ViewVC Help

View File | Revision Log | Show Annotations | View Changeset | Root Listing

root/group/trunk/OOPSE/libmdtools/do_Forces.F90

Comparing trunk/OOPSE/libmdtools/do_Forces.F90 (file contents):
Revision 441 by chuckv, Tue Apr 1 16:50:14 2003 UTC vs.
Revision 1208 by gezelter, Fri May 28 15:21:37 2004 UTC

#	Line 4 | Line 4
4
5		!! @author Charles F. Vardeman II
6		!! @author Matthew Meineke
7	<	!! @version $Id: do_Forces.F90,v 1.7 2003-04-01 16:50:14 chuckv Exp $, $Date: 2003-04-01 16:50:14 $, $Name: not supported by cvs2svn $, $Revision: 1.7 $
7	>	!! @version $Id: do_Forces.F90,v 1.66 2004-05-28 15:21:37 gezelter Exp $, $Date: 2004-05-28 15:21:37 $, $Name: not supported by cvs2svn $, $Revision: 1.66 $
8
9		module do_Forces
10		use force_globals
11		use simulation
12		use definitions
13		use atype_module
14	+	use switcheroo
15		use neighborLists
16		use lj
17		use sticky_pair
18		use dipole_dipole
19	+	use charge_charge
20		use reaction_field
21		use gb_pair
22	+	use vector_class
23	+	use eam
24	+	use status
25		#ifdef IS_MPI
26		use mpiSimulation
27		#endif
#	Line 26 | Line 31 | module do_Forces
31
32		#define __FORTRAN90
33		#include "fForceField.h"
34	+	#include "fSwitchingFunction.h"
35
36	<	logical, save :: do_forces_initialized = .false.
36	>	INTEGER, PARAMETER:: PREPAIR_LOOP = 1
37	>	INTEGER, PARAMETER:: PAIR_LOOP    = 2
38	>
39	>	logical, save :: haveRlist = .false.
40	>	logical, save :: haveNeighborList = .false.
41	>	logical, save :: havePolicies = .false.
42	>	logical, save :: haveSIMvariables = .false.
43	>	logical, save :: havePropertyMap = .false.
44	>	logical, save :: haveSaneForceField = .false.
45		logical, save :: FF_uses_LJ
46		logical, save :: FF_uses_sticky
47	+	logical, save :: FF_uses_charges
48		logical, save :: FF_uses_dipoles
49		logical, save :: FF_uses_RF
50		logical, save :: FF_uses_GB
51		logical, save :: FF_uses_EAM
52	+	logical, save :: SIM_uses_LJ
53	+	logical, save :: SIM_uses_sticky
54	+	logical, save :: SIM_uses_charges
55	+	logical, save :: SIM_uses_dipoles
56	+	logical, save :: SIM_uses_RF
57	+	logical, save :: SIM_uses_GB
58	+	logical, save :: SIM_uses_EAM
59	+	logical, save :: SIM_requires_postpair_calc
60	+	logical, save :: SIM_requires_prepair_calc
61	+	logical, save :: SIM_uses_directional_atoms
62	+	logical, save :: SIM_uses_PBC
63	+	logical, save :: SIM_uses_molecular_cutoffs
64
65	+	real(kind=dp), save :: rlist, rlistsq
66	+
67		public :: init_FF
68		public :: do_force_loop
69	+	public :: setRlistDF
70
71	+	#ifdef PROFILE
72	+	public :: getforcetime
73	+	real, save :: forceTime = 0
74	+	real :: forceTimeInitial, forceTimeFinal
75	+	integer :: nLoops
76	+	#endif
77	+
78	+	type :: Properties
79	+	logical :: is_lj     = .false.
80	+	logical :: is_sticky = .false.
81	+	logical :: is_dp     = .false.
82	+	logical :: is_gb     = .false.
83	+	logical :: is_eam    = .false.
84	+	logical :: is_charge = .false.
85	+	real(kind=DP) :: charge = 0.0_DP
86	+	real(kind=DP) :: dipole_moment = 0.0_DP
87	+	end type Properties
88	+
89	+	type(Properties), dimension(:),allocatable :: PropertyMap
90	+
91		contains
92
93	+	subroutine setRlistDF( this_rlist )
94	+
95	+	real(kind=dp) :: this_rlist
96	+
97	+	rlist = this_rlist
98	+	rlistsq = rlist * rlist
99	+
100	+	haveRlist = .true.
101	+
102	+	end subroutine setRlistDF
103	+
104	+	subroutine createPropertyMap(status)
105	+	integer :: nAtypes
106	+	integer :: status
107	+	integer :: i
108	+	logical :: thisProperty
109	+	real (kind=DP) :: thisDPproperty
110	+
111	+	status = 0
112	+
113	+	nAtypes = getSize(atypes)
114	+
115	+	if (nAtypes == 0) then
116	+	status = -1
117	+	return
118	+	end if
119	+
120	+	if (.not. allocated(PropertyMap)) then
121	+	allocate(PropertyMap(nAtypes))
122	+	endif
123	+
124	+	do i = 1, nAtypes
125	+	call getElementProperty(atypes, i, "is_LJ", thisProperty)
126	+	PropertyMap(i)%is_LJ = thisProperty
127	+
128	+	call getElementProperty(atypes, i, "is_Charge", thisProperty)
129	+	PropertyMap(i)%is_Charge = thisProperty
130	+
131	+	if (thisProperty) then
132	+	call getElementProperty(atypes, i, "charge", thisDPproperty)
133	+	PropertyMap(i)%charge = thisDPproperty
134	+	endif
135	+
136	+	call getElementProperty(atypes, i, "is_DP", thisProperty)
137	+	PropertyMap(i)%is_DP = thisProperty
138	+
139	+	if (thisProperty) then
140	+	call getElementProperty(atypes, i, "dipole_moment", thisDPproperty)
141	+	PropertyMap(i)%dipole_moment = thisDPproperty
142	+	endif
143	+
144	+	call getElementProperty(atypes, i, "is_Sticky", thisProperty)
145	+	PropertyMap(i)%is_Sticky = thisProperty
146	+	call getElementProperty(atypes, i, "is_GB", thisProperty)
147	+	PropertyMap(i)%is_GB = thisProperty
148	+	call getElementProperty(atypes, i, "is_EAM", thisProperty)
149	+	PropertyMap(i)%is_EAM = thisProperty
150	+	end do
151	+
152	+	havePropertyMap = .true.
153	+
154	+	end subroutine createPropertyMap
155	+
156	+	subroutine setSimVariables()
157	+	SIM_uses_LJ = SimUsesLJ()
158	+	SIM_uses_sticky = SimUsesSticky()
159	+	SIM_uses_charges = SimUsesCharges()
160	+	SIM_uses_dipoles = SimUsesDipoles()
161	+	SIM_uses_RF = SimUsesRF()
162	+	SIM_uses_GB = SimUsesGB()
163	+	SIM_uses_EAM = SimUsesEAM()
164	+	SIM_requires_postpair_calc = SimRequiresPostpairCalc()
165	+	SIM_requires_prepair_calc = SimRequiresPrepairCalc()
166	+	SIM_uses_directional_atoms = SimUsesDirectionalAtoms()
167	+	SIM_uses_PBC = SimUsesPBC()
168	+	!SIM_uses_molecular_cutoffs = SimUsesMolecularCutoffs()
169	+
170	+	haveSIMvariables = .true.
171	+
172	+	return
173	+	end subroutine setSimVariables
174	+
175	+	subroutine doReadyCheck(error)
176	+	integer, intent(out) :: error
177	+
178	+	integer :: myStatus
179	+
180	+	error = 0
181	+
182	+	if (.not. havePropertyMap) then
183	+
184	+	myStatus = 0
185	+
186	+	call createPropertyMap(myStatus)
187	+
188	+	if (myStatus .ne. 0) then
189	+	write(default_error, *) 'createPropertyMap failed in do_Forces!'
