mirror of
https://github.com/LCPQ/quantum_package
synced 2024-12-25 05:43:47 +01:00
reduced bandwidth and checkpoint updates
This commit is contained in:
Yann Garniron
parent
a865a842d2
commit
4394c04728
@ -43,7 +43,6 @@ subroutine run_wf
|
||||
! Selection
|
||||
! ---------
|
||||
|
||||
print *, 'dress'
|
||||
call zmq_get_psi(zmq_to_qp_run_socket,1,energy,N_states)
|
||||
|
||||
PROVIDE psi_bilinear_matrix_columns_loc psi_det_alpha_unique psi_det_beta_unique
|
||||
|
||||
@ -134,6 +134,7 @@ subroutine dress_collector(zmq_socket_pull, E, relative_error, delta, delta_s2,
|
||||
implicit none
|
||||
|
||||
|
||||
integer, parameter :: delta_loc_N = 2
|
||||
|
||||
integer(ZMQ_PTR), intent(in) :: zmq_socket_pull
|
||||
integer, intent(in) :: istate
|
||||
@ -144,7 +145,7 @@ subroutine dress_collector(zmq_socket_pull, E, relative_error, delta, delta_s2,
|
||||
|
||||
double precision, intent(out) :: delta(N_states, N_det)
|
||||
double precision, intent(out) :: delta_s2(N_states, N_det)
|
||||
double precision, allocatable :: delta_loc(:,:,:), delta_det(:,:,:,:)
|
||||
double precision, allocatable :: delta_loc(:,:,:,:), delta_det(:,:,:,:)
|
||||
double precision, allocatable :: dress_detail(:,:)
|
||||
double precision :: dress_mwen(N_states)
|
||||
integer(ZMQ_PTR),external :: new_zmq_to_qp_run_socket
|
||||
@ -154,7 +155,7 @@ subroutine dress_collector(zmq_socket_pull, E, relative_error, delta, delta_s2,
|
||||
|
||||
integer :: more
|
||||
integer :: i, j, k, i_state, N
|
||||
integer :: task_id, ind
|
||||
integer :: task_id, ind, inds(delta_loc_N)
|
||||
double precision, save :: time0 = -1.d0
|
||||
double precision :: time, timeLast, old_tooth
|
||||
double precision, external :: omp_get_wtime
|
||||
@ -162,12 +163,17 @@ subroutine dress_collector(zmq_socket_pull, E, relative_error, delta, delta_s2,
|
||||
integer, allocatable :: parts_to_get(:)
|
||||
logical, allocatable :: actually_computed(:)
|
||||
integer :: total_computed
|
||||
integer :: delta_loc_cur
|
||||
double precision :: fac(delta_loc_N) , wei(delta_loc_N)
|
||||
logical :: ok
|
||||
|
||||
delta_loc_cur = 1
|
||||
|
||||
delta = 0d0
|
||||
delta_s2 = 0d0
|
||||
allocate(delta_det(N_states, N_det, 0:comb_teeth+1, 2))
|
||||
allocate(cp(N_states, N_det, N_cp, 2), dress_detail(N_states, N_det))
|
||||
allocate(delta_loc(N_states, N_det, 2))
|
||||
allocate(delta_loc(N_states, N_det, 2, delta_loc_N))
|
||||
dress_detail = 0d0
|
||||
delta_det = 0d0
|
||||
cp = 0d0
|
||||
@ -196,58 +202,102 @@ subroutine dress_collector(zmq_socket_pull, E, relative_error, delta, delta_s2,
|
||||
cur_cp = 0
|
||||
old_cur_cp = 0
|
||||
logical :: loop
|
||||
integer :: felem, felem_loc
|
||||
loop = .true.
|
||||
|
||||
felem = N_det+1
|
||||
pullLoop : do while (loop)
|
||||
call pull_dress_results(zmq_socket_pull, ind, delta_loc, task_id)
|
||||
call pull_dress_results(zmq_socket_pull, ind, delta_loc(1,1,1,delta_loc_cur), task_id, felem_loc)
|
||||
felem = min(felem_loc, felem)
|
||||
dress_mwen(:) = 0d0
|
||||
|
||||
!!!!! A VERIFIER !!!!!
