Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Fix distance nb #91

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Fix distance nb #91

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f773334
fixed permuational invariance bug for order>2 distance_nb descriptors
albapa Sep 16, 2024
a4ba0a0
add xml version check
albapa Sep 17, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
55 changes: 39 additions & 16 deletions descriptors.f95
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3046,6 +3046,12 @@ subroutine distance_Nb_initialise(this,args_str,error)
endif
endif

if( this%compact_clusters .and. this%order > 2 .and. xml_version < 1726500349 ) then
RAISE_ERROR("distance_Nb_initialise: potential (compact_clusters="//this%compact_clusters//" order="//
this%order//") was generated using an earlier version of GAP ("//xml_version//
"), and is affected by https://github.com/libAtoms/QUIP/issues/669 which has since been fixed. Retrain your model with a recent version.", error)
endif


allocate(this%Z(this%order))
default_Z = ""
Expand Down Expand Up @@ -10037,6 +10043,7 @@ subroutine distance_Nb_calc(this,at,descriptor_out,do_descriptor,do_grad_descrip
integer :: d, n_descriptors, n_cross, i_desc, i_data, i, j, ii, jj, kk, ll, &
iConnectivity, n_index
integer, dimension(3) :: s_i, s_j
integer, dimension(this%order) :: new_order
real(dp) :: r_ij, fcut_connectivity
real(dp), dimension(3) :: dfcut_connectivity
real(dp), dimension(3) :: d_ij
Expand Down Expand Up @@ -10129,6 +10136,22 @@ subroutine distance_Nb_calc(this,at,descriptor_out,do_descriptor,do_grad_descrip
call distance_Nb_calc_get_clusters(this,at,atoms_in_descriptors=atoms_in_descriptors,error=error)
endif

do i_desc = 1, n_descriptors
if( this%order > 1 ) then
new_order = -1
do ii = 1, this%order
i = atoms_in_descriptors(1,ii,i_desc)
do jj = 1, this%order
if( at%Z(i) == this%Z(jj) .and. new_order(jj) == -1 ) then
new_order(jj) = ii
exit
endif
enddo
enddo
atoms_in_descriptors(:,:,i_desc) = atoms_in_descriptors(:,new_order,i_desc)
endif
enddo

allocate(fcut_pair(this%order,this%order))
if( my_do_grad_descriptor ) then
allocate(dfcut_pair(this%order,this%order), directions(3,this%order,this%order))
Expand Down Expand Up @@ -10172,26 +10195,26 @@ subroutine distance_Nb_calc(this,at,descriptor_out,do_descriptor,do_grad_descrip

descriptor_out%x(i_desc)%covariance_cutoff = 1.0_dp

do jj = 2, this%order
descriptor_out%x(i_desc)%covariance_cutoff = descriptor_out%x(i_desc)%covariance_cutoff * fcut_pair(jj,1)
do ii = 1, this%order
do jj = ii+1, this%order
descriptor_out%x(i_desc)%covariance_cutoff = descriptor_out%x(i_desc)%covariance_cutoff * fcut_pair(jj,ii)
enddo
enddo

if( my_do_grad_descriptor ) then
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,1) = 0.0_dp
do kk = 2, this%order
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,kk) = 1.0_dp
do jj = 2, this%order
if( jj == kk ) then
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,kk) = &
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,kk) * dfcut_pair(jj,1) * (-directions(:,jj,1))
else
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,kk) = &
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,kk) * fcut_pair(jj,1)
endif
if( descriptor_out%x(i_desc)%covariance_cutoff > 0.0_dp ) then
do ii = 1, this%order
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,ii) = 0.0_dp
do jj = 1, this%order
if( ii /= jj ) descriptor_out%x(i_desc)%grad_covariance_cutoff(:,ii) = &
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,ii) - &
descriptor_out%x(i_desc)%covariance_cutoff / fcut_pair(jj,ii) * dfcut_pair(jj,ii) * directions(:,ii,jj)
enddo
enddo
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,1) = &
descriptor_out%x(i_desc)%grad_covariance_cutoff(:,1) - descriptor_out%x(i_desc)%grad_covariance_cutoff(:,kk)
enddo
else
descriptor_out%x(i_desc)%grad_covariance_cutoff = 0.0_dp
endif

endif


Expand Down