190	+	error = -1
191	+	return
192	+	endif
193	+	endif
194	+
195	+	if (.not. haveSIMvariables) then
196	+	call setSimVariables()
197	+	endif
198	+
199	+	if (.not. haveRlist) then
200	+	write(default_error, *) 'rList has not been set in do_Forces!'
201	+	error = -1
202	+	return
203	+	endif
204	+
205	+	if (SIM_uses_LJ .and. FF_uses_LJ) then
206	+	if (.not. havePolicies) then
207	+	write(default_error, *) 'LJ mixing Policies have not been set in do_Forces!'
208	+	error = -1
209	+	return
210	+	endif
211	+	endif
212	+
213	+	if (.not. haveNeighborList) then
214	+	write(default_error, *) 'neighbor list has not been initialized in do_Forces!'
215	+	error = -1
216	+	return
217	+	end if
218	+
219	+	if (.not. haveSaneForceField) then
220	+	write(default_error, *) 'Force Field is not sane in do_Forces!'
221	+	error = -1
222	+	return
223	+	end if
224	+
225	+	#ifdef IS_MPI
226	+	if (.not. isMPISimSet()) then
227	+	write(default_error,*) "ERROR: mpiSimulation has not been initialized!"
228	+	error = -1
229	+	return
230	+	endif
231	+	#endif
232	+	return
233	+	end subroutine doReadyCheck
234	+
235	+
236		subroutine init_FF(LJMIXPOLICY, use_RF_c, thisStat)
237
238		integer, intent(in) :: LJMIXPOLICY
#	Line 64 | Line 257 | contains
257
258		FF_uses_LJ = .false.
259		FF_uses_sticky = .false.
260	+	FF_uses_charges = .false.
261		FF_uses_dipoles = .false.
262		FF_uses_GB = .false.
263		FF_uses_EAM = .false.
264
265		call getMatchingElementList(atypes, "is_LJ", .true., nMatches, MatchList)
266		if (nMatches .gt. 0) FF_uses_LJ = .true.
267	<
267	>
268	>	call getMatchingElementList(atypes, "is_Charge", .true., nMatches, MatchList)
269	>	if (nMatches .gt. 0) FF_uses_charges = .true.
270	>
271		call getMatchingElementList(atypes, "is_DP", .true., nMatches, MatchList)
272		if (nMatches .gt. 0) FF_uses_dipoles = .true.
273
#	Line 84 | Line 281 | contains
281		call getMatchingElementList(atypes, "is_EAM", .true., nMatches, MatchList)
282		if (nMatches .gt. 0) FF_uses_EAM = .true.
283
284	+	!! Assume sanity (for the sake of argument)
285	+	haveSaneForceField = .true.
286	+
287		!! check to make sure the FF_uses_RF setting makes sense
288
289		if (FF_uses_dipoles) then
90	–	rrf = getRrf()
91	–	rt = getRt()
92	–	call initialize_dipole(rrf, rt)
290		if (FF_uses_RF) then
291		dielect = getDielect()
292	<	call initialize_rf(rrf, rt, dielect)
292	>	call initialize_rf(dielect)
293		endif
294		else
295		if (FF_uses_RF) then
296		write(default_error,*) 'Using Reaction Field with no dipoles?  Huh?'
297		thisStat = -1
298	+	haveSaneForceField = .false.
299		return
300		endif
301	<	endif
301	>	endif
302
303		if (FF_uses_LJ) then
304
107	–	call getRcut(rcut)
108	–
305		select case (LJMIXPOLICY)
306		case (LB_MIXING_RULE)
307	<	call init_lj_FF(LB_MIXING_RULE, rcut, my_status)
307	>	call init_lj_FF(LB_MIXING_RULE, my_status)
308		case (EXPLICIT_MIXING_RULE)
309	<	call init_lj_FF(EXPLICIT_MIXING_RULE, rcut, my_status)
309	>	call init_lj_FF(EXPLICIT_MIXING_RULE, my_status)
310		case default
311		write(default_error,*) 'unknown LJ Mixing Policy!'
312		thisStat = -1
313	+	haveSaneForceField = .false.
314		return
315		end select
316		if (my_status /= 0) then
317		thisStat = -1
318	+	haveSaneForceField = .false.
319		return
320		end if
321	+	havePolicies = .true.
322		endif
323
324		if (FF_uses_sticky) then
325		call check_sticky_FF(my_status)
326		if (my_status /= 0) then
327		thisStat = -1
328	+	haveSaneForceField = .false.
329		return
330		end if
331		endif
332	<
332	>
333	>
334	>	if (FF_uses_EAM) then
335	>	call init_EAM_FF(my_status)
336	>	if (my_status /= 0) then
337	>	write(default_error, *) "init_EAM_FF returned a bad status"
338	>	thisStat = -1
339	>	haveSaneForceField = .false.
340	>	return
341	>	end if
342	>	endif
343	>
344		if (FF_uses_GB) then
345		call check_gb_pair_FF(my_status)
346		if (my_status .ne. 0) then
347		thisStat = -1
348	+	haveSaneForceField = .false.
349		return
350		endif
351		endif
352
353		if (FF_uses_GB .and. FF_uses_LJ) then
354		endif
355	+	if (.not. haveNeighborList) then
356	+	!! Create neighbor lists
357	+	call expandNeighborList(nLocal, my_status)
358	+	if (my_Status /= 0) then
359	+	write(default_error,*) "SimSetup: ExpandNeighborList returned error."
360	+	thisStat = -1
361	+	return
362	+	endif
363	+	haveNeighborList = .true.
364	+	endif
365
366	<
367	<	do_forces_initialized = .true.
146	<
366	>
367	>
368		end subroutine init_FF
369
370
371		!! Does force loop over i,j pairs. Calls do_pair to calculates forces.
372		!------------------------------------------------------------->
373	<	subroutine do_force_loop(q, A, u_l, f, t, tau, pot, do_pot_c, do_stress_c, &
374	<	error)
373	>	subroutine do_force_loop(q, q_group, A, u_l, f, t, tau, pot, &
374	>	do_pot_c, do_stress_c, error)
375		!! Position array provided by C, dimensioned by getNlocal
376	<	real ( kind = dp ), dimension(3,getNlocal()) :: q
376	>	real ( kind = dp ), dimension(3, nLocal) :: q
377	>	!! molecular center-of-mass position array
378	>	real ( kind = dp ), dimension(3, nGroups) :: q_group
379		!! Rotation Matrix for each long range particle in simulation.