|
||||
do i_state=1,N_states
|
||||
do i=1, N_det
|
||||
dress_mwen(i_state) += delta_loc(i_state, i, 1) * psi_coef(i, i_state)
|
||||
end do
|
||||
end do
|
||||
|
||||
dress_detail(:, ind) += dress_mwen(:)
|
||||
do j=1,N_cp !! optimizable
|
||||
if(cps(ind, j) > 0d0) then
|
||||
if(tooth_of_det(ind) < cp_first_tooth(j)) stop "coef on supposedely deterministic det"
|
||||
double precision :: fac
|
||||
integer :: toothMwen
|
||||
logical :: fracted
|
||||
fac = cps(ind, j) / cps_N(j) * dress_weight_inv(ind) * comb_step
|
||||
cp(1:N_states,1:N_det,j,1) += delta_loc(1:N_states,1:N_det,1) * fac
|
||||
cp(1:N_states,1:N_det,j,2) += delta_loc(1:N_states,1:N_det,2) * fac
|
||||
end if
|
||||
end do
|
||||
toothMwen = tooth_of_det(ind)
|
||||
fracted = (toothMwen /= 0)
|
||||
if(fracted) fracted = (ind == first_det_of_teeth(toothMwen))
|
||||
|
||||
if(fracted) then
|
||||
delta_det(1:N_states,1:N_det,toothMwen-1, 1) = delta_det(1:N_states,1:N_det,toothMwen-1, 1) + delta_loc(1:N_states,1:N_det,1) * (1d0-fractage(toothMwen))
|
||||
delta_det(1:N_states,1:N_det,toothMwen-1, 2) = delta_det(1:N_states,1:N_det,toothMwen-1, 2) + delta_loc(1:N_states,1:N_det,2) * (1d0-fractage(toothMwen))
|
||||
delta_det(1:N_states,1:N_det,toothMwen , 1) = delta_det(1:N_states,1:N_det,toothMwen , 1) + delta_loc(1:N_states,1:N_det,1) * (fractage(toothMwen))
|
||||
delta_det(1:N_states,1:N_det,toothMwen , 2) = delta_det(1:N_states,1:N_det,toothMwen , 2) + delta_loc(1:N_states,1:N_det,2) * (fractage(toothMwen))
|
||||
else
|
||||
delta_det(1:N_states,1:N_det,toothMwen , 1) = delta_det(1:N_states,1:N_det,toothMwen , 1) + delta_loc(1:N_states,1:N_det,1)
|
||||
delta_det(1:N_states,1:N_det,toothMwen , 2) = delta_det(1:N_states,1:N_det,toothMwen , 2) + delta_loc(1:N_states,1:N_det,2)
|
||||
end if
|
||||
|
||||
parts_to_get(ind) -= 1
|
||||
if(parts_to_get(ind) == 0) then
|
||||
actually_computed(ind) = .true.
|
||||
total_computed += 1
|
||||
end if
|
||||
|
||||
|
||||
integer, external :: zmq_delete_tasks
|
||||
|
||||
if (zmq_delete_tasks(zmq_to_qp_run_socket,zmq_socket_pull,task_id,1,more) == -1) then
|
||||
stop 'Unable to delete tasks'
|
||||
endif
|
||||
if(more == 0) loop = .false.
|
||||
|
||||
|
||||
do i_state=1,N_states
|
||||
do i=1, N_det
|
||||
dress_mwen(i_state) += delta_loc(i_state, i, 1, delta_loc_cur) * psi_coef(i, i_state)
|
||||
end do
|
||||
end do
|
||||
|
||||
dress_detail(:, ind) += dress_mwen(:)
|
||||
wei(delta_loc_cur) = dress_weight_inv(ind)
|
||||
inds(delta_loc_cur) = ind
|
||||
|
||||
if(delta_loc_cur == delta_loc_N .or. .not. loop) then
|
||||
do j=1,N_cp !! optimizable
|
||||
fac = 0d0
|
||||
ok = .false.
|
||||
|
||||
do i=1,delta_loc_cur
|
||||
!fac(i) = cps(inds(i), j) / cps_N(j) * wei(i) * comb_step
|
||||
fac(i) = cps(inds(i), j) * wei(i) * comb_step
|
||||
if(fac(i) /= 0d0) ok = .true.
