LCPQ/qmcchem
10
1
Fork 1
mirror of https://gitlab.com/scemama/qmcchem.git synced 2024-12-21 11:53:30 +01:00
Code Releases Activity

Add routines for CI optimization

Browse Source
This commit is contained in:
Anthony Scemama 2020-07-17 13:05:34 +02:00
parent a4f49f04c5
commit 5db05a2aee
1 changed files with 144 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

150
src/PROPERTIES/properties_ci.irp.f
View File

@ -39,10 +39,10 @@ BEGIN_PROVIDER [ double precision, ci_h_psidet, (size_ci_h_psidet) ]
T += det_alpha_grad_lapl(4,l,i)*det_beta_value (j)
enddo
do l=elec_beta_num+1,elec_num
T += det_beta_grad_lapl (4,l,j)*det_alpha_value(i)
T += det_beta_grad_lapl (4,l,j)*det_alpha_value(i)
enddo
ci_h_psidet(k) = -0.5d0*T + E_pot * det_alpha_value(i)*det_beta_value (j)
ci_h_psidet(k) *= psidet_inv
ci_h_psidet(k) *= psidet_inv
enddo
ci_h_psidet_min = min(ci_h_psidet_min,minval(ci_h_psidet))
@ -54,10 +54,9 @@ END_PROVIDER
BEGIN_PROVIDER [ double precision, ci_overlap_matrix, (size_ci_overlap_matrix) ]
implicit none
BEGIN_DOC
! < det(i) | det(j) >
! < det(i) | det(j) >
! < det(i) |H| det(j) >
!
! Dimensions : det_num
! Dimensions : det_num*det_num
END_DOC
integer :: i, j, k, l, m, n
@ -70,7 +69,7 @@ BEGIN_PROVIDER [ double precision, ci_overlap_matrix, (size_ci_overlap_matrix) ]
do l=1,det_num
m = det_coef_matrix_rows(l)
n = det_coef_matrix_columns(l)
ci_overlap_matrix(l) = det_alpha_value(m)*det_beta_value(n) * f
ci_overlap_matrix( det_num*(k-1) + l) = det_alpha_value(m)*det_beta_value(n) * f
enddo
enddo
@ -79,3 +78,142 @@ BEGIN_PROVIDER [ double precision, ci_overlap_matrix, (size_ci_overlap_matrix) ]
SOFT_TOUCH ci_overlap_matrix_min ci_overlap_matrix_max
END_PROVIDER
BEGIN_PROVIDER [ double precision, ci_h_matrix, (size_ci_h_matrix) ]
implicit none
BEGIN_DOC
! < det(i) |H| det(j) >
!
! Dimensions : det_num*det_num
END_DOC
integer :: i, j, k, l, m, n, e
double precision :: f, g, h, T, V
do l=1,det_num
m = det_coef_matrix_rows(l)
n = det_coef_matrix_columns(l)
! Lapl D
g = 0.d0
do e=1,elec_alpha_num
g += det_alpha_grad_lapl(4,e,m) * det_beta_value (n)
enddo
do e=elec_beta_num+1,elec_num
g += det_alpha_value(m) * det_beta_grad_lapl(4,e,n)
enddo
T = g
! D (Lapl J)/J
g = 0.d0
do e=1,elec_num
g += jast_lapl_jast_inv(e)
enddo
T += det_alpha_value(m) * det_beta_value(n) * g
! 2 (grad D).(Grad J)/J
g = 0.d0
do e=1,elec_alpha_num
g += &
det_alpha_grad_lapl(1,e,m) * jast_grad_jast_inv_x(e) + &
det_alpha_grad_lapl(2,e,m) * jast_grad_jast_inv_y(e) + &
det_alpha_grad_lapl(3,e,m) * jast_grad_jast_inv_z(e)
enddo
h = 0.d0
do e=elec_alpha_num+1,elec_num
h += &