380	<	real( kind = dp), dimension(9,getNlocal()) :: A
380	>	real( kind = dp), dimension(9, nLocal) :: A
381		!! Unit vectors for dipoles (lab frame)
382	<	real( kind = dp ), dimension(3,getNlocal()) :: u_l
382	>	real( kind = dp ), dimension(3,nLocal) :: u_l
383		!! Force array provided by C, dimensioned by getNlocal
384	<	real ( kind = dp ), dimension(3,getNlocal()) :: f
384	>	real ( kind = dp ), dimension(3,nLocal) :: f
385		!! Torsion array provided by C, dimensioned by getNlocal
386	<	real( kind = dp ), dimension(3,getNlocal()) :: t
386	>	real( kind = dp ), dimension(3,nLocal) :: t
387	>
388		!! Stress Tensor
389		real( kind = dp), dimension(9) :: tau
390		real ( kind = dp ) :: pot
391		logical ( kind = 2) :: do_pot_c, do_stress_c
392		logical :: do_pot
393		logical :: do_stress
394	+	logical :: in_switching_region
395		#ifdef IS_MPI
396		real( kind = DP ) :: pot_local
397	<	integer :: nrow
398	<	integer :: ncol
397	>	integer :: nAtomsInRow
398	>	integer :: nAtomsInCol
399	>	integer :: nprocs
400	>	integer :: nGroupsInRow
401	>	integer :: nGroupsInCol
402		#endif
175	–	integer :: nlocal
403		integer :: natoms
404		logical :: update_nlist
405	<	integer :: i, j, jbeg, jend, jnab
405	>	integer :: i, j, jstart, jend, jnab
406	>	integer :: istart, iend
407	>	integer :: ia, jb, atom1, atom2
408		integer :: nlist
409	<	real( kind = DP ) ::  rijsq, rlistsq, rcutsq, rlist, rcut
410	<	real(kind=dp),dimension(3) :: d
409	>	real( kind = DP ) :: ratmsq, rgrpsq, rgrp, vpair, vij
410	>	real( kind = DP ) :: sw, dswdr, swderiv, mf
411	>	real(kind=dp),dimension(3) :: d_atm, d_grp, fpair, fij
412		real(kind=dp) :: rfpot, mu_i, virial
413	<	integer :: me_i
413	>	integer :: me_i, me_j, n_in_i, n_in_j
414		logical :: is_dp_i
415		integer :: neighborListSize
416		integer :: listerror, error
417		integer :: localError
418	+	integer :: propPack_i, propPack_j
419	+	integer :: loopStart, loopEnd, loop
420
421	+	real(kind=dp) :: listSkin = 1.0
422	+
423		!! initialize local variables
424	<
424	>
425		#ifdef IS_MPI
426		pot_local = 0.0_dp
427	<	nlocal = getNlocal()
428	<	nrow   = getNrow(plan_row)
429	<	ncol   = getNcol(plan_col)
427	>	nAtomsInRow   = getNatomsInRow(plan_atom_row)
428	>	nAtomsInCol   = getNatomsInCol(plan_atom_col)
429	>	nGroupsInRow  = getNgroupsInRow(plan_group_row)
430	>	nGroupsInCol  = getNgroupsInCol(plan_group_col)
431	>	write(*,*) nAtomsInRow, nAtomsInCol, nGroupsInRow, nGroupsInCol
432	>	write(*,*) pot_local
433		#else
197	–	nlocal = getNlocal()
434		natoms = nlocal
435		#endif
200	–
201	–	call getRcut(rcut,rc2=rcutsq)
202	–	call getRlist(rlist,rlistsq)
436
437	<	call check_initialization(localError)
437	>	call doReadyCheck(localError)
438		if ( localError .ne. 0 ) then
439	+	call handleError("do_force_loop", "Not Initialized")
440		error = -1
441		return
442		end if
443		call zero_work_arrays()
444	<
444	>
445		do_pot = do_pot_c
446		do_stress = do_stress_c
447	<
447	>
448		! Gather all information needed by all force loops:
449
450		#ifdef IS_MPI
451	+
452	+	call gather(q, q_Row, plan_atom_row_3d)
453	+	call gather(q, q_Col, plan_atom_col_3d)
454
455	<	call gather(q,q_Row,plan_row3d)
456	<	call gather(q,q_Col,plan_col3d)
455	>	call gather(q_group, q_group_Row, plan_group_row_3d)
456	>	call gather(q_group, q_group_Col, plan_group_col_3d)
457
458	<	if (FF_UsesDirectionalAtoms() .and. SimUsesDirectionalAtoms()) then
459	<	call gather(u_l,u_l_Row,plan_row3d)
460	<	call gather(u_l,u_l_Col,plan_col3d)
458	>	if (FF_UsesDirectionalAtoms() .and. SIM_uses_directional_atoms) then
459	>	call gather(u_l, u_l_Row, plan_atom_row_3d)
460	>	call gather(u_l, u_l_Col, plan_atom_col_3d)
461
462	<	call gather(A,A_Row,plan_row_rotation)
463	<	call gather(A,A_Col,plan_col_rotation)
462	>	call gather(A, A_Row, plan_atom_row_rotation)
463	>	call gather(A, A_Col, plan_atom_col_rotation)
464		endif
465
466		#endif
467
468	<	if (FF_RequiresPrepairCalc() .and. SimRequiresPrepairCalc()) then
469	<	!! See if we need to update neighbor lists
470	<	call checkNeighborList(nlocal, q, rcut, rlist, update_nlist)
471	<	!! if_mpi_gather_stuff_for_prepair
472	<	!! do_prepair_loop_if_needed
473	<	!! if_mpi_scatter_stuff_from_prepair
474	<	!! if_mpi_gather_stuff_from_prepair_to_main_loop
468	>	!! Begin force loop timing:
469	>	#ifdef PROFILE
470	>	call cpu_time(forceTimeInitial)
471	>	nloops = nloops + 1
472	>	#endif
473	>
474	>	loopEnd = PAIR_LOOP
475	>	if (FF_RequiresPrepairCalc() .and. SIM_requires_prepair_calc) then
476	>	loopStart = PREPAIR_LOOP
477		else
478	<	!! See if we need to update neighbor lists
240	<	call checkNeighborList(nlocal, q, rcut, rlist, update_nlist)
478	>	loopStart = PAIR_LOOP
479		endif
480	<
480	>
481	>	do loop = loopStart, loopEnd
482	>	write(*,*) 'doign loop ', loop
483	>
484	>	! See if we need to update neighbor lists
485	>	! (but only on the first time through):
486	>	if (loop .eq. loopStart) then
487		#ifdef IS_MPI
488	<
489	<	if (update_nlist) then
488	>	call checkNeighborList(nGroupsInRow, q_group_row, listSkin, &
489	>	update_nlist)
490	>	#else
491	>	call checkNeighborList(nGroups, q_group, listSkin, &
492	>	update_nlist)
493	>	#endif
494	>	endif
495
496	<	!! save current configuration, construct neighbor list,
497	<	!! and calculate forces
498	<	call saveNeighborList(q)
496	>	if (update_nlist) then
497	>	!! save current configuration and construct neighbor list
498	>	#ifdef IS_MPI
499	>	call saveNeighborList(nGroupsInRow, q_group_row)
500	>	#else
501	>	call saveNeighborList(nGroups, q_group)
502	>	#endif
503	>	neighborListSize = size(list)
504	>	nlist = 0
505	>	endif
506
507	<	neighborListSize = size(list)
508	<	nlist = 0
509	<
510	<	do i = 1, nrow
511	<	point(i) = nlist + 1
507	>	istart = 1
508	>	#ifdef IS_MPI
509	>	iend = nGroupsInRow
510	>	#else
511	>	iend = nGroups - 1
512	>	#endif
513	>	outer: do i = istart, iend
514	>
515	>	write(*,*) 'doing ', i
516
517	<	inner: do j = 1, ncol
518	<
519	<	if (skipThisPair(i,j)) cycle inner
520	<
521	<	call get_interatomic_vector(q_Row(:,i), q_Col(:,j), d, rijsq)
522	<
523	<	if (rijsq <  rlistsq) then
524	<
525	<	nlist = nlist + 1
526	<
527	<	if (nlist > neighborListSize) then
528	<	call expandNeighborList(nlocal, listerror)
529	<	if (listerror /= 0) then
530	<	error = -1
531	<	write(DEFAULT_ERROR,*) "ERROR: nlist > list size and max allocations exceeded."
