, ONLY: dp,&
85 : int_8
86 : USE libint_2c_3c, ONLY: libint_potential_type
87 : USE machine, ONLY: m_flush,&
88 : m_walltime
89 : USE mathconstants, ONLY: fourpi
90 : USE message_passing, ONLY: mp_cart_type,&
91 : mp_para_env_release,&
92 : mp_para_env_type
93 : USE mp2_eri, ONLY: integrate_set_2c
94 : USE mp2_eri_gpw, ONLY: calc_potential_gpw,&
95 : cleanup_gpw,&
96 : prepare_gpw,&
97 : virial_gpw_potential
98 : USE mp2_types, ONLY: mp2_type
99 : USE orbital_pointers, ONLY: ncoset
100 : USE parallel_gemm_api, ONLY: parallel_gemm
101 : USE particle_methods, ONLY: get_particle_set
102 : USE particle_types, ONLY: particle_type
103 : USE pw_env_types, ONLY: pw_env_get,&
104 : pw_env_type
105 : USE pw_methods, ONLY: pw_axpy,&
106 : pw_copy,&
107 : pw_integral_ab,&
108 : pw_scale,&
109 : pw_transfer,&
110 : pw_zero
111 : USE pw_poisson_methods, ONLY: pw_poisson_solve
112 : USE pw_poisson_types, ONLY: pw_poisson_type
113 : USE pw_pool_types, ONLY: pw_pool_type
114 : USE pw_types, ONLY: pw_c1d_gs_type,&
115 : pw_r3d_rs_type
116 : USE qs_collocate_density, ONLY: calculate_rho_elec,&
117 : collocate_function
118 : USE qs_core_matrices, ONLY: core_matrices,&
119 : kinetic_energy_matrix
120 : USE qs_density_matrices, ONLY: calculate_whz_matrix
121 : USE qs_environment_types, ONLY: get_qs_env,&
122 : qs_environment_type,&
123 : set_qs_env
124 : USE qs_force_types, ONLY: qs_force_type
125 : USE qs_fxc, ONLY: qs_fxc_create
126 : USE qs_integral_utils, ONLY: basis_set_list_setup
127 : USE qs_integrate_potential, ONLY: integrate_pgf_product,&
128 : integrate_v_core_rspace,&
129 : integrate_v_rspace
130 : USE qs_interactions, ONLY: init_interaction_radii_orb_basis
131 : USE qs_kind_types, ONLY: qs_kind_type
132 : USE qs_ks_methods, ONLY: calc_rho_tot_gspace
133 : USE qs_ks_reference, ONLY: ks_ref_potential
134 : USE qs_ks_types, ONLY: set_ks_env
135 : USE qs_linres_types, ONLY: linres_control_type
136 : USE qs_matrix_w, ONLY: compute_matrix_w
137 : USE qs_mo_types, ONLY: get_mo_set,&
138 : mo_set_type
139 : USE qs_neighbor_list_types, ONLY: neighbor_list_set_p_type,&
140 : release_neighbor_list_sets
141 : USE qs_overlap, ONLY: build_overlap_matrix
142 : USE qs_p_env_methods, ONLY: p_env_create,&
143 : p_env_psi0_changed
144 : USE qs_p_env_types, ONLY: p_env_release,&
145 : qs_p_env_type
146 : USE qs_rho_atom_types, ONLY: rho_atom_type
147 : USE qs_rho_types, ONLY: qs_rho_create,&
148 : qs_rho_get,&
149 : qs_rho_set,&
150 : qs_rho_type
151 : USE qs_tensors, ONLY: &
152 : build_2c_derivatives, build_2c_integrals, build_2c_neighbor_lists, build_3c_derivatives, &
153 : build_3c_neighbor_lists, calc_2c_virial, calc_3c_virial, compress_tensor, &
154 : decompress_tensor, get_tensor_occupancy, neighbor_list_3c_destroy
155 : USE qs_tensors_types, ONLY: create_2c_tensor,&
156 : create_3c_tensor,&
157 : create_tensor_batches,&
158 : distribution_3d_create,&
159 : distribution_3d_type,&
160 : neighbor_list_3c_type
161 : USE realspace_grid_types, ONLY: map_gaussian_here,&
162 : realspace_grid_type
163 : USE response_solver, ONLY: response_equation_new
164 : USE rpa_im_time, ONLY: compute_mat_dm_global,&
165 : propagator_sector_occupied,&
166 : propagator_sector_virtual
167 : USE rpa_im_time_force_types, ONLY: im_time_force_type
168 : USE rs_pw_interface, ONLY: potential_pw2rs
169 : USE task_list_types, ONLY: task_list_type
170 : USE time_frequency_grids, ONLY: time_frequency_grid_type
171 : USE virial_types, ONLY: virial_type
172 : #include "./base/base_uses.f90"
173 :
174 : IMPLICIT NONE
175 :
176 : PRIVATE
177 :
178 : CHARACTER(len=*), PARAMETER, PRIVATE :: moduleN = 'rpa_im_time_force_methods'
179 :
180 : PUBLIC :: init_im_time_forces, calc_laplace_loop_forces, calc_post_loop_forces, &
181 : keep_initial_quad, calc_rpa_loop_forces
182 :
183 : CONTAINS
184 :
185 : ! **************************************************************************************************
186 : !> \brief Initializes and pre-calculates all needed tensors for the forces
187 : !> \param force_data ...
188 : !> \param fm_matrix_PQ ...
189 : !> \param t_3c_M the 3-center M tensor to be used as a template
190 : !> \param unit_nr ...
191 : !> \param mp2_env ...
192 : !> \param qs_env ...
193 : ! **************************************************************************************************
194 50 : SUBROUTINE init_im_time_forces(force_data, fm_matrix_PQ, t_3c_M, unit_nr, mp2_env, qs_env)
195 :
196 : TYPE(im_time_force_type), INTENT(INOUT) :: force_data
197 : TYPE(cp_fm_type), INTENT(IN) :: fm_matrix_PQ
198 : TYPE(dbt_type), INTENT(INOUT) :: t_3c_M
199 : INTEGER, INTENT(IN) :: unit_nr
200 : TYPE(mp2_type) :: mp2_env
201 : TYPE(qs_environment_type), POINTER :: qs_env
202 :
203 : CHARACTER(LEN=*), PARAMETER :: routineN = 'init_im_time_forces'
204 :
205 : INTEGER :: handle, i_mem, i_xyz, ibasis, ispin, &
206 : n_dependent, n_mem, n_rep, natom, &
207 : nkind, nspins
208 : INTEGER(int_8) :: nze, nze_tot
209 50 : INTEGER, ALLOCATABLE, DIMENSION(:) :: dist1, dist2, dist_AO_1, dist_AO_2, &
210 50 : dist_RI, dummy_end, dummy_start, &
211 100 : end_blocks, sizes_AO, sizes_RI, &
212 50 : start_blocks
213 : INTEGER, DIMENSION(2) :: pdims_t2c
214 : INTEGER, DIMENSION(3) :: nblks_total, pcoord, pdims, pdims_t3c
215 100 : INTEGER, DIMENSION(:), POINTER :: col_bsize, row_bsize
216 : LOGICAL :: do_periodic, use_virial
217 : REAL(dp) :: compression_factor, eps_pgf_orb, &
218 : eps_pgf_orb_old, memory, occ
219 : TYPE(cell_type), POINTER :: cell
220 : TYPE(cp_blacs_env_type), POINTER :: blacs_env
221 : TYPE(dbcsr_distribution_type) :: dbcsr_dist
222 100 : TYPE(dbcsr_p_type), DIMENSION(:), POINTER :: matrix_s, rho_ao
223 : TYPE(dbcsr_type) :: dbcsr_work, dbcsr_work2, dbcsr_work3
224 100 : TYPE(dbcsr_type), DIMENSION(1) :: t_2c_int_tmp
225 350 : TYPE(dbcsr_type), DIMENSION(1, 3) :: t_2c_der_tmp
226 250 : TYPE(dbt_pgrid_type) :: pgrid_t2c, pgrid_t3c
227 1000 : TYPE(dbt_type) :: t_2c_template, t_2c_tmp, t_3c_template
228 50 : TYPE(dbt_type), ALLOCATABLE, DIMENSION(:, :, :) :: t_3c_der_AO_prv, t_3c_der_RI_prv
229 : TYPE(dft_control_type), POINTER :: dft_control
230 : TYPE(distribution_2d_type), POINTER :: dist_2d
231 : TYPE(distribution_3d_type) :: dist_3d, dist_vir
232 : TYPE(gto_basis_set_p_type), ALLOCATABLE, &
233 50 : DIMENSION(:), TARGET :: basis_set_ao, basis_set_ri_aux
234 : TYPE(gto_basis_set_type), POINTER :: orb_basis, ri_basis
235 : TYPE(libint_potential_type) :: identity_pot
236 50 : TYPE(mp_cart_type) :: mp_comm_t3c, mp_comm_vir
237 : TYPE(mp_para_env_type), POINTER :: para_env
238 : TYPE(neighbor_list_3c_type) :: nl_3c
239 : TYPE(neighbor_list_set_p_type), DIMENSION(:), &
240 50 : POINTER :: nl_2c
241 50 : TYPE(particle_type), DIMENSION(:), POINTER :: particle_set
242 50 : TYPE(qs_kind_type), DIMENSION(:), POINTER :: qs_kind_set
243 : TYPE(qs_rho_type), POINTER :: rho
244 : TYPE(section_vals_type), POINTER :: qs_section
245 : TYPE(virial_type), POINTER :: virial
246 :
247 50 : NULLIFY (dft_control, para_env, particle_set, qs_kind_set, dist_2d, nl_2c, blacs_env, matrix_s, &
248 50 : rho, rho_ao, cell, qs_section, orb_basis, ri_basis, virial)
249 :
250 50 : CALL cite_reference(Bussy2023)
251 :
252 50 : CALL timeset(routineN, handle)
253 :
254 : CALL get_qs_env(qs_env, natom=natom, nkind=nkind, dft_control=dft_control, para_env=para_env, &
255 50 : particle_set=particle_set, qs_kind_set=qs_kind_set, cell=cell, virial=virial)
256 50 : IF (dft_control%qs_control%gapw) THEN
257 0 : CPABORT("Low-scaling RPA/SOS-MP2 forces only available with GPW")
258 : END IF
259 :
260 50 : use_virial = virial%pv_availability .AND. (.NOT. virial%pv_numer)
261 :
262 50 : do_periodic = .FALSE.
263 128 : IF (ANY(cell%perd == 1)) do_periodic = .TRUE.
264 50 : force_data%do_periodic = do_periodic
265 :
266 : !Dealing with the 3-center derivatives
267 50 : pdims_t3c = 0
268 50 : CALL dbt_pgrid_create(para_env, pdims_t3c, pgrid_t3c)
269 :
270 : !Make sure we use the proper QS EPS_PGF_ORB values
271 50 : qs_section => section_vals_get_subs_vals(qs_env%input, "DFT%QS")
272 50 : CALL section_vals_val_get(qs_section, "EPS_PGF_ORB", n_rep_val=n_rep)
273 50 : IF (n_rep /= 0) THEN
274 0 : CALL section_vals_val_get(qs_section, "EPS_PGF_ORB", r_val=eps_pgf_orb)
275 : ELSE
276 50 : CALL section_vals_val_get(qs_section, "EPS_DEFAULT", r_val=eps_pgf_orb)
277 50 : eps_pgf_orb = SQRT(eps_pgf_orb)
278 : END IF
279 50 : eps_pgf_orb_old = dft_control%qs_control%eps_pgf_orb
280 :
281 200 : ALLOCATE (sizes_RI(natom), sizes_AO(natom))
282 400 : ALLOCATE (basis_set_ri_aux(nkind), basis_set_ao(nkind))
283 50 : CALL basis_set_list_setup(basis_set_ri_aux, "RI_AUX", qs_kind_set)
284 50 : CALL get_particle_set(particle_set, qs_kind_set, nsgf=sizes_RI, basis=basis_set_ri_aux)
285 50 : CALL basis_set_list_setup(basis_set_ao, "ORB", qs_kind_set)
286 50 : CALL get_particle_set(particle_set, qs_kind_set, nsgf=sizes_AO, basis=basis_set_ao)
287 :
288 150 : DO ibasis = 1, SIZE(basis_set_ao)
289 100 : orb_basis => basis_set_ao(ibasis)%gto_basis_set
290 100 : CALL init_interaction_radii_orb_basis(orb_basis, eps_pgf_orb)
291 100 : ri_basis => basis_set_ri_aux(ibasis)%gto_basis_set
292 150 : CALL init_interaction_radii_orb_basis(ri_basis, eps_pgf_orb)
293 : END DO
294 :
295 : CALL create_3c_tensor(t_3c_template, dist_RI, dist_AO_1, dist_AO_2, pgrid_t3c, &
296 50 : sizes_RI, sizes_AO, sizes_AO, map1=[1], map2=[2, 3], name="der (RI AO | AO)")
297 :
298 1550 : ALLOCATE (t_3c_der_RI_prv(1, 1, 3), t_3c_der_AO_prv(1, 1, 3))
299 200 : DO i_xyz = 1, 3
300 150 : CALL dbt_create(t_3c_template, t_3c_der_RI_prv(1, 1, i_xyz))
301 200 : CALL dbt_create(t_3c_template, t_3c_der_AO_prv(1, 1, i_xyz))
302 : END DO
303 :
304 50 : IF (use_virial) THEN
305 52 : ALLOCATE (force_data%t_3c_virial, force_data%t_3c_virial_split)
306 4 : CALL dbt_create(t_3c_template, force_data%t_3c_virial)
307 4 : CALL dbt_create(t_3c_M, force_data%t_3c_virial_split)
308 : END IF
309 50 : CALL dbt_destroy(t_3c_template)
310 :
311 50 : CALL dbt_mp_environ_pgrid(pgrid_t3c, pdims, pcoord)
312 50 : CALL mp_comm_t3c%create(pgrid_t3c%mp_comm_2d, 3, pdims)
313 : CALL distribution_3d_create(dist_3d, dist_RI, dist_AO_1, dist_AO_2, &
314 50 : nkind, particle_set, mp_comm_t3c, own_comm=.TRUE.)
315 :
316 : !In case of virial, we need to store the 3c_nl
317 50 : IF (use_virial) THEN
318 4 : ALLOCATE (force_data%nl_3c)
319 4 : CALL mp_comm_vir%create(pgrid_t3c%mp_comm_2d, 3, pdims)
320 : CALL distribution_3d_create(dist_vir, dist_RI, dist_AO_1, dist_AO_2, &
321 4 : nkind, particle_set, mp_comm_vir, own_comm=.TRUE.)
322 : CALL build_3c_neighbor_lists(force_data%nl_3c, basis_set_ri_aux, basis_set_ao, basis_set_ao, &
323 : dist_vir, mp2_env%ri_metric, "RPA_3c_nl", qs_env, op_pos=1, &
324 4 : sym_jk=.FALSE., own_dist=.TRUE.)
325 : END IF
326 :
327 : CALL build_3c_neighbor_lists(nl_3c, basis_set_ri_aux, basis_set_ao, basis_set_ao, dist_3d, &
328 : mp2_env%ri_metric, "RPA_3c_nl", qs_env, op_pos=1, sym_jk=.TRUE., &
329 50 : own_dist=.TRUE.)
330 50 : DEALLOCATE (dist_RI, dist_AO_1, dist_AO_2)
331 :
332 : !Prepare the resulting 3c tensors in the format of t_3c_M for compatible traces: (RI|AO AO), split blocks
333 50 : CALL dbt_get_info(t_3c_M, nblks_total=nblks_total)
334 250 : ALLOCATE (force_data%bsizes_RI_split(nblks_total(1)), force_data%bsizes_AO_split(nblks_total(2)))
335 50 : CALL dbt_get_info(t_3c_M, blk_size_1=force_data%bsizes_RI_split, blk_size_2=force_data%bsizes_AO_split)
336 200 : DO i_xyz = 1, 3
337 150 : CALL dbt_create(t_3c_M, force_data%t_3c_der_RI(i_xyz))
338 200 : CALL dbt_create(t_3c_M, force_data%t_3c_der_AO(i_xyz))
339 : END DO
340 :
341 : !Keep track of atom index corresponding to split blocks
342 100 : ALLOCATE (force_data%idx_to_at_RI(nblks_total(1)))
343 50 : CALL get_idx_to_atom(force_data%idx_to_at_RI, force_data%bsizes_RI_split, sizes_RI)
344 :
345 100 : ALLOCATE (force_data%idx_to_at_AO(nblks_total(2)))
346 50 : CALL get_idx_to_atom(force_data%idx_to_at_AO, force_data%bsizes_AO_split, sizes_AO)
347 :
348 50 : n_mem = mp2_env%ri_rpa_im_time%cut_memory
349 50 : CALL create_tensor_batches(sizes_RI, n_mem, dummy_start, dummy_end, start_blocks, end_blocks)
350 50 : DEALLOCATE (dummy_start, dummy_end)
351 :
352 212300 : ALLOCATE (force_data%t_3c_der_AO_comp(n_mem, 3), force_data%t_3c_der_RI_comp(n_mem, 3))
353 1100 : ALLOCATE (force_data%t_3c_der_AO_ind(n_mem, 3), force_data%t_3c_der_RI_ind(n_mem, 3))
354 :
355 50 : memory = 0.0_dp
356 50 : nze_tot = 0
357 150 : DO i_mem = 1, n_mem
358 : CALL build_3c_derivatives(t_3c_der_RI_prv, t_3c_der_AO_prv, mp2_env%ri_rpa_im_time%eps_filter, &
359 : qs_env, nl_3c, basis_set_ri_aux, basis_set_ao, basis_set_ao, &
360 : mp2_env%ri_metric, der_eps=mp2_env%ri_rpa_im_time%eps_filter, op_pos=1, &
361 300 : bounds_i=[start_blocks(i_mem), end_blocks(i_mem)])
362 :
363 450 : DO i_xyz = 1, 3
364 300 : CALL dbt_copy(t_3c_der_RI_prv(1, 1, i_xyz), force_data%t_3c_der_RI(i_xyz), move_data=.TRUE.)
365 300 : CALL dbt_filter(force_data%t_3c_der_RI(i_xyz), mp2_env%ri_rpa_im_time%eps_filter)
366 300 : CALL get_tensor_occupancy(force_data%t_3c_der_RI(i_xyz), nze, occ)
367 300 : nze_tot = nze_tot + nze
368 :
369 300 : CALL alloc_containers(force_data%t_3c_der_RI_comp(i_mem, i_xyz), 1)
370 : CALL compress_tensor(force_data%t_3c_der_RI(i_xyz), force_data%t_3c_der_RI_ind(i_mem, i_xyz)%ind, &
371 300 : force_data%t_3c_der_RI_comp(i_mem, i_xyz), mp2_env%ri_rpa_im_time%eps_compress, memory)
372 300 : CALL dbt_clear(force_data%t_3c_der_RI(i_xyz))
373 :
374 300 : CALL dbt_copy(t_3c_der_AO_prv(1, 1, i_xyz), force_data%t_3c_der_AO(i_xyz), move_data=.TRUE.)
375 300 : CALL dbt_filter(force_data%t_3c_der_AO(i_xyz), mp2_env%ri_rpa_im_time%eps_filter)
376 300 : CALL get_tensor_occupancy(force_data%t_3c_der_AO(i_xyz), nze, occ)
377 300 : nze_tot = nze_tot + nze
378 :
379 300 : CALL alloc_containers(force_data%t_3c_der_AO_comp(i_mem, i_xyz), 1)
380 : CALL compress_tensor(force_data%t_3c_der_AO(i_xyz), force_data%t_3c_der_AO_ind(i_mem, i_xyz)%ind, &
381 300 : force_data%t_3c_der_AO_comp(i_mem, i_xyz), mp2_env%ri_rpa_im_time%eps_compress, memory)
382 1000 : CALL dbt_clear(force_data%t_3c_der_AO(i_xyz))
383 : END DO
384 : END DO
385 50 : CALL neighbor_list_3c_destroy(nl_3c)
386 200 : DO i_xyz = 1, 3
387 150 : CALL dbt_destroy(t_3c_der_RI_prv(1, 1, i_xyz))
388 200 : CALL dbt_destroy(t_3c_der_AO_prv(1, 1, i_xyz))
389 : END DO
390 :
391 50 : CALL para_env%sum(memory)
392 50 : compression_factor = REAL(nze_tot, dp)*1.0E-06_dp*8.0_dp/memory
393 50 : IF (unit_nr > 0) THEN
394 : WRITE (UNIT=unit_nr, FMT="((T3,A,T66,F11.2,A4))") &
395 25 : "MEMORY_INFO| Memory for 3-center derivatives (compressed):", memory, ' MiB'
396 :
397 : WRITE (UNIT=unit_nr, FMT="((T3,A,T60,F21.2))") &
398 25 : "MEMORY_INFO| Compression factor: ", compression_factor
399 : END IF
400 :
401 : !Dealing with the 2-center derivatives
402 50 : CALL get_qs_env(qs_env, distribution_2d=dist_2d, blacs_env=blacs_env, matrix_s=matrix_s)
403 50 : CALL cp_dbcsr_dist2d_to_dist(dist_2d, dbcsr_dist)
404 150 : ALLOCATE (row_bsize(SIZE(sizes_RI)))
405 100 : ALLOCATE (col_bsize(SIZE(sizes_RI)))
406 212 : row_bsize(:) = sizes_RI(:)
407 212 : col_bsize(:) = sizes_RI(:)
408 :
409 50 : pdims_t2c = 0
410 50 : CALL dbt_pgrid_create(para_env, pdims_t2c, pgrid_t2c)
411 : CALL create_2c_tensor(t_2c_template, dist1, dist2, pgrid_t2c, force_data%bsizes_RI_split, &
412 50 : force_data%bsizes_RI_split, name='(RI| RI)')
413 50 : DEALLOCATE (dist1, dist2)
414 :
415 50 : CALL dbcsr_create(t_2c_int_tmp(1), "(P|Q) RPA", dbcsr_dist, dbcsr_type_symmetric, row_bsize, col_bsize)
416 200 : DO i_xyz = 1, 3
417 : CALL dbcsr_create(t_2c_der_tmp(1, i_xyz), "(P|Q) RPA der", dbcsr_dist, &
418 200 : dbcsr_type_antisymmetric, row_bsize, col_bsize)
419 : END DO
420 :
421 50 : IF (use_virial) THEN
422 4 : ALLOCATE (force_data%RI_virial_pot, force_data%RI_virial_met)
423 : CALL dbcsr_create(force_data%RI_virial_pot, "RI_virial", dbcsr_dist, &
424 4 : dbcsr_type_no_symmetry, row_bsize, col_bsize)
425 : CALL dbcsr_create(force_data%RI_virial_met, "RI_virial", dbcsr_dist, &
426 4 : dbcsr_type_no_symmetry, row_bsize, col_bsize)
427 : END IF
428 :
429 : ! Main (P|Q) integrals and derivatives
430 : ! Integrals are passed as a full matrix => convert to DBCSR
431 50 : CALL dbcsr_create(dbcsr_work, template=t_2c_int_tmp(1))
432 50 : CALL copy_fm_to_dbcsr(fm_matrix_PQ, dbcsr_work)
433 :
434 : ! We need the +/- square root of (P|Q)
435 50 : CALL dbcsr_create(dbcsr_work2, template=t_2c_int_tmp(1))
436 50 : CALL dbcsr_create(dbcsr_work3, template=t_2c_int_tmp(1))
437 50 : CALL dbcsr_copy(dbcsr_work2, dbcsr_work)
438 50 : CALL cp_dbcsr_power(dbcsr_work, -0.5_dp, 1.0E-7_dp, n_dependent, para_env, blacs_env) !1.0E-7 ev qunenching thresh
439 :
440 : ! Transfer to tensor format with split blocks
441 50 : CALL dbt_create(dbcsr_work, t_2c_tmp)
442 50 : CALL dbt_copy_matrix_to_tensor(dbcsr_work, t_2c_tmp)
443 50 : CALL dbt_create(t_2c_template, force_data%t_2c_pot_msqrt)
444 50 : CALL dbt_copy(t_2c_tmp, force_data%t_2c_pot_msqrt, move_data=.TRUE.)
445 50 : CALL dbt_filter(force_data%t_2c_pot_msqrt, mp2_env%ri_rpa_im_time%eps_filter)
446 :
447 50 : CALL dbcsr_multiply('N', 'N', 1.0_dp, dbcsr_work2, dbcsr_work, 0.0_dp, dbcsr_work3)
448 50 : CALL dbt_copy_matrix_to_tensor(dbcsr_work3, t_2c_tmp)
449 50 : CALL dbt_create(t_2c_template, force_data%t_2c_pot_psqrt)
450 50 : CALL dbt_copy(t_2c_tmp, force_data%t_2c_pot_psqrt, move_data=.TRUE.)
451 50 : CALL dbt_filter(force_data%t_2c_pot_psqrt, mp2_env%ri_rpa_im_time%eps_filter)
452 50 : CALL dbt_destroy(t_2c_tmp)
453 50 : CALL dbcsr_release(dbcsr_work2)
454 50 : CALL dbcsr_release(dbcsr_work3)
455 50 : CALL dbcsr_clear(dbcsr_work)
456 :
457 : ! Deal with the 2c potential derivatives. Only precompute if not in PBCs
458 50 : IF (.NOT. do_periodic) THEN
459 : CALL build_2c_neighbor_lists(nl_2c, basis_set_ri_aux, basis_set_ri_aux, mp2_env%potential_parameter, &
460 26 : "RPA_2c_nl_pot", qs_env, sym_ij=.TRUE., dist_2d=dist_2d)
461 : CALL build_2c_derivatives(t_2c_der_tmp, mp2_env%ri_rpa_im_time%eps_filter, qs_env, nl_2c, &
462 26 : basis_set_ri_aux, basis_set_ri_aux, mp2_env%potential_parameter)
463 26 : CALL release_neighbor_list_sets(nl_2c)
464 :
465 104 : DO i_xyz = 1, 3
466 78 : CALL dbt_create(t_2c_der_tmp(1, i_xyz), t_2c_tmp)
467 78 : CALL dbt_copy_matrix_to_tensor(t_2c_der_tmp(1, i_xyz), t_2c_tmp)
468 78 : CALL dbt_create(t_2c_template, force_data%t_2c_der_pot(i_xyz))
469 78 : CALL dbt_copy(t_2c_tmp, force_data%t_2c_der_pot(i_xyz), move_data=.TRUE.)
470 78 : CALL dbt_filter(force_data%t_2c_der_pot(i_xyz), mp2_env%ri_rpa_im_time%eps_filter)
471 78 : CALL dbt_destroy(t_2c_tmp)
472 104 : CALL dbcsr_clear(t_2c_der_tmp(1, i_xyz))
473 : END DO
474 :
475 26 : IF (use_virial) THEN
476 : CALL build_2c_neighbor_lists(force_data%nl_2c_pot, basis_set_ri_aux, basis_set_ri_aux, &
477 : mp2_env%potential_parameter, "RPA_2c_nl_pot", qs_env, &
478 0 : sym_ij=.FALSE., dist_2d=dist_2d)
479 : END IF
480 : END IF
481 : ! Create a G_PQ matrix to collect the terms for the force trace in the periodic case
482 50 : CALL dbcsr_create(force_data%G_PQ, "G_PQ", dbcsr_dist, dbcsr_type_no_symmetry, row_bsize, col_bsize)
483 :
484 : ! we need the RI metric derivatives and the inverse of the integrals
485 : CALL build_2c_neighbor_lists(nl_2c, basis_set_ri_aux, basis_set_ri_aux, mp2_env%ri_metric, &
486 50 : "RPA_2c_nl_metric", qs_env, sym_ij=.TRUE., dist_2d=dist_2d)
487 : CALL build_2c_integrals(t_2c_int_tmp, mp2_env%ri_rpa_im_time%eps_filter, qs_env, nl_2c, &
488 50 : basis_set_ri_aux, basis_set_ri_aux, mp2_env%ri_metric)
489 : CALL build_2c_derivatives(t_2c_der_tmp, mp2_env%ri_rpa_im_time%eps_filter, qs_env, nl_2c, &
490 50 : basis_set_ri_aux, basis_set_ri_aux, mp2_env%ri_metric)
491 50 : CALL release_neighbor_list_sets(nl_2c)
492 :
493 50 : IF (use_virial) THEN
494 : CALL build_2c_neighbor_lists(force_data%nl_2c_met, basis_set_ri_aux, basis_set_ri_aux, &
495 : mp2_env%ri_metric, "RPA_2c_nl_metric", qs_env, sym_ij=.FALSE., &
496 4 : dist_2d=dist_2d)
497 : END IF
498 :
499 50 : CALL dbcsr_copy(dbcsr_work, t_2c_int_tmp(1))
500 50 : CALL cp_dbcsr_cholesky_decompose(dbcsr_work, para_env=para_env, blacs_env=blacs_env)
501 50 : CALL cp_dbcsr_cholesky_invert(dbcsr_work, para_env=para_env, blacs_env=blacs_env, uplo_to_full=.TRUE.)
502 :
503 50 : CALL dbt_create(dbcsr_work, t_2c_tmp)
504 50 : CALL dbt_copy_matrix_to_tensor(dbcsr_work, t_2c_tmp)
505 50 : CALL dbt_create(t_2c_template, force_data%t_2c_inv_metric)
506 50 : CALL dbt_copy(t_2c_tmp, force_data%t_2c_inv_metric, move_data=.TRUE.)
507 50 : CALL dbt_filter(force_data%t_2c_inv_metric, mp2_env%ri_rpa_im_time%eps_filter)
508 50 : CALL dbt_destroy(t_2c_tmp)
509 50 : CALL dbcsr_clear(dbcsr_work)
510 50 : CALL dbcsr_clear(t_2c_int_tmp(1))
511 :
512 200 : DO i_xyz = 1, 3
513 150 : CALL dbt_create(t_2c_der_tmp(1, i_xyz), t_2c_tmp)
514 150 : CALL dbt_copy_matrix_to_tensor(t_2c_der_tmp(1, i_xyz), t_2c_tmp)
515 150 : CALL dbt_create(t_2c_template, force_data%t_2c_der_metric(i_xyz))
516 150 : CALL dbt_copy(t_2c_tmp, force_data%t_2c_der_metric(i_xyz), move_data=.TRUE.)
517 150 : CALL dbt_filter(force_data%t_2c_der_metric(i_xyz), mp2_env%ri_rpa_im_time%eps_filter)
518 150 : CALL dbt_destroy(t_2c_tmp)
519 200 : CALL dbcsr_clear(t_2c_der_tmp(1, i_xyz))
520 : END DO
521 :
522 : !Pre-calculate matrix K = metric^-1 * V^0.5
523 50 : CALL dbt_create(t_2c_template, force_data%t_2c_K)
524 : CALL dbt_contract(1.0_dp, force_data%t_2c_inv_metric, force_data%t_2c_pot_psqrt, &
525 : 0.0_dp, force_data%t_2c_K, &
526 : contract_1=[2], notcontract_1=[1], &
527 : contract_2=[1], notcontract_2=[2], &
528 50 : map_1=[1], map_2=[2], filter_eps=mp2_env%ri_rpa_im_time%eps_filter)
529 :
530 : ! Finally, we need the overlap matrix derivative and the inverse of the integrals
531 50 : CALL dbt_destroy(t_2c_template)
532 50 : CALL dbcsr_release(dbcsr_work)
533 50 : CALL dbcsr_release(t_2c_int_tmp(1))
534 200 : DO i_xyz = 1, 3
535 200 : CALL dbcsr_release(t_2c_der_tmp(1, i_xyz))
536 : END DO
537 :
538 50 : DEALLOCATE (row_bsize, col_bsize)
539 150 : ALLOCATE (row_bsize(SIZE(sizes_AO)))
540 100 : ALLOCATE (col_bsize(SIZE(sizes_AO)))
541 212 : row_bsize(:) = sizes_AO(:)
542 212 : col_bsize(:) = sizes_AO(:)
543 :
544 : CALL create_2c_tensor(t_2c_template, dist1, dist2, pgrid_t2c, force_data%bsizes_AO_split, &
545 50 : force_data%bsizes_AO_split, name='(AO| AO)')
546 50 : DEALLOCATE (dist1, dist2)
547 :
548 200 : DO i_xyz = 1, 3
549 : CALL dbcsr_create(t_2c_der_tmp(1, i_xyz), "(P|Q) RPA der", dbcsr_dist, &
550 200 : dbcsr_type_antisymmetric, row_bsize, col_bsize)
551 : END DO
552 :
553 50 : identity_pot%potential_type = do_potential_id
554 : CALL build_2c_neighbor_lists(nl_2c, basis_set_ao, basis_set_ao, identity_pot, &
555 50 : "RPA_2c_nl_metric", qs_env, sym_ij=.TRUE., dist_2d=dist_2d)
556 : CALL build_2c_derivatives(t_2c_der_tmp, mp2_env%ri_rpa_im_time%eps_filter, qs_env, nl_2c, &
557 50 : basis_set_ao, basis_set_ao, identity_pot)
558 50 : CALL release_neighbor_list_sets(nl_2c)
559 :
560 50 : IF (use_virial) THEN
561 : CALL build_2c_neighbor_lists(force_data%nl_2c_ovlp, basis_set_ao, basis_set_ao, identity_pot, &
562 4 : "RPA_2c_nl_metric", qs_env, sym_ij=.FALSE., dist_2d=dist_2d)
563 : END IF
564 :
565 50 : CALL dbcsr_create(force_data%inv_ovlp, template=matrix_s(1)%matrix)
566 50 : CALL dbcsr_copy(force_data%inv_ovlp, matrix_s(1)%matrix)
567 50 : CALL cp_dbcsr_cholesky_decompose(force_data%inv_ovlp, para_env=para_env, blacs_env=blacs_env)
568 50 : CALL cp_dbcsr_cholesky_invert(force_data%inv_ovlp, para_env=para_env, blacs_env=blacs_env, uplo_to_full=.TRUE.)