det_beta_grad_lapl(1,e,n) * jast_grad_jast_inv_x(e) + &
det_beta_grad_lapl(2,e,n) * jast_grad_jast_inv_y(e) + &
det_beta_grad_lapl(3,e,n) * jast_grad_jast_inv_z(e)
enddo
T += 2.d0*( g * det_beta_value(n) + h * det_alpha_value(m) )
g = det_alpha_value(m)*det_beta_value(n)
V = E_pot* g
do e=1,elec_alpha_num
V -= pseudo_non_local(e)* g
V += det_alpha_pseudo(e,m) * det_beta_value(n)
enddo
do e=elec_alpha_num+1,elec_num
V -= pseudo_non_local(e)* g
V += det_alpha_value(m) * det_beta_pseudo(e,n)
enddo
f = -0.5d0*T + V
f *= psidet_inv * psidet_inv
do k=1,det_num
i = det_coef_matrix_rows(k)
j = det_coef_matrix_columns(k)
ci_h_matrix( det_num*(l-1) + k) = f * &
det_alpha_value(i)*det_beta_value (j)
enddo
enddo
ci_h_matrix_min = min(ci_h_matrix_min,minval(ci_h_matrix))
ci_h_matrix_max = max(ci_h_matrix_max,maxval(ci_h_matrix))
SOFT_TOUCH ci_h_matrix_min ci_h_matrix_max
END_PROVIDER
BEGIN_PROVIDER [ double precision, ci_h_matrix_diag, (size_ci_h_matrix_diag) ]
implicit none
BEGIN_DOC
! < det(i) |H| det(j) >
!
! Dimensions : det_num
END_DOC
integer :: i, j, k, l, m, n, e
double precision :: f, g, h, T, V
do l=1,det_num
m = det_coef_matrix_rows(l)
n = det_coef_matrix_columns(l)
! Lapl D
g = 0.d0
do e=1,elec_alpha_num
g += det_alpha_grad_lapl(4,e,m) * det_beta_value (n)
enddo
do e=elec_beta_num+1,elec_num
g += det_alpha_value(m) * det_beta_grad_lapl(4,e,n)
enddo
T = g
! D (Lapl J)/J
g = 0.d0
do e=1,elec_num
g += jast_lapl_jast_inv(e)
enddo
T += det_alpha_value(m) * det_beta_value(n) * g
! 2 (grad D).(Grad J)/J
g = 0.d0
do e=1,elec_alpha_num
g += &
det_alpha_grad_lapl(1,e,m) * jast_grad_jast_inv_x(e) + &
det_alpha_grad_lapl(2,e,m) * jast_grad_jast_inv_y(e) + &
det_alpha_grad_lapl(3,e,m) * jast_grad_jast_inv_z(e)
enddo
h = 0.d0
do e=elec_alpha_num+1,elec_num
h += &
det_beta_grad_lapl(1,e,n) * jast_grad_jast_inv_x(e) + &
det_beta_grad_lapl(2,e,n) * jast_grad_jast_inv_y(e) + &
det_beta_grad_lapl(3,e,n) * jast_grad_jast_inv_z(e)
enddo
T += 2.d0*( g * det_beta_value(n) + h * det_alpha_value(m) )
g = det_alpha_value(m)*det_beta_value(n)
V = E_pot* g
do e=1,elec_alpha_num
V -= pseudo_non_local(e)* g
V += det_alpha_pseudo(e,m) * det_beta_value(n)
enddo
do e=elec_alpha_num+1,elec_num
V -= pseudo_non_local(e)* g
V += det_alpha_value(m) * det_beta_pseudo(e,n)
enddo
f = -0.5d0*T + V
f *= psidet_inv * psidet_inv
ci_h_matrix_diag(l) = f * &
det_alpha_value(m)*det_beta_value (n)
enddo
ci_h_matrix_diag_min = min(ci_h_matrix_diag_min,minval(ci_h_matrix_diag))
ci_h_matrix_diag_max = max(ci_h_matrix_diag_max,maxval(ci_h_matrix_diag))
SOFT_TOUCH ci_h_matrix_diag_min ci_h_matrix_diag_max
END_PROVIDER