532	<	return
533	<	end if
534	<	neighborListSize = size(list)
535	<	endif
536	<
537	<	list(nlist) = j
538	<
539	<	if (rijsq <  rcutsq) then
540	<	call do_pair(i, j, rijsq, d, do_pot, do_stress, &
281	<	u_l, A, f, t, pot_local)
282	<	endif
517	>	if (update_nlist) point(i) = nlist + 1
518	>
519	>	n_in_i = groupStartRow(i+1) - groupStartRow(i)
520	>
521	>	if (update_nlist) then
522	>	#ifdef IS_MPI
523	>	jstart = 1
524	>	jend = nGroupsInCol
525	>	#else
526	>	jstart = i+1
527	>	jend = nGroups
528	>	#endif
529	>	else
530	>	jstart = point(i)
531	>	jend = point(i+1) - 1
532	>	! make sure group i has neighbors
533	>	if (jstart .gt. jend) cycle outer
534	>	endif
535	>
536	>	do jnab = jstart, jend
537	>	if (update_nlist) then
538	>	j = jnab
539	>	else
540	>	j = list(jnab)
541		endif
284	–	enddo inner
285	–	enddo
542
543	<	point(nrow + 1) = nlist + 1
544	<
545	<	else  !! (of update_check)
290	<
291	<	! use the list to find the neighbors
292	<	do i = 1, nrow
293	<	JBEG = POINT(i)
294	<	JEND = POINT(i+1) - 1
295	<	! check thiat molecule i has neighbors
296	<	if (jbeg .le. jend) then
297	<
298	<	do jnab = jbeg, jend
299	<	j = list(jnab)
300	<
301	<	call get_interatomic_vector(q_Row(:,i), q_Col(:,j), d, rijsq)
302	<	call do_pair(i, j, rijsq, d, do_pot, do_stress, &
303	<	u_l, A, f, t, pot_local)
304	<
305	<	enddo
306	<	endif
307	<	enddo
308	<	endif
309	<
543	>	#ifdef IS_MPI
544	>	call get_interatomic_vector(q_group_Row(:,i), &
545	>	q_group_Col(:,j), d_grp, rgrpsq)
546		#else
547	<
548	<	if (update_nlist) then
549	<
314	<	! save current configuration, contruct neighbor list,
315	<	! and calculate forces
316	<	call saveNeighborList(q)
317	<
318	<	neighborListSize = size(list)
319	<
320	<	nlist = 0
321	<
322	<	do i = 1, natoms-1
323	<	point(i) = nlist + 1
324	<
325	<	inner: do j = i+1, natoms
326	<
327	<	if (skipThisPair(i,j))  cycle inner
328	<
329	<	call get_interatomic_vector(q(:,i), q(:,j), d, rijsq)
330	<
547	>	call get_interatomic_vector(q_group(:,i), &
548	>	q_group(:,j), d_grp, rgrpsq)
549	>	#endif
550
551	<	if (rijsq <  rlistsq) then
551	>	if (rgrpsq < rlistsq) then
552	>	if (update_nlist) then
553	>	nlist = nlist + 1
554	>
555	>	if (nlist > neighborListSize) then
556	>	#ifdef IS_MPI
557	>	call expandNeighborList(nGroupsInRow, listerror)
558	>	#else
559	>	call expandNeighborList(nGroups, listerror)
560	>	#endif
561	>	if (listerror /= 0) then
562	>	error = -1
563	>	write(DEFAULT_ERROR,*) "ERROR: nlist > list size and max allocations exceeded."
564	>	return
565	>	end if
566	>	neighborListSize = size(list)
567	>	endif
568	>
569	>	list(nlist) = j
570	>	endif
571
572	<	nlist = nlist + 1
573	<
574	<	if (nlist > neighborListSize) then
337	<	call expandNeighborList(natoms, listerror)
338	<	if (listerror /= 0) then
339	<	error = -1
340	<	write(DEFAULT_ERROR,*) "ERROR: nlist > list size and max allocations exceeded."
341	<	return
342	<	end if
343	<	neighborListSize = size(list)
572	>	if (loop .eq. PAIR_LOOP) then
573	>	vij = 0.0d0
574	>	fij(1:3) = 0.0d0
575		endif
576
577	<	list(nlist) = j
577	>	call get_switch(rgrpsq, sw, dswdr, rgrp, group_switch, &
578	>	in_switching_region)
579
580	<	if (rijsq <  rcutsq) then
581	<	call do_pair(i, j, rijsq, d, do_pot, do_stress, &
582	<	u_l, A, f, t, pot)
580	>	n_in_j = groupStartCol(j+1) - groupStartCol(j)
581	>
582	>	do ia = groupStartRow(i), groupStartRow(i+1)-1
583	>
584	>	atom1 = groupListRow(ia)
585	>
586	>	inner: do jb = groupStartCol(j), groupStartCol(j+1)-1
587	>
588	>	atom2 = groupListCol(jb)
589	>
590	>	if (skipThisPair(atom1, atom2)) cycle inner
591	>
592	>	if ((n_in_i .eq. 1).and.(n_in_j .eq. 1)) then
593	>	d_atm(1:3) = d_grp(1:3)
594	>	ratmsq = rgrpsq
595	>	else
596	>	#ifdef IS_MPI
597	>	call get_interatomic_vector(q_Row(:,atom1), &
598	>	q_Col(:,atom2), d_atm, ratmsq)
599	>	#else
600	>	call get_interatomic_vector(q(:,atom1), &
601	>	q(:,atom2), d_atm, ratmsq)
602	>	#endif
603	>	endif
604	>	if (loop .eq. PREPAIR_LOOP) then
605	>	#ifdef IS_MPI
606	>	call do_prepair(atom1, atom2, ratmsq, d_atm, sw, &
607	>	rgrpsq, d_grp, do_pot, do_stress, &
608	>	u_l, A, f, t, pot_local)
609	>	#else
610	>	call do_prepair(atom1, atom2, ratmsq, d_atm, sw, &
611	>	rgrpsq, d_grp, do_pot, do_stress, &
612	>	u_l, A, f, t, pot)
613	>	#endif
614	>	else
615	>	#ifdef IS_MPI
616	>	call do_pair(atom1, atom2, ratmsq, d_atm, sw, &
617	>	do_pot, &
618	>	u_l, A, f, t, pot_local, vpair, fpair)
619	>	#else
620	>	call do_pair(atom1, atom2, ratmsq, d_atm, sw, &
621	>	do_pot,  &
622	>	u_l, A, f, t, pot, vpair, fpair)
623	>	#endif
624	>	vij = vij + vpair
625	>	fij(1:3) = fij(1:3) + fpair(1:3)
626	>	endif
627	>	enddo inner
628	>	enddo
629	>
630	>	if (loop .eq. PAIR_LOOP) then
631	>	if (in_switching_region) then
632	>	swderiv = vij*dswdr/rgrp
633	>	fij(1) = fij(1) + swderiv*d_grp(1)
634	>	fij(2) = fij(2) + swderiv*d_grp(2)
635	>	fij(3) = fij(3) + swderiv*d_grp(3)
636	>
637	>	do ia=groupStartRow(i), groupStartRow(i+1)-1
638	>	atom1=groupListRow(ia)
639	>	mf = mfactRow(atom1)
640	>	#ifdef IS_MPI
641	>	f_Row(1,atom1) = f_Row(1,atom1) + swderiv*d_grp(1)*mf
642	>	f_Row(2,atom1) = f_Row(2,atom1) + swderiv*d_grp(2)*mf
643	>	f_Row(3,atom1) = f_Row(3,atom1) + swderiv*d_grp(3)*mf
644	>	#else
645	>	f(1,atom1) = f(1,atom1) + swderiv*d_grp(1)*mf
646	>	f(2,atom1) = f(2,atom1) + swderiv*d_grp(2)*mf
647	>	f(3,atom1) = f(3,atom1) + swderiv*d_grp(3)*mf
648	>	#endif
649	>	enddo
650	>
651	>	do jb=groupStartCol(j), groupStartCol(j+1)-1
652	>	atom2=groupListCol(jb)
653	>	mf = mfactCol(atom2)
654	>	#ifdef IS_MPI
655	>	f_Col(1,atom2) = f_Col(1,atom2) - swderiv*d_grp(1)*mf
656	>	f_Col(2,atom2) = f_Col(2,atom2) - swderiv*d_grp(2)*mf
657	>	f_Col(3,atom2) = f_Col(3,atom2) - swderiv*d_grp(3)*mf
658	>	#else
659	>	f(1,atom2) = f(1,atom2) - swderiv*d_grp(1)*mf
660	>	f(2,atom2) = f(2,atom2) - swderiv*d_grp(2)*mf
661	>	f(3,atom2) = f(3,atom2) - swderiv*d_grp(3)*mf
662	>	#endif
663	>	enddo
664	>	endif
665	>
666	>	if (do_stress) call add_stress_tensor(d_grp, fij)
667		endif
668	<	endif
669	<	enddo inner
670	<	enddo
668	>	end if
669	>	enddo
670	>	enddo outer
671
672	<	point(natoms) = nlist + 1
673	<
674	<	else !! (update)
675	<
676	<	! use the list to find the neighbors
677	<	do i = 1, natoms-1
678	<	JBEG = POINT(i)
679	<	JEND = POINT(i+1) - 1
680	<	! check thiat molecule i has neighbors
681	<	if (jbeg .le. jend) then
682	<
367	<	do jnab = jbeg, jend
368	<	j = list(jnab)
369	<
370	<	call get_interatomic_vector(q(:,i), q(:,j), d, rijsq)
371	<	call do_pair(i, j, rijsq, d, do_pot, do_stress, &
372	<	u_l, A, f, t, pot)
373	<
374	<	enddo
672	>	if (update_nlist) then
673	>	#ifdef IS_MPI
674	>	point(nGroupsInRow + 1) = nlist + 1
675	>	#else
676	>	point(nGroups) = nlist + 1
677	>	#endif
678	>	if (loop .eq. PREPAIR_LOOP) then
679	>	! we just did the neighbor list update on the first
680	>	! pass, so we don't need to do it
681	>	! again on the second pass
682	>	update_nlist = .false.