|
||||
end do
|
||||
|
||||
if(ok) then
|
||||
do i=felem,N_det
|
||||
cp(:,i,j,1) += delta_loc(:,i,1,1) * fac(1) &
|
||||
+ delta_loc(:,i,1,2) * fac(2)
|
||||
!+ delta_loc(:,i,1,3) * fac(3) &
|
||||
!+ delta_loc(:,i,1,4) * fac(4) &
|
||||
!+ delta_loc(:,i,1,5) * fac(5) &
|
||||
!+ delta_loc(:,i,1,6) * fac(6) &
|
||||
!+ delta_loc(:,i,1,7) * fac(7) &
|
||||
!+ delta_loc(:,i,1,8) * fac(8)
|
||||
|
||||
cp(:,i,j,1) += delta_loc(:,i,2,1) * fac(1) &
|
||||
+ delta_loc(:,i,2,2) * fac(2)
|
||||
!+ delta_loc(:,i,2,3) * fac(3) &
|
||||
!+ delta_loc(:,i,2,4) * fac(4) &
|
||||
!+ delta_loc(:,i,2,5) * fac(5) &
|
||||
!+ delta_loc(:,i,2,6) * fac(6) &
|
||||
!+ delta_loc(:,i,2,7) * fac(7) &
|
||||
!+ delta_loc(:,i,2,8) * fac(8)
|
||||
end do
|
||||
!cp(1:N_states,indi:N_det,j,1) += delta_loc(1:N_states,indi:N_det,1) * fac
|
||||
!cp(1:N_states,indi:N_det,j,2) += delta_loc(1:N_states,indi:N_det,2) * fac
|
||||
end if
|
||||
end do
|
||||
|
||||
do i=1,delta_loc_cur
|
||||
logical :: fracted
|
||||
integer :: toothMwen
|
||||
ind = inds(i)
|
||||
|
||||
toothMwen = tooth_of_det(ind)
|
||||
fracted = (toothMwen /= 0)
|
||||
if(fracted) fracted = (ind == first_det_of_teeth(toothMwen))
|
||||
|
||||
if(fracted) then
|
||||
delta_det(1:N_states,felem:N_det,toothMwen-1, 1) = delta_det(1:N_states,felem:N_det,toothMwen-1, 1) + delta_loc(felem:N_states,1:N_det,1,i) * (1d0-fractage(toothMwen))
|
||||
delta_det(1:N_states,felem:N_det,toothMwen-1, 2) = delta_det(1:N_states,felem:N_det,toothMwen-1, 2) + delta_loc(felem:N_states,1:N_det,2,i) * (1d0-fractage(toothMwen))
|
||||
delta_det(1:N_states,felem:N_det,toothMwen , 1) = delta_det(1:N_states,felem:N_det,toothMwen , 1) + delta_loc(felem:N_states,1:N_det,1,i) * (fractage(toothMwen))
|
||||
delta_det(1:N_states,felem:N_det,toothMwen , 2) = delta_det(1:N_states,felem:N_det,toothMwen , 2) + delta_loc(felem:N_states,1:N_det,2,i) * (fractage(toothMwen))
|
||||
else
|
||||
delta_det(1:N_states,felem:N_det,toothMwen , 1) = delta_det(1:N_states,felem:N_det,toothMwen , 1) + delta_loc(1:N_states,felem:N_det,1,i)
|
||||
delta_det(1:N_states,felem:N_det,toothMwen , 2) = delta_det(1:N_states,felem:N_det,toothMwen , 2) + delta_loc(1:N_states,felem:N_det,2,i)
|
||||
end if
|
||||
|
||||
parts_to_get(ind) -= 1
|
||||
if(parts_to_get(ind) == 0) then
|
||||
actually_computed(ind) = .true.