569 :
570 200 : DO i_xyz = 1, 3
571 150 : CALL dbt_create(t_2c_der_tmp(1, i_xyz), t_2c_tmp)
572 150 : CALL dbt_copy_matrix_to_tensor(t_2c_der_tmp(1, i_xyz), t_2c_tmp)
573 150 : CALL dbt_create(t_2c_template, force_data%t_2c_der_ovlp(i_xyz))
574 150 : CALL dbt_copy(t_2c_tmp, force_data%t_2c_der_ovlp(i_xyz), move_data=.TRUE.)
575 150 : CALL dbt_filter(force_data%t_2c_der_ovlp(i_xyz), mp2_env%ri_rpa_im_time%eps_filter)
576 150 : CALL dbt_destroy(t_2c_tmp)
577 200 : CALL dbcsr_clear(t_2c_der_tmp(1, i_xyz))
578 : END DO
579 :
580 : !Create the rest of the 2-center AO tensors
581 50 : nspins = dft_control%nspins
582 324 : ALLOCATE (force_data%P_virt(nspins), force_data%P_occ(nspins))
583 274 : ALLOCATE (force_data%sum_YP_tau(nspins), force_data%sum_O_tau(nspins))
584 112 : DO ispin = 1, nspins
585 62 : ALLOCATE (force_data%P_virt(ispin)%matrix, force_data%P_occ(ispin)%matrix)
586 62 : ALLOCATE (force_data%sum_YP_tau(ispin)%matrix, force_data%sum_O_tau(ispin)%matrix)
587 62 : CALL dbcsr_create(force_data%P_virt(ispin)%matrix, template=matrix_s(1)%matrix)
588 62 : CALL dbcsr_create(force_data%P_occ(ispin)%matrix, template=matrix_s(1)%matrix)
589 62 : CALL dbcsr_create(force_data%sum_O_tau(ispin)%matrix, template=matrix_s(1)%matrix)
590 62 : CALL dbcsr_create(force_data%sum_YP_tau(ispin)%matrix, template=matrix_s(1)%matrix)
591 :
592 62 : CALL dbcsr_copy(force_data%sum_O_tau(ispin)%matrix, matrix_s(1)%matrix)
593 62 : CALL dbcsr_copy(force_data%sum_YP_tau(ispin)%matrix, matrix_s(1)%matrix)
594 :
595 62 : CALL dbcsr_set(force_data%sum_O_tau(ispin)%matrix, 0.0_dp)
596 112 : CALL dbcsr_set(force_data%sum_YP_tau(ispin)%matrix, 0.0_dp)
597 : END DO
598 :
599 : !Populate the density matrices: 1 = P_virt*S +P_occ*S ==> P_virt = S^-1 - P_occ
600 50 : CALL get_qs_env(qs_env, rho=rho)
601 50 : CALL qs_rho_get(rho, rho_ao=rho_ao)
602 50 : CALL dbcsr_copy(force_data%P_occ(1)%matrix, rho_ao(1)%matrix)
603 50 : IF (nspins == 1) THEN
604 38 : CALL dbcsr_scale(force_data%P_occ(1)%matrix, 0.5_dp) !because double occupency
605 : ELSE
606 12 : CALL dbcsr_copy(force_data%P_occ(2)%matrix, rho_ao(2)%matrix)
607 : END IF
608 112 : DO ispin = 1, nspins
609 62 : CALL dbcsr_copy(force_data%P_virt(ispin)%matrix, force_data%inv_ovlp)
610 112 : CALL dbcsr_add(force_data%P_virt(ispin)%matrix, force_data%P_occ(ispin)%matrix, 1.0_dp, -1.0_dp)
611 : END DO
612 :
613 150 : DO ibasis = 1, SIZE(basis_set_ao)
614 100 : orb_basis => basis_set_ao(ibasis)%gto_basis_set
615 100 : CALL init_interaction_radii_orb_basis(orb_basis, eps_pgf_orb_old)
616 100 : ri_basis => basis_set_ri_aux(ibasis)%gto_basis_set
617 150 : CALL init_interaction_radii_orb_basis(ri_basis, eps_pgf_orb_old)
618 : END DO
619 :
620 50 : CALL dbt_destroy(t_2c_template)
621 50 : CALL dbcsr_release(dbcsr_work)
622 200 : DO i_xyz = 1, 3
623 200 : CALL dbcsr_release(t_2c_der_tmp(1, i_xyz))
624 : END DO
625 50 : DEALLOCATE (row_bsize, col_bsize)
626 50 : CALL dbt_pgrid_destroy(pgrid_t3c)
627 50 : CALL dbt_pgrid_destroy(pgrid_t2c)
628 50 : CALL dbcsr_distribution_release(dbcsr_dist)
629 50 : CALL timestop(handle)
630 :
631 600 : END SUBROUTINE init_im_time_forces
632 :
633 : ! **************************************************************************************************
634 : !> \brief Updates the cubic-scaling SOS-Laplace-MP2 contribution to the forces at each quadrature point
635 : !> \param force_data ...
636 : !> \param mat_P_omega ...
637 : !> \param t_3c_M ...
638 : !> \param t_3c_O ...
639 : !> \param t_3c_O_compressed ...
640 : !> \param t_3c_O_ind ...
641 : !> \param cfm_mo_coeff ...
642 : !> \param homo ...
643 : !> \param starts_array_mc ...
644 : !> \param ends_array_mc ...
645 : !> \param starts_array_mc_block ...
646 : !> \param ends_array_mc_block ...
647 : !> \param nmo ...
648 : !> \param Eigenval ...
649 : !> \param grid ...
650 : !> \param cut_memory ...
651 : !> \param Pspin ...
652 : !> \param Qspin ...
653 : !> \param open_shell ...
654 : !> \param unit_nr ...
655 : !> \param dbcsr_time ...
656 : !> \param dbcsr_nflop ...
657 : !> \param mp2_env ...
658 : !> \param qs_env ...
659 : !> \note In open-shell, we need to take Q from one spin, and everything from the other
660 : ! **************************************************************************************************
661 130 : SUBROUTINE calc_laplace_loop_forces(force_data, mat_P_omega, t_3c_M, t_3c_O, t_3c_O_compressed, &
662 52 : t_3c_O_ind, cfm_mo_coeff, homo, starts_array_mc, ends_array_mc, &
663 26 : starts_array_mc_block, ends_array_mc_block, &
664 26 : nmo, Eigenval, grid, cut_memory, Pspin, Qspin, &
665 : open_shell, unit_nr, dbcsr_time, dbcsr_nflop, mp2_env, qs_env)
666 :
667 : TYPE(im_time_force_type), INTENT(INOUT) :: force_data
668 : TYPE(dbcsr_p_type), DIMENSION(:, :), INTENT(INOUT) :: mat_P_omega
669 : TYPE(dbt_type), INTENT(INOUT) :: t_3c_M, t_3c_O
670 : TYPE(hfx_compression_type), DIMENSION(:) :: t_3c_O_compressed
671 : TYPE(block_ind_type), DIMENSION(:), INTENT(INOUT) :: t_3c_O_ind
672 : TYPE(cp_cfm_type), DIMENSION(:), INTENT(IN) :: cfm_mo_coeff
673 : INTEGER, DIMENSION(:), INTENT(IN) :: homo, starts_array_mc, ends_array_mc, &
674 : starts_array_mc_block, &
675 : ends_array_mc_block
676 : INTEGER, INTENT(IN) :: nmo
677 : REAL(KIND=dp), DIMENSION(:, :), INTENT(IN) :: Eigenval
678 : TYPE(time_frequency_grid_type), INTENT(IN) :: grid
679 : INTEGER, INTENT(IN) :: cut_memory, Pspin, Qspin
680 : LOGICAL, INTENT(IN) :: open_shell
681 : INTEGER, INTENT(IN) :: unit_nr
682 : REAL(dp), INTENT(INOUT) :: dbcsr_time
683 : INTEGER(int_8), INTENT(INOUT) :: dbcsr_nflop
684 : TYPE(mp2_type) :: mp2_env
685 : TYPE(qs_environment_type), POINTER :: qs_env
686 :
687 : CHARACTER(LEN=*), PARAMETER :: routineN = 'calc_laplace_loop_forces'
688 :
689 : INTEGER :: dummy_int, handle, handle2, i_mem, i_xyz, ibasis, ispin, j_xyz, jquad, k_xyz, &
690 : n_mem_RI, n_rep, natom, nkind, nspins, num_integ_points, unit_nr_dbcsr
691 : INTEGER(int_8) :: flop, nze, nze_ddint, nze_der_AO, &
692 : nze_der_RI, nze_KQK
693 26 : INTEGER, ALLOCATABLE, DIMENSION(:) :: atom_of_kind, batch_blk_end, &
694 26 : batch_blk_start, batch_end_RI, &
695 26 : batch_start_RI, kind_of, mc_ranges, &
696 26 : mc_ranges_RI
697 26 : INTEGER, DIMENSION(:, :), POINTER :: dummy_ptr
698 : LOGICAL :: memory_info, use_virial
699 : REAL(dp) :: eps_filter, eps_pgf_orb, &
700 : eps_pgf_orb_old, fac, occ, occ_ddint, &
701 : occ_der_AO, occ_der_RI, occ_KQK, &
702 : omega, pref, t1, t2, tau
703 : REAL(dp), DIMENSION(3, 3) :: work_virial, work_virial_ovlp
704 26 : TYPE(atomic_kind_type), DIMENSION(:), POINTER :: atomic_kind_set
705 : TYPE(cell_type), POINTER :: cell
706 26 : TYPE(dbcsr_p_type), DIMENSION(:), POINTER :: matrix_ks, matrix_s
707 26 : TYPE(dbcsr_p_type), DIMENSION(:, :, :), POINTER :: propagator
708 : TYPE(dbcsr_type) :: dbcsr_work1, dbcsr_work2, dbcsr_work3, &
709 : exp_occ, exp_virt, R_occ, R_virt, &
710 : virial_ovlp, Y_1, Y_2
711 1274 : TYPE(dbt_type) :: t_2c_AO, t_2c_RI, t_2c_RI_2, t_2c_tmp, t_3c_0, t_3c_1, t_3c_3, t_3c_4, &
712 1274 : t_3c_5, t_3c_6, t_3c_7, t_3c_8, t_3c_help_1, t_3c_help_2, t_3c_ints, t_3c_sparse, &
713 1456 : t_3c_work, t_dm_occ, t_dm_virt, t_KQKT, t_M_occ, t_M_virt, t_Q, t_R_occ, t_R_virt
714 26 : TYPE(dbt_type), ALLOCATABLE, DIMENSION(:) :: t_P
715 : TYPE(dft_control_type), POINTER :: dft_control
716 : TYPE(gto_basis_set_p_type), ALLOCATABLE, &
717 26 : DIMENSION(:), TARGET :: basis_set_ao, basis_set_ri_aux
718 : TYPE(gto_basis_set_type), POINTER :: orb_basis, ri_basis
719 : TYPE(libint_potential_type) :: identity_pot
720 : TYPE(mp_para_env_type), POINTER :: para_env
721 26 : TYPE(particle_type), DIMENSION(:), POINTER :: particle_set
722 26 : TYPE(qs_force_type), DIMENSION(:), POINTER :: force
723 26 : TYPE(qs_kind_type), DIMENSION(:), POINTER :: qs_kind_set
724 : TYPE(section_vals_type), POINTER :: qs_section
725 : TYPE(virial_type), POINTER :: virial
726 :
727 26 : NULLIFY (matrix_s, dummy_ptr, atomic_kind_set, force, matrix_s, matrix_ks)
728 26 : NULLIFY (dft_control, virial, particle_set, cell, para_env, orb_basis, ri_basis, qs_section)
729 26 : NULLIFY (qs_kind_set)
730 :
731 26 : CALL timeset(routineN, handle)
732 :
733 26 : num_integ_points = SIZE(grid%imaginary_time)
734 :
735 26 : NULLIFY (propagator)
736 :
737 : CALL get_qs_env(qs_env, matrix_s=matrix_s, natom=natom, atomic_kind_set=atomic_kind_set, &
738 : force=force, matrix_ks=matrix_ks, dft_control=dft_control, virial=virial, &
739 : particle_set=particle_set, cell=cell, para_env=para_env, nkind=nkind, &
740 26 : qs_kind_set=qs_kind_set)
741 26 : eps_filter = mp2_env%ri_rpa_im_time%eps_filter
742 26 : nspins = dft_control%nspins
743 :
744 26 : memory_info = mp2_env%ri_rpa_im_time%memory_info
745 26 : IF (memory_info) THEN
746 0 : unit_nr_dbcsr = unit_nr
747 : ELSE
748 26 : unit_nr_dbcsr = 0
749 : END IF
750 :
751 26 : use_virial = virial%pv_availability .AND. (.NOT. virial%pv_numer)
752 :
753 26 : IF (use_virial) virial%pv_calculate = .TRUE.
754 :
755 26 : IF (use_virial) THEN
756 2 : qs_section => section_vals_get_subs_vals(qs_env%input, "DFT%QS")
757 2 : CALL section_vals_val_get(qs_section, "EPS_PGF_ORB", n_rep_val=n_rep)
758 2 : IF (n_rep /= 0) THEN
759 0 : CALL section_vals_val_get(qs_section, "EPS_PGF_ORB", r_val=eps_pgf_orb)
760 : ELSE
761 2 : CALL section_vals_val_get(qs_section, "EPS_DEFAULT", r_val=eps_pgf_orb)
762 2 : eps_pgf_orb = SQRT(eps_pgf_orb)
763 : END IF
764 2 : eps_pgf_orb_old = dft_control%qs_control%eps_pgf_orb
765 :
766 16 : ALLOCATE (basis_set_ri_aux(nkind), basis_set_ao(nkind))
767 2 : CALL basis_set_list_setup(basis_set_ri_aux, "RI_AUX", qs_kind_set)
768 2 : CALL basis_set_list_setup(basis_set_ao, "ORB", qs_kind_set)
769 :
770 8 : DO ibasis = 1, SIZE(basis_set_ao)
771 4 : orb_basis => basis_set_ao(ibasis)%gto_basis_set
772 4 : CALL init_interaction_radii_orb_basis(orb_basis, eps_pgf_orb)
773 4 : ri_basis => basis_set_ri_aux(ibasis)%gto_basis_set
774 6 : CALL init_interaction_radii_orb_basis(ri_basis, eps_pgf_orb)
775 : END DO
776 : END IF
777 :
778 : !We follow the general logic of the compute_mat_P_omega routine
779 268 : ALLOCATE (t_P(nspins))
780 26 : CALL dbt_create(force_data%t_2c_K, t_2c_RI)
781 26 : CALL dbt_create(force_data%t_2c_K, t_2c_RI_2)
782 26 : CALL dbt_create(force_data%t_2c_der_ovlp(1), t_2c_AO)
783 :
784 26 : CALL get_atomic_kind_set(atomic_kind_set, kind_of=kind_of, atom_of_kind=atom_of_kind)
785 :
786 : ! Always do the batching of the MO on mu and sigma, such that it is consistent between
787 : ! the occupied and the virtual quantities
788 78 : ALLOCATE (mc_ranges(cut_memory + 1))
789 78 : mc_ranges(:cut_memory) = starts_array_mc_block(:)
790 26 : mc_ranges(cut_memory + 1) = ends_array_mc_block(cut_memory) + 1
791 :
792 : ! Also need some batching on the RI, because it loses sparsity at some point
793 26 : n_mem_RI = cut_memory
794 : CALL create_tensor_batches(force_data%bsizes_RI_split, n_mem_RI, batch_start_RI, batch_end_RI, &
795 26 : batch_blk_start, batch_blk_end)
796 78 : ALLOCATE (mc_ranges_RI(n_mem_RI + 1))
797 78 : mc_ranges_RI(1:n_mem_RI) = batch_blk_start(1:n_mem_RI)
798 26 : mc_ranges_RI(n_mem_RI + 1) = batch_blk_end(n_mem_RI) + 1
799 26 : DEALLOCATE (batch_blk_start, batch_blk_end)
800 :
801 : !Pre-allocate all required tensors and matrices
802 60 : DO ispin = 1, nspins
803 60 : CALL dbt_create(t_2c_RI, t_P(ispin))
804 : END DO
805 26 : CALL dbt_create(t_2c_RI, t_Q)
806 26 : CALL dbt_create(t_2c_RI, t_KQKT)
807 26 : CALL dbt_create(t_2c_AO, t_dm_occ)
808 26 : CALL dbt_create(t_2c_AO, t_dm_virt)
809 :
810 : !note: t_3c_O and t_3c_M have different mappings (map_1d, map_2d)
811 26 : CALL dbt_create(t_3c_O, t_M_occ)
812 26 : CALL dbt_create(t_3c_O, t_M_virt)
813 26 : CALL dbt_create(t_3c_O, t_3c_0)
814 :
815 26 : CALL dbt_create(t_3c_O, t_3c_1)
816 26 : CALL dbt_create(t_3c_O, t_3c_3)
817 26 : CALL dbt_create(t_3c_O, t_3c_4)
818 26 : CALL dbt_create(t_3c_O, t_3c_5)
819 26 : CALL dbt_create(t_3c_M, t_3c_6)
820 26 : CALL dbt_create(t_3c_M, t_3c_7)
821 26 : CALL dbt_create(t_3c_M, t_3c_8)
822 26 : CALL dbt_create(t_3c_M, t_3c_sparse)
823 26 : CALL dbt_create(t_3c_O, t_3c_help_1)
824 26 : CALL dbt_create(t_3c_O, t_3c_help_2)
825 26 : CALL dbt_create(t_2c_AO, t_R_occ)
826 26 : CALL dbt_create(t_2c_AO, t_R_virt)
827 26 : CALL dbt_create(t_3c_M, t_3c_ints)
828 26 : CALL dbt_create(t_3c_M, t_3c_work)
829 :
830 : !Pre-define the sparsity of t_3c_4 as a function of the derivatives
831 26 : occ_der_AO = 0; nze_der_AO = 0
832 26 : occ_der_RI = 0; nze_der_RI = 0
833 104 : DO i_xyz = 1, 3
834 260 : DO i_mem = 1, cut_memory
835 : CALL decompress_tensor(force_data%t_3c_der_RI(i_xyz), force_data%t_3c_der_RI_ind(i_mem, i_xyz)%ind, &
836 156 : force_data%t_3c_der_RI_comp(i_mem, i_xyz), mp2_env%ri_rpa_im_time%eps_compress)
837 156 : CALL get_tensor_occupancy(force_data%t_3c_der_RI(i_xyz), nze, occ)
838 156 : occ_der_RI = occ_der_RI + occ
839 156 : nze_der_RI = nze_der_RI + nze
840 156 : CALL dbt_copy(force_data%t_3c_der_RI(i_xyz), t_3c_sparse, summation=.TRUE., move_data=.TRUE.)
841 :
842 : CALL decompress_tensor(force_data%t_3c_der_AO(i_xyz), force_data%t_3c_der_AO_ind(i_mem, i_xyz)%ind, &
843 156 : force_data%t_3c_der_AO_comp(i_mem, i_xyz), mp2_env%ri_rpa_im_time%eps_compress)
844 156 : CALL get_tensor_occupancy(force_data%t_3c_der_AO(i_xyz), nze, occ)
845 156 : occ_der_AO = occ_der_AO + occ
846 156 : nze_der_AO = nze_der_AO + nze
847 156 : CALL dbt_copy(force_data%t_3c_der_AO(i_xyz), t_3c_sparse, order=[1, 3, 2], summation=.TRUE.)
848 546 : CALL dbt_copy(force_data%t_3c_der_AO(i_xyz), t_3c_sparse, summation=.TRUE., move_data=.TRUE.)
849 : END DO
850 : END DO
851 26 : occ_der_RI = occ_der_RI/3.0_dp
852 26 : occ_der_AO = occ_der_AO/3.0_dp
853 26 : nze_der_RI = nze_der_RI/3
854 26 : nze_der_AO = nze_der_AO/3
855 :
856 26 : CALL dbcsr_create(R_occ, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
857 26 : CALL dbcsr_create(R_virt, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
858 26 : CALL dbcsr_create(dbcsr_work1, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
859 26 : CALL dbcsr_create(dbcsr_work2, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
860 26 : CALL dbcsr_create(dbcsr_work3, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
861 26 : CALL dbcsr_create(exp_occ, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
862 26 : CALL dbcsr_create(exp_virt, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
863 26 : IF (use_virial) CALL dbcsr_create(virial_ovlp, template=dbcsr_work1)
864 :
865 26 : CALL dbt_batched_contract_init(t_3c_0, batch_range_2=mc_ranges, batch_range_3=mc_ranges)
866 26 : CALL dbt_batched_contract_init(t_3c_1, batch_range_2=mc_ranges, batch_range_3=mc_ranges)
867 26 : CALL dbt_batched_contract_init(t_3c_3, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
868 26 : CALL dbt_batched_contract_init(t_M_occ, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
869 26 : CALL dbt_batched_contract_init(t_M_virt, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
870 :
871 26 : CALL dbt_batched_contract_init(t_3c_ints, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
872 26 : CALL dbt_batched_contract_init(t_3c_work, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
873 :
874 : CALL dbt_batched_contract_init(t_3c_4, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
875 26 : batch_range_3=mc_ranges)
876 : CALL dbt_batched_contract_init(t_3c_5, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
877 26 : batch_range_3=mc_ranges)
878 : CALL dbt_batched_contract_init(t_3c_6, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
879 26 : batch_range_3=mc_ranges)
880 : CALL dbt_batched_contract_init(t_3c_7, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
881 26 : batch_range_3=mc_ranges)
882 : CALL dbt_batched_contract_init(t_3c_8, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
883 26 : batch_range_3=mc_ranges)
884 : CALL dbt_batched_contract_init(t_3c_sparse, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
885 26 : batch_range_3=mc_ranges)
886 :
887 26 : work_virial = 0.0_dp
888 26 : work_virial_ovlp = 0.0_dp
889 104 : DO jquad = 1, num_integ_points
890 78 : tau = grid%imaginary_time(jquad)
891 78 : omega = grid%time_weights_at_zero_frequency(jquad)
892 78 : fac = -2.0_dp*omega*mp2_env%scale_S
893 78 : IF (open_shell) fac = 0.5_dp*fac
894 78 : occ_ddint = 0; nze_ddint = 0
895 :
896 78 : CALL para_env%sync()
897 78 : t1 = m_walltime()
898 :
899 : !Deal with the force contributions where there is no explicit 3-center quantities, i.e. the
900 : !forces due to the metric and potential derivatives
901 180 : DO ispin = 1, nspins
902 102 : CALL dbt_create(mat_P_omega(jquad, ispin)%matrix, t_2c_tmp)
903 102 : CALL dbt_copy_matrix_to_tensor(mat_P_omega(jquad, ispin)%matrix, t_2c_tmp)
904 102 : CALL dbt_copy(t_2c_tmp, t_P(ispin), move_data=.TRUE.)
905 102 : CALL dbt_filter(t_P(ispin), eps_filter)
906 180 : CALL dbt_destroy(t_2c_tmp)
907 : END DO
908 :
909 : !Q = K^T*P*K, open-shell: Q is from one spin, everything else from the other
910 : CALL dbt_contract(1.0_dp, t_P(Qspin), force_data%t_2c_K, 0.0_dp, t_2c_RI, &
911 : contract_1=[2], notcontract_1=[1], &
912 : contract_2=[1], notcontract_2=[2], &
913 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
914 78 : flop=flop, unit_nr=unit_nr_dbcsr)
915 78 : dbcsr_nflop = dbcsr_nflop + flop
916 : CALL dbt_contract(1.0_dp, force_data%t_2c_K, t_2c_RI, 0.0_dp, t_Q, &
917 : contract_1=[1], notcontract_1=[2], &
918 : contract_2=[1], notcontract_2=[2], &
919 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
920 78 : flop=flop, unit_nr=unit_nr_dbcsr)
921 78 : dbcsr_nflop = dbcsr_nflop + flop
922 78 : CALL dbt_clear(t_2c_RI)
923 :
924 : CALL perform_2c_ops(force, t_KQKT, force_data, fac, t_Q, t_P(Pspin), t_2c_RI, t_2c_RI_2, &
925 78 : use_virial, atom_of_kind, kind_of, eps_filter, dbcsr_nflop, unit_nr_dbcsr)
926 78 : CALL get_tensor_occupancy(t_KQKT, nze_KQK, occ_KQK)
927 :
928 : !Calculate the pseudo-density matrix in tensor form. There are a few useless arguments for SOS-MP2
929 : CALL compute_mat_dm_global(grid, nmo, cfm_mo_coeff(Pspin), homo(Pspin), propagator, &
930 : matrix_s, Pspin, Eigenval(:, Pspin), 0.0_dp, eps_filter, &
931 : mp2_env%ri_rpa_im_time%memory_info, unit_nr, &
932 78 : jquad, .FALSE., .FALSE., qs_env, dummy_int, dummy_ptr, para_env)
933 :
934 78 : CALL dbt_create(propagator(propagator_sector_occupied, jquad, 1)%matrix, t_2c_tmp)
935 78 : CALL dbt_copy_matrix_to_tensor(propagator(propagator_sector_occupied, jquad, 1)%matrix, t_2c_tmp)
936 78 : CALL dbt_copy(t_2c_tmp, t_dm_occ, move_data=.TRUE.)
937 78 : CALL dbt_filter(t_dm_occ, eps_filter)
938 78 : CALL dbt_destroy(t_2c_tmp)
939 :
940 78 : CALL dbt_create(propagator(propagator_sector_virtual, jquad, 1)%matrix, t_2c_tmp)
941 78 : CALL dbt_copy_matrix_to_tensor(propagator(propagator_sector_virtual, jquad, 1)%matrix, t_2c_tmp)
942 78 : CALL dbt_copy(t_2c_tmp, t_dm_virt, move_data=.TRUE.)
943 78 : CALL dbt_filter(t_dm_virt, eps_filter)
944 78 : CALL dbt_destroy(t_2c_tmp)
945 :
946 : !Deal with the 3-center quantities.
947 : CALL perform_3c_ops(force, t_R_occ, t_R_virt, force_data, fac, cut_memory, n_mem_RI, &
948 : t_KQKT, t_dm_occ, t_dm_virt, t_3c_O, t_3c_M, t_M_occ, t_M_virt, t_3c_0, t_3c_1, &
949 : t_3c_3, t_3c_4, t_3c_5, t_3c_6, t_3c_7, t_3c_8, t_3c_sparse, t_3c_help_1, t_3c_help_2, &
950 : t_3c_ints, t_3c_work, starts_array_mc, ends_array_mc, batch_start_RI, &
951 : batch_end_RI, t_3c_O_compressed, t_3c_O_ind, use_virial, &
952 : atom_of_kind, kind_of, eps_filter, occ_ddint, nze_ddint, dbcsr_nflop, &
953 78 : unit_nr_dbcsr, mp2_env)
954 :
955 78 : CALL timeset(routineN//"_dbcsr", handle2)
956 : !We go back to DBCSR matrices from now on
957 : !Note: R matrices are in fact symmetric, but use a normal type for convenience
958 78 : CALL dbt_create(matrix_s(1)%matrix, t_2c_tmp)
959 78 : CALL dbt_copy(t_R_occ, t_2c_tmp, move_data=.TRUE.)
960 78 : CALL dbt_copy_tensor_to_matrix(t_2c_tmp, R_occ)
961 :
962 78 : CALL dbt_copy(t_R_virt, t_2c_tmp, move_data=.TRUE.)
963 78 : CALL dbt_copy_tensor_to_matrix(t_2c_tmp, R_virt)
964 :
965 : !Iteratively calculate the Y1 and Y2 matrices
966 : CALL dbcsr_multiply('N', 'N', tau, force_data%P_occ(Pspin)%matrix, &
967 78 : matrix_ks(Pspin)%matrix, 0.0_dp, dbcsr_work1)
968 78 : CALL build_Y_matrix(Y_1, dbcsr_work1, force_data%P_occ(Pspin)%matrix, R_virt, eps_filter)
969 78 : CALL matrix_exponential(exp_occ, dbcsr_work1, 1.0_dp, 1.0_dp, eps_filter)
970 :
971 : CALL dbcsr_multiply('N', 'N', -tau, force_data%P_virt(Pspin)%matrix, &
972 78 : matrix_ks(Pspin)%matrix, 0.0_dp, dbcsr_work1)
973 78 : CALL build_Y_matrix(Y_2, dbcsr_work1, force_data%P_virt(Pspin)%matrix, R_occ, eps_filter)
974 78 : CALL matrix_exponential(exp_virt, dbcsr_work1, 1.0_dp, 1.0_dp, eps_filter)
975 :
976 : !The force contribution coming from [-S^-1*(e^-tau*P_virt*F)^T*R_occ*S^-1
977 : ! +tau*S^-1*Y_2^T*F*S^-1] * der_S
978 78 : CALL dbcsr_multiply('N', 'N', 1.0_dp, R_occ, force_data%inv_ovlp, 0.0_dp, dbcsr_work1)
979 78 : CALL dbcsr_multiply('T', 'N', 1.0_dp, exp_virt, dbcsr_work1, 0.0_dp, dbcsr_work3)
980 78 : CALL dbcsr_multiply('N', 'N', 1.0_dp, force_data%inv_ovlp, dbcsr_work3, 0.0_dp, dbcsr_work2)
981 :
982 78 : CALL dbcsr_multiply('N', 'T', tau, force_data%inv_ovlp, Y_2, 0.0_dp, dbcsr_work3)
983 78 : CALL dbcsr_multiply('N', 'N', 1.0_dp, dbcsr_work3, matrix_ks(Pspin)%matrix, 0.0_dp, dbcsr_work1)
984 78 : CALL dbcsr_multiply('N', 'N', 1.0_dp, dbcsr_work1, force_data%inv_ovlp, 0.0_dp, dbcsr_work3)
985 :
986 78 : CALL dbcsr_add(dbcsr_work2, dbcsr_work3, 1.0_dp, -1.0_dp)
987 :
988 78 : CALL dbt_copy_matrix_to_tensor(dbcsr_work2, t_2c_tmp)
989 78 : CALL dbt_copy(t_2c_tmp, t_2c_AO, move_data=.TRUE.)
990 :
991 78 : pref = -1.0_dp*fac
992 : CALL get_2c_der_force(force, t_2c_AO, force_data%t_2c_der_ovlp, atom_of_kind, &
993 78 : kind_of, force_data%idx_to_at_AO, pref, do_ovlp=.TRUE.)
994 :
995 78 : IF (use_virial) CALL dbcsr_add(virial_ovlp, dbcsr_work2, 1.0_dp, pref)
996 :
997 : !The final contribution from Tr[(tau*Y_1*P_occ - tau*Y_2*P_virt) * der_F]
998 : CALL dbcsr_multiply('N', 'N', tau*fac, Y_1, force_data%P_occ(Pspin)%matrix, 1.0_dp, &
999 78 : force_data%sum_YP_tau(Pspin)%matrix, retain_sparsity=.TRUE.)
1000 : CALL dbcsr_multiply('N', 'N', -tau*fac, Y_2, force_data%P_virt(Pspin)%matrix, 1.0_dp, &
1001 78 : force_data%sum_YP_tau(Pspin)%matrix, retain_sparsity=.TRUE.)
1002 :
1003 : !Build-up the RHS of the response equation.
1004 78 : pref = -omega*mp2_env%scale_S
1005 : CALL dbcsr_multiply('N', 'N', pref, R_virt, exp_occ, 1.0_dp, &
1006 78 : force_data%sum_O_tau(Pspin)%matrix, retain_sparsity=.TRUE.)
1007 : CALL dbcsr_multiply('N', 'N', -pref, R_occ, exp_virt, 1.0_dp, &
1008 78 : force_data%sum_O_tau(Pspin)%matrix, retain_sparsity=.TRUE.)