683		endif
684	<	enddo
685	<	endif
684	>	endif
685	>
686	>	if (loop .eq. PREPAIR_LOOP) then
687	>	call do_preforce(nlocal, pot)
688	>	endif
689	>
690	>	enddo
691
692	<	#endif
692	>	!! Do timing
693	>	#ifdef PROFILE
694	>	call cpu_time(forceTimeFinal)
695	>	forceTime = forceTime + forceTimeFinal - forceTimeInitial
696	>	#endif
697
381	–	! phew, done with main loop.
382	–
698		#ifdef IS_MPI
699		!!distribute forces
700	<
700	>
701		f_temp = 0.0_dp
702	<	call scatter(f_Row,f_temp,plan_row3d)
702	>	call scatter(f_Row,f_temp,plan_atom_row_3d)
703		do i = 1,nlocal
704		f(1:3,i) = f(1:3,i) + f_temp(1:3,i)
705		end do
706	<
706	>
707		f_temp = 0.0_dp
708	<	call scatter(f_Col,f_temp,plan_col3d)
708	>	call scatter(f_Col,f_temp,plan_atom_col_3d)
709		do i = 1,nlocal
710		f(1:3,i) = f(1:3,i) + f_temp(1:3,i)
711		end do
712
713	<	if (FF_UsesDirectionalAtoms() .and. SimUsesDirectionalAtoms()) then
713	>	if (FF_UsesDirectionalAtoms() .and. SIM_uses_directional_atoms) then
714		t_temp = 0.0_dp
715	<	call scatter(t_Row,t_temp,plan_row3d)
715	>	call scatter(t_Row,t_temp,plan_atom_row_3d)
716		do i = 1,nlocal
717		t(1:3,i) = t(1:3,i) + t_temp(1:3,i)
718		end do
719		t_temp = 0.0_dp
720	<	call scatter(t_Col,t_temp,plan_col3d)
720	>	call scatter(t_Col,t_temp,plan_atom_col_3d)
721
722		do i = 1,nlocal
723		t(1:3,i) = t(1:3,i) + t_temp(1:3,i)
#	Line 411 | Line 726 | contains
726
727		if (do_pot) then
728		! scatter/gather pot_row into the members of my column
729	<	call scatter(pot_Row, pot_Temp, plan_row)
730	<
729	>	call scatter(pot_Row, pot_Temp, plan_atom_row)
730	>
731		! scatter/gather pot_local into all other procs
732		! add resultant to get total pot
733		do i = 1, nlocal
#	Line 420 | Line 735 | contains
735		enddo
736
737		pot_Temp = 0.0_DP
738	<
739	<	call scatter(pot_Col, pot_Temp, plan_col)
738	>
739	>	call scatter(pot_Col, pot_Temp, plan_atom_col)
740		do i = 1, nlocal
741		pot_local = pot_local + pot_Temp(i)
742		enddo
743	<
744	<	endif
743	>
744	>	endif
745		#endif
746	<
747	<	if (FF_RequiresPostpairCalc() .and. SimRequiresPostpairCalc()) then
746	>
747	>	if (FF_RequiresPostpairCalc() .and. SIM_requires_postpair_calc) then
748
749	<	if (FF_uses_RF .and. SimUsesRF()) then
749	>	if (FF_uses_RF .and. SIM_uses_RF) then
750
751		#ifdef IS_MPI
752	<	call scatter(rf_Row,rf,plan_row3d)
753	<	call scatter(rf_Col,rf_Temp,plan_col3d)
752	>	call scatter(rf_Row,rf,plan_atom_row_3d)
753	>	call scatter(rf_Col,rf_Temp,plan_atom_col_3d)
754		do i = 1,nlocal
755		rf(1:3,i) = rf(1:3,i) + rf_Temp(1:3,i)
756		end do
757		#endif
758
759	<	do i = 1, getNlocal()
760	<
759	>	do i = 1, nLocal
760	>
761		rfpot = 0.0_DP
762		#ifdef IS_MPI
763		me_i = atid_row(i)
764		#else
765		me_i = atid(i)
766		#endif
767	<	call getElementProperty(atypes, me_i, "is_DP", is_DP_i)
768	<	if ( is_DP_i ) then
769	<	call getElementProperty(atypes, me_i, "dipole_moment", mu_i)
767	>
768	>	if (PropertyMap(me_i)%is_DP) then
769	>
770	>	mu_i = PropertyMap(me_i)%dipole_moment
771	>
772		!! The reaction field needs to include a self contribution
773		!! to the field:
774	<	call accumulate_self_rf(i, mu_i, u_l)
774	>	call accumulate_self_rf(i, mu_i, u_l)
775		!! Get the reaction field contribution to the
776		!! potential and torques:
777		call reaction_field_final(i, mu_i, u_l, rfpot, t, do_pot)
#	Line 468 | Line 785 | contains
785		enddo
786		endif
787		endif
788	<
789	<
788	>
789	>
790		#ifdef IS_MPI
791	<
791	>
792		if (do_pot) then
793		pot = pot + pot_local
794		!! we assume the c code will do the allreduce to get the total potential
795		!! we could do it right here if we needed to...