|
||||
total_computed += 1
|
||||
end if
|
||||
end do
|
||||
felem = N_det+1
|
||||
delta_loc_cur = 1
|
||||
else
|
||||
delta_loc_cur += 1
|
||||
cycle
|
||||
end if
|
||||
|
||||
|
||||
|
||||
time = omp_get_wtime()
|
||||
|
||||
if((time - timeLast > 2d0) .or. (.not. loop)) then
|
||||
@ -304,17 +354,21 @@ subroutine dress_collector(zmq_socket_pull, E, relative_error, delta, delta_s2,
|
||||
end if
|
||||
end do pullLoop
|
||||
|
||||
delta (1:N_states,1:N_det) = 0d0
|
||||
delta_s2(1:N_states,1:N_det) = 0d0
|
||||
|
||||
if(total_computed == N_det_generators) then
|
||||
delta (1:N_states,1:N_det) = 0d0
|
||||
delta_s2(1:N_states,1:N_det) = 0d0
|
||||
do i=comb_teeth+1,0,-1
|
||||
delta (1:N_states,1:N_det) = delta (1:N_states,1:N_det) + delta_det(1:N_states,1:N_det,i,1)
|
||||
delta_s2(1:N_states,1:N_det) = delta_s2(1:N_states,1:N_det) + delta_det(1:N_states,1:N_det,i,2)
|
||||
end do
|
||||
else
|
||||
|
||||
delta (1:N_states,1:N_det) = cp(1:N_states,1:N_det,cur_cp,1)
|
||||
delta_s2(1:N_states,1:N_det) = cp(1:N_states,1:N_det,cur_cp,2)
|
||||
do i=1,cur_cp
|
||||
delta (1:N_states,1:N_det) += cp(1:N_states,1:N_det,i,1)
|
||||
delta_s2(1:N_states,1:N_det) += cp(1:N_states,1:N_det,i,2)
|
||||
end do
|
||||
delta (1:N_states,1:N_det) = delta(1:N_states,1:N_det) / cps_N(cur_cp)
|
||||
delta_s2(1:N_states,1:N_det) = delta_s2(1:N_states,1:N_det) / cps_N(cur_cp)
|
||||
do i=cp_first_tooth(cur_cp)-1,0,-1
|
||||
delta (1:N_states,1:N_det) = delta (1:N_states,1:N_det) + delta_det(1:N_states,1:N_det,i,1)
|
||||
delta_s2(1:N_states,1:N_det) = delta_s2(1:N_states,1:N_det) + delta_det(1:N_states,1:N_det,i,2)
|
||||
@ -363,7 +417,7 @@ end function
|
||||
! gen_per_cp : number of generators per checkpoint
|
||||
END_DOC
|
||||
comb_teeth = 64
|
||||
N_cps_max = 64
|
||||
N_cps_max = 32
|
||||
gen_per_cp = (N_det_generators / N_cps_max) + 1
|
||||
END_PROVIDER
|
||||
|
||||
@ -455,13 +509,19 @@ END_PROVIDER
|
||||
end if
|
||||
end do
|
||||
end do
|
||||
cps(:, N_cp) = 0d0
|
||||
cp_first_tooth(N_cp) = comb_teeth+1
|
||||
|
||||
iorder = -1
|
||||
do i=1,N_cp-1
|
||||
call isort(dress_jobs(first_cp(i)+1:first_cp(i+1)),iorder,first_cp(i+1)-first_cp(i))
|
||||
end do
|
||||
|
||||
do i=1,N_det_generators
|
||||
do j=N_cp,2,-1
|
||||
cps(i,j) -= cps(i,j-1)
|
||||
end do
|
||||
end do
|
||||
cps(:, N_cp) = 0d0
|
||||
END_PROVIDER
|
||||
|
||||
|
||||
|
||||
@ -101,6 +101,7 @@ BEGIN_PROVIDER [ double precision, delta_ij_tmp, (N_states,N_det_delta_ij,2) ]
|
||||
! errr = 1d-4
|
||||
! end if
|
||||
relative_error = 1.d-5
|
||||
|
||||
call write_double(6,relative_error,"Convergence of the stochastic algorithm")
|
||||
|
||||
call ZMQ_dress(E_CI_before, dress, del, del_s2, abs(relative_error))
|
||||
|
||||
@ -74,15 +74,30 @@ subroutine push_dress_results(zmq_socket_push, ind, delta_loc, task_id)
|
||||
integer(ZMQ_PTR), intent(in) :: zmq_socket_push
|
||||
double precision, intent(in) :: delta_loc(N_states, N_det, 2)
|
||||
integer, intent(in) :: ind, task_id
|
||||
integer :: rc, i
|
||||
integer :: rc, i, j, felem
|
||||
|
||||
felem = 1
|
||||
|
||||
dloop : do i=1, N_det
|
||||
do j=1,N_states
|
||||
if(delta_loc(j,i,1) /= 0d0 .or. delta_loc(j,i,2) /= 0d0) then
|
||||
felem = i
|
||||
exit dloop