1009 : CALL dbcsr_multiply('N', 'N', pref*tau, matrix_ks(Pspin)%matrix, Y_1, 1.0_dp, &
1010 78 : force_data%sum_O_tau(Pspin)%matrix, retain_sparsity=.TRUE.)
1011 : CALL dbcsr_multiply('N', 'N', pref*tau, matrix_ks(Pspin)%matrix, Y_2, 1.0_dp, &
1012 78 : force_data%sum_O_tau(Pspin)%matrix, retain_sparsity=.TRUE.)
1013 :
1014 78 : CALL timestop(handle2)
1015 :
1016 : !Print some info
1017 78 : CALL para_env%sync()
1018 78 : t2 = m_walltime()
1019 78 : dbcsr_time = dbcsr_time + t2 - t1
1020 :
1021 78 : IF (unit_nr > 0) THEN
1022 : WRITE (unit_nr, '(/T3,A,1X,I3,A)') &
1023 39 : 'RPA_LOW_SCALING_INFO| Info for time point', jquad, ' (gradients)'
1024 : WRITE (unit_nr, '(T6,A,T56,F25.6)') &
1025 39 : 'Execution time (s):', t2 - t1
1026 : WRITE (unit_nr, '(T6,A,T63,ES7.1,1X,A1,1X,F7.3,A1)') &
1027 39 : 'Occupancy of 3c AO derivs:', REAL(nze_der_AO, dp), '/', occ_der_AO*100, '%'
1028 : WRITE (unit_nr, '(T6,A,T63,ES7.1,1X,A1,1X,F7.3,A1)') &
1029 39 : 'Occupancy of 3c RI derivs:', REAL(nze_der_RI, dp), '/', occ_der_RI*100, '%'
1030 : WRITE (unit_nr, '(T6,A,T63,ES7.1,1X,A1,1X,F7.3,A1)') &
1031 39 : 'Occupancy of the Docc * Dvirt * 3c-int tensor', REAL(nze_ddint, dp), '/', occ_ddint*100, '%'
1032 : WRITE (unit_nr, '(T6,A,T63,ES7.1,1X,A1,1X,F7.3,A1)') &
1033 39 : 'Occupancy of KQK^T 2c-tensor:', REAL(nze_KQK, dp), '/', occ_KQK*100, '%'
1034 39 : CALL m_flush(unit_nr)
1035 : END IF
1036 :
1037 : !intermediate clean-up
1038 78 : CALL dbcsr_release(Y_1)
1039 78 : CALL dbcsr_release(Y_2)
1040 416 : CALL dbt_destroy(t_2c_tmp)
1041 : END DO !jquad
1042 :
1043 26 : CALL dbt_batched_contract_finalize(t_3c_0)
1044 26 : CALL dbt_batched_contract_finalize(t_3c_1)
1045 26 : CALL dbt_batched_contract_finalize(t_3c_3)
1046 26 : CALL dbt_batched_contract_finalize(t_M_occ)
1047 26 : CALL dbt_batched_contract_finalize(t_M_virt)
1048 :
1049 26 : CALL dbt_batched_contract_finalize(t_3c_ints)
1050 26 : CALL dbt_batched_contract_finalize(t_3c_work)
1051 :
1052 26 : CALL dbt_batched_contract_finalize(t_3c_4)
1053 26 : CALL dbt_batched_contract_finalize(t_3c_5)
1054 26 : CALL dbt_batched_contract_finalize(t_3c_6)
1055 26 : CALL dbt_batched_contract_finalize(t_3c_7)
1056 26 : CALL dbt_batched_contract_finalize(t_3c_8)
1057 26 : CALL dbt_batched_contract_finalize(t_3c_sparse)
1058 :
1059 : !Calculate the 2c and 3c contributions to the virial
1060 26 : IF (use_virial) THEN
1061 2 : CALL dbt_copy(force_data%t_3c_virial_split, force_data%t_3c_virial, move_data=.TRUE.)
1062 : CALL calc_3c_virial(work_virial, force_data%t_3c_virial, 1.0_dp, qs_env, force_data%nl_3c, &
1063 : basis_set_ri_aux, basis_set_ao, basis_set_ao, mp2_env%ri_metric, &
1064 2 : der_eps=mp2_env%ri_rpa_im_time%eps_filter, op_pos=1)
1065 :
1066 : CALL calc_2c_virial(work_virial, force_data%RI_virial_met, 1.0_dp, qs_env, force_data%nl_2c_met, &
1067 2 : basis_set_ri_aux, basis_set_ri_aux, mp2_env%ri_metric)
1068 2 : CALL dbcsr_clear(force_data%RI_virial_met)
1069 :
1070 2 : IF (.NOT. force_data%do_periodic) THEN
1071 : CALL calc_2c_virial(work_virial, force_data%RI_virial_pot, 1.0_dp, qs_env, force_data%nl_2c_pot, &
1072 0 : basis_set_ri_aux, basis_set_ri_aux, mp2_env%potential_parameter)
1073 0 : CALL dbcsr_clear(force_data%RI_virial_pot)
1074 : END IF
1075 :
1076 2 : identity_pot%potential_type = do_potential_id
1077 : CALL calc_2c_virial(work_virial_ovlp, virial_ovlp, 1.0_dp, qs_env, force_data%nl_2c_ovlp, &
1078 2 : basis_set_ao, basis_set_ao, identity_pot)
1079 2 : CALL dbcsr_release(virial_ovlp)
1080 :
1081 8 : DO k_xyz = 1, 3
1082 26 : DO j_xyz = 1, 3
1083 78 : DO i_xyz = 1, 3
1084 : virial%pv_mp2(i_xyz, j_xyz) = virial%pv_mp2(i_xyz, j_xyz) &
1085 54 : - work_virial(i_xyz, k_xyz)*cell%hmat(j_xyz, k_xyz)
1086 : virial%pv_overlap(i_xyz, j_xyz) = virial%pv_overlap(i_xyz, j_xyz) &
1087 54 : - work_virial_ovlp(i_xyz, k_xyz)*cell%hmat(j_xyz, k_xyz)
1088 : virial%pv_virial(i_xyz, j_xyz) = virial%pv_virial(i_xyz, j_xyz) &
1089 : - work_virial(i_xyz, k_xyz)*cell%hmat(j_xyz, k_xyz) &
1090 72 : - work_virial_ovlp(i_xyz, k_xyz)*cell%hmat(j_xyz, k_xyz)
1091 : END DO
1092 : END DO
1093 : END DO
1094 : END IF
1095 :
1096 : !Calculate the periodic contributions of (P|Q) to the force and the virial
1097 26 : work_virial = 0.0_dp
1098 26 : IF (force_data%do_periodic) THEN
1099 10 : IF (mp2_env%eri_method == do_eri_gpw) THEN
1100 6 : CALL get_2c_gpw_forces(force_data%G_PQ, force, work_virial, use_virial, mp2_env, qs_env)
1101 4 : ELSE IF (mp2_env%eri_method == do_eri_mme) THEN
1102 4 : CALL get_2c_mme_forces(force_data%G_PQ, force, mp2_env, qs_env)
1103 4 : IF (use_virial) CPABORT("Stress tensor not available with MME intrgrals")
1104 : ELSE
1105 0 : CPABORT("Periodic case not possible with OS integrals")
1106 : END IF
1107 10 : CALL dbcsr_clear(force_data%G_PQ)
1108 : END IF
1109 :
1110 26 : IF (use_virial) THEN
1111 26 : virial%pv_mp2 = virial%pv_mp2 + work_virial
1112 26 : virial%pv_virial = virial%pv_virial + work_virial
1113 2 : virial%pv_calculate = .FALSE.
1114 :
1115 6 : DO ibasis = 1, SIZE(basis_set_ao)
1116 4 : orb_basis => basis_set_ao(ibasis)%gto_basis_set
1117 4 : CALL init_interaction_radii_orb_basis(orb_basis, eps_pgf_orb_old)
1118 4 : ri_basis => basis_set_ri_aux(ibasis)%gto_basis_set
1119 6 : CALL init_interaction_radii_orb_basis(ri_basis, eps_pgf_orb_old)
1120 : END DO
1121 : END IF
1122 :
1123 : !clean-up
1124 26 : IF (ASSOCIATED(dummy_ptr)) DEALLOCATE (dummy_ptr)
1125 60 : DO ispin = 1, nspins
1126 60 : CALL dbt_destroy(t_P(ispin))
1127 : END DO
1128 26 : CALL dbt_destroy(t_3c_0)
1129 26 : CALL dbt_destroy(t_3c_1)
1130 26 : CALL dbt_destroy(t_3c_3)
1131 26 : CALL dbt_destroy(t_3c_4)
1132 26 : CALL dbt_destroy(t_3c_5)
1133 26 : CALL dbt_destroy(t_3c_6)
1134 26 : CALL dbt_destroy(t_3c_7)
1135 26 : CALL dbt_destroy(t_3c_8)
1136 26 : CALL dbt_destroy(t_3c_sparse)
1137 26 : CALL dbt_destroy(t_3c_help_1)
1138 26 : CALL dbt_destroy(t_3c_help_2)
1139 26 : CALL dbt_destroy(t_3c_ints)
1140 26 : CALL dbt_destroy(t_3c_work)
1141 26 : CALL dbt_destroy(t_R_occ)
1142 26 : CALL dbt_destroy(t_R_virt)
1143 26 : CALL dbt_destroy(t_dm_occ)
1144 26 : CALL dbt_destroy(t_dm_virt)
1145 26 : CALL dbt_destroy(t_Q)
1146 26 : CALL dbt_destroy(t_KQKT)
1147 26 : CALL dbt_destroy(t_M_occ)
1148 26 : CALL dbt_destroy(t_M_virt)
1149 26 : CALL dbcsr_release(R_occ)
1150 26 : CALL dbcsr_release(R_virt)
1151 26 : CALL dbcsr_release(dbcsr_work1)
1152 26 : CALL dbcsr_release(dbcsr_work2)
1153 26 : CALL dbcsr_release(dbcsr_work3)
1154 26 : CALL dbcsr_release(exp_occ)
1155 26 : CALL dbcsr_release(exp_virt)
1156 :
1157 26 : CALL dbt_destroy(t_2c_RI)
1158 26 : CALL dbt_destroy(t_2c_RI_2)
1159 26 : CALL dbt_destroy(t_2c_AO)
1160 26 : CALL dbcsr_deallocate_matrix_set(propagator)
1161 :
1162 26 : CALL timestop(handle)
1163 :
1164 112 : END SUBROUTINE calc_laplace_loop_forces
1165 :
1166 : ! **************************************************************************************************
1167 : !> \brief Updates the cubic-scaling RPA contribution to the forces at each quadrature point. This
1168 : !> routine is adapted from the corresponding Laplace SOS-MP2 loop force one.
1169 : !> \param force_data ...
1170 : !> \param mat_P_omega ...
1171 : !> \param t_3c_M ...
1172 : !> \param t_3c_O ...
1173 : !> \param t_3c_O_compressed ...
1174 : !> \param t_3c_O_ind ...
1175 : !> \param cfm_mo_coeff ...
1176 : !> \param homo ...
1177 : !> \param starts_array_mc ...
1178 : !> \param ends_array_mc ...
1179 : !> \param starts_array_mc_block ...
1180 : !> \param ends_array_mc_block ...
1181 : !> \param nmo ...
1182 : !> \param Eigenval ...
1183 : !> \param e_fermi ...
1184 : !> \param grid ...
1185 : !> \param cut_memory ...
1186 : !> \param ispin ...
1187 : !> \param open_shell ...
1188 : !> \param unit_nr ...
1189 : !> \param dbcsr_time ...
1190 : !> \param dbcsr_nflop ...
1191 : !> \param mp2_env ...
1192 : !> \param qs_env ...
1193 : ! **************************************************************************************************
1194 180 : SUBROUTINE calc_rpa_loop_forces(force_data, mat_P_omega, t_3c_M, t_3c_O, t_3c_O_compressed, &
1195 72 : t_3c_O_ind, cfm_mo_coeff, homo, starts_array_mc, ends_array_mc, &
1196 36 : starts_array_mc_block, ends_array_mc_block, &
1197 36 : nmo, Eigenval, e_fermi, grid, cut_memory, ispin, open_shell, unit_nr, dbcsr_time, &
1198 : dbcsr_nflop, mp2_env, qs_env)
1199 :
1200 : TYPE(im_time_force_type), INTENT(INOUT) :: force_data
1201 : TYPE(dbcsr_p_type), DIMENSION(:, :), INTENT(INOUT) :: mat_P_omega
1202 : TYPE(dbt_type), INTENT(INOUT) :: t_3c_M, t_3c_O
1203 : TYPE(hfx_compression_type), DIMENSION(:) :: t_3c_O_compressed
1204 : TYPE(block_ind_type), DIMENSION(:), INTENT(INOUT) :: t_3c_O_ind
1205 : TYPE(cp_cfm_type), DIMENSION(:), INTENT(IN) :: cfm_mo_coeff
1206 : INTEGER, DIMENSION(:), INTENT(IN) :: homo, starts_array_mc, ends_array_mc, &
1207 : starts_array_mc_block, &
1208 : ends_array_mc_block
1209 : INTEGER, INTENT(IN) :: nmo
1210 : REAL(KIND=dp), DIMENSION(:, :), INTENT(IN) :: Eigenval
1211 : REAL(KIND=dp), INTENT(IN) :: e_fermi
1212 : TYPE(time_frequency_grid_type), INTENT(IN) :: grid
1213 : INTEGER, INTENT(IN) :: cut_memory, ispin
1214 : LOGICAL, INTENT(IN) :: open_shell
1215 : INTEGER, INTENT(IN) :: unit_nr
1216 : REAL(dp), INTENT(INOUT) :: dbcsr_time
1217 : INTEGER(int_8), INTENT(INOUT) :: dbcsr_nflop
1218 : TYPE(mp2_type) :: mp2_env
1219 : TYPE(qs_environment_type), POINTER :: qs_env
1220 :
1221 : CHARACTER(LEN=*), PARAMETER :: routineN = 'calc_rpa_loop_forces'
1222 :
1223 : INTEGER :: dummy_int, handle, handle2, i_mem, i_xyz, ibasis, iquad, j_xyz, jquad, k_xyz, &
1224 : n_mem_RI, n_rep, natom, nkind, nspins, num_integ_points, unit_nr_dbcsr
1225 : INTEGER(int_8) :: flop, nze, nze_ddint, nze_der_AO, &
1226 : nze_der_RI, nze_KBK
1227 36 : INTEGER, ALLOCATABLE, DIMENSION(:) :: atom_of_kind, batch_blk_end, &
1228 36 : batch_blk_start, batch_end_RI, &
1229 36 : batch_start_RI, kind_of, mc_ranges, &
1230 36 : mc_ranges_RI
1231 36 : INTEGER, DIMENSION(:, :), POINTER :: dummy_ptr
1232 : LOGICAL :: memory_info, use_virial
1233 : REAL(dp) :: eps_filter, eps_pgf_orb, eps_pgf_orb_old, fac, occ, occ_ddint, occ_der_AO, &
1234 : occ_der_RI, occ_KBK, omega, pref, spin_fac, t1, t2, tau, weight
1235 : REAL(dp), DIMENSION(3, 3) :: work_virial, work_virial_ovlp
1236 36 : TYPE(atomic_kind_type), DIMENSION(:), POINTER :: atomic_kind_set
1237 : TYPE(cell_type), POINTER :: cell
1238 : TYPE(cp_blacs_env_type), POINTER :: blacs_env
1239 36 : TYPE(dbcsr_p_type), DIMENSION(:), POINTER :: mat_P_tau, matrix_ks, matrix_s
1240 36 : TYPE(dbcsr_p_type), DIMENSION(:, :, :), POINTER :: propagator
1241 : TYPE(dbcsr_type) :: dbcsr_work1, dbcsr_work2, dbcsr_work3, &
1242 : dbcsr_work_symm, exp_occ, exp_virt, &
1243 : R_occ, R_virt, virial_ovlp, Y_1, Y_2
1244 1764 : TYPE(dbt_type) :: t_2c_AO, t_2c_RI, t_2c_RI_2, t_2c_tmp, t_3c_0, t_3c_1, t_3c_3, t_3c_4, &
1245 1764 : t_3c_5, t_3c_6, t_3c_7, t_3c_8, t_3c_help_1, t_3c_help_2, t_3c_ints, t_3c_sparse, &
1246 2016 : t_3c_work, t_dm_occ, t_dm_virt, t_KBKT, t_M_occ, t_M_virt, t_P, t_R_occ, t_R_virt
1247 36 : TYPE(dbt_type), ALLOCATABLE, DIMENSION(:) :: t_B
1248 : TYPE(dft_control_type), POINTER :: dft_control
1249 : TYPE(gto_basis_set_p_type), ALLOCATABLE, &
1250 36 : DIMENSION(:), TARGET :: basis_set_ao, basis_set_ri_aux
1251 : TYPE(gto_basis_set_type), POINTER :: orb_basis, ri_basis
1252 : TYPE(libint_potential_type) :: identity_pot
1253 : TYPE(mp_para_env_type), POINTER :: para_env
1254 36 : TYPE(particle_type), DIMENSION(:), POINTER :: particle_set
1255 36 : TYPE(qs_force_type), DIMENSION(:), POINTER :: force
1256 36 : TYPE(qs_kind_type), DIMENSION(:), POINTER :: qs_kind_set
1257 : TYPE(section_vals_type), POINTER :: qs_section
1258 : TYPE(virial_type), POINTER :: virial
1259 :
1260 36 : NULLIFY (matrix_s, dummy_ptr, atomic_kind_set, force, matrix_s, matrix_ks)
1261 36 : NULLIFY (dft_control, virial, particle_set, cell, blacs_env, para_env, orb_basis, ri_basis)
1262 36 : NULLIFY (qs_kind_set)
1263 :
1264 36 : CALL timeset(routineN, handle)
1265 :
1266 36 : num_integ_points = SIZE(grid%imaginary_time)
1267 :
1268 36 : NULLIFY (propagator)
1269 :
1270 : CALL get_qs_env(qs_env, matrix_s=matrix_s, natom=natom, atomic_kind_set=atomic_kind_set, &
1271 : force=force, matrix_ks=matrix_ks, dft_control=dft_control, virial=virial, &
1272 : particle_set=particle_set, cell=cell, blacs_env=blacs_env, para_env=para_env, &
1273 36 : qs_kind_set=qs_kind_set, nkind=nkind)
1274 36 : eps_filter = mp2_env%ri_rpa_im_time%eps_filter
1275 36 : nspins = dft_control%nspins
1276 :
1277 36 : memory_info = mp2_env%ri_rpa_im_time%memory_info
1278 36 : IF (memory_info) THEN
1279 0 : unit_nr_dbcsr = unit_nr
1280 : ELSE
1281 36 : unit_nr_dbcsr = 0
1282 : END IF
1283 :
1284 36 : use_virial = virial%pv_availability .AND. (.NOT. virial%pv_numer)
1285 :
1286 36 : IF (use_virial) virial%pv_calculate = .TRUE.
1287 :
1288 36 : IF (use_virial) THEN
1289 2 : qs_section => section_vals_get_subs_vals(qs_env%input, "DFT%QS")
1290 2 : CALL section_vals_val_get(qs_section, "EPS_PGF_ORB", n_rep_val=n_rep)
1291 2 : IF (n_rep /= 0) THEN
1292 0 : CALL section_vals_val_get(qs_section, "EPS_PGF_ORB", r_val=eps_pgf_orb)
1293 : ELSE
1294 2 : CALL section_vals_val_get(qs_section, "EPS_DEFAULT", r_val=eps_pgf_orb)
1295 2 : eps_pgf_orb = SQRT(eps_pgf_orb)
1296 : END IF
1297 2 : eps_pgf_orb_old = dft_control%qs_control%eps_pgf_orb
1298 :
1299 16 : ALLOCATE (basis_set_ri_aux(nkind), basis_set_ao(nkind))
1300 2 : CALL basis_set_list_setup(basis_set_ri_aux, "RI_AUX", qs_kind_set)
1301 2 : CALL basis_set_list_setup(basis_set_ao, "ORB", qs_kind_set)
1302 :
1303 8 : DO ibasis = 1, SIZE(basis_set_ao)
1304 4 : orb_basis => basis_set_ao(ibasis)%gto_basis_set
1305 4 : CALL init_interaction_radii_orb_basis(orb_basis, eps_pgf_orb)
1306 4 : ri_basis => basis_set_ri_aux(ibasis)%gto_basis_set
1307 6 : CALL init_interaction_radii_orb_basis(ri_basis, eps_pgf_orb)
1308 : END DO
1309 : END IF
1310 :
1311 : !We follow the general logic of the compute_mat_P_omega routine
1312 36 : CALL dbt_create(force_data%t_2c_K, t_2c_RI)
1313 36 : CALL dbt_create(force_data%t_2c_K, t_2c_RI_2)
1314 36 : CALL dbt_create(force_data%t_2c_der_ovlp(1), t_2c_AO)
1315 :
1316 36 : CALL get_atomic_kind_set(atomic_kind_set, kind_of=kind_of, atom_of_kind=atom_of_kind)
1317 :
1318 : ! Always do the batching of the MO on mu and sigma, such that it is consistent between
1319 : ! the occupied and the virtual quantities
1320 108 : ALLOCATE (mc_ranges(cut_memory + 1))
1321 108 : mc_ranges(:cut_memory) = starts_array_mc_block(:)
1322 36 : mc_ranges(cut_memory + 1) = ends_array_mc_block(cut_memory) + 1
1323 :
1324 : ! Also need some batching on the RI, because it loses sparsity at some point
1325 36 : n_mem_RI = cut_memory
1326 : CALL create_tensor_batches(force_data%bsizes_RI_split, n_mem_RI, batch_start_RI, batch_end_RI, &
1327 36 : batch_blk_start, batch_blk_end)
1328 108 : ALLOCATE (mc_ranges_RI(n_mem_RI + 1))
1329 108 : mc_ranges_RI(1:n_mem_RI) = batch_blk_start(1:n_mem_RI)
1330 36 : mc_ranges_RI(n_mem_RI + 1) = batch_blk_end(n_mem_RI) + 1
1331 36 : DEALLOCATE (batch_blk_start, batch_blk_end)
1332 :
1333 : !Pre-allocate all required tensors and matrices
1334 36 : CALL dbt_create(t_2c_RI, t_P)
1335 36 : CALL dbt_create(t_2c_RI, t_KBKT)
1336 36 : CALL dbt_create(t_2c_AO, t_dm_occ)
1337 36 : CALL dbt_create(t_2c_AO, t_dm_virt)
1338 :
1339 : !note: t_3c_O and t_3c_M have different mappings (map_1d, map_2d)
1340 36 : CALL dbt_create(t_3c_O, t_M_occ)
1341 36 : CALL dbt_create(t_3c_O, t_M_virt)
1342 36 : CALL dbt_create(t_3c_O, t_3c_0)
1343 :
1344 36 : CALL dbt_create(t_3c_O, t_3c_1)
1345 36 : CALL dbt_create(t_3c_O, t_3c_3)
1346 36 : CALL dbt_create(t_3c_O, t_3c_4)
1347 36 : CALL dbt_create(t_3c_O, t_3c_5)
1348 36 : CALL dbt_create(t_3c_M, t_3c_6)
1349 36 : CALL dbt_create(t_3c_M, t_3c_7)
1350 36 : CALL dbt_create(t_3c_M, t_3c_8)
1351 36 : CALL dbt_create(t_3c_M, t_3c_sparse)
1352 36 : CALL dbt_create(t_3c_O, t_3c_help_1)
1353 36 : CALL dbt_create(t_3c_O, t_3c_help_2)
1354 36 : CALL dbt_create(t_2c_AO, t_R_occ)
1355 36 : CALL dbt_create(t_2c_AO, t_R_virt)
1356 36 : CALL dbt_create(t_3c_M, t_3c_ints)
1357 36 : CALL dbt_create(t_3c_M, t_3c_work)
1358 :
1359 : !Before entring the loop, need to compute the 2c tensors B = (1 + Q(w))^-1 - 1, for each
1360 : !frequency grid point, before doing the transformation to the time grid
1361 416 : ALLOCATE (t_B(num_integ_points))
1362 128 : DO jquad = 1, num_integ_points
1363 128 : CALL dbt_create(t_2c_RI, t_B(jquad))
1364 : END DO
1365 :
1366 200 : ALLOCATE (mat_P_tau(num_integ_points))
1367 128 : DO jquad = 1, num_integ_points
1368 92 : ALLOCATE (mat_P_tau(jquad)%matrix)
1369 128 : CALL dbcsr_create(mat_P_tau(jquad)%matrix, template=mat_P_omega(jquad, ispin)%matrix)
1370 : END DO
1371 :
1372 36 : CALL dbcsr_create(dbcsr_work_symm, template=force_data%G_PQ, matrix_type=dbcsr_type_symmetric)
1373 36 : CALL dbt_create(dbcsr_work_symm, t_2c_tmp)
1374 :
1375 : !loop over freqeuncies
1376 128 : DO iquad = 1, num_integ_points
1377 92 : omega = grid%frequency(iquad)
1378 :
1379 : !calculate (1 + Q(w))^-1 - 1 for the given freq.
1380 : !Always take spin alpha (get 2*alpha in closed shell, and alpha+beta in open-shell)
1381 92 : CALL dbcsr_copy(dbcsr_work_symm, mat_P_omega(iquad, 1)%matrix)
1382 92 : CALL dbt_copy_matrix_to_tensor(dbcsr_work_symm, t_2c_tmp)
1383 92 : CALL dbt_copy(t_2c_tmp, t_2c_RI, move_data=.TRUE.)
1384 :
1385 : CALL dbt_contract(1.0_dp, t_2c_RI, force_data%t_2c_K, 0.0_dp, t_2c_RI_2, &
1386 : contract_1=[2], notcontract_1=[1], &
1387 : contract_2=[1], notcontract_2=[2], &
1388 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
1389 92 : flop=flop, unit_nr=unit_nr_dbcsr)
1390 92 : dbcsr_nflop = dbcsr_nflop + flop
1391 : CALL dbt_contract(1.0_dp, force_data%t_2c_K, t_2c_RI_2, 0.0_dp, t_2c_RI, &
1392 : contract_1=[1], notcontract_1=[2], &
1393 : contract_2=[1], notcontract_2=[2], &
1394 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
1395 92 : flop=flop, unit_nr=unit_nr_dbcsr)
1396 92 : CALL dbt_copy(t_2c_RI, t_2c_tmp, move_data=.TRUE.)
1397 92 : CALL dbt_copy_tensor_to_matrix(t_2c_tmp, dbcsr_work_symm)
1398 92 : CALL dbcsr_add_on_diag(dbcsr_work_symm, 1.0_dp)
1399 :
1400 92 : CALL cp_dbcsr_cholesky_decompose(dbcsr_work_symm, para_env=para_env, blacs_env=blacs_env)
1401 92 : CALL cp_dbcsr_cholesky_invert(dbcsr_work_symm, para_env=para_env, blacs_env=blacs_env, uplo_to_full=.TRUE.)
1402 :
1403 92 : CALL dbcsr_add_on_diag(dbcsr_work_symm, -1.0_dp)
1404 :
1405 372 : DO jquad = 1, num_integ_points
1406 244 : tau = grid%imaginary_time(jquad)
1407 :
1408 : !the P matrix to time.
1409 244 : weight = grid%cosine_frequency_to_time_weights(jquad, iquad)*COS(tau*omega)
1410 244 : IF (open_shell) THEN
1411 64 : IF (ispin == 1) THEN
1412 : !mat_P_omega contains the sum of alpha and beta spin => we only want alpha
1413 32 : CALL dbcsr_add(mat_P_tau(jquad)%matrix, mat_P_omega(iquad, 1)%matrix, 1.0_dp, weight)
1414 32 : CALL dbcsr_add(mat_P_tau(jquad)%matrix, mat_P_omega(iquad, 2)%matrix, 1.0_dp, -weight)
1415 : ELSE
1416 32 : CALL dbcsr_add(mat_P_tau(jquad)%matrix, mat_P_omega(iquad, 2)%matrix, 1.0_dp, weight)
1417 : END IF
1418 : ELSE
1419 : !factor 0.5 because originam matrix Q is scaled by 2 in RPA (spin)
1420 180 : weight = 0.5_dp*weight
1421 180 : CALL dbcsr_add(mat_P_tau(jquad)%matrix, mat_P_omega(iquad, 1)%matrix, 1.0_dp, weight)
1422 : END IF
1423 :
1424 : !convert B matrix to time
1425 : weight = grid%cosine_time_to_frequency_weights(iquad, jquad)*COS(tau*omega)* &
1426 244 : grid%frequency_weights(iquad)
1427 244 : CALL dbt_copy_matrix_to_tensor(dbcsr_work_symm, t_2c_tmp)
1428 244 : CALL dbt_scale(t_2c_tmp, weight)
1429 336 : CALL dbt_copy(t_2c_tmp, t_B(jquad), summation=.TRUE., move_data=.TRUE.)
1430 : END DO
1431 : END DO
1432 36 : CALL dbt_destroy(t_2c_tmp)
1433 36 : CALL dbcsr_release(dbcsr_work_symm)
1434 36 : CALL dbt_clear(t_2c_RI)
1435 36 : CALL dbt_clear(t_2c_RI_2)
1436 :
1437 : !Pre-define the sparsity of t_3c_4 as a function of the derivatives
1438 36 : occ_der_AO = 0; nze_der_AO = 0
1439 36 : occ_der_RI = 0; nze_der_RI = 0
1440 144 : DO i_xyz = 1, 3
1441 360 : DO i_mem = 1, cut_memory
1442 : CALL decompress_tensor(force_data%t_3c_der_RI(i_xyz), force_data%t_3c_der_RI_ind(i_mem, i_xyz)%ind, &
1443 216 : force_data%t_3c_der_RI_comp(i_mem, i_xyz), mp2_env%ri_rpa_im_time%eps_compress)
1444 216 : CALL get_tensor_occupancy(force_data%t_3c_der_RI(i_xyz), nze, occ)
1445 216 : occ_der_RI = occ_der_RI + occ
1446 216 : nze_der_RI = nze_der_RI + nze
1447 216 : CALL dbt_copy(force_data%t_3c_der_RI(i_xyz), t_3c_sparse, summation=.TRUE., move_data=.TRUE.)
1448 :
1449 : CALL decompress_tensor(force_data%t_3c_der_AO(i_xyz), force_data%t_3c_der_AO_ind(i_mem, i_xyz)%ind, &
1450 216 : force_data%t_3c_der_AO_comp(i_mem, i_xyz), mp2_env%ri_rpa_im_time%eps_compress)
1451 216 : CALL get_tensor_occupancy(force_data%t_3c_der_AO(i_xyz), nze, occ)
1452 216 : occ_der_AO = occ_der_AO + occ
1453 216 : nze_der_AO = nze_der_AO + nze
1454 216 : CALL dbt_copy(force_data%t_3c_der_AO(i_xyz), t_3c_sparse, order=[1, 3, 2], summation=.TRUE.)
1455 756 : CALL dbt_copy(force_data%t_3c_der_AO(i_xyz), t_3c_sparse, summation=.TRUE., move_data=.TRUE.)
1456 : END DO
1457 : END DO
1458 36 : occ_der_RI = occ_der_RI/3.0_dp
1459 36 : occ_der_AO = occ_der_AO/3.0_dp
1460 36 : nze_der_RI = nze_der_RI/3
1461 36 : nze_der_AO = nze_der_AO/3
1462 :
1463 36 : CALL dbcsr_create(R_occ, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
1464 36 : CALL dbcsr_create(R_virt, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
1465 36 : CALL dbcsr_create(dbcsr_work_symm, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_symmetric)
1466 36 : CALL dbcsr_create(dbcsr_work1, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
1467 36 : CALL dbcsr_create(dbcsr_work2, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
1468 36 : CALL dbcsr_create(dbcsr_work3, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
1469 36 : CALL dbcsr_create(exp_occ, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
1470 36 : CALL dbcsr_create(exp_virt, template=matrix_s(1)%matrix, matrix_type=dbcsr_type_no_symmetry)
1471 36 : IF (use_virial) CALL dbcsr_create(virial_ovlp, template=dbcsr_work1)
1472 :
1473 36 : CALL dbt_batched_contract_init(t_3c_0, batch_range_2=mc_ranges, batch_range_3=mc_ranges)
1474 36 : CALL dbt_batched_contract_init(t_3c_1, batch_range_2=mc_ranges, batch_range_3=mc_ranges)
1475 36 : CALL dbt_batched_contract_init(t_3c_3, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
1476 36 : CALL dbt_batched_contract_init(t_M_occ, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
1477 36 : CALL dbt_batched_contract_init(t_M_virt, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
1478 :
1479 36 : CALL dbt_batched_contract_init(t_3c_ints, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
1480 36 : CALL dbt_batched_contract_init(t_3c_work, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges)
1481 :
1482 : CALL dbt_batched_contract_init(t_3c_4, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
1483 36 : batch_range_3=mc_ranges)
1484 : CALL dbt_batched_contract_init(t_3c_5, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
1485 36 : batch_range_3=mc_ranges)
1486 : CALL dbt_batched_contract_init(t_3c_6, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
1487 36 : batch_range_3=mc_ranges)
1488 : CALL dbt_batched_contract_init(t_3c_7, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
1489 36 : batch_range_3=mc_ranges)
1490 : CALL dbt_batched_contract_init(t_3c_8, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
1491 36 : batch_range_3=mc_ranges)
1492 : CALL dbt_batched_contract_init(t_3c_sparse, batch_range_1=mc_ranges_RI, batch_range_2=mc_ranges, &
1493 36 : batch_range_3=mc_ranges)
1494 :
1495 36 : fac = 1.0_dp/fourpi*mp2_env%ri_rpa%scale_rpa
1496 36 : IF (open_shell) fac = 0.5_dp*fac
1497 :
1498 36 : work_virial = 0.0_dp
1499 36 : work_virial_ovlp = 0.0_dp
1500 128 : DO jquad = 1, num_integ_points
1501 92 : tau = grid%imaginary_time(jquad)
1502 92 : occ_ddint = 0; nze_ddint = 0
1503 :
1504 92 : CALL para_env%sync()
1505 92 : t1 = m_walltime()
1506 :
1507 : !Deal with the force contributions where there is no explicit 3-center quantities, i.e. the
1508 : !forces due to the metric and potential derivatives
1509 92 : CALL dbt_create(mat_P_tau(jquad)%matrix, t_2c_tmp)
1510 92 : CALL dbt_copy_matrix_to_tensor(mat_P_tau(jquad)%matrix, t_2c_tmp)
1511 92 : CALL dbt_copy(t_2c_tmp, t_P, move_data=.TRUE.)