796		endif
797	<
797	>
798		if (do_stress) then
799	<	call mpi_allreduce(tau, tau_Temp,9,mpi_double_precision,mpi_sum, &
799	>	call mpi_allreduce(tau_Temp, tau, 9,mpi_double_precision,mpi_sum, &
800		mpi_comm_world,mpi_err)
801	<	call mpi_allreduce(virial, virial_Temp,1,mpi_double_precision,mpi_sum, &
801	>	call mpi_allreduce(virial_Temp, virial,1,mpi_double_precision,mpi_sum, &
802		mpi_comm_world,mpi_err)
803		endif
804	<
804	>
805		#else
806	<
806	>
807		if (do_stress) then
808		tau = tau_Temp
809		virial = virial_Temp
810		endif
494	–
495	–	#endif
811
812	+	#endif
813	+
814		end subroutine do_force_loop
815	+
816	+	subroutine do_pair(i, j, rijsq, d, sw, do_pot, &
817	+	u_l, A, f, t, pot, vpair, fpair)
818
819	<	subroutine do_pair(i, j, rijsq, d, do_pot, do_stress, u_l, A, f, t, pot)
819	>	real( kind = dp ) :: pot, vpair, sw
820	>	real( kind = dp ), dimension(3) :: fpair
821	>	real( kind = dp ), dimension(nLocal)   :: mfact
822	>	real( kind = dp ), dimension(3,nLocal) :: u_l
823	>	real( kind = dp ), dimension(9,nLocal) :: A
824	>	real( kind = dp ), dimension(3,nLocal) :: f
825	>	real( kind = dp ), dimension(3,nLocal) :: t
826
827	<	real( kind = dp ) :: pot
502	<	real( kind = dp ), dimension(:,:) :: u_l
503	<	real (kind=dp), dimension(:,:) :: A
504	<	real (kind=dp), dimension(:,:) :: f
505	<	real (kind=dp), dimension(:,:) :: t
506	<
507	<	logical, intent(inout) :: do_pot, do_stress
827	>	logical, intent(inout) :: do_pot
828		integer, intent(in) :: i, j
829	<	real ( kind = dp ), intent(inout)    :: rijsq
829	>	real ( kind = dp ), intent(inout) :: rijsq
830		real ( kind = dp )                :: r
831		real ( kind = dp ), intent(inout) :: d(3)
512	–	logical :: is_LJ_i, is_LJ_j
513	–	logical :: is_DP_i, is_DP_j
514	–	logical :: is_GB_i, is_GB_j
515	–	logical :: is_Sticky_i, is_Sticky_j
832		integer :: me_i, me_j
833
834		r = sqrt(rijsq)
835	+	vpair = 0.0d0
836	+	fpair(1:3) = 0.0d0
837
838		#ifdef IS_MPI
839	<
839	>	if (AtomRowToGlobal(i) .eq. AtomColToGlobal(j)) then
840	>	write(0,*) 'do_pair is doing', i , j, AtomRowToGlobal(i), AtomColToGlobal(j)
841	>	endif
842		me_i = atid_row(i)
843		me_j = atid_col(j)
524	–
844		#else
526	–
845		me_i = atid(i)
846		me_j = atid(j)
529	–
847		#endif
848	+
849	+	if (FF_uses_LJ .and. SIM_uses_LJ) then
850	+
851	+	if ( PropertyMap(me_i)%is_LJ .and. PropertyMap(me_j)%is_LJ ) then
852	+	!write(*,*) 'calling lj with'
853	+	!write(*,*) i, j, r, rijsq
854	+	!write(*,'(3es12.3)') d(1), d(2), d(3)
855	+	!write(*,'(3es12.3)') sw, vpair, pot
856	+	!write(*,*)
857
858	<	if (FF_uses_LJ .and. SimUsesLJ()) then
859	<	call getElementProperty(atypes, me_i, "is_LJ", is_LJ_i)
860	<	call getElementProperty(atypes, me_j, "is_LJ", is_LJ_j)
535	<
536	<	if ( is_LJ_i .and. is_LJ_j ) &
537	<	call do_lj_pair(i, j, d, r, rijsq, pot, f, do_pot, do_stress)
858	>	call do_lj_pair(i, j, d, r, rijsq, sw, vpair, fpair, pot, f, do_pot)
859	>	endif
860	>
861		endif
862	<
863	<	if (FF_uses_dipoles .and. SimUsesDipoles()) then
541	<	call getElementProperty(atypes, me_i, "is_DP", is_DP_i)
542	<	call getElementProperty(atypes, me_j, "is_DP", is_DP_j)
862	>
863	>	if (FF_uses_charges .and. SIM_uses_charges) then
864
865	<	if ( is_DP_i .and. is_DP_j ) then
866	<
546	<	call do_dipole_pair(i, j, d, r, rijsq, pot, u_l, f, t, &
547	<	do_pot, do_stress)
548	<	if (FF_uses_RF .and. SimUsesRF()) then
549	<	call accumulate_rf(i, j, r, u_l)
550	<	call rf_correct_forces(i, j, d, r, u_l, f, do_stress)
551	<	endif
552	<
865	>	if (PropertyMap(me_i)%is_Charge .and. PropertyMap(me_j)%is_Charge) then
866	>	call do_charge_pair(i, j, d, r, rijsq, sw, vpair, fpair, pot, f, do_pot)
867		endif
868	+
869		endif
870	+
871	+	if (FF_uses_dipoles .and. SIM_uses_dipoles) then
872	+
873	+	if ( PropertyMap(me_i)%is_DP .and. PropertyMap(me_j)%is_DP) then
874	+	call do_dipole_pair(i, j, d, r, rijsq, sw, vpair, fpair, pot, u_l, f, t, &
875	+	do_pot)
876	+	if (FF_uses_RF .and. SIM_uses_RF) then
877	+	call accumulate_rf(i, j, r, u_l, sw)
878	+	call rf_correct_forces(i, j, d, r, u_l, sw, f, fpair)
879	+	endif
880	+	endif
881
882	<	if (FF_uses_Sticky .and. SimUsesSticky()) then
882	>	endif
883
884	<	call getElementProperty(atypes, me_i, "is_Sticky", is_Sticky_i)
559	<	call getElementProperty(atypes, me_j, "is_Sticky", is_Sticky_j)
884	>	if (FF_uses_Sticky .and. SIM_uses_sticky) then
885
886	<	if ( is_Sticky_i .and. is_Sticky_j ) then
887	<	call do_sticky_pair(i, j, d, r, rijsq, A, pot, f, t, &
888	<	do_pot, do_stress)
886	>	if ( PropertyMap(me_i)%is_Sticky .and. PropertyMap(me_j)%is_Sticky) then
887	>	call do_sticky_pair(i, j, d, r, rijsq, sw, vpair, fpair, pot, A, f, t, &
888	>	do_pot)
889		endif
890	+
891		endif
892
893
894	<	if (FF_uses_GB .and. SimUsesGB()) then
894	>	if (FF_uses_GB .and. SIM_uses_GB) then
895	>
896	>	if ( PropertyMap(me_i)%is_GB .and. PropertyMap(me_j)%is_GB) then
897	>	call do_gb_pair(i, j, d, r, rijsq, sw, vpair, fpair, pot, u_l, f, t, &
898	>	do_pot)
899	>	endif
900
901	<	call getElementProperty(atypes, me_i, "is_GB", is_GB_i)
571	<	call getElementProperty(atypes, me_j, "is_GB", is_GB_j)
901	>	endif
902
903	<	if ( is_GB_i .and. is_GB_j ) then
904	<	call do_gb_pair(i, j, d, r, rijsq, u_l, pot, f, t, &
905	<	do_pot, do_stress)
903	>	if (FF_uses_EAM .and. SIM_uses_EAM) then
904	>
905	>	if ( PropertyMap(me_i)%is_EAM .and. PropertyMap(me_j)%is_EAM) then
906	>	call do_eam_pair(i, j, d, r, rijsq, sw, vpair, fpair, pot, f, &
907	>	do_pot)
908		endif
909	+
910		endif
911	<
911	>
912		end subroutine do_pair
913
914	+	subroutine do_prepair(i, j, rijsq, d, sw, rcijsq, dc, &
915	+	do_pot, do_stress, u_l, A, f, t, pot)
916
917	<	subroutine get_interatomic_vector(q_i, q_j, d, r_sq)
918	<
919	<	real (kind = dp), dimension(3) :: q_i
920	<	real (kind = dp), dimension(3) :: q_j
921	<	real ( kind = dp ), intent(out) :: r_sq
922	<	real( kind = dp ) :: d(3)
923	<	real( kind = dp ) :: d_old(3)
924	<	d(1:3) = q_i(1:3) - q_j(1:3)
925	<	d_old = d
926	<	! Wrap back into periodic box if necessary
927	<	if ( SimUsesPBC() ) then
928	<
929	<	d(1:3) = d(1:3) - box(1:3) * sign(1.0_dp,d(1:3)) * &
930	<	int(abs(d(1:3)/box(1:3)) + 0.5_dp)
931	<
932	<	endif
598	<	r_sq = dot_product(d,d)
599	<
600	<	end subroutine get_interatomic_vector
917	>	real( kind = dp ) :: pot, sw
918	>	real( kind = dp ), dimension(3,nLocal) :: u_l
919	>	real (kind=dp), dimension(9,nLocal) :: A
920	>	real (kind=dp), dimension(3,nLocal) :: f
921	>	real (kind=dp), dimension(3,nLocal) :: t
922	>
923	>	logical, intent(inout) :: do_pot, do_stress
924	>	integer, intent(in) :: i, j
925	>	real ( kind = dp ), intent(inout)    :: rijsq, rcijsq
926	>	real ( kind = dp )                :: r, rc
927	>	real ( kind = dp ), intent(inout) :: d(3), dc(3)
928	>
929	>	logical :: is_EAM_i, is_EAM_j
930	>
931	>	integer :: me_i, me_j
932	>
933
934	<	subroutine check_initialization(error)
935	<	integer, intent(out) :: error
936	<
937	<	error = 0
938	<	! Make sure we are properly initialized.