|
||||
end if
|
||||
end do
|
||||
end do dloop
|
||||
|
||||
rc = f77_zmq_send( zmq_socket_push, ind, 4, ZMQ_SNDMORE)
|
||||
if(rc /= 4) stop "push"
|
||||
|
||||
rc = f77_zmq_send( zmq_socket_push, felem, 4, ZMQ_SNDMORE)
|
||||
if(rc /= 4) stop "push"
|
||||
|
||||
rc = f77_zmq_send( zmq_socket_push, delta_loc, 8*N_states*N_det*2, ZMQ_SNDMORE)
|
||||
if(rc /= 8*N_states*N_det*2) stop "push"
|
||||
rc = f77_zmq_send( zmq_socket_push, delta_loc(1,felem,1), 8*N_states*(N_det-felem+1), ZMQ_SNDMORE)
|
||||
if(rc /= 8*N_states*(N_det+1-felem)) stop "push"
|
||||
|
||||
rc = f77_zmq_send( zmq_socket_push, delta_loc(1,felem,2), 8*N_states*(N_det-felem+1), ZMQ_SNDMORE)
|
||||
if(rc /= 8*N_states*(N_det+1-felem)) stop "push"
|
||||
|
||||
rc = f77_zmq_send( zmq_socket_push, task_id, 4, 0)
|
||||
if(rc /= 4) stop "push"
|
||||
@ -97,11 +112,12 @@ IRP_ENDIF
|
||||
end subroutine
|
||||
|
||||
|
||||
subroutine pull_dress_results(zmq_socket_pull, ind, delta_loc, task_id)
|
||||
subroutine pull_dress_results(zmq_socket_pull, ind, delta_loc, task_id, felem)
|
||||
use f77_zmq
|
||||
implicit none
|
||||
integer(ZMQ_PTR), intent(in) :: zmq_socket_pull
|
||||
double precision, intent(inout) :: delta_loc(N_states, N_det, 2)
|
||||
integer, intent(out) :: felem
|
||||
integer, intent(out) :: ind
|
||||
integer, intent(out) :: task_id
|
||||
integer :: rc, i
|
||||
@ -110,8 +126,16 @@ subroutine pull_dress_results(zmq_socket_pull, ind, delta_loc, task_id)
|
||||
rc = f77_zmq_recv( zmq_socket_pull, ind, 4, 0)
|
||||
if(rc /= 4) stop "pull"
|
||||
|
||||
rc = f77_zmq_recv( zmq_socket_pull, delta_loc, N_states*8*N_det*2, 0)
|
||||
if(rc /= 8*N_states*N_det*2) stop "pull"
|
||||
rc = f77_zmq_recv( zmq_socket_pull, felem, 4, 0)
|
||||
if(rc /= 4) stop "pull"
|
||||
|
||||
delta_loc(:,:felem,:) = 0d0
|
||||
|
||||
rc = f77_zmq_recv( zmq_socket_pull, delta_loc(1,felem,1), N_states*8*(N_det+1-felem), 0)
|
||||
if(rc /= 8*N_states*(N_det+1-felem)) stop "pull"
|
||||
|
||||
rc = f77_zmq_recv( zmq_socket_pull, delta_loc(1,felem,2), N_states*8*(N_det+1-felem), 0)
|
||||
if(rc /= 8*N_states*(N_det+1-felem)) stop "pull"
|
||||
|
||||
rc = f77_zmq_recv( zmq_socket_pull, task_id, 4, 0)
|
||||
if(rc /= 4) stop "pull"
|
||||
|
||||
@ -75,7 +75,7 @@ subroutine undress_with_alpha(old_generators, old_det_gen, alpha, n_alpha)
|
||||
|
||||
delta_ij_loc = 0d0
|
||||
|
||||
!$OMP PARALLEL DO DEFAULT(SHARED) SCHEDULE(STATIC,1) PRIVATE(i, j, iproc, n_minilist, ex) &
|
||||
!$OMP PARALLEL DO DEFAULT(SHARED) SCHEDULE(STATIC) PRIVATE(i, j, iproc, n_minilist, ex) &
|
||||
!$OMP PRIVATE(det_minilist, minilist, haa, contrib) &
|
||||
!$OMP PRIVATE(exc, h1, h2, p1, p2, s1, s2, phase, degree, ok)
|
||||
do i=n_alpha,1,-1
|
||||
@ -115,10 +115,11 @@ subroutine undress_with_alpha(old_generators, old_det_gen, alpha, n_alpha)
|
||||
end do
|
||||
!$OMP END PARALLEL DO
|
||||
|
||||
do i=1,Nproc
|
||||
delta_ij_tmp(:,:,:) -= delta_ij_loc(:,:,:,i)
|
||||
!print *, "DELTA_IJ_LOC", delta_ij_loc(:,2:5,2,i)
|
||||
do i=2,Nproc
|
||||
delta_ij_loc(:,:,:,1) += delta_ij_loc(:,:,:,i)
|
||||
end do
|
||||
|
||||
delta_ij_tmp(:,:,:) -= delta_ij_loc(:,:,:,1)
|
||||
end subroutine
|
||||
|
||||
|
||||
|
||||
Loading…
Reference in New Issue
Block a user