1512 92 : CALL dbt_filter(t_P, eps_filter)
1513 92 : CALL dbt_destroy(t_2c_tmp)
1514 :
1515 : CALL perform_2c_ops(force, t_KBKT, force_data, fac, t_B(jquad), t_P, t_2c_RI, t_2c_RI_2, &
1516 92 : use_virial, atom_of_kind, kind_of, eps_filter, dbcsr_nflop, unit_nr_dbcsr)
1517 92 : CALL get_tensor_occupancy(t_KBKT, nze_KBK, occ_KBK)
1518 :
1519 : !Calculate the pseudo-density matrix in tensor form. There are a few useless arguments for SOS-MP2
1520 : CALL compute_mat_dm_global(grid, nmo, cfm_mo_coeff(ispin), homo(ispin), propagator, &
1521 : matrix_s, ispin, Eigenval(:, ispin), e_fermi, eps_filter, &
1522 : mp2_env%ri_rpa_im_time%memory_info, unit_nr, &
1523 92 : jquad, .FALSE., .FALSE., qs_env, dummy_int, dummy_ptr, para_env)
1524 :
1525 92 : CALL dbt_create(propagator(propagator_sector_occupied, jquad, 1)%matrix, t_2c_tmp)
1526 92 : CALL dbt_copy_matrix_to_tensor(propagator(propagator_sector_occupied, jquad, 1)%matrix, t_2c_tmp)
1527 92 : CALL dbt_copy(t_2c_tmp, t_dm_occ, move_data=.TRUE.)
1528 92 : CALL dbt_filter(t_dm_occ, eps_filter)
1529 92 : CALL dbt_destroy(t_2c_tmp)
1530 :
1531 92 : CALL dbt_create(propagator(propagator_sector_virtual, jquad, 1)%matrix, t_2c_tmp)
1532 92 : CALL dbt_copy_matrix_to_tensor(propagator(propagator_sector_virtual, jquad, 1)%matrix, t_2c_tmp)
1533 92 : CALL dbt_copy(t_2c_tmp, t_dm_virt, move_data=.TRUE.)
1534 92 : CALL dbt_filter(t_dm_virt, eps_filter)
1535 92 : CALL dbt_destroy(t_2c_tmp)
1536 :
1537 : !Deal with the 3-center quantities.
1538 : CALL perform_3c_ops(force, t_R_occ, t_R_virt, force_data, fac, cut_memory, n_mem_RI, &
1539 : t_KBKT, t_dm_occ, t_dm_virt, t_3c_O, t_3c_M, t_M_occ, t_M_virt, t_3c_0, t_3c_1, &
1540 : t_3c_3, t_3c_4, t_3c_5, t_3c_6, t_3c_7, t_3c_8, t_3c_sparse, t_3c_help_1, t_3c_help_2, &
1541 : t_3c_ints, t_3c_work, starts_array_mc, ends_array_mc, batch_start_RI, &
1542 : batch_end_RI, t_3c_O_compressed, t_3c_O_ind, use_virial, &
1543 : atom_of_kind, kind_of, eps_filter, occ_ddint, nze_ddint, dbcsr_nflop, &
1544 92 : unit_nr_dbcsr, mp2_env)
1545 :
1546 92 : CALL timeset(routineN//"_dbcsr", handle2)
1547 : !We go back to DBCSR matrices from now on
1548 : !Note: R matrices are in fact symmetric, but use a normal type for convenience
1549 92 : CALL dbt_create(matrix_s(1)%matrix, t_2c_tmp)
1550 92 : CALL dbt_copy(t_R_occ, t_2c_tmp, move_data=.TRUE.)
1551 92 : CALL dbt_copy_tensor_to_matrix(t_2c_tmp, R_occ)
1552 :
1553 92 : CALL dbt_copy(t_R_virt, t_2c_tmp, move_data=.TRUE.)
1554 92 : CALL dbt_copy_tensor_to_matrix(t_2c_tmp, R_virt)
1555 :
1556 : !Iteratively calculate the Y1 and Y2 matrices
1557 92 : CALL dbcsr_copy(dbcsr_work_symm, matrix_ks(ispin)%matrix)
1558 92 : CALL dbcsr_add(dbcsr_work_symm, matrix_s(1)%matrix, 1.0_dp, -e_fermi)
1559 : CALL dbcsr_multiply('N', 'N', tau, force_data%P_occ(ispin)%matrix, &
1560 92 : dbcsr_work_symm, 0.0_dp, dbcsr_work1)
1561 92 : CALL build_Y_matrix(Y_1, dbcsr_work1, force_data%P_occ(ispin)%matrix, R_virt, eps_filter)
1562 92 : CALL matrix_exponential(exp_occ, dbcsr_work1, 1.0_dp, 1.0_dp, eps_filter)
1563 :
1564 : CALL dbcsr_multiply('N', 'N', -tau, force_data%P_virt(ispin)%matrix, &
1565 92 : dbcsr_work_symm, 0.0_dp, dbcsr_work1)
1566 92 : CALL build_Y_matrix(Y_2, dbcsr_work1, force_data%P_virt(ispin)%matrix, R_occ, eps_filter)
1567 92 : CALL matrix_exponential(exp_virt, dbcsr_work1, 1.0_dp, 1.0_dp, eps_filter)
1568 :
1569 : !The force contribution coming from [-S^-1*(e^-tau*P_virt*F)^T*R_occ*S^-1
1570 : ! +tau*S^-1*Y_2^T*F*S^-1] * der_S
1571 : !as well as -tau*e_fermi*Y_1*P^occ + tau*e_fermi*Y_2*P^virt
1572 92 : CALL dbcsr_multiply('N', 'N', 1.0_dp, R_occ, force_data%inv_ovlp, 0.0_dp, dbcsr_work1)
1573 92 : CALL dbcsr_multiply('T', 'N', 1.0_dp, exp_virt, dbcsr_work1, 0.0_dp, dbcsr_work3)
1574 92 : CALL dbcsr_multiply('N', 'N', 1.0_dp, force_data%inv_ovlp, dbcsr_work3, 0.0_dp, dbcsr_work2)
1575 :
1576 92 : CALL dbcsr_multiply('N', 'T', tau, force_data%inv_ovlp, Y_2, 0.0_dp, dbcsr_work3)
1577 92 : CALL dbcsr_multiply('N', 'N', 1.0_dp, dbcsr_work3, dbcsr_work_symm, 0.0_dp, dbcsr_work1)
1578 92 : CALL dbcsr_multiply('N', 'N', -1.0_dp, dbcsr_work1, force_data%inv_ovlp, 1.0_dp, dbcsr_work2)
1579 :
1580 92 : CALL dbcsr_multiply('N', 'T', tau*e_fermi, force_data%P_occ(ispin)%matrix, Y_1, 1.0_dp, dbcsr_work2)
1581 92 : CALL dbcsr_multiply('N', 'T', -tau*e_fermi, force_data%P_virt(ispin)%matrix, Y_2, 1.0_dp, dbcsr_work2)
1582 :
1583 92 : CALL dbt_copy_matrix_to_tensor(dbcsr_work2, t_2c_tmp)
1584 92 : CALL dbt_copy(t_2c_tmp, t_2c_AO, move_data=.TRUE.)
1585 :
1586 92 : pref = -1.0_dp*fac
1587 : CALL get_2c_der_force(force, t_2c_AO, force_data%t_2c_der_ovlp, atom_of_kind, &
1588 92 : kind_of, force_data%idx_to_at_AO, pref, do_ovlp=.TRUE.)
1589 :
1590 92 : IF (use_virial) CALL dbcsr_add(virial_ovlp, dbcsr_work2, 1.0_dp, pref)
1591 :
1592 : !The final contribution from Tr[(tau*Y_1*P_occ - tau*Y_2*P_virt) * der_F]
1593 : CALL dbcsr_multiply('N', 'N', fac*tau, Y_1, force_data%P_occ(ispin)%matrix, 1.0_dp, &
1594 92 : force_data%sum_YP_tau(ispin)%matrix, retain_sparsity=.TRUE.)
1595 : CALL dbcsr_multiply('N', 'N', -fac*tau, Y_2, force_data%P_virt(ispin)%matrix, 1.0_dp, &
1596 92 : force_data%sum_YP_tau(ispin)%matrix, retain_sparsity=.TRUE.)
1597 :
1598 92 : spin_fac = 0.5_dp*fac
1599 92 : IF (open_shell) spin_fac = 2.0_dp*spin_fac
1600 : !Build-up the RHS of the response equation.
1601 : CALL dbcsr_multiply('N', 'N', 1.0_dp*spin_fac, R_virt, exp_occ, 1.0_dp, &
1602 92 : force_data%sum_O_tau(ispin)%matrix, retain_sparsity=.TRUE.)
1603 : CALL dbcsr_multiply('N', 'N', -1.0_dp*spin_fac, R_occ, exp_virt, 1.0_dp, &
1604 92 : force_data%sum_O_tau(ispin)%matrix, retain_sparsity=.TRUE.)
1605 : CALL dbcsr_multiply('N', 'N', tau*spin_fac, dbcsr_work_symm, Y_1, 1.0_dp, &
1606 92 : force_data%sum_O_tau(ispin)%matrix, retain_sparsity=.TRUE.)
1607 : CALL dbcsr_multiply('N', 'N', tau*spin_fac, dbcsr_work_symm, Y_2, 1.0_dp, &
1608 92 : force_data%sum_O_tau(ispin)%matrix, retain_sparsity=.TRUE.)
1609 :
1610 92 : CALL timestop(handle2)
1611 :
1612 : !Print some info
1613 92 : CALL para_env%sync()
1614 92 : t2 = m_walltime()
1615 92 : dbcsr_time = dbcsr_time + t2 - t1
1616 :
1617 92 : IF (unit_nr > 0) THEN
1618 : WRITE (unit_nr, '(/T3,A,1X,I3,A)') &
1619 46 : 'RPA_LOW_SCALING_INFO| Info for time point', jquad, ' (gradients)'
1620 : WRITE (unit_nr, '(T6,A,T56,F25.6)') &
1621 46 : 'Time:', t2 - t1
1622 : WRITE (unit_nr, '(T6,A,T63,ES7.1,1X,A1,1X,F7.3,A1)') &
1623 46 : 'Occupancy of 3c AO derivs:', REAL(nze_der_AO, dp), '/', occ_der_AO*100, '%'
1624 : WRITE (unit_nr, '(T6,A,T63,ES7.1,1X,A1,1X,F7.3,A1)') &
1625 46 : 'Occupancy of 3c RI derivs:', REAL(nze_der_RI, dp), '/', occ_der_RI*100, '%'
1626 : WRITE (unit_nr, '(T6,A,T63,ES7.1,1X,A1,1X,F7.3,A1)') &
1627 46 : 'Occupancy of the Docc * Dvirt * 3c-int tensor', REAL(nze_ddint, dp), '/', occ_ddint*100, '%'
1628 : WRITE (unit_nr, '(T6,A,T63,ES7.1,1X,A1,1X,F7.3,A1)') &
1629 46 : 'Occupancy of KBK^T 2c-tensor:', REAL(nze_KBK, dp), '/', occ_KBK*100, '%'
1630 46 : CALL m_flush(unit_nr)
1631 : END IF
1632 :
1633 : !intermediate clean-up
1634 92 : CALL dbcsr_release(Y_1)
1635 92 : CALL dbcsr_release(Y_2)
1636 496 : CALL dbt_destroy(t_2c_tmp)
1637 :
1638 : END DO !jquad
1639 :
1640 36 : CALL dbt_batched_contract_finalize(t_3c_0)
1641 36 : CALL dbt_batched_contract_finalize(t_3c_1)
1642 36 : CALL dbt_batched_contract_finalize(t_3c_3)
1643 36 : CALL dbt_batched_contract_finalize(t_M_occ)
1644 36 : CALL dbt_batched_contract_finalize(t_M_virt)
1645 :
1646 36 : CALL dbt_batched_contract_finalize(t_3c_ints)
1647 36 : CALL dbt_batched_contract_finalize(t_3c_work)
1648 :
1649 36 : CALL dbt_batched_contract_finalize(t_3c_4)
1650 36 : CALL dbt_batched_contract_finalize(t_3c_5)
1651 36 : CALL dbt_batched_contract_finalize(t_3c_6)
1652 36 : CALL dbt_batched_contract_finalize(t_3c_7)
1653 36 : CALL dbt_batched_contract_finalize(t_3c_8)
1654 36 : CALL dbt_batched_contract_finalize(t_3c_sparse)
1655 :
1656 : !Calculate the 2c and 3c contributions to the virial
1657 36 : IF (use_virial) THEN
1658 2 : CALL dbt_copy(force_data%t_3c_virial_split, force_data%t_3c_virial, move_data=.TRUE.)
1659 : CALL calc_3c_virial(work_virial, force_data%t_3c_virial, 1.0_dp, qs_env, force_data%nl_3c, &
1660 : basis_set_ri_aux, basis_set_ao, basis_set_ao, mp2_env%ri_metric, &
1661 2 : der_eps=mp2_env%ri_rpa_im_time%eps_filter, op_pos=1)
1662 :
1663 : CALL calc_2c_virial(work_virial, force_data%RI_virial_met, 1.0_dp, qs_env, force_data%nl_2c_met, &
1664 2 : basis_set_ri_aux, basis_set_ri_aux, mp2_env%ri_metric)
1665 2 : CALL dbcsr_clear(force_data%RI_virial_met)
1666 :
1667 2 : IF (.NOT. force_data%do_periodic) THEN
1668 : CALL calc_2c_virial(work_virial, force_data%RI_virial_pot, 1.0_dp, qs_env, force_data%nl_2c_pot, &
1669 0 : basis_set_ri_aux, basis_set_ri_aux, mp2_env%potential_parameter)
1670 0 : CALL dbcsr_clear(force_data%RI_virial_pot)
1671 : END IF
1672 :
1673 2 : identity_pot%potential_type = do_potential_id
1674 : CALL calc_2c_virial(work_virial_ovlp, virial_ovlp, 1.0_dp, qs_env, force_data%nl_2c_ovlp, &
1675 2 : basis_set_ao, basis_set_ao, identity_pot)
1676 2 : CALL dbcsr_release(virial_ovlp)
1677 :
1678 8 : DO k_xyz = 1, 3
1679 26 : DO j_xyz = 1, 3
1680 78 : DO i_xyz = 1, 3
1681 : virial%pv_mp2(i_xyz, j_xyz) = virial%pv_mp2(i_xyz, j_xyz) &
1682 54 : - work_virial(i_xyz, k_xyz)*cell%hmat(j_xyz, k_xyz)
1683 : virial%pv_overlap(i_xyz, j_xyz) = virial%pv_overlap(i_xyz, j_xyz) &
1684 54 : - work_virial_ovlp(i_xyz, k_xyz)*cell%hmat(j_xyz, k_xyz)
1685 : virial%pv_virial(i_xyz, j_xyz) = virial%pv_virial(i_xyz, j_xyz) &
1686 : - work_virial(i_xyz, k_xyz)*cell%hmat(j_xyz, k_xyz) &
1687 72 : - work_virial_ovlp(i_xyz, k_xyz)*cell%hmat(j_xyz, k_xyz)
1688 : END DO
1689 : END DO
1690 : END DO
1691 : END IF
1692 :
1693 : !Calculate the periodic contributions of (P|Q) to the force and the virial
1694 36 : work_virial = 0.0_dp
1695 36 : IF (force_data%do_periodic) THEN
1696 18 : IF (mp2_env%eri_method == do_eri_gpw) THEN
1697 6 : CALL get_2c_gpw_forces(force_data%G_PQ, force, work_virial, use_virial, mp2_env, qs_env)
1698 12 : ELSE IF (mp2_env%eri_method == do_eri_mme) THEN
1699 12 : CALL get_2c_mme_forces(force_data%G_PQ, force, mp2_env, qs_env)
1700 12 : IF (use_virial) CPABORT("Stress tensor not available with MME intrgrals")
1701 : ELSE
1702 0 : CPABORT("Periodic case not possible with OS integrals")
1703 : END IF
1704 18 : CALL dbcsr_clear(force_data%G_PQ)
1705 : END IF
1706 :
1707 36 : IF (use_virial) THEN
1708 26 : virial%pv_mp2 = virial%pv_mp2 + work_virial
1709 26 : virial%pv_virial = virial%pv_virial + work_virial
1710 2 : virial%pv_calculate = .FALSE.
1711 :
1712 6 : DO ibasis = 1, SIZE(basis_set_ao)
1713 4 : orb_basis => basis_set_ao(ibasis)%gto_basis_set
1714 4 : CALL init_interaction_radii_orb_basis(orb_basis, eps_pgf_orb_old)
1715 4 : ri_basis => basis_set_ri_aux(ibasis)%gto_basis_set
1716 6 : CALL init_interaction_radii_orb_basis(ri_basis, eps_pgf_orb_old)
1717 : END DO
1718 : END IF
1719 :
1720 : !clean-up
1721 36 : IF (ASSOCIATED(dummy_ptr)) DEALLOCATE (dummy_ptr)
1722 128 : DO jquad = 1, num_integ_points
1723 128 : CALL dbt_destroy(t_B(jquad))
1724 : END DO
1725 36 : CALL dbt_destroy(t_P)
1726 36 : CALL dbt_destroy(t_3c_0)
1727 36 : CALL dbt_destroy(t_3c_1)
1728 36 : CALL dbt_destroy(t_3c_3)
1729 36 : CALL dbt_destroy(t_3c_4)
1730 36 : CALL dbt_destroy(t_3c_5)
1731 36 : CALL dbt_destroy(t_3c_6)
1732 36 : CALL dbt_destroy(t_3c_7)
1733 36 : CALL dbt_destroy(t_3c_8)
1734 36 : CALL dbt_destroy(t_3c_sparse)
1735 36 : CALL dbt_destroy(t_3c_help_1)
1736 36 : CALL dbt_destroy(t_3c_help_2)
1737 36 : CALL dbt_destroy(t_3c_ints)
1738 36 : CALL dbt_destroy(t_3c_work)
1739 36 : CALL dbt_destroy(t_R_occ)
1740 36 : CALL dbt_destroy(t_R_virt)
1741 36 : CALL dbt_destroy(t_dm_occ)
1742 36 : CALL dbt_destroy(t_dm_virt)
1743 36 : CALL dbt_destroy(t_KBKT)
1744 36 : CALL dbt_destroy(t_M_occ)
1745 36 : CALL dbt_destroy(t_M_virt)
1746 36 : CALL dbcsr_release(R_occ)
1747 36 : CALL dbcsr_release(R_virt)
1748 36 : CALL dbcsr_release(dbcsr_work_symm)
1749 36 : CALL dbcsr_release(dbcsr_work1)
1750 36 : CALL dbcsr_release(dbcsr_work2)
1751 36 : CALL dbcsr_release(dbcsr_work3)
1752 36 : CALL dbcsr_release(exp_occ)
1753 36 : CALL dbcsr_release(exp_virt)
1754 :
1755 36 : CALL dbt_destroy(t_2c_RI)
1756 36 : CALL dbt_destroy(t_2c_RI_2)
1757 36 : CALL dbt_destroy(t_2c_AO)
1758 36 : CALL dbcsr_deallocate_matrix_set(propagator)
1759 36 : CALL dbcsr_deallocate_matrix_set(mat_P_tau)
1760 :
1761 36 : CALL timestop(handle)
1762 :
1763 200 : END SUBROUTINE calc_rpa_loop_forces
1764 :
1765 : ! **************************************************************************************************
1766 : !> \brief This subroutines performs the 2c tensor operations that are common accros low-scaling RPA
1767 : !> and SOS-MP2, including forces and virial
1768 : !> \param force ...
1769 : !> \param t_KBKT returns the 2c tensor product of K*B*K^T
1770 : !> \param force_data ...
1771 : !> \param fac ...
1772 : !> \param t_B depending on RPA or SOS-MP2, t_B contains (1 + Q)^-1 - 1 or simply Q, respectively
1773 : !> \param t_P ...
1774 : !> \param t_2c_RI ...
1775 : !> \param t_2c_RI_2 ...
1776 : !> \param use_virial ...
1777 : !> \param atom_of_kind ...
1778 : !> \param kind_of ...
1779 : !> \param eps_filter ...
1780 : !> \param dbcsr_nflop ...
1781 : !> \param unit_nr_dbcsr ...
1782 : ! **************************************************************************************************
1783 170 : SUBROUTINE perform_2c_ops(force, t_KBKT, force_data, fac, t_B, t_P, t_2c_RI, t_2c_RI_2, use_virial, &
1784 170 : atom_of_kind, kind_of, eps_filter, dbcsr_nflop, unit_nr_dbcsr)
1785 :
1786 : TYPE(qs_force_type), DIMENSION(:), POINTER :: force
1787 : TYPE(dbt_type), INTENT(INOUT) :: t_KBKT
1788 : TYPE(im_time_force_type), INTENT(INOUT) :: force_data
1789 : REAL(dp), INTENT(IN) :: fac
1790 : TYPE(dbt_type), INTENT(INOUT) :: t_B, t_P, t_2c_RI, t_2c_RI_2
1791 : LOGICAL, INTENT(IN) :: use_virial
1792 : INTEGER, DIMENSION(:), INTENT(IN) :: atom_of_kind, kind_of
1793 : REAL(dp), INTENT(IN) :: eps_filter
1794 : INTEGER(int_8), INTENT(INOUT) :: dbcsr_nflop
1795 : INTEGER, INTENT(IN) :: unit_nr_dbcsr
1796 :
1797 : CHARACTER(LEN=*), PARAMETER :: routineN = 'perform_2c_ops'
1798 :
1799 : INTEGER :: handle
1800 : INTEGER(int_8) :: flop
1801 : REAL(dp) :: pref
1802 2890 : TYPE(dbt_type) :: t_2c_tmp, t_2c_virial
1803 :
1804 170 : CALL timeset(routineN, handle)
1805 :
1806 170 : IF (use_virial) CALL dbt_create(force_data%RI_virial_pot, t_2c_virial)
1807 :
1808 : !P^T*K*B + P*K*B^T (note we calculate and save K*B*K^T for later, and P=P^T)
1809 : CALL dbt_contract(1.0_dp, force_data%t_2c_K, t_B, 0.0_dp, t_2c_RI, &
1810 : contract_1=[2], notcontract_1=[1], &
1811 : contract_2=[1], notcontract_2=[2], &
1812 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
1813 170 : flop=flop, unit_nr=unit_nr_dbcsr)
1814 170 : dbcsr_nflop = dbcsr_nflop + flop
1815 :
1816 : CALL dbt_contract(1.0_dp, t_2c_RI, force_data%t_2c_K, 0.0_dp, t_KBKT, &
1817 : contract_1=[2], notcontract_1=[1], &
1818 : contract_2=[2], notcontract_2=[1], &
1819 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
1820 170 : flop=flop, unit_nr=unit_nr_dbcsr)
1821 170 : dbcsr_nflop = dbcsr_nflop + flop
1822 :
1823 : CALL dbt_contract(2.0_dp, t_P, t_2c_RI, 0.0_dp, t_2c_RI_2, & !t_2c_RI_2 holds P^T*K*B
1824 : contract_1=[2], notcontract_1=[1], &
1825 : contract_2=[1], notcontract_2=[2], &
1826 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
1827 170 : flop=flop, unit_nr=unit_nr_dbcsr)
1828 170 : dbcsr_nflop = dbcsr_nflop + flop
1829 170 : CALL dbt_clear(t_2c_RI)
1830 : !t_2c_RI_2 currently holds 2*P^T*K*B = P^T*K*B + P*K*B^T (because of symmetry)
1831 :
1832 : !For the metric contribution, we need S^-1*(P^T*K*B + P*K*B^T)*K^T
1833 : CALL dbt_contract(1.0_dp, force_data%t_2c_inv_metric, t_2c_RI_2, 0.0_dp, t_2c_RI, &
1834 : contract_1=[2], notcontract_1=[1], &
1835 : contract_2=[1], notcontract_2=[2], &
1836 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
1837 170 : flop=flop, unit_nr=unit_nr_dbcsr)
1838 170 : dbcsr_nflop = dbcsr_nflop + flop
1839 :
1840 : CALL dbt_contract(1.0_dp, t_2c_RI, force_data%t_2c_K, 0.0_dp, t_2c_RI_2, &
1841 : contract_1=[2], notcontract_1=[1], &
1842 : contract_2=[2], notcontract_2=[1], &
1843 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
1844 170 : flop=flop, unit_nr=unit_nr_dbcsr)
1845 170 : dbcsr_nflop = dbcsr_nflop + flop
1846 :
1847 : !Here we do the trace for the force
1848 170 : pref = -1.0_dp*fac
1849 : CALL get_2c_der_force(force, t_2c_RI_2, force_data%t_2c_der_metric, atom_of_kind, &
1850 170 : kind_of, force_data%idx_to_at_RI, pref, do_mp2=.TRUE.)
1851 170 : IF (use_virial) THEN
1852 12 : CALL dbt_copy(t_2c_RI_2, t_2c_virial)
1853 12 : CALL dbt_scale(t_2c_virial, pref)
1854 12 : CALL dbt_copy_tensor_to_matrix(t_2c_virial, force_data%RI_virial_met, summation=.TRUE.)
1855 12 : CALL dbt_clear(t_2c_virial)
1856 : END IF
1857 :
1858 : !For the potential contribution, we need S^-1*(P^T*K*B + P*K*B^T)*V^-0.5
1859 : !some of it is still in t_2c_RI: ( S^-1*(P^T*K*B + P*K*B^T) )
1860 : CALL dbt_contract(1.0_dp, t_2c_RI, force_data%t_2c_pot_msqrt, 0.0_dp, t_2c_RI_2, &
1861 : contract_1=[2], notcontract_1=[1], &
1862 : contract_2=[1], notcontract_2=[2], &
1863 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
1864 170 : flop=flop, unit_nr=unit_nr_dbcsr)
1865 170 : dbcsr_nflop = dbcsr_nflop + flop
1866 :
1867 : !Here we do the trace for the force. In the periodic case, we store the matrix in G_PQ for later
1868 170 : pref = 0.5_dp*fac
1869 170 : IF (force_data%do_periodic) THEN
1870 76 : CALL dbt_scale(t_2c_RI_2, pref)
1871 76 : CALL dbt_create(force_data%G_PQ, t_2c_tmp)
1872 76 : CALL dbt_copy(t_2c_RI_2, t_2c_tmp, move_data=.TRUE.)
1873 76 : CALL dbt_copy_tensor_to_matrix(t_2c_tmp, force_data%G_PQ, summation=.TRUE.)
1874 76 : CALL dbt_destroy(t_2c_tmp)
1875 : ELSE
1876 : CALL get_2c_der_force(force, t_2c_RI_2, force_data%t_2c_der_pot, atom_of_kind, &
1877 94 : kind_of, force_data%idx_to_at_RI, pref, do_mp2=.TRUE.)
1878 :
1879 94 : IF (use_virial) THEN
1880 0 : CALL dbt_copy(t_2c_RI_2, t_2c_virial)
1881 0 : CALL dbt_scale(t_2c_virial, pref)
1882 0 : CALL dbt_copy_tensor_to_matrix(t_2c_virial, force_data%RI_virial_pot, summation=.TRUE.)
1883 0 : CALL dbt_clear(t_2c_virial)
1884 : END IF
1885 : END IF
1886 :
1887 170 : CALL dbt_clear(t_2c_RI)
1888 170 : CALL dbt_clear(t_2c_RI_2)
1889 :
1890 170 : IF (use_virial) CALL dbt_destroy(t_2c_virial)
1891 :
1892 170 : CALL timestop(handle)
1893 :
1894 170 : END SUBROUTINE perform_2c_ops
1895 :
1896 : ! **************************************************************************************************
1897 : !> \brief This subroutines performs the 3c tensor operations that are common accros low-scaling RPA
1898 : !> and SOS-MP2, including forces and virial
1899 : !> \param force ...
1900 : !> \param t_R_occ ...
1901 : !> \param t_R_virt ...
1902 : !> \param force_data ...
1903 : !> \param fac ...
1904 : !> \param cut_memory ...
1905 : !> \param n_mem_RI ...
1906 : !> \param t_KBKT ...
1907 : !> \param t_dm_occ ...
1908 : !> \param t_dm_virt ...
1909 : !> \param t_3c_O ...
1910 : !> \param t_3c_M ...
1911 : !> \param t_M_occ ...
1912 : !> \param t_M_virt ...
1913 : !> \param t_3c_0 ...
1914 : !> \param t_3c_1 ...
1915 : !> \param t_3c_3 ...
1916 : !> \param t_3c_4 ...
1917 : !> \param t_3c_5 ...
1918 : !> \param t_3c_6 ...
1919 : !> \param t_3c_7 ...
1920 : !> \param t_3c_8 ...
1921 : !> \param t_3c_sparse ...
1922 : !> \param t_3c_help_1 ...
1923 : !> \param t_3c_help_2 ...
1924 : !> \param t_3c_ints ...
1925 : !> \param t_3c_work ...
1926 : !> \param starts_array_mc ...
1927 : !> \param ends_array_mc ...
1928 : !> \param batch_start_RI ...
1929 : !> \param batch_end_RI ...
1930 : !> \param t_3c_O_compressed ...
1931 : !> \param t_3c_O_ind ...
1932 : !> \param use_virial ...
1933 : !> \param atom_of_kind ...
1934 : !> \param kind_of ...
1935 : !> \param eps_filter ...
1936 : !> \param occ_ddint ...
1937 : !> \param nze_ddint ...
1938 : !> \param dbcsr_nflop ...
1939 : !> \param unit_nr_dbcsr ...
1940 : !> \param mp2_env ...
1941 : ! **************************************************************************************************
1942 170 : SUBROUTINE perform_3c_ops(force, t_R_occ, t_R_virt, force_data, fac, cut_memory, n_mem_RI, &
1943 : t_KBKT, t_dm_occ, t_dm_virt, t_3c_O, t_3c_M, t_M_occ, t_M_virt, t_3c_0, t_3c_1, &
1944 : t_3c_3, t_3c_4, t_3c_5, t_3c_6, t_3c_7, t_3c_8, t_3c_sparse, t_3c_help_1, t_3c_help_2, &
1945 170 : t_3c_ints, t_3c_work, starts_array_mc, ends_array_mc, batch_start_RI, &
1946 170 : batch_end_RI, t_3c_O_compressed, t_3c_O_ind, use_virial, &
1947 170 : atom_of_kind, kind_of, eps_filter, occ_ddint, nze_ddint, dbcsr_nflop, &
1948 : unit_nr_dbcsr, mp2_env)
1949 :
1950 : TYPE(qs_force_type), DIMENSION(:), POINTER :: force
1951 : TYPE(dbt_type), INTENT(INOUT) :: t_R_occ, t_R_virt
1952 : TYPE(im_time_force_type), INTENT(INOUT) :: force_data
1953 : REAL(dp), INTENT(IN) :: fac
1954 : INTEGER, INTENT(IN) :: cut_memory, n_mem_RI
1955 : TYPE(dbt_type), INTENT(INOUT) :: t_KBKT, t_dm_occ, t_dm_virt, t_3c_O, t_3c_M, t_M_occ, &
1956 : t_M_virt, t_3c_0, t_3c_1, t_3c_3, t_3c_4, t_3c_5, t_3c_6, t_3c_7, t_3c_8, t_3c_sparse, &
1957 : t_3c_help_1, t_3c_help_2, t_3c_ints, t_3c_work
1958 : INTEGER, DIMENSION(:), INTENT(IN) :: starts_array_mc, ends_array_mc, &
1959 : batch_start_RI, batch_end_RI
1960 : TYPE(hfx_compression_type), DIMENSION(:) :: t_3c_O_compressed
1961 : TYPE(block_ind_type), DIMENSION(:), INTENT(INOUT) :: t_3c_O_ind
1962 : LOGICAL, INTENT(IN) :: use_virial
1963 : INTEGER, DIMENSION(:), INTENT(IN) :: atom_of_kind, kind_of
1964 : REAL(dp), INTENT(IN) :: eps_filter
1965 : REAL(dp), INTENT(INOUT) :: occ_ddint
1966 : INTEGER(int_8), INTENT(INOUT) :: nze_ddint, dbcsr_nflop
1967 : INTEGER, INTENT(IN) :: unit_nr_dbcsr
1968 : TYPE(mp2_type) :: mp2_env
1969 :
1970 : CHARACTER(LEN=*), PARAMETER :: routineN = 'perform_3c_ops'
1971 :
1972 : INTEGER :: dummy_int, handle, handle2, i_mem, &
1973 : i_xyz, j_mem, k_mem
1974 : INTEGER(int_8) :: flop, nze
1975 : INTEGER, DIMENSION(2, 1) :: ibounds, jbounds, kbounds
1976 : INTEGER, DIMENSION(2, 2) :: bounds_2c
1977 : INTEGER, DIMENSION(2, 3) :: bounds_cpy
1978 : INTEGER, DIMENSION(3) :: bounds_3c
1979 : REAL(dp) :: memory, occ, pref
1980 170 : TYPE(block_ind_type), ALLOCATABLE, DIMENSION(:, :) :: blk_indices
1981 : TYPE(hfx_compression_type), ALLOCATABLE, &
1982 170 : DIMENSION(:, :) :: store_3c
1983 :
1984 170 : CALL timeset(routineN, handle)
1985 :
1986 170 : CALL dbt_get_info(t_3c_M, nfull_total=bounds_3c)
1987 :
1988 : !Pre-compute and compress KBK^T * (pq|R)
1989 360910 : ALLOCATE (store_3c(n_mem_RI, cut_memory))
1990 1700 : ALLOCATE (blk_indices(n_mem_RI, cut_memory))
1991 170 : memory = 0.0_dp
1992 170 : CALL timeset(routineN//"_pre_3c", handle2)
1993 : !temporarily build the full int 3c tensor
1994 170 : CALL dbt_copy(t_3c_O, t_3c_0)
1995 510 : DO i_mem = 1, cut_memory
1996 : CALL decompress_tensor(t_3c_O, t_3c_O_ind(i_mem)%ind, t_3c_O_compressed(i_mem), &
1997 340 : mp2_env%ri_rpa_im_time%eps_compress)
1998 340 : CALL dbt_copy(t_3c_O, t_3c_ints)
1999 340 : CALL dbt_copy(t_3c_O, t_3c_0, move_data=.TRUE., summation=.TRUE.)
2000 :
2001 1190 : DO k_mem = 1, n_mem_RI
2002 2040 : kbounds(:, 1) = [batch_start_RI(k_mem), batch_end_RI(k_mem)]
2003 :
2004 680 : CALL alloc_containers(store_3c(k_mem, i_mem), 1)
2005 :
2006 : !contract with KBK^T over the RI index and store
2007 680 : CALL dbt_batched_contract_init(t_KBKT)
2008 : CALL dbt_contract(1.0_dp, t_KBKT, t_3c_ints, 0.0_dp, t_3c_work, &
2009 : contract_1=[2], notcontract_1=[1], &
2010 : contract_2=[1], notcontract_2=[2, 3], &
2011 : map_1=[1], map_2=[2, 3], filter_eps=eps_filter, &
2012 680 : bounds_2=kbounds, flop=flop, unit_nr=unit_nr_dbcsr)
2013 680 : CALL dbt_batched_contract_finalize(t_KBKT)
2014 680 : dbcsr_nflop = dbcsr_nflop + flop
2015 :
2016 680 : CALL dbt_copy(t_3c_work, t_3c_M, move_data=.TRUE.)
2017 : CALL compress_tensor(t_3c_M, blk_indices(k_mem, i_mem)%ind, store_3c(k_mem, i_mem), &
2018 1020 : mp2_env%ri_rpa_im_time%eps_compress, memory)
2019 : END DO
2020 : END DO !i_mem
2021 170 : CALL dbt_clear(t_3c_M)
2022 170 : CALL dbt_copy(t_3c_M, t_3c_ints)
2023 170 : CALL timestop(handle2)
2024 :
2025 170 : CALL dbt_batched_contract_init(t_R_occ)
2026 170 : CALL dbt_batched_contract_init(t_R_virt)
2027 510 : DO i_mem = 1, cut_memory
2028 1020 : ibounds(:, 1) = [starts_array_mc(i_mem), ends_array_mc(i_mem)]
2029 :
2030 : !Compute the matrices M (integrals in t_3c_0)
2031 340 : CALL timeset(routineN//"_3c_M", handle2)
2032 340 : CALL dbt_batched_contract_init(t_dm_occ)
2033 : CALL dbt_contract(1.0_dp, t_3c_0, t_dm_occ, 0.0_dp, t_3c_1, &
2034 : contract_1=[3], notcontract_1=[1, 2], &
2035 : contract_2=[1], notcontract_2=[2], &
2036 : map_1=[1, 2], map_2=[3], filter_eps=eps_filter, &
2037 340 : bounds_3=ibounds, flop=flop, unit_nr=unit_nr_dbcsr)
2038 340 : dbcsr_nflop = dbcsr_nflop + flop
2039 340 : CALL dbt_batched_contract_finalize(t_dm_occ)
2040 340 : CALL dbt_copy(t_3c_1, t_M_occ, order=[1, 3, 2], move_data=.TRUE.)
2041 :
2042 340 : CALL dbt_batched_contract_init(t_dm_virt)
2043 : CALL dbt_contract(1.0_dp, t_3c_0, t_dm_virt, 0.0_dp, t_3c_1, &
2044 : contract_1=[3], notcontract_1=[1, 2], &
2045 : contract_2=[1], notcontract_2=[2], &
2046 : map_1=[1, 2], map_2=[3], filter_eps=eps_filter, &
2047 340 : bounds_3=ibounds, flop=flop, unit_nr=unit_nr_dbcsr)
2048 340 : dbcsr_nflop = dbcsr_nflop + flop
2049 340 : CALL dbt_batched_contract_finalize(t_dm_virt)
2050 340 : CALL dbt_copy(t_3c_1, t_M_virt, order=[1, 3, 2], move_data=.TRUE.)
2051 340 : CALL timestop(handle2)
2052 :
2053 : !Compute the R matrices
2054 340 : CALL timeset(routineN//"_3c_R", handle2)
2055 1020 : DO k_mem = 1, n_mem_RI
2056 : CALL decompress_tensor(t_3c_M, blk_indices(k_mem, i_mem)%ind, store_3c(k_mem, i_mem), &
2057 680 : mp2_env%ri_rpa_im_time%eps_compress)
2058 680 : CALL dbt_copy(t_3c_M, t_3c_3, move_data=.TRUE.)
2059 :
2060 : CALL dbt_contract(1.0_dp, t_M_occ, t_3c_3, 1.0_dp, t_R_occ, &
2061 : contract_1=[1, 2], notcontract_1=[3], &
2062 : contract_2=[1, 2], notcontract_2=[3], &
2063 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
2064 680 : flop=flop, unit_nr=unit_nr_dbcsr)
2065 680 : dbcsr_nflop = dbcsr_nflop + flop
2066 :
2067 : CALL dbt_contract(1.0_dp, t_M_virt, t_3c_3, 1.0_dp, t_R_virt, &
2068 : contract_1=[1, 2], notcontract_1=[3], &
2069 : contract_2=[1, 2], notcontract_2=[3], &
2070 : map_1=[1], map_2=[2], filter_eps=eps_filter, &
2071 680 : flop=flop, unit_nr=unit_nr_dbcsr)
2072 1020 : dbcsr_nflop = dbcsr_nflop + flop
2073 : END DO
2074 340 : CALL dbt_copy(t_3c_M, t_3c_3)
2075 340 : CALL dbt_copy(t_3c_M, t_M_virt)
2076 340 : CALL timestop(handle2)
2077 :
2078 340 : CALL dbt_copy(t_M_occ, t_3c_4, move_data=.TRUE.)
2079 :
2080 340 : IF (cut_memory > 0) CALL dbt_batched_contract_init(t_KBKT)
2081 1020 : DO j_mem = 1, cut_memory
2082 2040 : jbounds(:, 1) = [starts_array_mc(j_mem), ends_array_mc(j_mem)]
2083 :
2084 2040 : bounds_cpy(:, 1) = [1, bounds_3c(1)]
2085 2040 : bounds_cpy(:, 2) = [starts_array_mc(i_mem), ends_array_mc(i_mem)]
2086 2040 : bounds_cpy(:, 3) = [starts_array_mc(j_mem), ends_array_mc(j_mem)]
2087 680 : CALL dbt_copy(t_3c_sparse, t_3c_7, bounds=bounds_cpy)
2088 :
2089 680 : CALL dbt_batched_contract_init(t_dm_virt)
2090 2040 : DO k_mem = 1, n_mem_RI
2091 4080 : bounds_2c(:, 1) = [batch_start_RI(k_mem), batch_end_RI(k_mem)]
2092 4080 : bounds_2c(:, 2) = [starts_array_mc(i_mem), ends_array_mc(i_mem)]
2093 :
2094 1360 : CALL timeset(routineN//"_3c_dm", handle2)
2095 :
2096 : !Calculate (mu nu| P) * D_occ * D_virt
2097 : !Note: technically need M_occ*D_virt + M_virt*D_occ, but it is equivalent to 2*M_occ*D_virt
2098 : CALL dbt_contract(2.0_dp, t_3c_4, t_dm_virt, 0.0_dp, t_3c_5, &
2099 : contract_1=[3], notcontract_1=[1, 2], &
2100 : contract_2=[1], notcontract_2=[2], &
2101 : map_1=[1, 2], map_2=[3], filter_eps=eps_filter, &
2102 1360 : bounds_2=bounds_2c, bounds_3=jbounds, flop=flop, unit_nr=unit_nr_dbcsr)
2103 1360 : dbcsr_nflop = dbcsr_nflop + flop
2104 :
2105 1360 : CALL get_tensor_occupancy(t_3c_5, nze, occ)
2106 1360 : nze_ddint = nze_ddint + nze
2107 1360 : occ_ddint = occ_ddint + occ
2108 :
2109 : ! Skip the expensive KBK^T contraction when the intermediate block is empty
2110 1360 : IF (nze == 0) THEN
2111 0 : CALL dbt_clear(t_3c_5)
2112 0 : CYCLE
2113 : END IF
2114 :
2115 1360 : CALL dbt_copy(t_3c_5, t_3c_6, move_data=.TRUE.)
2116 1360 : CALL timestop(handle2)
2117 :
2118 : !Calculate the contraction of the above with K*B*K^T
2119 1360 : CALL timeset(routineN//"_3c_KBK", handle2)
2120 : CALL dbt_contract(1.0_dp, t_KBKT, t_3c_6, 0.0_dp, t_3c_7, &
2121 : contract_1=[2], notcontract_1=[1], &
2122 : contract_2=[1], notcontract_2=[2, 3], &
2123 : map_1=[1], map_2=[2, 3], &
2124 1360 : retain_sparsity=.TRUE., flop=flop, unit_nr=unit_nr_dbcsr)
2125 1360 : dbcsr_nflop = dbcsr_nflop + flop
2126 1360 : CALL timestop(handle2)
2127 6120 : CALL dbt_copy(t_3c_7, t_3c_8, summation=.TRUE.)
2128 :
2129 : END DO !k_mem
2130 1020 : CALL dbt_batched_contract_finalize(t_dm_virt)
2131 : END DO !j_mem
2132 340 : IF (cut_memory > 0) CALL dbt_batched_contract_finalize(t_KBKT)
2133 :
2134 340 : CALL dbt_copy(t_3c_8, t_3c_help_1, move_data=.TRUE.)
2135 :
2136 340 : pref = 1.0_dp*fac
2137 1020 : DO k_mem = 1, cut_memory
2138 2720 : DO i_xyz = 1, 3
2139 2040 : CALL dbt_clear(force_data%t_3c_der_RI(i_xyz))
2140 : CALL decompress_tensor(force_data%t_3c_der_RI(i_xyz), force_data%t_3c_der_RI_ind(k_mem, i_xyz)%ind, &
2141 2720 : force_data%t_3c_der_RI_comp(k_mem, i_xyz), mp2_env%ri_rpa_im_time%eps_compress)
2142 : END DO
2143 : CALL get_force_from_3c_trace(force, t_3c_help_1, force_data%t_3c_der_RI, atom_of_kind, kind_of, &
2144 1020 : force_data%idx_to_at_RI, pref, do_mp2=.TRUE., deriv_dim=1)
2145 : END DO
2146 :
2147 340 : IF (use_virial) THEN
2148 24 : CALL dbt_copy(t_3c_help_1, t_3c_help_2)
2149 24 : CALL dbt_scale(t_3c_help_2, pref)
2150 24 : CALL dbt_copy(t_3c_help_2, force_data%t_3c_virial_split, summation=.TRUE., move_data=.TRUE.)
2151 : END IF
2152 :
2153 340 : CALL dbt_copy(t_3c_help_1, t_3c_help_2)
2154 340 : CALL dbt_copy(t_3c_help_1, t_3c_help_2, order=[1, 3, 2], move_data=.TRUE., summation=.TRUE.)
2155 1020 : DO k_mem = 1, cut_memory
2156 2720 : DO i_xyz = 1, 3
2157 2040 : CALL dbt_clear(force_data%t_3c_der_AO(i_xyz))
2158 : CALL decompress_tensor(force_data%t_3c_der_AO(i_xyz), force_data%t_3c_der_AO_ind(k_mem, i_xyz)%ind, &
2159 2720 : force_data%t_3c_der_AO_comp(k_mem, i_xyz), mp2_env%ri_rpa_im_time%eps_compress)
2160 : END DO
2161 : CALL get_force_from_3c_trace(force, t_3c_help_2, force_data%t_3c_der_AO, atom_of_kind, kind_of, &
2162 1020 : force_data%idx_to_at_AO, pref, do_mp2=.TRUE., deriv_dim=3)
2163 : END DO
2164 :
2165 1190 : CALL dbt_clear(t_3c_help_2)
2166 : END DO !i_mem
2167 170 : CALL dbt_batched_contract_finalize(t_R_occ)
2168 170 : CALL dbt_batched_contract_finalize(t_R_virt)
2169 :
2170 510 : DO k_mem = 1, n_mem_RI
2171 1190 : DO i_mem = 1, cut_memory
2172 1020 : CALL dealloc_containers(store_3c(k_mem, i_mem), dummy_int)
2173 : END DO
2174 : END DO
2175 850 : DEALLOCATE (store_3c, blk_indices)
2176 :
2177 170 : CALL timestop(handle)
2178 :
2179 340 : END SUBROUTINE perform_3c_ops
2180 :
2181 : ! **************************************************************************************************
2182 : !> \brief All the forces that can be calculated after the loop on the Laplace quaradture, using
2183 : !> terms collected during the said loop. This inludes the z-vector equation and its reponse
2184 : !> forces, as well as the force coming from the trace with the derivative of the KS matrix
2185 : !> \param force_data ...
2186 : !> \param unit_nr ...
2187 : !> \param qs_env ...
2188 : ! **************************************************************************************************
2189 50 : SUBROUTINE calc_post_loop_forces(force_data, unit_nr, qs_env)
2190 :
2191 : TYPE(im_time_force_type), INTENT(INOUT) :: force_data
2192 : INTEGER, INTENT(IN) :: unit_nr
2193 : TYPE(qs_environment_type), POINTER :: qs_env
2194 :
2195 : CHARACTER(len=*), PARAMETER :: routineN = 'calc_post_loop_forces'
2196 :
2197 : INTEGER :: handle, ispin, nao, nao_aux, nocc, nspins
2198 : LOGICAL :: do_exx
2199 : REAL(dp) :: focc
2200 : TYPE(admm_type), POINTER :: admm_env
2201 : TYPE(cp_fm_struct_type), POINTER :: fm_struct
2202 50 : TYPE(cp_fm_type), ALLOCATABLE, DIMENSION(:) :: cpmos, mo_occ
2203 : TYPE(cp_fm_type), POINTER :: mo_coeff
2204 50 : TYPE(dbcsr_p_type), DIMENSION(:), POINTER :: dbcsr_p_work, matrix_p_mp2, &
2205 50 : matrix_p_mp2_admm, matrix_s, &
2206 50 : matrix_s_aux, work_admm, YP_admm
2207 : TYPE(dft_control_type), POINTER :: dft_control
2208 : TYPE(linres_control_type), POINTER :: linres_control
2209 50 : TYPE(mo_set_type), DIMENSION(:), POINTER :: mos
2210 : TYPE(qs_p_env_type), POINTER :: p_env
2211 : TYPE(section_vals_type), POINTER :: hfx_section, lr_section
2212 :
2213 50 : NULLIFY (linres_control, p_env, dft_control, matrix_s, mos, mo_coeff, fm_struct, lr_section, &
2214 50 : dbcsr_p_work, YP_admm, matrix_p_mp2, admm_env, work_admm, matrix_s_aux, matrix_p_mp2_admm)
2215 :
2216 50 : CALL timeset(routineN, handle)
2217 :
2218 50 : CALL get_qs_env(qs_env, dft_control=dft_control, matrix_s=matrix_s, mos=mos)
2219 50 : nspins = dft_control%nspins
2220 :
2221 : ! Setting up for the z-vector equation
2222 :
2223 : ! Initialize linres_control
2224 50 : lr_section => section_vals_get_subs_vals(qs_env%input, "DFT%XC%WF_CORRELATION%LOW_SCALING%CPHF")
2225 :
2226 50 : ALLOCATE (linres_control)
2227 50 : CALL section_vals_val_get(lr_section, "MAX_ITER", i_val=linres_control%max_iter)
2228 50 : CALL section_vals_val_get(lr_section, "EPS_CONV", r_val=linres_control%eps)
2229 50 : CALL section_vals_val_get(lr_section, "PRECONDITIONER", i_val=linres_control%preconditioner_type)
2230 50 : CALL section_vals_val_get(lr_section, "ENERGY_GAP", r_val=linres_control%energy_gap)
2231 :
2232 50 : linres_control%do_kernel = .TRUE.
2233 50 : linres_control%lr_triplet = .FALSE.
2234 50 : linres_control%converged = .FALSE.
2235 50 : linres_control%eps_filter = qs_env%mp2_env%ri_rpa_im_time%eps_filter
2236 :
2237 50 : CALL set_qs_env(qs_env, linres_control=linres_control)
2238 :
2239 50 : IF (unit_nr > 0) THEN
2240 25 : WRITE (unit_nr, *)
2241 25 : WRITE (unit_nr, '(T3,A)') 'MP2_CPHF| Iterative solution of Z-Vector equations'
2242 25 : WRITE (unit_nr, '(T3,A,T45,ES8.1)') 'MP2_CPHF| Convergence threshold:', linres_control%eps
2243 25 : WRITE (unit_nr, '(T3,A,T45,I8)') 'MP2_CPHF| Maximum number of iterations: ', linres_control%max_iter
2244 : END IF
2245 :
2246 350 : ALLOCATE (p_env)
2247 50 : CALL p_env_create(p_env, qs_env, orthogonal_orbitals=.TRUE., linres_control=linres_control)
2248 50 : CALL p_env_psi0_changed(p_env, qs_env)
2249 :
2250 : ! Matrix allocation
2251 50 : CALL dbcsr_allocate_matrix_set(p_env%p1, nspins)
2252 50 : CALL dbcsr_allocate_matrix_set(p_env%w1, nspins)
2253 50 : CALL dbcsr_allocate_matrix_set(dbcsr_p_work, nspins)
2254 112 : DO ispin = 1, nspins
2255 62 : ALLOCATE (p_env%p1(ispin)%matrix, p_env%w1(ispin)%matrix, dbcsr_p_work(ispin)%matrix)
2256 62 : CALL dbcsr_create(matrix=p_env%p1(ispin)%matrix, template=matrix_s(1)%matrix)
2257 62 : CALL dbcsr_create(matrix=p_env%w1(ispin)%matrix, template=matrix_s(1)%matrix)
2258 62 : CALL dbcsr_create(matrix=dbcsr_p_work(ispin)%matrix, template=matrix_s(1)%matrix)
2259 62 : CALL dbcsr_copy(p_env%p1(ispin)%matrix, matrix_s(1)%matrix)
2260 62 : CALL dbcsr_copy(p_env%w1(ispin)%matrix, matrix_s(1)%matrix)
2261 62 : CALL dbcsr_copy(dbcsr_p_work(ispin)%matrix, matrix_s(1)%matrix)
2262 62 : CALL dbcsr_set(p_env%p1(ispin)%matrix, 0.0_dp)
2263 62 : CALL dbcsr_set(p_env%w1(ispin)%matrix, 0.0_dp)
2264 112 : CALL dbcsr_set(dbcsr_p_work(ispin)%matrix, 0.0_dp)
2265 : END DO
2266 :
2267 50 : IF (dft_control%do_admm) THEN
2268 16 : CALL get_admm_env(qs_env%admm_env, matrix_s_aux_fit=matrix_s_aux)
2269 16 : CALL dbcsr_allocate_matrix_set(p_env%p1_admm, nspins)
2270 16 : CALL dbcsr_allocate_matrix_set(work_admm, nspins)
2271 36 : DO ispin = 1, nspins
2272 20 : ALLOCATE (p_env%p1_admm(ispin)%matrix, work_admm(ispin)%matrix)
2273 20 : CALL dbcsr_create(p_env%p1_admm(ispin)%matrix, template=matrix_s_aux(1)%matrix)
2274 20 : CALL dbcsr_copy(p_env%p1_admm(ispin)%matrix, matrix_s_aux(1)%matrix)
2275 20 : CALL dbcsr_set(p_env%p1_admm(ispin)%matrix, 0.0_dp)
2276 20 : CALL dbcsr_create(work_admm(ispin)%matrix, template=matrix_s_aux(1)%matrix)
2277 20 : CALL dbcsr_copy(work_admm(ispin)%matrix, matrix_s_aux(1)%matrix)
2278 36 : CALL dbcsr_set(work_admm(ispin)%matrix, 0.0_dp)
2279 : END DO
2280 : END IF
2281 :
2282 : ! Preparing the RHS of the z-vector equation
2283 50 : CALL prepare_for_response(force_data, qs_env)
2284 324 : ALLOCATE (cpmos(nspins), mo_occ(nspins))
2285 112 : DO ispin = 1, nspins
2286 62 : CALL get_mo_set(mo_set=mos(ispin), mo_coeff=mo_coeff, nao=nao, homo=nocc)
2287 62 : NULLIFY (fm_struct)
2288 : CALL cp_fm_struct_create(fm_struct, ncol_global=nocc, &
2289 62 : template_fmstruct=mo_coeff%matrix_struct)
2290 62 : CALL cp_fm_create(cpmos(ispin), fm_struct)
2291 62 : CALL cp_fm_set_all(cpmos(ispin), 0.0_dp)
2292 62 : CALL cp_fm_create(mo_occ(ispin), fm_struct)
2293 62 : CALL cp_fm_to_fm(mo_coeff, mo_occ(ispin), nocc)
2294 174 : CALL cp_fm_struct_release(fm_struct)
2295 : END DO
2296 :
2297 : ! in case of EXX, need to add the HF Hamiltonian to the RHS of the Z-vector equation
2298 : ! Strategy: we take the ks_matrix, remove the current xc contribution, and then add the RPA HF one
2299 50 : do_exx = .FALSE.
2300 50 : IF (qs_env%mp2_env%method == ri_rpa_method_gpw) THEN
2301 28 : hfx_section => section_vals_get_subs_vals(qs_env%input, "DFT%XC%WF_CORRELATION%RI_RPA%HF")
2302 28 : CALL section_vals_get(hfx_section, explicit=do_exx)
2303 : END IF
2304 :
2305 50 : IF (do_exx) THEN
2306 : CALL add_exx_to_rhs(rhs=force_data%sum_O_tau, &
2307 : qs_env=qs_env, &
2308 : ext_hfx_section=hfx_section, &
2309 : x_data=qs_env%mp2_env%ri_rpa%x_data, &
2310 : recalc_integrals=.FALSE., &
2311 : do_admm=qs_env%mp2_env%ri_rpa%do_admm, &
2312 : do_exx=do_exx, &
2313 18 : reuse_hfx=qs_env%mp2_env%ri_rpa%reuse_hfx)
2314 : END IF
2315 :
2316 50 : focc = 2.0_dp
2317 50 : IF (nspins == 1) focc = 4.0_dp
2318 112 : DO ispin = 1, nspins
2319 62 : CALL get_mo_set(mo_set=mos(ispin), mo_coeff=mo_coeff, homo=nocc)
2320 : CALL cp_dbcsr_sm_fm_multiply(force_data%sum_O_tau(ispin)%matrix, mo_occ(ispin), &
2321 : cpmos(ispin), nocc, &
2322 112 : alpha=focc, beta=0.0_dp)
2323 : END DO
2324 :
2325 : ! The z-vector equation and associated forces
2326 50 : CALL response_equation_new(qs_env, p_env, cpmos, unit_nr)
2327 :
2328 : ! Save the mp2 density matrix
2329 50 : CALL get_qs_env(qs_env, matrix_p_mp2=matrix_p_mp2)
2330 50 : IF (ASSOCIATED(matrix_p_mp2)) CALL dbcsr_deallocate_matrix_set(matrix_p_mp2)
2331 112 : DO ispin = 1, nspins
2332 62 : CALL dbcsr_copy(dbcsr_p_work(ispin)%matrix, p_env%p1(ispin)%matrix)
2333 112 : CALL dbcsr_add(dbcsr_p_work(ispin)%matrix, force_data%sum_YP_tau(ispin)%matrix, 1.0_dp, 1.0_dp)
2334 : END DO
2335 50 : CALL set_ks_env(qs_env%ks_env, matrix_p_mp2=dbcsr_p_work)
2336 :
2337 50 : IF (dft_control%do_admm) THEN
2338 16 : CALL dbcsr_allocate_matrix_set(YP_admm, nspins)
2339 16 : CALL get_qs_env(qs_env, matrix_p_mp2_admm=matrix_p_mp2_admm, admm_env=admm_env)
2340 16 : nao = admm_env%nao_orb
2341 16 : nao_aux = admm_env%nao_aux_fit
2342 16 : IF (ASSOCIATED(matrix_p_mp2_admm)) CALL dbcsr_deallocate_matrix_set(matrix_p_mp2_admm)
2343 36 : DO ispin = 1, nspins
2344 :
2345 : !sum_YP_tau in the auxiliary basis
2346 20 : CALL copy_dbcsr_to_fm(force_data%sum_YP_tau(ispin)%matrix, admm_env%work_orb_orb)
2347 : CALL parallel_gemm('N', 'N', nao_aux, nao, nao, 1.0_dp, admm_env%A, admm_env%work_orb_orb, &
2348 20 : 0.0_dp, admm_env%work_aux_orb)
2349 : CALL parallel_gemm('N', 'T', nao_aux, nao_aux, nao, 1.0_dp, admm_env%work_aux_orb, admm_env%A, &
2350 20 : 0.0_dp, admm_env%work_aux_aux)
2351 20 : CALL copy_fm_to_dbcsr(admm_env%work_aux_aux, work_admm(ispin)%matrix, keep_sparsity=.TRUE.)
2352 :
2353 : !save the admm representation od sum_YP_tau
2354 20 : ALLOCATE (YP_admm(ispin)%matrix)
2355 20 : CALL dbcsr_create(YP_admm(ispin)%matrix, template=work_admm(ispin)%matrix)
2356 20 : CALL dbcsr_copy(YP_admm(ispin)%matrix, work_admm(ispin)%matrix)
2357 :
2358 36 : CALL dbcsr_add(work_admm(ispin)%matrix, p_env%p1_admm(ispin)%matrix, 1.0_dp, 1.0_dp)
2359 :
2360 : END DO
2361 16 : CALL set_ks_env(qs_env%ks_env, matrix_p_mp2_admm=work_admm)
2362 : END IF
2363 :
2364 : !Calculate the response force and the force from the trace with F
2365 50 : CALL update_im_time_forces(p_env, force_data%sum_O_tau, force_data%sum_YP_tau, YP_admm, qs_env)
2366 :
2367 : !clean-up
2368 50 : IF (dft_control%do_admm) CALL dbcsr_deallocate_matrix_set(YP_admm)
2369 :
2370 50 : CALL cp_fm_release(cpmos)
2371 50 : CALL cp_fm_release(mo_occ)
2372 50 : CALL p_env_release(p_env)
2373 50 : DEALLOCATE (p_env)
2374 :
2375 50 : CALL timestop(handle)
2376 :
2377 100 : END SUBROUTINE calc_post_loop_forces
2378 :
2379 : ! **************************************************************************************************
2380 : !> \brief Prepares the RHS of the z-vector equation. Apply the xc and HFX kernel on the previously
2381 : !> stored sum_YP_tau density, and add it to the final force_data%sum_O_tau quantity
2382 : !> \param force_data ...
2383 : !> \param qs_env ...
2384 : ! **************************************************************************************************
2385 50 : SUBROUTINE prepare_for_response(force_data, qs_env)
2386 :
2387 : TYPE(im_time_force_type), INTENT(INOUT) :: force_data
2388 : TYPE(qs_environment_type), POINTER :: qs_env
2389 :
2390 : CHARACTER(LEN=*), PARAMETER :: routineN = 'prepare_for_response'
2391 :
2392 : INTEGER :: handle, ispin, nao, nao_aux, nspins
2393 : LOGICAL :: do_hfx, do_tau, do_tau_admm
2394 : REAL(dp) :: ehartree
2395 : TYPE(admm_type), POINTER :: admm_env
2396 50 : TYPE(dbcsr_p_type), DIMENSION(:), POINTER :: dbcsr_p_work, ker_tau_admm, matrix_s, &
2397 50 : matrix_s_aux, work_admm
2398 : TYPE(dbcsr_type) :: dbcsr_work
2399 : TYPE(dft_control_type), POINTER :: dft_control
2400 : TYPE(pw_c1d_gs_type) :: rhoz_tot_gspace, zv_hartree_gspace
2401 50 : TYPE(pw_c1d_gs_type), DIMENSION(:), POINTER :: rhoz_g
2402 : TYPE(pw_env_type), POINTER :: pw_env
2403 : TYPE(pw_poisson_type), POINTER :: poisson_env
2404 : TYPE(pw_pool_type), POINTER :: auxbas_pw_pool
2405 : TYPE(pw_r3d_rs_type) :: zv_hartree_rspace
2406 50 : TYPE(pw_r3d_rs_type), DIMENSION(:), POINTER :: rhoz_r, tauz_r, v_xc, v_xc_tau
2407 : TYPE(qs_rho_type), POINTER :: rho, rho_aux_fit, rhoz
2408 50 : TYPE(rho_atom_type), DIMENSION(:), POINTER :: rho0_atom_set, rho1_atom_set
2409 : TYPE(section_vals_type), POINTER :: hfx_section, xc_section
2410 : TYPE(task_list_type), POINTER :: task_list_aux_fit
2411 :
2412 50 : NULLIFY (pw_env, rhoz_r, rhoz_g, tauz_r, v_xc, v_xc_tau, &
2413 50 : poisson_env, auxbas_pw_pool, dft_control, admm_env, xc_section, rho, rho_aux_fit, &
2414 50 : task_list_aux_fit, ker_tau_admm, work_admm, dbcsr_p_work, matrix_s, hfx_section)
2415 50 : NULLIFY (rho0_atom_set, rho1_atom_set)
2416 :
2417 50 : CALL timeset(routineN, handle)
2418 :
2419 50 : CALL get_qs_env(qs_env, dft_control=dft_control, pw_env=pw_env, rho=rho, matrix_s=matrix_s)
2420 50 : CALL pw_env_get(pw_env, auxbas_pw_pool=auxbas_pw_pool, poisson_env=poisson_env)
2421 50 : nspins = dft_control%nspins
2422 :
2423 50 : CALL dbcsr_allocate_matrix_set(dbcsr_p_work, nspins)
2424 112 : DO ispin = 1, nspins
2425 62 : ALLOCATE (dbcsr_p_work(ispin)%matrix)
2426 62 : CALL dbcsr_create(matrix=dbcsr_p_work(ispin)%matrix, template=matrix_s(1)%matrix)
2427 62 : CALL dbcsr_copy(dbcsr_p_work(ispin)%matrix, matrix_s(1)%matrix)
2428 112 : CALL dbcsr_set(dbcsr_p_work(ispin)%matrix, 0.0_dp)
2429 : END DO
2430 :
2431 : !Apply the kernel on the density saved in force_data%sum_YP_tau
2432 374 : ALLOCATE (rhoz_r(nspins), rhoz_g(nspins))
2433 112 : DO ispin = 1, nspins
2434 62 : CALL auxbas_pw_pool%create_pw(rhoz_r(ispin))
2435 112 : CALL auxbas_pw_pool%create_pw(rhoz_g(ispin))
2436 : END DO
2437 50 : CALL auxbas_pw_pool%create_pw(rhoz_tot_gspace)
2438 50 : CALL auxbas_pw_pool%create_pw(zv_hartree_rspace)
2439 50 : CALL auxbas_pw_pool%create_pw(zv_hartree_gspace)
2440 :
2441 50 : CALL pw_zero(rhoz_tot_gspace)
2442 112 : DO ispin = 1, nspins
2443 : CALL calculate_rho_elec(ks_env=qs_env%ks_env, matrix_p=force_data%sum_YP_tau(ispin)%matrix, &
2444 62 : rho=rhoz_r(ispin), rho_gspace=rhoz_g(ispin))
2445 112 : CALL pw_axpy(rhoz_g(ispin), rhoz_tot_gspace)
2446 : END DO
2447 :
2448 : CALL pw_poisson_solve(poisson_env, rhoz_tot_gspace, ehartree, &
2449 50 : zv_hartree_gspace)
2450 :
2451 50 : CALL pw_transfer(zv_hartree_gspace, zv_hartree_rspace)
2452 50 : CALL pw_scale(zv_hartree_rspace, zv_hartree_rspace%pw_grid%dvol)
2453 :
2454 50 : CALL qs_rho_get(rho, tau_r_valid=do_tau)
2455 50 : IF (do_tau) THEN
2456 : BLOCK
2457 : TYPE(pw_c1d_gs_type) :: tauz_g
2458 24 : ALLOCATE (tauz_r(nspins))
2459 8 : CALL auxbas_pw_pool%create_pw(tauz_g)
2460 16 : DO ispin = 1, nspins
2461 8 : CALL auxbas_pw_pool%create_pw(tauz_r(ispin))
2462 :
2463 : CALL calculate_rho_elec(ks_env=qs_env%ks_env, matrix_p=force_data%sum_YP_tau(ispin)%matrix, &
2464 16 : rho=tauz_r(ispin), rho_gspace=tauz_g, compute_tau=.TRUE.)
2465 : END DO
2466 16 : CALL auxbas_pw_pool%give_back_pw(tauz_g)
2467 : END BLOCK
2468 : END IF
2469 :
2470 50 : IF (dft_control%do_admm) THEN
2471 16 : CALL get_qs_env(qs_env, admm_env=admm_env)
2472 16 : xc_section => admm_env%xc_section_primary
2473 : ELSE
2474 34 : xc_section => section_vals_get_subs_vals(qs_env%input, "DFT%XC")
2475 : END IF
2476 :
2477 : !Primary XC kernel
2478 50 : ALLOCATE (rhoz)
2479 50 : CALL qs_rho_create(rhoz)
2480 50 : IF (ASSOCIATED(rhoz_r)) THEN
2481 50 : CALL qs_rho_set(rhoz, rho_r=rhoz_r, rho_r_valid=.TRUE.)
2482 : END IF
2483 50 : IF (ASSOCIATED(rhoz_g)) THEN
2484 50 : CALL qs_rho_set(rhoz, rho_g=rhoz_g, rho_g_valid=.TRUE.)
2485 : END IF
2486 50 : IF (ASSOCIATED(tauz_r)) THEN
2487 8 : CALL qs_rho_set(rhoz, tau_r=tauz_r, tau_r_valid=.TRUE.)
2488 : END IF
2489 : !
2490 : CALL qs_fxc_create(qs_env, rho, rhoz, rho0_atom_set, xc_section, .FALSE., &
2491 50 : v_xc, v_xc_tau, rho1_atom_set)
2492 : !
2493 50 : DEALLOCATE (rhoz)
2494 :
2495 112 : DO ispin = 1, nspins
2496 62 : CALL pw_scale(v_xc(ispin), v_xc(ispin)%pw_grid%dvol)
2497 62 : CALL pw_axpy(zv_hartree_rspace, v_xc(ispin))
2498 : CALL integrate_v_rspace(qs_env=qs_env, &
2499 : v_rspace=v_xc(ispin), &
2500 : hmat=dbcsr_p_work(ispin), &
2501 62 : calculate_forces=.FALSE.)
2502 :
2503 112 : CALL auxbas_pw_pool%give_back_pw(v_xc(ispin))
2504 : END DO
2505 50 : CALL auxbas_pw_pool%give_back_pw(rhoz_tot_gspace)
2506 50 : CALL auxbas_pw_pool%give_back_pw(zv_hartree_rspace)
2507 50 : CALL auxbas_pw_pool%give_back_pw(zv_hartree_gspace)
2508 50 : DEALLOCATE (v_xc)
2509 :
2510 50 : IF (do_tau) THEN
2511 16 : DO ispin = 1, nspins
2512 8 : CALL pw_scale(v_xc_tau(ispin), v_xc_tau(ispin)%pw_grid%dvol)
2513 : CALL integrate_v_rspace(qs_env=qs_env, &
2514 : v_rspace=v_xc_tau(ispin), &
2515 : hmat=dbcsr_p_work(ispin), &
2516 : compute_tau=.TRUE., &
2517 8 : calculate_forces=.FALSE.)
2518 16 : CALL auxbas_pw_pool%give_back_pw(v_xc_tau(ispin))
2519 : END DO
2520 8 : DEALLOCATE (v_xc_tau)
2521 : END IF
2522 :
2523 : !Auxiliary xc kernel (admm)
2524 50 : IF (dft_control%do_admm) THEN
2525 16 : CALL get_qs_env(qs_env, admm_env=admm_env)
2526 : CALL get_admm_env(admm_env, matrix_s_aux_fit=matrix_s_aux, &
2527 16 : task_list_aux_fit=task_list_aux_fit, rho_aux_fit=rho_aux_fit)
2528 :
2529 16 : CALL qs_rho_get(rho_aux_fit, tau_r_valid=do_tau_admm)
2530 :
2531 16 : CALL dbcsr_allocate_matrix_set(work_admm, nspins)
2532 16 : CALL dbcsr_allocate_matrix_set(ker_tau_admm, nspins)
2533 36 : DO ispin = 1, nspins
2534 20 : ALLOCATE (work_admm(ispin)%matrix, ker_tau_admm(ispin)%matrix)
2535 20 : CALL dbcsr_create(work_admm(ispin)%matrix, template=matrix_s_aux(1)%matrix)
2536 20 : CALL dbcsr_copy(work_admm(ispin)%matrix, matrix_s_aux(1)%matrix)
2537 20 : CALL dbcsr_set(work_admm(ispin)%matrix, 0.0_dp)
2538 20 : CALL dbcsr_create(ker_tau_admm(ispin)%matrix, template=matrix_s_aux(1)%matrix)
2539 20 : CALL dbcsr_copy(ker_tau_admm(ispin)%matrix, matrix_s_aux(1)%matrix)
2540 36 : CALL dbcsr_set(ker_tau_admm(ispin)%matrix, 0.0_dp)
2541 : END DO
2542 :
2543 : !get the density in the auxuliary density
2544 16 : CPASSERT(ASSOCIATED(admm_env%work_orb_orb))
2545 16 : CPASSERT(ASSOCIATED(admm_env%work_aux_orb))
2546 16 : CPASSERT(ASSOCIATED(admm_env%work_aux_aux))
2547 16 : nao = admm_env%nao_orb
2548 16 : nao_aux = admm_env%nao_aux_fit
2549 36 : DO ispin = 1, nspins
2550 20 : CALL copy_dbcsr_to_fm(force_data%sum_YP_tau(ispin)%matrix, admm_env%work_orb_orb)
2551 : CALL parallel_gemm('N', 'N', nao_aux, nao, nao, 1.0_dp, admm_env%A, admm_env%work_orb_orb, &
2552 20 : 0.0_dp, admm_env%work_aux_orb)
2553 : CALL parallel_gemm('N', 'T', nao_aux, nao_aux, nao, 1.0_dp, admm_env%work_aux_orb, admm_env%A, &
2554 20 : 0.0_dp, admm_env%work_aux_aux)
2555 36 : CALL copy_fm_to_dbcsr(admm_env%work_aux_aux, ker_tau_admm(ispin)%matrix, keep_sparsity=.TRUE.)
2556 : END DO
2557 :
2558 16 : IF (.NOT. qs_env%admm_env%aux_exch_func == do_admm_aux_exch_func_none) THEN
2559 36 : DO ispin = 1, nspins
2560 20 : CALL pw_zero(rhoz_r(ispin))
2561 20 : CALL pw_zero(rhoz_g(ispin))
2562 : CALL calculate_rho_elec(ks_env=qs_env%ks_env, matrix_p=ker_tau_admm(ispin)%matrix, &
2563 : rho=rhoz_r(ispin), rho_gspace=rhoz_g(ispin), &
2564 36 : basis_type="AUX_FIT", task_list_external=task_list_aux_fit)
2565 : END DO
2566 :
2567 16 : IF (do_tau_admm) THEN
2568 : BLOCK
2569 : TYPE(pw_c1d_gs_type) :: tauz_g
2570 0 : CALL auxbas_pw_pool%create_pw(tauz_g)
2571 0 : DO ispin = 1, nspins
2572 0 : CALL pw_zero(tauz_r(ispin))
2573 : CALL calculate_rho_elec(ks_env=qs_env%ks_env, matrix_p=ker_tau_admm(ispin)%matrix, &
2574 : rho=tauz_r(ispin), rho_gspace=tauz_g, &
2575 : basis_type="AUX_FIT", task_list_external=task_list_aux_fit, &
2576 0 : compute_tau=.TRUE.)
2577 : END DO
2578 0 : CALL auxbas_pw_pool%give_back_pw(tauz_g)
2579 : END BLOCK
2580 : END IF
2581 :
2582 16 : xc_section => admm_env%xc_section_aux
2583 16 : ALLOCATE (rhoz)
2584 16 : CALL qs_rho_create(rhoz)
2585 16 : IF (ASSOCIATED(rhoz_r)) THEN
2586 16 : CALL qs_rho_set(rhoz, rho_r=rhoz_r, rho_r_valid=.TRUE.)
2587 : END IF
2588 16 : IF (ASSOCIATED(rhoz_g)) THEN
2589 16 : CALL qs_rho_set(rhoz, rho_g=rhoz_g, rho_g_valid=.TRUE.)
2590 : END IF
2591 16 : IF (ASSOCIATED(tauz_r)) THEN
2592 0 : CALL qs_rho_set(rhoz, tau_r=tauz_r, tau_r_valid=.TRUE.)
2593 : END IF
2594 : !
2595 : CALL qs_fxc_create(qs_env, rho_aux_fit, rhoz, rho0_atom_set, xc_section, .FALSE., &
2596 16 : v_xc, v_xc_tau, rho1_atom_set)
2597 : !
2598 16 : DEALLOCATE (rhoz)
2599 :
2600 36 : DO ispin = 1, nspins
2601 20 : CALL pw_scale(v_xc(ispin), v_xc(ispin)%pw_grid%dvol)
2602 : CALL integrate_v_rspace(qs_env=qs_env, &
2603 : v_rspace=v_xc(ispin), &
2604 : hmat=work_admm(ispin), &
2605 : calculate_forces=.FALSE., &
2606 : basis_type="AUX_FIT", &
2607 20 : task_list_external=task_list_aux_fit)
2608 36 : CALL auxbas_pw_pool%give_back_pw(v_xc(ispin))
2609 : END DO
2610 16 : DEALLOCATE (v_xc)
2611 :
2612 16 : IF (do_tau_admm) THEN
2613 0 : DO ispin = 1, nspins
2614 0 : CALL pw_scale(v_xc_tau(ispin), v_xc_tau(ispin)%pw_grid%dvol)
2615 : CALL integrate_v_rspace(qs_env=qs_env, &
2616 : v_rspace=v_xc_tau(ispin), &
2617 : hmat=work_admm(ispin), &
2618 : calculate_forces=.FALSE., &
2619 : basis_type="AUX_FIT", &
2620 : task_list_external=task_list_aux_fit, &
2621 0 : compute_tau=.TRUE.)
2622 0 : CALL auxbas_pw_pool%give_back_pw(v_xc_tau(ispin))
2623 : END DO
2624 0 : DEALLOCATE (v_xc_tau)
2625 : END IF
2626 : END IF !admm
2627 : END IF
2628 :
2629 112 : DO ispin = 1, nspins
2630 62 : CALL auxbas_pw_pool%give_back_pw(rhoz_r(ispin))
2631 112 : CALL auxbas_pw_pool%give_back_pw(rhoz_g(ispin))
2632 : END DO
2633 50 : DEALLOCATE (rhoz_r, rhoz_g)
2634 :
2635 50 : IF (do_tau) THEN
2636 16 : DO ispin = 1, nspins
2637 16 : CALL auxbas_pw_pool%give_back_pw(tauz_r(ispin))
2638 : END DO
2639 8 : DEALLOCATE (tauz_r)
2640 : END IF
2641 :
2642 : !HFX kernel
2643 50 : hfx_section => section_vals_get_subs_vals(qs_env%input, "DFT%XC%HF")
2644 50 : CALL section_vals_get(hfx_section, explicit=do_hfx)
2645 50 : IF (do_hfx) THEN
2646 32 : IF (dft_control%do_admm) THEN
2647 16 : CALL tddft_hfx_matrix(work_admm, ker_tau_admm, qs_env, .FALSE., .FALSE.)
2648 :
2649 : !Going back to primary basis
2650 16 : CALL dbcsr_create(dbcsr_work, template=dbcsr_p_work(1)%matrix)
2651 16 : CALL dbcsr_copy(dbcsr_work, dbcsr_p_work(1)%matrix)
2652 16 : CALL dbcsr_set(dbcsr_work, 0.0_dp)
2653 36 : DO ispin = 1, nspins
2654 20 : CALL copy_dbcsr_to_fm(work_admm(ispin)%matrix, admm_env%work_aux_aux)
2655 : CALL parallel_gemm('N', 'N', nao_aux, nao, nao_aux, 1.0_dp, admm_env%work_aux_aux, admm_env%A, &
2656 20 : 0.0_dp, admm_env%work_aux_orb)
2657 : CALL parallel_gemm('T', 'N', nao, nao, nao_aux, 1.0_dp, admm_env%A, admm_env%work_aux_orb, &
2658 20 : 0.0_dp, admm_env%work_orb_orb)
2659 20 : CALL copy_fm_to_dbcsr(admm_env%work_orb_orb, dbcsr_work, keep_sparsity=.TRUE.)
2660 36 : CALL dbcsr_add(dbcsr_p_work(ispin)%matrix, dbcsr_work, 1.0_dp, 1.0_dp)
2661 : END DO
2662 16 : CALL dbcsr_release(dbcsr_work)
2663 16 : CALL dbcsr_deallocate_matrix_set(ker_tau_admm)
2664 : ELSE
2665 16 : CALL tddft_hfx_matrix(dbcsr_p_work, force_data%sum_YP_tau, qs_env, .FALSE., .FALSE.)
2666 : END IF
2667 : END IF
2668 :
2669 112 : DO ispin = 1, nspins
2670 112 : CALL dbcsr_add(force_data%sum_O_tau(ispin)%matrix, dbcsr_p_work(ispin)%matrix, 1.0_dp, 1.0_dp)
2671 : END DO
2672 :
2673 50 : CALL dbcsr_deallocate_matrix_set(dbcsr_p_work)
2674 50 : CALL dbcsr_deallocate_matrix_set(work_admm)
2675 :
2676 50 : CALL timestop(handle)
2677 :
2678 250 : END SUBROUTINE prepare_for_response
2679 :
2680 : ! **************************************************************************************************
2681 : !> \brief Calculate the force and virial due to the (P|Q) GPW integral derivatives
2682 : !> \param G_PQ ...
2683 : !> \param force ...
2684 : !> \param h_stress ...
2685 : !> \param use_virial ...
2686 : !> \param mp2_env ...
2687 : !> \param qs_env ...
2688 : ! **************************************************************************************************
2689 12 : SUBROUTINE get_2c_gpw_forces(G_PQ, force, h_stress, use_virial, mp2_env, qs_env)
2690 :
2691 : TYPE(dbcsr_type), INTENT(INOUT) :: G_PQ
2692 : TYPE(qs_force_type), DIMENSION(:), POINTER :: force
2693 : REAL(dp), DIMENSION(3, 3), INTENT(INOUT) :: h_stress
2694 : LOGICAL, INTENT(IN) :: use_virial
2695 : TYPE(mp2_type), INTENT(INOUT) :: mp2_env
2696 : TYPE(qs_environment_type), POINTER :: qs_env
2697 :
2698 : CHARACTER(len=*), PARAMETER :: routineN = 'get_2c_gpw_forces'
2699 :
2700 : INTEGER :: atom_a, color, handle, i, i_RI, i_xyz, iatom, igrid_level, ikind, ipgf, iset, j, &
2701 : j_RI, jatom, lb_RI, n_RI, natom, ncoa, ncoms, nkind, nproc, nseta, o1, offset, pdims(2), &
2702 : sgfa, ub_RI
2703 24 : INTEGER, ALLOCATABLE, DIMENSION(:) :: atom_of_kind, iproc_map, kind_of, &
2704 12 : sizes_RI
2705 24 : INTEGER, DIMENSION(:), POINTER :: col_dist, la_max, la_min, npgfa, nsgfa, &
2706 12 : row_dist
2707 12 : INTEGER, DIMENSION(:, :), POINTER :: first_sgfa, pgrid
2708 : LOGICAL :: found, one_proc_group
2709 : REAL(dp) :: cutoff_old, radius, relative_cutoff_old
2710 12 : REAL(dp), ALLOCATABLE, DIMENSION(:) :: e_cutoff_old, wf_vector
2711 : REAL(dp), DIMENSION(3) :: force_a, force_b, ra
2712 : REAL(dp), DIMENSION(3, 3) :: my_virial_a, my_virial_b
2713 12 : REAL(KIND=dp), DIMENSION(:, :), POINTER :: h_tmp, I_ab, pab, pblock, sphi_a, zeta
2714 12 : TYPE(atomic_kind_type), DIMENSION(:), POINTER :: atomic_kind_set
2715 : TYPE(cell_type), POINTER :: cell
2716 : TYPE(dbcsr_distribution_type) :: dbcsr_dist
2717 : TYPE(dbcsr_type) :: tmp_G_PQ
2718 : TYPE(dft_control_type), POINTER :: dft_control
2719 : TYPE(gto_basis_set_p_type), ALLOCATABLE, &
2720 12 : DIMENSION(:), TARGET :: basis_set_ri_aux
2721 : TYPE(gto_basis_set_type), POINTER :: basis_set_a
2722 12 : TYPE(mo_set_type), DIMENSION(:), POINTER :: mos
2723 : TYPE(mp_para_env_type), POINTER :: para_env, para_env_ext
2724 : TYPE(neighbor_list_set_p_type), DIMENSION(:), &
2725 12 : POINTER :: sab_orb
2726 12 : TYPE(particle_type), DIMENSION(:), POINTER :: particle_set
2727 48 : TYPE(pw_c1d_gs_type) :: dvg(3), pot_g, rho_g, rho_g_copy
2728 : TYPE(pw_env_type), POINTER :: pw_env_ext
2729 : TYPE(pw_poisson_type), POINTER :: poisson_env
2730 : TYPE(pw_pool_type), POINTER :: auxbas_pw_pool
2731 : TYPE(pw_r3d_rs_type) :: psi_L, rho_r
2732 12 : TYPE(qs_kind_type), DIMENSION(:), POINTER :: qs_kind_set
2733 12 : TYPE(realspace_grid_type), DIMENSION(:), POINTER :: rs_v
2734 : TYPE(task_list_type), POINTER :: task_list_ext
2735 :
2736 12 : NULLIFY (sab_orb, task_list_ext, particle_set, qs_kind_set, dft_control, pw_env_ext, auxbas_pw_pool, &
2737 12 : poisson_env, atomic_kind_set, para_env, cell, rs_v, mos, basis_set_a)
2738 :
2739 12 : CALL timeset(routineN, handle)
2740 :
2741 : CALL get_qs_env(qs_env, dft_control=dft_control, para_env=para_env, sab_orb=sab_orb, &
2742 : natom=natom, nkind=nkind, qs_kind_set=qs_kind_set, particle_set=particle_set, &
2743 12 : mos=mos, cell=cell, atomic_kind_set=atomic_kind_set)
2744 :
2745 : !The idea is to use GPW to compute the integrals and derivatives. Because the potential needs
2746 : !to be calculated for each phi_j (column) of all AO pairs, and because that is expensive, we want
2747 : !to minimize the amount of time we do that. Therefore, we work with a special distribution, where
2748 : !each column of the resulting DBCSR matrix is mapped to a sub-communicator.
2749 :
2750 : !Try to get the optimal pdims (we want a grid that is flat: many cols, few rows)
2751 12 : IF (para_env%num_pe <= natom) THEN
2752 : pdims(1) = 1
2753 : pdims(2) = para_env%num_pe
2754 : ELSE
2755 0 : DO i = natom, 1, -1
2756 0 : IF (MODULO(para_env%num_pe, i) == 0) THEN
2757 0 : pdims(1) = para_env%num_pe/i
2758 0 : pdims(2) = i
2759 0 : EXIT
2760 : END IF
2761 : END DO
2762 : END IF
2763 :
2764 48 : ALLOCATE (row_dist(natom), col_dist(natom))
2765 48 : DO iatom = 1, natom
2766 48 : row_dist(iatom) = MODULO(iatom, pdims(1))
2767 : END DO
2768 48 : DO jatom = 1, natom
2769 48 : col_dist(jatom) = MODULO(jatom, pdims(2))
2770 : END DO
2771 :
2772 48 : ALLOCATE (pgrid(0:pdims(1) - 1, 0:pdims(2) - 1))
2773 12 : nproc = 0
2774 24 : DO i = 0, pdims(1) - 1
2775 48 : DO j = 0, pdims(2) - 1
2776 24 : pgrid(i, j) = nproc
2777 36 : nproc = nproc + 1
2778 : END DO
2779 : END DO
2780 :
2781 12 : CALL dbcsr_distribution_new(dbcsr_dist, group=para_env%get_handle(), pgrid=pgrid, row_dist=row_dist, col_dist=col_dist)
2782 :
2783 : !The temporary DBCSR integrals and derivatives matrices in this flat distribution
2784 12 : CALL dbcsr_create(tmp_G_PQ, template=G_PQ, matrix_type=dbcsr_type_no_symmetry, dist=dbcsr_dist)
2785 12 : CALL dbcsr_complete_redistribute(G_PQ, tmp_G_PQ)
2786 :
2787 84 : ALLOCATE (basis_set_ri_aux(nkind), sizes_RI(natom))
2788 12 : CALL basis_set_list_setup(basis_set_ri_aux, "RI_AUX", qs_kind_set)
2789 12 : CALL get_particle_set(particle_set, qs_kind_set, nsgf=sizes_RI, basis=basis_set_ri_aux)
2790 48 : n_RI = SUM(sizes_RI)
2791 :
2792 12 : one_proc_group = mp2_env%mp2_num_proc == 1
2793 12 : ALLOCATE (para_env_ext)
2794 12 : IF (one_proc_group) THEN
2795 : !one subgroup per proc
2796 4 : CALL para_env_ext%from_split(para_env, para_env%mepos)
2797 : ELSE
2798 : !Split the communicator accross the columns of the matrix
2799 8 : ncoms = MIN(pdims(2), para_env%num_pe/mp2_env%mp2_num_proc)
2800 16 : DO i = 0, pdims(1) - 1
2801 32 : DO j = 0, pdims(2) - 1
2802 24 : IF (pgrid(i, j) == para_env%mepos) color = MODULO(j + 1, ncoms)
2803 : END DO
2804 : END DO
2805 8 : CALL para_env_ext%from_split(para_env, color)
2806 : END IF
2807 :
2808 : !sab_orb and task_list_ext are essentially dummies
2809 : CALL prepare_gpw(qs_env, dft_control, e_cutoff_old, cutoff_old, relative_cutoff_old, para_env_ext, pw_env_ext, &
2810 12 : auxbas_pw_pool, poisson_env, task_list_ext, rho_r, rho_g, pot_g, psi_L, sab_orb)
2811 :
2812 12 : IF (use_virial) THEN
2813 4 : CALL auxbas_pw_pool%create_pw(rho_g_copy)
2814 16 : DO i_xyz = 1, 3
2815 16 : CALL auxbas_pw_pool%create_pw(dvg(i_xyz))
2816 : END DO
2817 : END IF
2818 :
2819 36 : ALLOCATE (wf_vector(n_RI))
2820 :
2821 12 : CALL get_atomic_kind_set(atomic_kind_set, kind_of=kind_of, atom_of_kind=atom_of_kind)
2822 :
2823 36 : ALLOCATE (iproc_map(natom))
2824 :
2825 : !Loop over the atomic blocks
2826 48 : DO jatom = 1, natom
2827 :
2828 : !Only calculate if on the correct sub-communicator/proc
2829 36 : IF (one_proc_group) THEN
2830 12 : iproc_map = 0
2831 48 : DO iatom = 1, natom
2832 48 : IF (pgrid(row_dist(iatom), col_dist(jatom)) == para_env%mepos) iproc_map(iatom) = 1
2833 : END DO
2834 30 : IF (.NOT. ANY(iproc_map == 1)) CYCLE
2835 : ELSE
2836 24 : IF (.NOT. MODULO(col_dist(jatom) + 1, ncoms) == color) CYCLE
2837 : END IF
2838 :
2839 60 : lb_RI = SUM(sizes_RI(1:jatom - 1))
2840 30 : ub_RI = lb_RI + sizes_RI(jatom)
2841 872 : DO j_RI = lb_RI + 1, ub_RI
2842 :
2843 830 : wf_vector = 0.0_dp
2844 830 : wf_vector(j_RI) = 1.0_dp
2845 :
2846 : CALL collocate_function(wf_vector, psi_L, rho_g, atomic_kind_set, qs_kind_set, cell, &
2847 : particle_set, pw_env_ext, dft_control%qs_control%eps_rho_rspace, &
2848 830 : basis_type="RI_AUX")
2849 :
2850 830 : IF (use_virial) THEN
2851 166 : CALL calc_potential_gpw(rho_r, rho_g, poisson_env, pot_g, mp2_env%potential_parameter, dvg)
2852 :
2853 166 : wf_vector = 0.0_dp
2854 664 : DO iatom = 1, natom
2855 : !only compute if i,j atom pair on correct proc
2856 498 : IF (one_proc_group) THEN
2857 498 : IF (.NOT. iproc_map(iatom) == 1) CYCLE
2858 : END IF
2859 :
2860 498 : CALL dbcsr_get_block_p(tmp_G_PQ, iatom, jatom, pblock, found)
2861 498 : IF (.NOT. found) CYCLE
2862 :
2863 996 : i_RI = SUM(sizes_RI(1:iatom - 1))
2864 14940 : wf_vector(i_RI + 1:i_RI + sizes_RI(iatom)) = pblock(:, j_RI - lb_RI)
2865 : END DO
2866 :
2867 166 : CALL pw_copy(rho_g, rho_g_copy)
2868 : CALL collocate_function(wf_vector, psi_L, rho_g, atomic_kind_set, qs_kind_set, cell, &
2869 : particle_set, pw_env_ext, dft_control%qs_control%eps_rho_rspace, &
2870 166 : basis_type="RI_AUX")
2871 :
2872 : CALL calc_potential_gpw(psi_L, rho_g, poisson_env, pot_g, mp2_env%potential_parameter, &
2873 166 : no_transfer=.TRUE.)
2874 : CALL virial_gpw_potential(rho_g_copy, pot_g, rho_g, dvg, h_stress, &
2875 166 : mp2_env%potential_parameter, para_env_ext)
2876 : ELSE
2877 664 : CALL calc_potential_gpw(rho_r, rho_g, poisson_env, pot_g, mp2_env%potential_parameter)
2878 : END IF
2879 :
2880 830 : NULLIFY (rs_v)
2881 830 : CALL pw_env_get(pw_env_ext, rs_grids=rs_v)
2882 830 : CALL potential_pw2rs(rs_v, rho_r, pw_env_ext)
2883 :
2884 3356 : DO iatom = 1, natom
2885 :
2886 : !only compute if i,j atom pair on correct proc
2887 2490 : IF (one_proc_group) THEN
2888 498 : IF (.NOT. iproc_map(iatom) == 1) CYCLE
2889 : END IF
2890 :
2891 2490 : force_a(:) = 0.0_dp
2892 2490 : force_b(:) = 0.0_dp
2893 2490 : IF (use_virial) THEN
2894 498 : my_virial_a = 0.0_dp
2895 498 : my_virial_b = 0.0_dp
2896 : END IF
2897 :
2898 2490 : ikind = kind_of(iatom)
2899 2490 : atom_a = atom_of_kind(iatom)
2900 :
2901 2490 : basis_set_a => basis_set_ri_aux(ikind)%gto_basis_set
2902 2490 : first_sgfa => basis_set_a%first_sgf
2903 2490 : la_max => basis_set_a%lmax
2904 2490 : la_min => basis_set_a%lmin
2905 2490 : nseta = basis_set_a%nset
2906 2490 : nsgfa => basis_set_a%nsgf_set
2907 2490 : sphi_a => basis_set_a%sphi
2908 2490 : zeta => basis_set_a%zet
2909 2490 : npgfa => basis_set_a%npgf
2910 :
2911 2490 : ra(:) = pbc(particle_set(iatom)%r, cell)
2912 :
2913 2490 : CALL dbcsr_get_block_p(tmp_G_PQ, iatom, jatom, pblock, found)
2914 2490 : IF (.NOT. found) CYCLE
2915 :
2916 : offset = 0
2917 15936 : DO iset = 1, nseta
2918 14442 : ncoa = npgfa(iset)*ncoset(la_max(iset))
2919 14442 : sgfa = first_sgfa(1, iset)
2920 :
2921 131472 : ALLOCATE (h_tmp(ncoa, 1)); h_tmp = 0.0_dp
2922 99102 : ALLOCATE (I_ab(nsgfa(iset), 1)); I_ab = 0.0_dp
2923 117030 : ALLOCATE (pab(ncoa, 1)); pab = 0.0_dp
2924 :
2925 97110 : I_ab(1:nsgfa(iset), 1) = 2.0_dp*pblock(offset + 1:offset + nsgfa(iset), j_RI - lb_RI)
2926 : CALL dgemm("N", "N", ncoa, 1, nsgfa(iset), 1.0_dp, sphi_a(1, sgfa), SIZE(sphi_a, 1), &
2927 14442 : I_ab(1, 1), nsgfa(iset), 0.0_dp, pab(1, 1), ncoa)
2928 :
2929 28884 : igrid_level = gaussian_gridlevel(pw_env_ext%gridlevel_info, MINVAL(zeta(:, iset)))
2930 :
2931 : ! The last three parameters are used to check whether a given function is within the own range.
2932 : ! Here, it is always the case, so let's enforce it because mod(0, 1)==0
2933 14442 : IF (map_gaussian_here(rs_v(igrid_level), cell%h_inv, ra, 0, 1, 0)) THEN
2934 28884 : DO ipgf = 1, npgfa(iset)
2935 14442 : o1 = (ipgf - 1)*ncoset(la_max(iset))
2936 14442 : igrid_level = gaussian_gridlevel(pw_env_ext%gridlevel_info, zeta(ipgf, iset))
2937 :
2938 : radius = exp_radius_very_extended(la_min=la_min(iset), la_max=la_max(iset), &
2939 : lb_min=0, lb_max=0, ra=ra, rb=ra, rp=ra, &
2940 : zetp=zeta(ipgf, iset), &
2941 : eps=dft_control%qs_control%eps_gvg_rspace, &
2942 14442 : prefactor=1.0_dp, cutoff=1.0_dp)
2943 :
2944 : CALL integrate_pgf_product( &
2945 : la_max=la_max(iset), zeta=zeta(ipgf, iset), la_min=la_min(iset), &
2946 : lb_max=0, zetb=0.0_dp, lb_min=0, &
2947 : ra=ra, rab=[0.0_dp, 0.0_dp, 0.0_dp], &
2948 : rsgrid=rs_v(igrid_level), &
2949 : hab=h_tmp, pab=pab, &
2950 : o1=o1, &
2951 : o2=0, &
2952 : radius=radius, &
2953 : calculate_forces=.TRUE., &
2954 : force_a=force_a, force_b=force_b, &
2955 28884 : use_virial=use_virial, my_virial_a=my_virial_a, my_virial_b=my_virial_b)
2956 :
2957 : END DO
2958 :
2959 : END IF
2960 :
2961 14442 : offset = offset + nsgfa(iset)
2962 15936 : DEALLOCATE (pab, h_tmp, I_ab)
2963 : END DO !iset
2964 :
2965 5976 : force(ikind)%mp2_non_sep(:, atom_a) = force(ikind)%mp2_non_sep(:, atom_a) + force_a + force_b
2966 10790 : IF (use_virial) h_stress = h_stress + my_virial_a + my_virial_b
2967 :
2968 : END DO !iatom
2969 : END DO !j_RI
2970 : END DO !jatom
2971 :
2972 12 : IF (use_virial) THEN
2973 4 : CALL auxbas_pw_pool%give_back_pw(rho_g_copy)
2974 16 : DO i_xyz = 1, 3
2975 16 : CALL auxbas_pw_pool%give_back_pw(dvg(i_xyz))
2976 : END DO
2977 : END IF
2978 :
2979 : CALL cleanup_gpw(qs_env, e_cutoff_old, cutoff_old, relative_cutoff_old, para_env_ext, pw_env_ext, &
2980 12 : task_list_ext, auxbas_pw_pool, rho_r, rho_g, pot_g, psi_L)
2981 :
2982 12 : CALL dbcsr_release(tmp_G_PQ)
2983 12 : CALL dbcsr_distribution_release(dbcsr_dist)
2984 12 : DEALLOCATE (col_dist, row_dist, pgrid)
2985 :
2986 12 : CALL mp_para_env_release(para_env_ext)
2987 :
2988 12 : CALL timestop(handle)
2989 :
2990 36 : END SUBROUTINE get_2c_gpw_forces
2991 :
2992 : ! **************************************************************************************************
2993 : !> \brief Calculate the forces due to the (P|Q) MME integral derivatives
2994 : !> \param G_PQ ...
2995 : !> \param force ...
2996 : !> \param mp2_env ...
2997 : !> \param qs_env ...
2998 : ! **************************************************************************************************
2999 16 : SUBROUTINE get_2c_mme_forces(G_PQ, force, mp2_env, qs_env)
3000 :
3001 : TYPE(dbcsr_type), INTENT(INOUT) :: G_PQ
3002 : TYPE(qs_force_type), DIMENSION(:), POINTER :: force
3003 : TYPE(mp2_type), INTENT(INOUT) :: mp2_env
3004 : TYPE(qs_environment_type), POINTER :: qs_env
3005 :
3006 : CHARACTER(len=*), PARAMETER :: routineN = 'get_2c_mme_forces'
3007 :
3008 : INTEGER :: atom_a, atom_b, G_count, handle, i_xyz, iatom, ikind, iset, jatom, jkind, jset, &
3009 : natom, nkind, nseta, nsetb, offset_hab_a, offset_hab_b, R_count, sgfa, sgfb
3010 16 : INTEGER, ALLOCATABLE, DIMENSION(:) :: atom_of_kind, kind_of
3011 16 : INTEGER, DIMENSION(:), POINTER :: la_max, la_min, lb_max, lb_min, npgfa, &
3012 16 : npgfb, nsgfa, nsgfb
3013 16 : INTEGER, DIMENSION(:, :), POINTER :: first_sgfa, first_sgfb
3014 : LOGICAL :: found
3015 : REAL(dp) :: new_force, pref
3016 16 : REAL(dp), ALLOCATABLE, DIMENSION(:, :) :: hab
3017 16 : REAL(dp), ALLOCATABLE, DIMENSION(:, :, :) :: hdab
3018 16 : REAL(dp), DIMENSION(:, :), POINTER :: pblock
3019 : REAL(KIND=dp), DIMENSION(3) :: ra, rb
3020 16 : REAL(KIND=dp), DIMENSION(:, :), POINTER :: sphi_a, sphi_b, zeta, zetb
3021 16 : TYPE(atomic_kind_type), DIMENSION(:), POINTER :: atomic_kind_set
3022 : TYPE(cell_type), POINTER :: cell
3023 : TYPE(dbcsr_iterator_type) :: iter
3024 : TYPE(gto_basis_set_p_type), ALLOCATABLE, &
3025 16 : DIMENSION(:), TARGET :: basis_set_ri_aux
3026 : TYPE(gto_basis_set_type), POINTER :: basis_set_a, basis_set_b
3027 : TYPE(mp_para_env_type), POINTER :: para_env
3028 16 : TYPE(particle_type), DIMENSION(:), POINTER :: particle_set
3029 16 : TYPE(qs_kind_type), DIMENSION(:), POINTER :: qs_kind_set
3030 :
3031 16 : NULLIFY (qs_kind_set, basis_set_a, basis_set_b, pblock, particle_set, &
3032 16 : cell, la_max, la_min, lb_min, npgfa, lb_max, npgfb, nsgfa, &
3033 16 : nsgfb, first_sgfa, first_sgfb, sphi_a, sphi_b, zeta, zetb, &
3034 16 : atomic_kind_set, para_env)
3035 :
3036 16 : CALL timeset(routineN, handle)
3037 :
3038 : CALL get_qs_env(qs_env, qs_kind_set=qs_kind_set, nkind=nkind, particle_set=particle_set, &
3039 16 : cell=cell, atomic_kind_set=atomic_kind_set, natom=natom, para_env=para_env)
3040 :
3041 16 : CALL get_atomic_kind_set(atomic_kind_set, kind_of=kind_of, atom_of_kind=atom_of_kind)
3042 :
3043 80 : ALLOCATE (basis_set_ri_aux(nkind))
3044 16 : CALL basis_set_list_setup(basis_set_ri_aux, "RI_AUX", qs_kind_set)
3045 :
3046 16 : G_count = 0; R_count = 0
3047 :
3048 16 : CALL dbcsr_iterator_start(iter, G_PQ)
3049 116 : DO WHILE (dbcsr_iterator_blocks_left(iter))
3050 :
3051 100 : CALL dbcsr_iterator_next_block(iter, row=iatom, column=jatom)
3052 100 : CALL dbcsr_get_block_p(G_PQ, iatom, jatom, pblock, found)
3053 100 : IF (.NOT. found) CYCLE
3054 100 : IF (iatom > jatom) CYCLE
3055 64 : pref = 2.0_dp
3056 64 : IF (iatom == jatom) pref = 1.0_dp
3057 :
3058 64 : ikind = kind_of(iatom)
3059 64 : jkind = kind_of(jatom)
3060 :
3061 64 : atom_a = atom_of_kind(iatom)
3062 64 : atom_b = atom_of_kind(jatom)
3063 :
3064 64 : basis_set_a => basis_set_ri_aux(ikind)%gto_basis_set
3065 64 : first_sgfa => basis_set_a%first_sgf
3066 64 : la_max => basis_set_a%lmax
3067 64 : la_min => basis_set_a%lmin
3068 64 : nseta = basis_set_a%nset
3069 64 : nsgfa => basis_set_a%nsgf_set
3070 64 : sphi_a => basis_set_a%sphi
3071 64 : zeta => basis_set_a%zet
3072 64 : npgfa => basis_set_a%npgf
3073 :
3074 64 : basis_set_b => basis_set_ri_aux(jkind)%gto_basis_set
3075 64 : first_sgfb => basis_set_b%first_sgf
3076 64 : lb_max => basis_set_b%lmax
3077 64 : lb_min => basis_set_b%lmin
3078 64 : nsetb = basis_set_b%nset
3079 64 : nsgfb => basis_set_b%nsgf_set
3080 64 : sphi_b => basis_set_b%sphi
3081 64 : zetb => basis_set_b%zet
3082 64 : npgfb => basis_set_b%npgf
3083 :
3084 64 : ra(:) = pbc(particle_set(iatom)%r, cell)
3085 64 : rb(:) = pbc(particle_set(jatom)%r, cell)
3086 :
3087 256 : ALLOCATE (hab(basis_set_a%nsgf, basis_set_b%nsgf))
3088 256 : ALLOCATE (hdab(3, basis_set_a%nsgf, basis_set_b%nsgf))
3089 64 : hab(:, :) = 0.0_dp
3090 64 : hdab(:, :, :) = 0.0_dp
3091 :
3092 64 : offset_hab_a = 0
3093 756 : DO iset = 1, nseta
3094 692 : sgfa = first_sgfa(1, iset)
3095 :
3096 692 : offset_hab_b = 0
3097 6340 : DO jset = 1, nsetb
3098 5648 : sgfb = first_sgfb(1, jset)
3099 :
3100 : CALL integrate_set_2c(mp2_env%eri_mme_param%par, mp2_env%potential_parameter, la_min(iset), &
3101 : la_max(iset), lb_min(jset), lb_max(jset), npgfa(iset), npgfb(jset), &
3102 : zeta(:, iset), zetb(:, jset), ra, rb, hab, nsgfa(iset), nsgfb(jset), &
3103 : offset_hab_a, offset_hab_b, 0, 0, sphi_a, sphi_b, sgfa, sgfb, &
3104 : nsgfa(iset), nsgfb(jset), do_eri_mme, hdab=hdab, &
3105 5648 : G_count=G_count, R_count=R_count)
3106 :
3107 6340 : offset_hab_b = offset_hab_b + nsgfb(jset)
3108 : END DO
3109 756 : offset_hab_a = offset_hab_a + nsgfa(iset)
3110 : END DO
3111 :
3112 256 : DO i_xyz = 1, 3
3113 143832 : new_force = pref*SUM(pblock(:, :)*hdab(i_xyz, :, :))
3114 192 : force(ikind)%mp2_non_sep(i_xyz, atom_a) = force(ikind)%mp2_non_sep(i_xyz, atom_a) + new_force
3115 256 : force(jkind)%mp2_non_sep(i_xyz, atom_b) = force(jkind)%mp2_non_sep(i_xyz, atom_b) - new_force
3116 : END DO
3117 :
3118 216 : DEALLOCATE (hab, hdab)
3119 : END DO
3120 16 : CALL dbcsr_iterator_stop(iter)
3121 :
3122 16 : CALL cp_eri_mme_update_local_counts(mp2_env%eri_mme_param, para_env, G_count_2c=G_count, R_count_2c=R_count)
3123 :
3124 16 : CALL timestop(handle)
3125 :
3126 48 : END SUBROUTINE get_2c_mme_forces
3127 :
3128 : ! **************************************************************************************************
3129 : !> \brief This routines gather all the force updates due to the response density and the trace with F
3130 : !> Also update the forces due to the SCF density for XC and exact exchange
3131 : !> \param p_env the p_env coming from the response calculation
3132 : !> \param matrix_hz the matrix going into the RHS of the response equation
3133 : !> \param matrix_p_F the density matrix with which we evaluate Trace[P*F]
3134 : !> \param matrix_p_F_admm ...
3135 : !> \param qs_env ...
3136 : !> \note very much inspired from the response_force routine in response_solver.F, especially for virial
3137 : ! **************************************************************************************************
3138 50 : SUBROUTINE update_im_time_forces(p_env, matrix_hz, matrix_p_F, matrix_p_F_admm, qs_env)
3139 :
3140 : TYPE(qs_p_env_type), POINTER :: p_env
3141 : TYPE(dbcsr_p_type), DIMENSION(:), POINTER :: matrix_hz, matrix_p_F, matrix_p_F_admm
3142 : TYPE(qs_environment_type), POINTER :: qs_env
3143 :
3144 : CHARACTER(len=*), PARAMETER :: routineN = 'update_im_time_forces'
3145 :
3146 : INTEGER :: handle, i, idens, ispin, n_rep_hf, nao, &
3147 : nao_aux, nder, nimages, nocc, nspins
3148 : LOGICAL :: do_exx, do_hfx, do_tau, do_tau_admm, &
3149 : use_virial
3150 : REAL(dp) :: dummy_real1, dummy_real2, ehartree, exc, &
3151 : focc
3152 : REAL(dp), DIMENSION(3, 3) :: h_stress, pv_loc
3153 : TYPE(admm_type), POINTER :: admm_env
3154 50 : TYPE(atomic_kind_type), DIMENSION(:), POINTER :: atomic_kind_set
3155 50 : TYPE(dbcsr_p_type), DIMENSION(:), POINTER :: current_density, current_density_admm, &
3156 50 : current_mat_h, matrix_p_mp2, matrix_p_mp2_admm, matrix_s, matrix_s_aux_fit, matrix_w, &
3157 50 : rho_ao, rho_ao_aux, scrm, scrm_admm
3158 50 : TYPE(dbcsr_p_type), DIMENSION(:, :), POINTER :: dbcsr_work_h, dbcsr_work_p, mpa2
3159 : TYPE(dbcsr_type) :: dbcsr_work
3160 : TYPE(dft_control_type), POINTER :: dft_control
3161 50 : TYPE(hfx_type), DIMENSION(:, :), POINTER :: x_data
3162 50 : TYPE(mo_set_type), DIMENSION(:), POINTER :: mos
3163 : TYPE(mp_para_env_type), POINTER :: para_env
3164 : TYPE(neighbor_list_set_p_type), DIMENSION(:), &
3165 50 : POINTER :: sab_orb, sac_ae, sac_ppl, sap_ppnl
3166 50 : TYPE(particle_type), DIMENSION(:), POINTER :: particle_set
3167 : TYPE(pw_c1d_gs_type) :: rho_tot_gspace, rhoz_tot_gspace, &
3168 : zv_hartree_gspace
3169 50 : TYPE(pw_c1d_gs_type), DIMENSION(:), POINTER :: rhoz_g
3170 : TYPE(pw_env_type), POINTER :: pw_env
3171 : TYPE(pw_poisson_type), POINTER :: poisson_env
3172 : TYPE(pw_pool_type), POINTER :: auxbas_pw_pool
3173 : TYPE(pw_r3d_rs_type) :: vh_rspace, vhxc_rspace, zv_hartree_rspace
3174 50 : TYPE(pw_r3d_rs_type), DIMENSION(:), POINTER :: rhoz_r, tauz_r, v_xc, v_xc_tau, &
3175 50 : vadmm_rspace, vtau_rspace, vxc_rspace
3176 50 : TYPE(qs_force_type), DIMENSION(:), POINTER :: force
3177 50 : TYPE(qs_kind_type), DIMENSION(:), POINTER :: qs_kind_set
3178 : TYPE(qs_rho_type), POINTER :: rho, rho_aux_fit, rhoz
3179 50 : TYPE(rho_atom_type), DIMENSION(:), POINTER :: rho0_atom_set, rho1_atom_set
3180 : TYPE(section_vals_type), POINTER :: hfx_section, xc_section
3181 : TYPE(task_list_type), POINTER :: task_list_aux_fit
3182 : TYPE(virial_type), POINTER :: virial
3183 :
3184 50 : NULLIFY (scrm, rho, dft_control, matrix_p_mp2, matrix_s, &
3185 50 : matrix_p_mp2_admm, admm_env, sab_orb, dbcsr_work_p, &
3186 50 : dbcsr_work_h, sac_ae, sac_ppl, sap_ppnl, force, virial, &
3187 50 : qs_kind_set, atomic_kind_set, particle_set, pw_env, poisson_env, &
3188 50 : auxbas_pw_pool, task_list_aux_fit, matrix_s_aux_fit, scrm_admm, &
3189 50 : rho_aux_fit, rho_ao_aux, x_data, hfx_section, xc_section, &
3190 50 : para_env, rhoz_g, rhoz_r, tauz_r, v_xc, v_xc_tau, vxc_rspace, &
3191 50 : vtau_rspace, vadmm_rspace, rho_ao, matrix_w)
3192 50 : NULLIFY (rho0_atom_set, rho1_atom_set)
3193 :
3194 50 : CALL timeset(routineN, handle)
3195 :
3196 : CALL get_qs_env(qs_env, rho=rho, dft_control=dft_control, matrix_s=matrix_s, admm_env=admm_env, &
3197 : sab_orb=sab_orb, sac_ae=sac_ae, sac_ppl=sac_ppl, sap_ppnl=sap_ppnl, force=force, &
3198 : virial=virial, particle_set=particle_set, qs_kind_set=qs_kind_set, &
3199 50 : atomic_kind_set=atomic_kind_set, x_data=x_data, para_env=para_env)
3200 50 : nspins = dft_control%nspins
3201 :
3202 50 : use_virial = virial%pv_availability .AND. (.NOT. virial%pv_numer)
3203 50 : IF (use_virial) virial%pv_calculate = .TRUE.
3204 :
3205 : !Whether we replace the force/energy of SCF XC with HF in RPA
3206 50 : do_exx = .FALSE.
3207 50 : IF (qs_env%mp2_env%method == ri_rpa_method_gpw) THEN
3208 28 : hfx_section => section_vals_get_subs_vals(qs_env%input, "DFT%XC%WF_CORRELATION%RI_RPA%HF")
3209 28 : CALL section_vals_get(hfx_section, explicit=do_exx)
3210 : END IF
3211 :
3212 : !Get the mp2 density matrix which is p_env%p1 + matrix_p_F
3213 50 : CALL get_qs_env(qs_env, matrix_p_mp2=matrix_p_mp2, matrix_p_mp2_admm=matrix_p_mp2_admm)
3214 :
3215 : !The kinetic term (only response density)
3216 50 : NULLIFY (scrm)
3217 50 : mpa2(1:nspins, 1:1) => matrix_p_mp2(1:nspins)
3218 : CALL kinetic_energy_matrix(qs_env, matrix_t=scrm, matrix_p=mpa2, &
3219 : matrix_name="KINETIC ENERGY MATRIX", &
3220 : basis_type="ORB", &
3221 50 : sab_orb=sab_orb, calculate_forces=.TRUE.)
3222 50 : CALL dbcsr_deallocate_matrix_set(scrm)
3223 :
3224 : !The pseudo-potential terms (only reponse density)
3225 50 : CALL dbcsr_allocate_matrix_set(scrm, nspins)
3226 112 : DO ispin = 1, nspins
3227 62 : ALLOCATE (scrm(ispin)%matrix)
3228 62 : CALL dbcsr_create(scrm(ispin)%matrix, template=matrix_s(1)%matrix)
3229 62 : CALL dbcsr_copy(scrm(ispin)%matrix, matrix_s(1)%matrix)
3230 112 : CALL dbcsr_set(scrm(ispin)%matrix, 0.0_dp)
3231 : END DO
3232 :
3233 50 : nder = 1
3234 50 : nimages = 1
3235 424 : ALLOCATE (dbcsr_work_p(nspins, 1), dbcsr_work_h(nspins, 1))
3236 112 : DO ispin = 1, nspins
3237 62 : dbcsr_work_p(ispin, 1)%matrix => matrix_p_mp2(ispin)%matrix
3238 112 : dbcsr_work_h(ispin, 1)%matrix => scrm(ispin)%matrix
3239 : END DO
3240 :
3241 50 : CALL core_matrices(qs_env, dbcsr_work_h, dbcsr_work_p, .TRUE., nder)
3242 :
3243 50 : DEALLOCATE (dbcsr_work_p, dbcsr_work_h)
3244 :
3245 50 : IF (use_virial) THEN
3246 4 : h_stress = 0.0_dp
3247 52 : virial%pv_xc = 0.0_dp
3248 4 : NULLIFY (vxc_rspace, vtau_rspace, vadmm_rspace)
3249 : CALL ks_ref_potential(qs_env, vh_rspace, vxc_rspace, vtau_rspace, vadmm_rspace, &
3250 4 : dummy_real1, dummy_real2, h_stress)
3251 52 : virial%pv_ehartree = virial%pv_ehartree + h_stress/REAL(para_env%num_pe, dp)
3252 52 : virial%pv_virial = virial%pv_virial + h_stress/REAL(para_env%num_pe, dp)
3253 4 : IF (.NOT. do_exx) THEN
3254 : !if RPA EXX, then do not consider XC virial (replaced by RPA%HF virial)
3255 52 : virial%pv_exc = virial%pv_exc - virial%pv_xc
3256 52 : virial%pv_virial = virial%pv_virial - virial%pv_xc
3257 : END IF
3258 : ELSE
3259 46 : CALL ks_ref_potential(qs_env, vh_rspace, vxc_rspace, vtau_rspace, vadmm_rspace, dummy_real1, dummy_real2)
3260 : END IF
3261 50 : do_tau = ASSOCIATED(vtau_rspace)
3262 :
3263 : !Core forces from the SCF
3264 50 : CALL integrate_v_core_rspace(vh_rspace, qs_env)
3265 :
3266 : !The Hartree-xc potential term, P*dVHxc (mp2 + SCF density x deriv of the SCF potential)
3267 : !Get the total density
3268 50 : CALL qs_rho_get(rho, rho_ao=rho_ao)
3269 112 : DO ispin = 1, nspins
3270 112 : CALL dbcsr_add(rho_ao(ispin)%matrix, matrix_p_mp2(ispin)%matrix, 1.0_dp, 1.0_dp)
3271 : END DO
3272 :
3273 50 : CALL get_qs_env(qs_env, pw_env=pw_env)
3274 : CALL pw_env_get(pw_env, auxbas_pw_pool=auxbas_pw_pool, &
3275 50 : poisson_env=poisson_env)
3276 50 : CALL auxbas_pw_pool%create_pw(vhxc_rspace)
3277 :
3278 98 : IF (use_virial) pv_loc = virial%pv_virial
3279 :
3280 50 : IF (do_exx) THEN
3281 : !Only want response XC contribution, but SCF+response Hartree contribution
3282 44 : DO ispin = 1, nspins
3283 : !Hartree
3284 26 : CALL pw_transfer(vh_rspace, vhxc_rspace)
3285 : CALL integrate_v_rspace(v_rspace=vhxc_rspace, &
3286 : hmat=scrm(ispin), pmat=rho_ao(ispin), &
3287 26 : qs_env=qs_env, calculate_forces=.TRUE.)
3288 : !XC
3289 26 : CALL pw_transfer(vxc_rspace(ispin), vhxc_rspace)
3290 : CALL integrate_v_rspace(v_rspace=vhxc_rspace, &
3291 : hmat=scrm(ispin), pmat=matrix_p_mp2(ispin), &
3292 26 : qs_env=qs_env, calculate_forces=.TRUE.)
3293 44 : IF (do_tau) THEN
3294 : CALL integrate_v_rspace(v_rspace=vtau_rspace(ispin), &
3295 : hmat=scrm(ispin), pmat=matrix_p_mp2(ispin), &
3296 0 : qs_env=qs_env, calculate_forces=.TRUE., compute_tau=.TRUE.)
3297 : END IF
3298 : END DO
3299 : ELSE
3300 68 : DO ispin = 1, nspins
3301 36 : CALL pw_transfer(vh_rspace, vhxc_rspace)
3302 36 : CALL pw_axpy(vxc_rspace(ispin), vhxc_rspace)
3303 : CALL integrate_v_rspace(v_rspace=vhxc_rspace, &
3304 : hmat=scrm(ispin), pmat=rho_ao(ispin), &
3305 36 : qs_env=qs_env, calculate_forces=.TRUE.)
3306 68 : IF (do_tau) THEN
3307 : CALL integrate_v_rspace(v_rspace=vtau_rspace(ispin), &
3308 : hmat=scrm(ispin), pmat=rho_ao(ispin), &
3309 8 : qs_env=qs_env, calculate_forces=.TRUE., compute_tau=.TRUE.)
3310 : END IF
3311 : END DO
3312 : END IF
3313 50 : CALL auxbas_pw_pool%give_back_pw(vhxc_rspace)
3314 :
3315 98 : IF (use_virial) virial%pv_ehartree = virial%pv_ehartree + (virial%pv_virial - pv_loc)
3316 :
3317 : !The admm projection contribution (mp2 + SCF densities). If EXX, then only mp2 density
3318 50 : IF (dft_control%do_admm) THEN
3319 : CALL get_admm_env(admm_env, task_list_aux_fit=task_list_aux_fit, rho_aux_fit=rho_aux_fit, &
3320 16 : matrix_s_aux_fit=matrix_s_aux_fit)
3321 16 : CALL qs_rho_get(rho_aux_fit, rho_ao=rho_ao_aux)
3322 16 : CALL dbcsr_allocate_matrix_set(scrm_admm, nspins)
3323 36 : DO ispin = 1, nspins
3324 20 : ALLOCATE (scrm_admm(ispin)%matrix)
3325 20 : CALL dbcsr_create(scrm_admm(ispin)%matrix, template=matrix_s_aux_fit(1)%matrix)
3326 20 : CALL dbcsr_copy(scrm_admm(ispin)%matrix, matrix_s_aux_fit(1)%matrix)
3327 36 : CALL dbcsr_set(scrm_admm(ispin)%matrix, 0.0_dp)
3328 : END DO
3329 :
3330 64 : IF (use_virial) pv_loc = virial%pv_virial
3331 16 : IF (.NOT. qs_env%admm_env%aux_exch_func == do_admm_aux_exch_func_none) THEN
3332 36 : DO ispin = 1, nspins
3333 36 : IF (do_exx) THEN
3334 : CALL integrate_v_rspace(v_rspace=vadmm_rspace(ispin), &
3335 : hmat=scrm_admm(ispin), pmat=matrix_p_mp2_admm(ispin), &
3336 : qs_env=qs_env, calculate_forces=.TRUE., &
3337 8 : basis_type="AUX_FIT", task_list_external=task_list_aux_fit)
3338 : ELSE
3339 12 : CALL dbcsr_add(rho_ao_aux(ispin)%matrix, matrix_p_mp2_admm(ispin)%matrix, 1.0_dp, 1.0_dp)
3340 : CALL integrate_v_rspace(v_rspace=vadmm_rspace(ispin), &
3341 : hmat=scrm_admm(ispin), pmat=rho_ao_aux(ispin), &
3342 : qs_env=qs_env, calculate_forces=.TRUE., &
3343 12 : basis_type="AUX_FIT", task_list_external=task_list_aux_fit)
3344 12 : CALL dbcsr_add(rho_ao_aux(ispin)%matrix, matrix_p_mp2_admm(ispin)%matrix, 1.0_dp, -1.0_dp)
3345 : END IF
3346 : END DO
3347 : END IF
3348 64 : IF (use_virial) virial%pv_ehartree = virial%pv_ehartree + (virial%pv_virial - pv_loc)
3349 :
3350 16 : CALL tddft_hfx_matrix(scrm_admm, rho_ao_aux, qs_env, .FALSE., .FALSE.)
3351 :
3352 16 : IF (do_exx) THEN
3353 4 : CALL admm_projection_derivative(qs_env, scrm_admm, matrix_p_mp2)
3354 : ELSE
3355 12 : CALL admm_projection_derivative(qs_env, scrm_admm, rho_ao)
3356 : END IF
3357 : END IF
3358 :
3359 : !The exact-exchange term (mp2 + SCF densities)
3360 50 : xc_section => section_vals_get_subs_vals(qs_env%input, "DFT%XC")
3361 50 : hfx_section => section_vals_get_subs_vals(xc_section, "HF")
3362 50 : CALL section_vals_get(hfx_section, explicit=do_hfx)
3363 :
3364 50 : IF (do_hfx) THEN
3365 32 : CALL section_vals_get(hfx_section, n_repetition=n_rep_hf)
3366 32 : CPASSERT(n_rep_hf == 1)
3367 80 : IF (use_virial) virial%pv_fock_4c = 0.0_dp
3368 :
3369 : !In case of EXX, only want to response HFX forces, as the SCF will change according to RI_RPA%HF
3370 32 : IF (do_exx) THEN
3371 8 : IF (dft_control%do_admm) THEN
3372 4 : CALL get_admm_env(admm_env, rho_aux_fit=rho_aux_fit)
3373 4 : CALL qs_rho_get(rho_aux_fit, rho_ao=rho_ao_aux, rho_ao_kp=dbcsr_work_p)
3374 4 : IF (x_data(1, 1)%do_hfx_ri) THEN
3375 :
3376 : CALL hfx_ri_update_forces(qs_env, x_data(1, 1)%ri_data, nspins, &
3377 : x_data(1, 1)%general_parameter%fraction, &
3378 : rho_ao=dbcsr_work_p, rho_ao_resp=matrix_p_mp2_admm, &
3379 0 : use_virial=use_virial, resp_only=.TRUE.)
3380 : ELSE
3381 : CALL derivatives_four_center(qs_env, dbcsr_work_p, matrix_p_mp2_admm, hfx_section, para_env, &
3382 4 : 1, use_virial, resp_only=.TRUE.)
3383 : END IF
3384 : ELSE
3385 8 : DO ispin = 1, nspins
3386 8 : CALL dbcsr_add(rho_ao(ispin)%matrix, matrix_p_mp2(ispin)%matrix, 1.0_dp, -1.0_dp)
3387 : END DO
3388 4 : CALL qs_rho_get(rho, rho_ao_kp=dbcsr_work_p)
3389 4 : IF (x_data(1, 1)%do_hfx_ri) THEN
3390 :
3391 : CALL hfx_ri_update_forces(qs_env, x_data(1, 1)%ri_data, nspins, &
3392 : x_data(1, 1)%general_parameter%fraction, &
3393 : rho_ao=dbcsr_work_p, rho_ao_resp=matrix_p_mp2, &
3394 0 : use_virial=use_virial, resp_only=.TRUE.)
3395 : ELSE
3396 : CALL derivatives_four_center(qs_env, dbcsr_work_p, matrix_p_mp2, hfx_section, para_env, &
3397 4 : 1, use_virial, resp_only=.TRUE.)
3398 : END IF
3399 8 : DO ispin = 1, nspins
3400 8 : CALL dbcsr_add(rho_ao(ispin)%matrix, matrix_p_mp2(ispin)%matrix, 1.0_dp, 1.0_dp)
3401 : END DO
3402 : END IF !admm
3403 :
3404 : ELSE !No Exx
3405 24 : IF (dft_control%do_admm) THEN
3406 12 : CALL get_admm_env(admm_env, rho_aux_fit=rho_aux_fit)
3407 12 : CALL qs_rho_get(rho_aux_fit, rho_ao=rho_ao_aux, rho_ao_kp=dbcsr_work_p)
3408 24 : DO ispin = 1, nspins
3409 24 : CALL dbcsr_add(rho_ao_aux(ispin)%matrix, matrix_p_mp2_admm(ispin)%matrix, 1.0_dp, 1.0_dp)
3410 : END DO
3411 12 : IF (x_data(1, 1)%do_hfx_ri) THEN
3412 :
3413 : CALL hfx_ri_update_forces(qs_env, x_data(1, 1)%ri_data, nspins, &
3414 : x_data(1, 1)%general_parameter%fraction, &
3415 : rho_ao=dbcsr_work_p, rho_ao_resp=matrix_p_mp2_admm, &
3416 0 : use_virial=use_virial, resp_only=.FALSE.)
3417 : ELSE
3418 : CALL derivatives_four_center(qs_env, dbcsr_work_p, matrix_p_mp2_admm, hfx_section, para_env, &
3419 12 : 1, use_virial, resp_only=.FALSE.)
3420 : END IF
3421 24 : DO ispin = 1, nspins
3422 24 : CALL dbcsr_add(rho_ao_aux(ispin)%matrix, matrix_p_mp2_admm(ispin)%matrix, 1.0_dp, -1.0_dp)
3423 : END DO
3424 : ELSE
3425 12 : CALL qs_rho_get(rho, rho_ao_kp=dbcsr_work_p)
3426 12 : IF (x_data(1, 1)%do_hfx_ri) THEN
3427 :
3428 : CALL hfx_ri_update_forces(qs_env, x_data(1, 1)%ri_data, nspins, &
3429 : x_data(1, 1)%general_parameter%fraction, &
3430 : rho_ao=dbcsr_work_p, rho_ao_resp=matrix_p_mp2, &
3431 0 : use_virial=use_virial, resp_only=.FALSE.)
3432 : ELSE
3433 : CALL derivatives_four_center(qs_env, dbcsr_work_p, matrix_p_mp2, hfx_section, para_env, &
3434 12 : 1, use_virial, resp_only=.FALSE.)
3435 : END IF
3436 : END IF
3437 : END IF !do_exx
3438 :
3439 32 : IF (use_virial) THEN
3440 52 : virial%pv_exx = virial%pv_exx - virial%pv_fock_4c
3441 52 : virial%pv_virial = virial%pv_virial - virial%pv_fock_4c
3442 : END IF
3443 : END IF
3444 :
3445 : !retrieve the SCF density
3446 50 : CALL qs_rho_get(rho, rho_ao=rho_ao)
3447 112 : DO ispin = 1, nspins
3448 112 : CALL dbcsr_add(rho_ao(ispin)%matrix, matrix_p_mp2(ispin)%matrix, 1.0_dp, -1.0_dp)
3449 : END DO
3450 :
3451 : !From here, we need to do everything twice. Once for the response density, and once for the
3452 : !density that is used for the trace Tr[P*F]. The reason is that the former is needed for the
3453 : !eventual overlap contribution from matrix_wz
3454 : !Only with the mp2 density
3455 :
3456 436 : ALLOCATE (current_density(nspins), current_mat_h(nspins), current_density_admm(nspins))
3457 150 : DO idens = 1, 2
3458 224 : DO ispin = 1, nspins
3459 224 : IF (idens == 1) THEN
3460 62 : current_density(ispin)%matrix => matrix_p_F(ispin)%matrix
3461 62 : current_mat_h(ispin)%matrix => scrm(ispin)%matrix
3462 62 : IF (dft_control%do_admm) current_density_admm(ispin)%matrix => matrix_p_F_admm(ispin)%matrix
3463 : ELSE
3464 62 : current_density(ispin)%matrix => p_env%p1(ispin)%matrix
3465 62 : current_mat_h(ispin)%matrix => matrix_hz(ispin)%matrix
3466 62 : IF (dft_control%do_admm) current_density_admm(ispin)%matrix => p_env%p1_admm(ispin)%matrix
3467 : END IF
3468 : END DO
3469 :
3470 : !The core-denstiy derivative
3471 748 : ALLOCATE (rhoz_r(nspins), rhoz_g(nspins))
3472 224 : DO ispin = 1, nspins
3473 124 : CALL auxbas_pw_pool%create_pw(rhoz_r(ispin))
3474 224 : CALL auxbas_pw_pool%create_pw(rhoz_g(ispin))
3475 : END DO
3476 100 : CALL auxbas_pw_pool%create_pw(rhoz_tot_gspace)
3477 100 : CALL auxbas_pw_pool%create_pw(zv_hartree_rspace)
3478 100 : CALL auxbas_pw_pool%create_pw(zv_hartree_gspace)
3479 :
3480 100 : CALL pw_zero(rhoz_tot_gspace)
3481 224 : DO ispin = 1, nspins
3482 : CALL calculate_rho_elec(ks_env=qs_env%ks_env, matrix_p=current_density(ispin)%matrix, &
3483 124 : rho=rhoz_r(ispin), rho_gspace=rhoz_g(ispin))
3484 224 : CALL pw_axpy(rhoz_g(ispin), rhoz_tot_gspace)
3485 : END DO
3486 :
3487 100 : IF (use_virial) THEN
3488 :
3489 8 : CALL get_qs_env(qs_env, rho=rho)
3490 8 : CALL auxbas_pw_pool%create_pw(rho_tot_gspace)
3491 :
3492 8 : CALL calc_rho_tot_gspace(rho_tot_gspace, qs_env, rho)
3493 :
3494 8 : h_stress(:, :) = 0.0_dp
3495 : CALL pw_poisson_solve(poisson_env, &
3496 : density=rhoz_tot_gspace, &
3497 : ehartree=ehartree, &
3498 : vhartree=zv_hartree_gspace, &
3499 : h_stress=h_stress, &
3500 8 : aux_density=rho_tot_gspace)
3501 :
3502 8 : CALL auxbas_pw_pool%give_back_pw(rho_tot_gspace)
3503 :
3504 : !Green contribution
3505 104 : virial%pv_ehartree = virial%pv_ehartree + 2.0_dp*h_stress/REAL(para_env%num_pe, dp)
3506 104 : virial%pv_virial = virial%pv_virial + 2.0_dp*h_stress/REAL(para_env%num_pe, dp)
3507 :
3508 : ELSE
3509 : CALL pw_poisson_solve(poisson_env, rhoz_tot_gspace, ehartree, &
3510 92 : zv_hartree_gspace)
3511 : END IF
3512 :
3513 100 : CALL pw_transfer(zv_hartree_gspace, zv_hartree_rspace)
3514 100 : CALL pw_scale(zv_hartree_rspace, zv_hartree_rspace%pw_grid%dvol)
3515 100 : CALL integrate_v_core_rspace(zv_hartree_rspace, qs_env)
3516 :
3517 100 : IF (do_tau) THEN
3518 : BLOCK
3519 : TYPE(pw_c1d_gs_type) :: tauz_g
3520 16 : CALL auxbas_pw_pool%create_pw(tauz_g)
3521 48 : ALLOCATE (tauz_r(nspins))
3522 32 : DO ispin = 1, nspins
3523 16 : CALL auxbas_pw_pool%create_pw(tauz_r(ispin))
3524 :
3525 : CALL calculate_rho_elec(ks_env=qs_env%ks_env, matrix_p=current_density(ispin)%matrix, &
3526 32 : rho=tauz_r(ispin), rho_gspace=tauz_g, compute_tau=.TRUE.)
3527 : END DO
3528 16 : CALL auxbas_pw_pool%give_back_pw(tauz_g)
3529 : END BLOCK
3530 : END IF
3531 :
3532 : !Volume contribution to the virial
3533 100 : IF (use_virial) THEN
3534 : !Volume contribution
3535 : exc = 0.0_dp
3536 16 : DO ispin = 1, nspins
3537 : exc = exc + pw_integral_ab(rhoz_r(ispin), vxc_rspace(ispin))/ &
3538 16 : vxc_rspace(ispin)%pw_grid%dvol
3539 : END DO
3540 8 : IF (ASSOCIATED(vtau_rspace)) THEN
3541 0 : DO ispin = 1, nspins
3542 : exc = exc + pw_integral_ab(tauz_r(ispin), vtau_rspace(ispin))/ &
3543 0 : vtau_rspace(ispin)%pw_grid%dvol
3544 : END DO
3545 : END IF
3546 32 : DO i = 1, 3
3547 24 : virial%pv_ehartree(i, i) = virial%pv_ehartree(i, i) - 4.0_dp*ehartree/REAL(para_env%num_pe, dp)
3548 24 : virial%pv_exc(i, i) = virial%pv_exc(i, i) - exc/REAL(para_env%num_pe, dp)
3549 : virial%pv_virial(i, i) = virial%pv_virial(i, i) - 4.0_dp*ehartree/REAL(para_env%num_pe, dp) &
3550 32 : - exc/REAL(para_env%num_pe, dp)
3551 : END DO
3552 : END IF
3553 :
3554 : !The xc-kernel term.
3555 100 : IF (dft_control%do_admm) THEN
3556 32 : CALL get_qs_env(qs_env, admm_env=admm_env)
3557 32 : xc_section => admm_env%xc_section_primary
3558 : ELSE
3559 68 : xc_section => section_vals_get_subs_vals(qs_env%input, "DFT%XC")
3560 : END IF
3561 :
3562 196 : IF (use_virial) virial%pv_xc = 0.0_dp
3563 :
3564 100 : ALLOCATE (rhoz)
3565 100 : CALL qs_rho_create(rhoz)
3566 100 : IF (ASSOCIATED(rhoz_r)) THEN
3567 100 : CALL qs_rho_set(rhoz, rho_r=rhoz_r, rho_r_valid=.TRUE.)
3568 : END IF
3569 100 : IF (ASSOCIATED(rhoz_g)) THEN
3570 100 : CALL qs_rho_set(rhoz, rho_g=rhoz_g, rho_g_valid=.TRUE.)
3571 : END IF
3572 100 : IF (ASSOCIATED(tauz_r)) THEN
3573 16 : CALL qs_rho_set(rhoz, tau_r=tauz_r, tau_r_valid=.TRUE.)
3574 : END IF
3575 : !
3576 : CALL qs_fxc_create(qs_env, rho, rhoz, rho0_atom_set, xc_section, .FALSE., &
3577 : v_xc, v_xc_tau, rho1_atom_set, &
3578 100 : compute_virial=use_virial, virial_xc=virial%pv_xc)
3579 : !
3580 100 : DEALLOCATE (rhoz)
3581 :
3582 100 : IF (use_virial) THEN
3583 104 : virial%pv_exc = virial%pv_exc + virial%pv_xc
3584 104 : virial%pv_virial = virial%pv_virial + virial%pv_xc
3585 :
3586 104 : pv_loc = virial%pv_virial
3587 : END IF
3588 :
3589 100 : CALL qs_rho_get(rho, rho_ao_kp=dbcsr_work_p)
3590 224 : DO ispin = 1, nspins
3591 124 : CALL pw_scale(v_xc(ispin), v_xc(ispin)%pw_grid%dvol)
3592 124 : CALL pw_axpy(zv_hartree_rspace, v_xc(ispin))
3593 : CALL integrate_v_rspace(qs_env=qs_env, &
3594 : v_rspace=v_xc(ispin), &
3595 : hmat=current_mat_h(ispin), &
3596 : pmat=dbcsr_work_p(ispin, 1), &
3597 124 : calculate_forces=.TRUE.)
3598 224 : CALL auxbas_pw_pool%give_back_pw(v_xc(ispin))
3599 : END DO
3600 100 : CALL auxbas_pw_pool%give_back_pw(rhoz_tot_gspace)
3601 100 : CALL auxbas_pw_pool%give_back_pw(zv_hartree_rspace)
3602 100 : CALL auxbas_pw_pool%give_back_pw(zv_hartree_gspace)
3603 100 : DEALLOCATE (v_xc)
3604 :
3605 100 : IF (do_tau) THEN
3606 32 : DO ispin = 1, nspins
3607 16 : CALL pw_scale(v_xc_tau(ispin), v_xc_tau(ispin)%pw_grid%dvol)
3608 : CALL integrate_v_rspace(qs_env=qs_env, &
3609 : v_rspace=v_xc_tau(ispin), &
3610 : hmat=current_mat_h(ispin), &
3611 : pmat=dbcsr_work_p(ispin, 1), &
3612 : compute_tau=.TRUE., &
3613 16 : calculate_forces=.TRUE.)
3614 32 : CALL auxbas_pw_pool%give_back_pw(v_xc_tau(ispin))
3615 : END DO
3616 16 : DEALLOCATE (v_xc_tau)
3617 : END IF
3618 :
3619 196 : IF (use_virial) virial%pv_ehartree = virial%pv_ehartree + (virial%pv_virial - pv_loc)
3620 :
3621 100 : IF (do_hfx) THEN
3622 64 : IF (dft_control%do_admm) THEN
3623 72 : DO ispin = 1, nspins
3624 72 : CALL dbcsr_set(scrm_admm(ispin)%matrix, 0.0_dp)
3625 : END DO
3626 32 : CALL qs_rho_get(rho_aux_fit, tau_r_valid=do_tau_admm)
3627 :
3628 32 : IF (.NOT. admm_env%aux_exch_func == do_admm_aux_exch_func_none) THEN
3629 32 : CALL get_admm_env(admm_env, rho_aux_fit=rho_aux_fit)
3630 72 : DO ispin = 1, nspins
3631 40 : CALL pw_zero(rhoz_r(ispin))
3632 40 : CALL pw_zero(rhoz_g(ispin))
3633 : CALL calculate_rho_elec(ks_env=qs_env%ks_env, matrix_p=current_density_admm(ispin)%matrix, &
3634 : rho=rhoz_r(ispin), rho_gspace=rhoz_g(ispin), &
3635 72 : basis_type="AUX_FIT", task_list_external=task_list_aux_fit)
3636 : END DO
3637 :
3638 32 : IF (do_tau_admm) THEN
3639 : BLOCK
3640 : TYPE(pw_c1d_gs_type) :: tauz_g
3641 0 : CALL auxbas_pw_pool%create_pw(tauz_g)
3642 0 : DO ispin = 1, nspins
3643 0 : CALL pw_zero(tauz_r(ispin))
3644 : CALL calculate_rho_elec(ks_env=qs_env%ks_env, matrix_p=current_density(ispin)%matrix, &
3645 : rho=tauz_r(ispin), rho_gspace=tauz_g, &
3646 : basis_type="AUX_FIT", task_list_external=task_list_aux_fit, &
3647 0 : compute_tau=.TRUE.)
3648 : END DO
3649 0 : CALL auxbas_pw_pool%give_back_pw(tauz_g)
3650 : END BLOCK
3651 : END IF
3652 :
3653 : !Volume contribution to the virial
3654 32 : IF (use_virial) THEN
3655 : exc = 0.0_dp
3656 16 : DO ispin = 1, nspins
3657 : exc = exc + pw_integral_ab(rhoz_r(ispin), vadmm_rspace(ispin))/ &
3658 16 : vadmm_rspace(ispin)%pw_grid%dvol
3659 : END DO
3660 32 : DO i = 1, 3
3661 24 : virial%pv_exc(i, i) = virial%pv_exc(i, i) - exc/REAL(para_env%num_pe, dp)
3662 32 : virial%pv_virial(i, i) = virial%pv_virial(i, i) - exc/REAL(para_env%num_pe, dp)
3663 : END DO
3664 :
3665 104 : virial%pv_xc = 0.0_dp
3666 : END IF
3667 :
3668 32 : xc_section => admm_env%xc_section_aux
3669 32 : ALLOCATE (rhoz)
3670 32 : CALL qs_rho_create(rhoz)
3671 32 : IF (ASSOCIATED(rhoz_r)) THEN
3672 32 : CALL qs_rho_set(rhoz, rho_r=rhoz_r, rho_r_valid=.TRUE.)
3673 : END IF
3674 32 : IF (ASSOCIATED(rhoz_g)) THEN
3675 32 : CALL qs_rho_set(rhoz, rho_g=rhoz_g, rho_g_valid=.TRUE.)
3676 : END IF
3677 32 : IF (ASSOCIATED(tauz_r)) THEN
3678 0 : CALL qs_rho_set(rhoz, tau_r=tauz_r, tau_r_valid=.TRUE.)
3679 : END IF
3680 : !
3681 : CALL qs_fxc_create(qs_env, rho_aux_fit, rhoz, rho0_atom_set, xc_section, .FALSE., &
3682 : v_xc, v_xc_tau, rho1_atom_set, &
3683 32 : compute_virial=use_virial, virial_xc=virial%pv_xc)
3684 : !
3685 32 : DEALLOCATE (rhoz)
3686 :
3687 32 : IF (use_virial) THEN
3688 104 : virial%pv_exc = virial%pv_exc + virial%pv_xc
3689 104 : virial%pv_virial = virial%pv_virial + virial%pv_xc
3690 :
3691 104 : pv_loc = virial%pv_virial
3692 : END IF
3693 :
3694 32 : CALL qs_rho_get(rho_aux_fit, rho_ao_kp=dbcsr_work_p)
3695 72 : DO ispin = 1, nspins
3696 40 : CALL pw_scale(v_xc(ispin), v_xc(ispin)%pw_grid%dvol)
3697 : CALL integrate_v_rspace(qs_env=qs_env, &
3698 : v_rspace=v_xc(ispin), &
3699 : hmat=scrm_admm(ispin), &
3700 : pmat=dbcsr_work_p(ispin, 1), &
3701 : calculate_forces=.TRUE., &
3702 : basis_type="AUX_FIT", &
3703 40 : task_list_external=task_list_aux_fit)
3704 72 : CALL auxbas_pw_pool%give_back_pw(v_xc(ispin))
3705 : END DO
3706 32 : DEALLOCATE (v_xc)
3707 :
3708 32 : IF (do_tau_admm) THEN
3709 0 : DO ispin = 1, nspins
3710 0 : CALL pw_scale(v_xc_tau(ispin), v_xc_tau(ispin)%pw_grid%dvol)
3711 : CALL integrate_v_rspace(qs_env=qs_env, &
3712 : v_rspace=v_xc_tau(ispin), &
3713 : hmat=scrm_admm(ispin), &
3714 : pmat=dbcsr_work_p(ispin, 1), &
3715 : calculate_forces=.TRUE., &
3716 : basis_type="AUX_FIT", &
3717 : task_list_external=task_list_aux_fit, &
3718 0 : compute_tau=.TRUE.)
3719 0 : CALL auxbas_pw_pool%give_back_pw(v_xc_tau(ispin))
3720 : END DO
3721 0 : DEALLOCATE (v_xc_tau)
3722 : END IF
3723 :
3724 128 : IF (use_virial) virial%pv_ehartree = virial%pv_ehartree + (virial%pv_virial - pv_loc)
3725 : END IF
3726 :
3727 32 : CALL tddft_hfx_matrix(scrm_admm, current_density_admm, qs_env, .FALSE., .FALSE.)
3728 :
3729 32 : CALL qs_rho_get(rho, rho_ao_kp=dbcsr_work_p)
3730 32 : CALL admm_projection_derivative(qs_env, scrm_admm, dbcsr_work_p(:, 1))
3731 :
3732 : !If response density, need to get matrix_hz contribution
3733 32 : CALL dbcsr_create(dbcsr_work, template=matrix_s(1)%matrix)
3734 32 : IF (idens == 2) THEN
3735 16 : nao = admm_env%nao_orb
3736 16 : nao_aux = admm_env%nao_aux_fit
3737 36 : DO ispin = 1, nspins
3738 20 : CALL dbcsr_copy(dbcsr_work, matrix_hz(ispin)%matrix)
3739 20 : CALL dbcsr_set(dbcsr_work, 0.0_dp)
3740 :
3741 : CALL cp_dbcsr_sm_fm_multiply(scrm_admm(ispin)%matrix, admm_env%A, &
3742 20 : admm_env%work_aux_orb, nao)
3743 : CALL parallel_gemm('T', 'N', nao, nao, nao_aux, &
3744 : 1.0_dp, admm_env%A, admm_env%work_aux_orb, 0.0_dp, &
3745 20 : admm_env%work_orb_orb)
3746 20 : CALL copy_fm_to_dbcsr(admm_env%work_orb_orb, dbcsr_work, keep_sparsity=.TRUE.)
3747 36 : CALL dbcsr_add(matrix_hz(ispin)%matrix, dbcsr_work, 1.0_dp, 1.0_dp)
3748 : END DO
3749 : END IF
3750 :
3751 32 : CALL dbcsr_release(dbcsr_work)
3752 : ELSE !no admm
3753 :
3754 : !Need the contribution to matrix_hz as well
3755 32 : IF (idens == 2) THEN
3756 16 : CALL tddft_hfx_matrix(matrix_hz, current_density, qs_env, .FALSE., .FALSE.)
3757 : END IF
3758 : END IF !admm
3759 : END IF !do_hfx
3760 :
3761 224 : DO ispin = 1, nspins
3762 124 : CALL auxbas_pw_pool%give_back_pw(rhoz_r(ispin))
3763 224 : CALL auxbas_pw_pool%give_back_pw(rhoz_g(ispin))
3764 : END DO
3765 100 : DEALLOCATE (rhoz_r, rhoz_g)
3766 :
3767 250 : IF (do_tau) THEN
3768 32 : DO ispin = 1, nspins
3769 32 : CALL auxbas_pw_pool%give_back_pw(tauz_r(ispin))
3770 : END DO
3771 16 : DEALLOCATE (tauz_r)
3772 : END IF
3773 : END DO !idens
3774 50 : CALL dbcsr_deallocate_matrix_set(scrm_admm)
3775 :
3776 50 : DEALLOCATE (current_density, current_mat_h, current_density_admm)
3777 50 : CALL dbcsr_deallocate_matrix_set(scrm)
3778 :
3779 : !The energy weighted and overlap term. ONLY with the response density
3780 50 : focc = 2.0_dp
3781 50 : IF (nspins == 2) focc = 1.0_dp
3782 50 : CALL get_qs_env(qs_env, mos=mos)
3783 112 : DO ispin = 1, nspins
3784 62 : CALL get_mo_set(mo_set=mos(ispin), homo=nocc)
3785 : CALL calculate_whz_matrix(mos(ispin)%mo_coeff, matrix_hz(ispin)%matrix, &
3786 112 : p_env%w1(ispin)%matrix, focc, nocc)
3787 : END DO
3788 50 : IF (nspins == 2) CALL dbcsr_add(p_env%w1(1)%matrix, p_env%w1(2)%matrix, 1.0_dp, 1.0_dp)
3789 :
3790 : !Add to it the SCF W matrix, except if EXX (because taken care of by HF response)
3791 50 : IF (.NOT. do_exx) THEN
3792 32 : CALL compute_matrix_w(qs_env, calc_forces=.TRUE.)
3793 32 : CALL get_qs_env(qs_env, matrix_w=matrix_w)
3794 32 : CALL dbcsr_add(p_env%w1(1)%matrix, matrix_w(1)%matrix, 1.0_dp, 1.0_dp)
3795 32 : IF (nspins == 2) CALL dbcsr_add(p_env%w1(1)%matrix, matrix_w(2)%matrix, 1.0_dp, 1.0_dp)
3796 : END IF
3797 :
3798 50 : NULLIFY (scrm)
3799 : CALL build_overlap_matrix(qs_env%ks_env, matrix_s=scrm, &
3800 : matrix_name="OVERLAP MATRIX", &
3801 : basis_type_a="ORB", basis_type_b="ORB", &
3802 : sab_nl=sab_orb, calculate_forces=.TRUE., &
3803 50 : matrix_p=p_env%w1(1)%matrix)
3804 :
3805 50 : IF (.NOT. do_exx) THEN
3806 32 : CALL dbcsr_add(p_env%w1(1)%matrix, matrix_w(1)%matrix, 1.0_dp, -1.0_dp)
3807 32 : IF (nspins == 2) CALL dbcsr_add(p_env%w1(1)%matrix, matrix_w(2)%matrix, 1.0_dp, -1.0_dp)
3808 68 : DO ispin = 1, nspins
3809 68 : CALL dbcsr_set(matrix_w(ispin)%matrix, 0.0_dp)
3810 : END DO
3811 : END IF
3812 :
3813 50 : IF (nspins == 2) CALL dbcsr_add(p_env%w1(1)%matrix, p_env%w1(2)%matrix, 1.0_dp, -1.0_dp)
3814 50 : CALL dbcsr_deallocate_matrix_set(scrm)
3815 :
3816 50 : IF (use_virial) virial%pv_calculate = .FALSE.
3817 :
3818 : !clean-up
3819 50 : CALL auxbas_pw_pool%give_back_pw(vh_rspace)
3820 :
3821 112 : DO ispin = 1, nspins
3822 62 : CALL auxbas_pw_pool%give_back_pw(vxc_rspace(ispin))
3823 62 : IF (ASSOCIATED(vtau_rspace)) THEN
3824 8 : CALL auxbas_pw_pool%give_back_pw(vtau_rspace(ispin))
3825 : END IF
3826 112 : IF (ASSOCIATED(vadmm_rspace)) THEN
3827 20 : CALL auxbas_pw_pool%give_back_pw(vadmm_rspace(ispin))
3828 : END IF
3829 : END DO
3830 50 : DEALLOCATE (vxc_rspace)
3831 50 : IF (ASSOCIATED(vtau_rspace)) DEALLOCATE (vtau_rspace)
3832 50 : IF (ASSOCIATED(vadmm_rspace)) DEALLOCATE (vadmm_rspace)
3833 :
3834 50 : CALL timestop(handle)
3835 :
3836 100 : END SUBROUTINE update_im_time_forces
3837 :
3838 : ! **************************************************************************************************
3839 : !> \brief Iteratively builds the matrix Y = sum_k Y_k until convergence, where
3840 : !> Y_k = 1/k*2^n (A/2^n) Y_k-1 + 1/k!*2^n * PR(n) * (A/2^n)^(k-1)
3841 : !> n is chosen such that the norm of A is < 1 (and e^A converges fast)
3842 : !> PR(n) = e^(A/2^n)*PR(n-1) + PR(n-1)*e^(A/2^n), PR(0) = P*R^T
3843 : !> \param Y ...
3844 : !> \param A ...
3845 : !> \param P ...
3846 : !> \param R ...
3847 : !> \param filter_eps ...
3848 : ! **************************************************************************************************
3849 340 : SUBROUTINE build_Y_matrix(Y, A, P, R, filter_eps)
3850 :
3851 : TYPE(dbcsr_type), INTENT(OUT) :: Y
3852 : TYPE(dbcsr_type), INTENT(INOUT) :: A, P, R
3853 : REAL(dp), INTENT(IN) :: filter_eps
3854 :
3855 : CHARACTER(LEN=*), PARAMETER :: routineN = 'build_Y_matrix'
3856 :
3857 : INTEGER :: handle, k, n
3858 : REAL(dp) :: norm_scalar, threshold
3859 : TYPE(dbcsr_type) :: A2n, exp_A2n, PRn, work, work2, Yk
3860 :
3861 340 : CALL timeset(routineN, handle)
3862 :
3863 340 : threshold = 1.0E-16_dp
3864 :
3865 : !Find n such that norm(A) < 1 and we insure convergence of the exponential
3866 340 : norm_scalar = dbcsr_frobenius_norm(A)
3867 :
3868 : !checked: result invariant with value of n
3869 340 : n = 1
3870 466 : DO
3871 806 : IF ((norm_scalar/2.0_dp**n) < 1.0_dp) EXIT
3872 466 : n = n + 1
3873 : END DO
3874 :
3875 : !Calculate PR(n) recursively
3876 340 : CALL dbcsr_create(PRn, template=A, matrix_type=dbcsr_type_no_symmetry)
3877 340 : CALL dbcsr_create(work, template=A, matrix_type=dbcsr_type_no_symmetry)
3878 340 : CALL dbcsr_multiply('N', 'N', 1.0_dp, P, R, 0.0_dp, work, filter_eps=filter_eps)
3879 340 : CALL dbcsr_create(exp_A2n, template=A, matrix_type=dbcsr_type_no_symmetry)
3880 :
3881 1146 : DO k = 1, n
3882 806 : CALL matrix_exponential(exp_A2n, A, 1.0_dp, 0.5_dp**k, threshold)
3883 806 : CALL dbcsr_multiply('N', 'N', 1.0_dp, exp_A2n, work, 0.0_dp, PRn, filter_eps=filter_eps)
3884 806 : CALL dbcsr_multiply('N', 'N', 1.0_dp, work, exp_A2n, 1.0_dp, PRn, filter_eps=filter_eps)
3885 1146 : CALL dbcsr_copy(work, PRn)
3886 : END DO
3887 340 : CALL dbcsr_release(exp_A2n)
3888 :
3889 : !Calculate Y iteratively, until convergence
3890 340 : CALL dbcsr_create(A2n, template=A, matrix_type=dbcsr_type_no_symmetry)
3891 340 : CALL dbcsr_copy(A2n, A)
3892 340 : CALL dbcsr_scale(A2n, 0.5_dp**n)
3893 340 : CALL dbcsr_create(Y, template=A, matrix_type=dbcsr_type_no_symmetry)
3894 340 : CALL dbcsr_create(Yk, template=A, matrix_type=dbcsr_type_no_symmetry)
3895 340 : CALL dbcsr_create(work2, template=A, matrix_type=dbcsr_type_no_symmetry)
3896 :
3897 : !k=1
3898 340 : CALL dbcsr_scale(PRn, 0.5_dp**n)
3899 340 : CALL dbcsr_copy(work, PRn)
3900 340 : CALL dbcsr_copy(work2, PRn)
3901 340 : CALL dbcsr_add(Y, PRn, 1.0_dp, 1.0_dp)
3902 :
3903 340 : k = 1
3904 1908 : DO
3905 2248 : k = k + 1
3906 2248 : CALL dbcsr_multiply('N', 'N', 1.0_dp/REAL(k, dp), A2n, work, 0.0_dp, Yk, filter_eps=filter_eps)
3907 2248 : CALL dbcsr_multiply('N', 'N', 1.0_dp/REAL(k, dp), work2, A2n, 0.0_dp, PRn, filter_eps=filter_eps)
3908 :
3909 2248 : CALL dbcsr_add(Yk, PRn, 1.0_dp, 1.0_dp)
3910 2248 : CALL dbcsr_add(Y, Yk, 1.0_dp, 1.0_dp)
3911 :
3912 2248 : IF (dbcsr_frobenius_norm(Yk) < threshold) EXIT
3913 1908 : CALL dbcsr_copy(work, Yk)
3914 1908 : CALL dbcsr_copy(work2, PRn)
3915 : END DO
3916 :
3917 340 : CALL dbcsr_release(work)
3918 340 : CALL dbcsr_release(work2)
3919 340 : CALL dbcsr_release(PRn)
3920 340 : CALL dbcsr_release(A2n)
3921 340 : CALL dbcsr_release(Yk)
3922 :
3923 340 : CALL timestop(handle)
3924 :
3925 340 : END SUBROUTINE build_Y_matrix
3926 :
3927 : ! **************************************************************************************************
3928 : !> \brief Overwrites the "optimal" Laplace quadrature with that of the first step
3929 : !> \param grid ...
3930 : !> \param do_laplace ...
3931 : !> \param do_im_time ...
3932 : !> \param unit_nr ...
3933 : !> \param qs_env ...
3934 : ! **************************************************************************************************
3935 214 : SUBROUTINE keep_initial_quad(grid, do_laplace, do_im_time, unit_nr, qs_env)
3936 :
3937 : TYPE(time_frequency_grid_type), INTENT(INOUT) :: grid
3938 : LOGICAL, INTENT(IN) :: do_laplace, do_im_time
3939 : INTEGER, INTENT(IN) :: unit_nr
3940 : TYPE(qs_environment_type), POINTER :: qs_env
3941 :
3942 : INTEGER :: jquad
3943 :
3944 214 : IF (do_laplace .OR. do_im_time) THEN
3945 172 : IF (.NOT. ALLOCATED(qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%imaginary_time)) THEN
3946 1410 : ALLOCATE (qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%imaginary_time, SOURCE=grid%imaginary_time)
3947 0 : ALLOCATE (qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%time_weights_at_zero_frequency, &
3948 1410 : SOURCE=grid%time_weights_at_zero_frequency)
3949 : ELSE
3950 : !If weights already stored, we overwrite the new ones
3951 152 : grid%imaginary_time(:) = qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%imaginary_time(:)
3952 : grid%time_weights_at_zero_frequency(:) = &
3953 194 : qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%time_weights_at_zero_frequency(:)
3954 : END IF
3955 : END IF
3956 214 : IF (.NOT. do_laplace) THEN
3957 150 : IF (.NOT. ALLOCATED(qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%frequency)) THEN
3958 1348 : ALLOCATE (qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%frequency, SOURCE=grid%frequency)
3959 0 : ALLOCATE (qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%frequency_weights, &
3960 1348 : SOURCE=grid%frequency_weights)
3961 122 : IF (do_im_time) THEN
3962 0 : ALLOCATE (qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%cosine_time_to_frequency_weights, &
3963 14260 : SOURCE=grid%cosine_time_to_frequency_weights)
3964 0 : ALLOCATE (qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%cosine_frequency_to_time_weights, &
3965 14260 : SOURCE=grid%cosine_frequency_to_time_weights)
3966 : END IF
3967 : ELSE
3968 120 : grid%frequency(:) = qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%frequency(:)
3969 120 : grid%frequency_weights(:) = qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%frequency_weights(:)
3970 28 : IF (do_im_time) THEN
3971 : grid%cosine_time_to_frequency_weights(:, :) = &
3972 184 : qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%cosine_time_to_frequency_weights(:, :)
3973 : grid%cosine_frequency_to_time_weights(:, :) = &
3974 184 : qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%cosine_frequency_to_time_weights(:, :)
3975 : END IF
3976 : END IF
3977 : END IF
3978 214 : IF (unit_nr > 0) THEN
3979 : !Printing order same as in mp2_grids.F for consistency
3980 107 : IF (ALLOCATED(qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%frequency) .AND. (.NOT. do_laplace)) THEN
3981 : WRITE (UNIT=unit_nr, FMT="(T3,A,T75,i6)") &
3982 75 : "MINIMAX_INFO| Number of integration points:", SIZE(grid%frequency)
3983 : WRITE (UNIT=unit_nr, FMT="(T3,A,T54,A,T72,A)") &
3984 75 : "MINIMAX_INFO| Minimax params (freq grid, scaled):", "Weights", "Abscissas"
3985 612 : DO jquad = 1, SIZE(grid%frequency)
3986 : WRITE (UNIT=unit_nr, FMT="(T41,F20.10,F20.10)") &
3987 612 : grid%frequency_weights(jquad), grid%frequency(jquad)
3988 : END DO
3989 75 : CALL m_flush(unit_nr)
3990 : END IF
3991 107 : IF (ALLOCATED(qs_env%mp2_env%ri_rpa_im_time%time_frequency_grid%imaginary_time)) THEN
3992 : WRITE (UNIT=unit_nr, FMT="(T3,A,T75,i6)") &
3993 86 : "MINIMAX_INFO| Number of integration points:", SIZE(grid%imaginary_time)
3994 : WRITE (UNIT=unit_nr, FMT="(T3,A,T54,A,T72,A)") &
3995 86 : "MINIMAX_INFO| Minimax params (time grid, scaled):", "Weights", "Abscissas"
3996 645 : DO jquad = 1, SIZE(grid%imaginary_time)
3997 : WRITE (UNIT=unit_nr, FMT="(T41,F20.10,F20.10)") &
3998 645 : grid%time_weights_at_zero_frequency(jquad), grid%imaginary_time(jquad)
3999 : END DO
4000 86 : CALL m_flush(unit_nr)
4001 : END IF
4002 : END IF
4003 :
4004 214 : END SUBROUTINE keep_initial_quad
4005 :
4006 : END MODULE rpa_im_time_force_methods
|