607	<	if (.not. do_forces_initialized) then
608	<	error = -1
609	<	return
934	>	r = sqrt(rijsq)
935	>	if (SIM_uses_molecular_cutoffs) then
936	>	rc = sqrt(rcijsq)
937	>	else
938	>	rc = r
939		endif
940	+
941
942		#ifdef IS_MPI
943	<	if (.not. isMPISimSet()) then
944	<	write(default_error,*) "ERROR: mpiSimulation has not been initialized!"
945	<	error = -1
946	<	return
947	<	endif
948	<	#endif
949	<
950	<	return
951	<	end subroutine check_initialization
952	<
953	<
954	<	subroutine zero_work_arrays()
625	<
626	<	#ifdef IS_MPI
627	<
628	<	q_Row = 0.0_dp
629	<	q_Col = 0.0_dp
630	<
631	<	u_l_Row = 0.0_dp
632	<	u_l_Col = 0.0_dp
633	<
634	<	A_Row = 0.0_dp
635	<	A_Col = 0.0_dp
636	<
637	<	f_Row = 0.0_dp
638	<	f_Col = 0.0_dp
639	<	f_Temp = 0.0_dp
640	<
641	<	t_Row = 0.0_dp
642	<	t_Col = 0.0_dp
643	<	t_Temp = 0.0_dp
644	<
645	<	pot_Row = 0.0_dp
646	<	pot_Col = 0.0_dp
647	<	pot_Temp = 0.0_dp
648	<
649	<	rf_Row = 0.0_dp
650	<	rf_Col = 0.0_dp
651	<	rf_Temp = 0.0_dp
652	<
653	<	#endif
654	<
655	<	rf = 0.0_dp
656	<	tau_Temp = 0.0_dp
657	<	virial_Temp = 0.0_dp
658	<
659	<	end subroutine zero_work_arrays
660	<
661	<	function skipThisPair(atom1, atom2) result(skip_it)
662	<	integer, intent(in) :: atom1
663	<	integer, intent(in), optional :: atom2
664	<	logical :: skip_it
665	<	integer :: unique_id_1, unique_id_2
666	<	integer :: me_i,me_j
667	<	integer :: i
668	<
669	<	skip_it = .false.
670	<
671	<	!! there are a number of reasons to skip a pair or a particle
672	<	!! mostly we do this to exclude atoms who are involved in short
673	<	!! range interactions (bonds, bends, torsions), but we also need
674	<	!! to exclude some overcounted interactions that result from
675	<	!! the parallel decomposition
676	<
677	<	#ifdef IS_MPI
678	<	!! in MPI, we have to look up the unique IDs for each atom
679	<	unique_id_1 = tagRow(atom1)
680	<	#else
681	<	!! in the normal loop, the atom numbers are unique
682	<	unique_id_1 = atom1
683	<	#endif
684	<
685	<	!! We were called with only one atom, so just check the global exclude
686	<	!! list for this atom
687	<	if (.not. present(atom2)) then
688	<	do i = 1, nExcludes_global
689	<	if (excludesGlobal(i) == unique_id_1) then
690	<	skip_it = .true.
691	<	return
692	<	end if
693	<	end do
694	<	return
695	<	end if
696	<
697	<	#ifdef IS_MPI
698	<	unique_id_2 = tagColumn(atom2)
699	<	#else
700	<	unique_id_2 = atom2
943	>	if (AtomRowToGlobal(i) .eq. AtomColToGlobal(j)) then
944	>	write(0,*) 'do_prepair is doing', i , j, AtomRowToGlobal(i), AtomColToGlobal(j)
945	>	endif
946	>
947	>	me_i = atid_row(i)
948	>	me_j = atid_col(j)
949	>
950	>	#else
951	>
952	>	me_i = atid(i)
953	>	me_j = atid(j)
954	>
955		#endif
956	+
957	+	if (FF_uses_EAM .and. SIM_uses_EAM) then
958	+
959	+	if (PropertyMap(me_i)%is_EAM .and. PropertyMap(me_j)%is_EAM) &
960	+	call calc_EAM_prepair_rho(i, j, d, r, rijsq )
961	+
962	+	endif
963	+
964	+	end subroutine do_prepair
965	+
966	+
967	+	subroutine do_preforce(nlocal,pot)
968	+	integer :: nlocal
969	+	real( kind = dp ) :: pot
970	+
971	+	if (FF_uses_EAM .and. SIM_uses_EAM) then
972	+	call calc_EAM_preforce_Frho(nlocal,pot)
973	+	endif
974	+
975	+
976	+	end subroutine do_preforce
977	+
978	+
979	+	subroutine get_interatomic_vector(q_i, q_j, d, r_sq)
980	+
981	+	real (kind = dp), dimension(3) :: q_i
982	+	real (kind = dp), dimension(3) :: q_j
983	+	real ( kind = dp ), intent(out) :: r_sq
984	+	real( kind = dp ) :: d(3), scaled(3)
985	+	integer i
986	+
987	+	d(1:3) = q_j(1:3) - q_i(1:3)
988	+
989	+	! Wrap back into periodic box if necessary
990	+	if ( SIM_uses_PBC ) then
991	+
992	+	if( .not.boxIsOrthorhombic ) then
993	+	! calc the scaled coordinates.
994	+
995	+	scaled = matmul(HmatInv, d)
996	+
997	+	! wrap the scaled coordinates
998	+
999	+	scaled = scaled  - anint(scaled)
1000	+
1001	+
1002	+	! calc the wrapped real coordinates from the wrapped scaled
1003	+	! coordinates
1004	+
1005	+	d = matmul(Hmat,scaled)
1006	+
1007	+	else
1008	+	! calc the scaled coordinates.
1009	+
1010	+	do i = 1, 3
1011	+	scaled(i) = d(i) * HmatInv(i,i)
1012	+
1013	+	! wrap the scaled coordinates
1014	+
1015	+	scaled(i) = scaled(i) - anint(scaled(i))
1016	+
1017	+	! calc the wrapped real coordinates from the wrapped scaled
1018	+	! coordinates
1019	+
1020	+	d(i) = scaled(i)*Hmat(i,i)
1021	+	enddo
1022	+	endif
1023	+
1024	+	endif
1025	+
1026	+	r_sq = dot_product(d,d)
1027	+
1028	+	end subroutine get_interatomic_vector
1029	+
1030	+	subroutine zero_work_arrays()
1031	+
1032	+	#ifdef IS_MPI
1033	+
1034	+	q_Row = 0.0_dp
1035	+	q_Col = 0.0_dp
1036
1037	+	q_group_Row = 0.0_dp
1038	+	q_group_Col = 0.0_dp
1039	+
1040	+	u_l_Row = 0.0_dp
1041	+	u_l_Col = 0.0_dp
1042	+
1043	+	A_Row = 0.0_dp
1044	+	A_Col = 0.0_dp
1045	+
1046	+	f_Row = 0.0_dp
1047	+	f_Col = 0.0_dp
1048	+	f_Temp = 0.0_dp
1049	+
1050	+	t_Row = 0.0_dp
1051	+	t_Col = 0.0_dp
1052	+	t_Temp = 0.0_dp
1053	+
1054	+	pot_Row = 0.0_dp
1055	+	pot_Col = 0.0_dp
1056	+	pot_Temp = 0.0_dp
1057	+
1058	+	rf_Row = 0.0_dp
1059	+	rf_Col = 0.0_dp
1060	+	rf_Temp = 0.0_dp
1061	+
1062	+	#endif
1063	+
1064	+	if (FF_uses_EAM .and. SIM_uses_EAM) then
1065	+	call clean_EAM()
1066	+	endif
1067	+
1068	+	rf = 0.0_dp
1069	+	tau_Temp = 0.0_dp
1070	+	virial_Temp = 0.0_dp
1071	+	end subroutine zero_work_arrays
1072	+
1073	+	function skipThisPair(atom1, atom2) result(skip_it)
1074	+	integer, intent(in) :: atom1
1075	+	integer, intent(in), optional :: atom2
1076	+	logical :: skip_it
1077	+	integer :: unique_id_1, unique_id_2
1078	+	integer :: me_i,me_j
1079	+	integer :: i
1080	+
1081	+	skip_it = .false.
1082	+
1083	+	!! there are a number of reasons to skip a pair or a particle
1084	+	!! mostly we do this to exclude atoms who are involved in short
1085	+	!! range interactions (bonds, bends, torsions), but we also need
1086	+	!! to exclude some overcounted interactions that result from
1087	+	!! the parallel decomposition
1088	+
1089		#ifdef IS_MPI
1090	<	!! this situation should only arise in MPI simulations
1091	<	if (unique_id_1 == unique_id_2) then
1092	<	skip_it = .true.
1093	<	return
1094	<	end if
709	<
710	<	!! this prevents us from doing the pair on multiple processors
711	<	if (unique_id_1 < unique_id_2) then
712	<	if (mod(unique_id_1 + unique_id_2,2) == 0) then
713	<	skip_it = .true.
714	<	return
715	<	endif
716	<	else
717	<	if (mod(unique_id_1 + unique_id_2,2) == 1) then
718	<	skip_it = .true.
719	<	return
720	<	endif
721	<	endif
1090	>	!! in MPI, we have to look up the unique IDs for each atom
1091	>	unique_id_1 = AtomRowToGlobal(atom1)
1092	>	#else
1093	>	!! in the normal loop, the atom numbers are unique
1094	>	unique_id_1 = atom1
1095		#endif
1096	+
1097	+	!! We were called with only one atom, so just check the global exclude
1098	+	!! list for this atom
1099	+	if (.not. present(atom2)) then
1100	+	do i = 1, nExcludes_global
1101	+	if (excludesGlobal(i) == unique_id_1) then
1102	+	skip_it = .true.
1103	+	return
1104	+	end if
1105	+	end do
1106	+	return
1107	+	end if
1108	+
1109	+	#ifdef IS_MPI
1110	+	unique_id_2 = AtomColToGlobal(atom2)
1111	+	#else
1112	+	unique_id_2 = atom2
1113	+	#endif
1114	+
1115	+	#ifdef IS_MPI
1116	+	!! this situation should only arise in MPI simulations
1117	+	if (unique_id_1 == unique_id_2) then
1118	+	skip_it = .true.
1119	+	return
1120	+	end if
1121	+
1122	+	!! this prevents us from doing the pair on multiple processors
1123	+	if (unique_id_1 < unique_id_2) then
1124	+	if (mod(unique_id_1 + unique_id_2,2) == 0) then
1125	+	skip_it = .true.
1126	+	return
1127	+	endif
1128	+	else
1129	+	if (mod(unique_id_1 + unique_id_2,2) == 1) then
1130	+	skip_it = .true.
1131	+	return
1132	+	endif
1133	+	endif
1134	+	#endif
1135	+
1136	+	!! the rest of these situations can happen in all simulations:
1137	+	do i = 1, nExcludes_global
1138	+	if ((excludesGlobal(i) == unique_id_1) .or. &
1139	+	(excludesGlobal(i) == unique_id_2)) then
1140	+	skip_it = .true.
1141	+	return
1142	+	endif
1143	+	enddo
1144	+
1145	+	do i = 1, nSkipsForAtom(atom1)
1146	+	if (skipsForAtom(atom1, i) .eq. unique_id_2) then
1147	+	skip_it = .true.
1148	+	return
1149	+	endif
1150	+	end do
1151	+
1152	+	return
1153	+	end function skipThisPair
1154
1155	<	!! the rest of these situations can happen in all simulations:
1156	<	do i = 1, nExcludes_global
1157	<	if ((excludesGlobal(i) == unique_id_1) .or. &
1158	<	(excludesGlobal(i) == unique_id_2)) then
1159	<	skip_it = .true.
1160	<	return
1161	<	endif
1162	<	enddo
1155	>	function FF_UsesDirectionalAtoms() result(doesit)
1156	>	logical :: doesit
1157	>	doesit = FF_uses_dipoles .or. FF_uses_sticky .or. &
1158	>	FF_uses_GB .or. FF_uses_RF
1159	>	end function FF_UsesDirectionalAtoms
1160	>
1161	>	function FF_RequiresPrepairCalc() result(doesit)
1162	>	logical :: doesit
1163	>	doesit = FF_uses_EAM
1164	>	end function FF_RequiresPrepairCalc
1165	>
1166	>	function FF_RequiresPostpairCalc() result(doesit)
1167	>	logical :: doesit
1168	>	doesit = FF_uses_RF
1169	>	end function FF_RequiresPostpairCalc
1170	>
1171	>	#ifdef PROFILE
1172	>	function getforcetime() result(totalforcetime)
1173	>	real(kind=dp) :: totalforcetime
1174	>	totalforcetime = forcetime
1175	>	end function getforcetime
1176	>	#endif
1177	>
1178	>	!! This cleans componets of force arrays belonging only to fortran
1179
1180	<	do i = 1, nExcludes_local
1181	<	if (excludesLocal(1,i) == unique_id_1) then
1182	<	if (excludesLocal(2,i) == unique_id_2) then
1183	<	skip_it = .true.
1184	<	return
1185	<	endif
1186	<	else
1187	<	if (excludesLocal(1,i) == unique_id_2) then
1188	<	if (excludesLocal(2,i) == unique_id_1) then
1189	<	skip_it = .true.
1190	<	return
1191	<	endif
1192	<	endif
1193	<	endif
1194	<	end do
1195	<
1196	<	return
1197	<	end function skipThisPair
1198	<
1199	<	function FF_UsesDirectionalAtoms() result(doesit)
1200	<	logical :: doesit
1201	<	doesit = FF_uses_dipoles .or. FF_uses_sticky .or. &
1202	<	FF_uses_GB .or. FF_uses_RF
1203	<	end function FF_UsesDirectionalAtoms
757	<
758	<	function FF_RequiresPrepairCalc() result(doesit)
759	<	logical :: doesit
760	<	doesit = FF_uses_EAM
761	<	end function FF_RequiresPrepairCalc
762	<
763	<	function FF_RequiresPostpairCalc() result(doesit)
764	<	logical :: doesit
765	<	doesit = FF_uses_RF
766	<	end function FF_RequiresPostpairCalc
767	<
1180	>	subroutine add_stress_tensor(dpair, fpair)
1181	>
1182	>	real( kind = dp ), dimension(3), intent(in) :: dpair, fpair
1183	>
1184	>	! because the d vector is the rj - ri vector, and
1185	>	! because fx, fy, fz are the force on atom i, we need a
1186	>	! negative sign here:
1187	>
1188	>	tau_Temp(1) = tau_Temp(1) - dpair(1) * fpair(1)
1189	>	tau_Temp(2) = tau_Temp(2) - dpair(1) * fpair(2)
1190	>	tau_Temp(3) = tau_Temp(3) - dpair(1) * fpair(3)
1191	>	tau_Temp(4) = tau_Temp(4) - dpair(2) * fpair(1)
1192	>	tau_Temp(5) = tau_Temp(5) - dpair(2) * fpair(2)
1193	>	tau_Temp(6) = tau_Temp(6) - dpair(2) * fpair(3)
1194	>	tau_Temp(7) = tau_Temp(7) - dpair(3) * fpair(1)
1195	>	tau_Temp(8) = tau_Temp(8) - dpair(3) * fpair(2)
1196	>	tau_Temp(9) = tau_Temp(9) - dpair(3) * fpair(3)
1197	>
1198	>	!write(*,'(6es12.3)')  fpair(1:3), tau_Temp(1), tau_Temp(5), tau_temp(9)
1199	>	virial_Temp = virial_Temp + &
1200	>	(tau_Temp(1) + tau_Temp(5) + tau_Temp(9))
1201	>
1202	>	end subroutine add_stress_tensor
1203	>
1204		end module do_Forces
1205	+

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines