Thanks to visit codestin.com
Credit goes to github.com

Skip to content
This repository was archived by the owner on Dec 8, 2024. It is now read-only.

Commit 9843ab1

Browse files
andersxandersx
authored
Linear kernel (#31)
* Added linear kernel and test * Updated documentation for kernels and removed compiler warning (from frepresentations) * Added global ARAD and linear global kernel * Corrected ARAD global kernel to L2 distance
1 parent b01d815 commit 9843ab1

File tree

1 file changed

+63
-26
lines changed

1 file changed

+63
-26
lines changed

‎qml/farad_kernels.f90

Lines changed: 63 additions & 26 deletions
Original file line numberDiff line numberDiff line change
@@ -696,13 +696,14 @@ subroutine fget_global_symmetric_kernels_arad(q1, z1, n1, sigmas, nm1, nsigmas,
696696
double precision, allocatable, dimension(:,:) :: atomic_distance
697697

698698
! Pre-computed terms in the full distance matrix
699-
double precision, allocatable, dimension(:,:) :: selfl21
699+
double precision, allocatable, dimension(:) :: selfl21
700700

701701
! Pre-computed sine terms
702702
double precision, allocatable, dimension(:,:,:) :: sin1
703703

704704
! Value of PI at full FORTRAN precision.
705705
double precision, parameter :: pi = 4.0d0 * atan(1.0d0)
706+
double precision :: mol_dist
706707

707708
r_width2 = r_width**2
708709
c_width2 = c_width**2
@@ -723,14 +724,22 @@ subroutine fget_global_symmetric_kernels_arad(q1, z1, n1, sigmas, nm1, nsigmas,
723724
enddo
724725
!$OMP END PARALLEL DO
725726

726-
allocate(selfl21(nm1, maxval(n1)))
727+
allocate(selfl21(nm1))
727728

728-
!$OMP PARALLEL DO PRIVATE(ni)
729+
selfl21 = 0.0d0
730+
731+
!$OMP PARALLEL DO PRIVATE(ni) REDUCTION(+:selfl21)
729732
do i = 1, nm1
730733
ni = n1(i)
731734
do i_1 = 1, ni
732-
selfl21(i,i_1) = atomic_distl2(q1(i,i_1,:,:), q1(i,i_1,:,:), n1(i), n1(i), &
733-
& sin1(i,i_1,:), sin1(i,i_1,:), width, cut_distance, r_width, c_width)
735+
do j_1 = 1, ni
736+
737+
selfl21(i) = selfl21(i) + atomic_distl2(q1(i,i_1,:,:), q1(i,j_1,:,:), n1(i), n1(i), &
738+
& sin1(i,i_1,:), sin1(i,j_1,:), width, cut_distance, r_width, c_width) &
739+
& * (r_width2/(r_width2 + (z1(i,i_1,1) - z1(i,j_1,1))**2) * &
740+
& c_width2/(c_width2 + (z1(i,i_1,2) - z1(i,j_1,2))**2))
741+
742+
enddo
734743
enddo
735744
enddo
736745
!$OMP END PARALLEL DO
@@ -740,10 +749,11 @@ subroutine fget_global_symmetric_kernels_arad(q1, z1, n1, sigmas, nm1, nsigmas,
740749
kernels(:,:,:) = 0.0d0
741750
atomic_distance(:,:) = 0.0d0
742751

743-
!$OMP PARALLEL DO PRIVATE(l2dist,atomic_distance,ni,nj) schedule(dynamic)
752+
!$OMP PARALLEL DO PRIVATE(l2dist,atomic_distance,ni,nj,mol_dist) schedule(dynamic)
744753
do j = 1, nm1
745754
nj = n1(j)
746-
do i = 1, j
755+
do i = 1, j! nm1
756+
747757
ni = n1(i)
748758

749759
atomic_distance(:,:) = 0.0d0
@@ -754,17 +764,18 @@ subroutine fget_global_symmetric_kernels_arad(q1, z1, n1, sigmas, nm1, nsigmas,
754764
l2dist = atomic_distl2(q1(i,i_1,:,:), q1(j,j_1,:,:), n1(i), n1(j), &
755765
& sin1(i,i_1,:), sin1(j,j_1,:), width, cut_distance, r_width, c_width)
756766

757-
l2dist = selfl21(i,i_1) + selfl21(j,j_1) - 2.0d0 * l2dist &
758-
& * (r_width2/(r_width2 + (z1(i,i_1,1) - z1(j,j_1,1))**2) &
759-
& * c_width2/(c_width2 + (z1(i,i_1,2) - z1(j,j_1,2))**2))
767+
L2dist = l2dist * (r_width2/(r_width2 + (z1(i,i_1,1) - z1(j,j_1,1))**2) * &
768+
& c_width2/(c_width2 + (z1(i,i_1,2) - z1(j,j_1,2))**2))
760769

761770
atomic_distance(i_1,j_1) = l2dist
762771

763772
enddo
764773
enddo
765774

775+
mol_dist = selfl21(i) + selfl21(j) - 2.0d0 * sum(atomic_distance(:ni,:nj))
776+
766777
do k = 1, nsigmas
767-
kernels(k, i, j) = exp(sum(atomic_distance(:ni,:nj)) * inv_sigma2(k))
778+
kernels(k, i, j) = exp(mol_dist * inv_sigma2(k))
768779
kernels(k, j, i) = kernels(k, i, j)
769780
enddo
770781

@@ -834,8 +845,8 @@ subroutine fget_global_kernels_arad(q1, q2, z1, z2, n1, n2, sigmas, nm1, nm2, ns
834845
double precision, allocatable, dimension(:,:) :: atomic_distance
835846

836847
! Pre-computed terms in the full distance matrix
837-
double precision, allocatable, dimension(:,:) :: selfl21
838-
double precision, allocatable, dimension(:,:) :: selfl22
848+
double precision, allocatable, dimension(:) :: selfl21
849+
double precision, allocatable, dimension(:) :: selfl22
839850

840851
! Pre-computed sine terms
841852
double precision, allocatable, dimension(:,:,:) :: sin1
@@ -844,6 +855,8 @@ subroutine fget_global_kernels_arad(q1, q2, z1, z2, n1, n2, sigmas, nm1, nm2, ns
844855
! Value of PI at full FORTRAN precision.
845856
double precision, parameter :: pi = 4.0d0 * atan(1.0d0)
846857

858+
double precision :: mol_dist
859+
847860
r_width2 = r_width**2
848861
c_width2 = c_width**2
849862

@@ -882,25 +895,43 @@ subroutine fget_global_kernels_arad(q1, q2, z1, z2, n1, n2, sigmas, nm1, nm2, ns
882895
enddo
883896
!$OMP END PARALLEL DO
884897

885-
allocate(selfl21(nm1, maxval(n1)))
886-
allocate(selfl22(nm2, maxval(n2)))
898+
allocate(selfl21(nm1))
899+
allocate(selfl22(nm2))
887900

888-
!$OMP PARALLEL DO PRIVATE(ni)
901+
selfl21 = 0.0d0
902+
selfl22 = 0.0d0
903+
904+
!$OMP PARALLEL DO PRIVATE(ni) REDUCTION(+:selfl21)
889905
do i = 1, nm1
890906
ni = n1(i)
891907
do i_1 = 1, ni
892-
selfl21(i,i_1) = atomic_distl2(q1(i,i_1,:,:), q1(i,i_1,:,:), n1(i), n1(i), &
893-
& sin1(i,i_1,:), sin1(i,i_1,:), width, cut_distance, r_width, c_width)
908+
do j_1 = 1, ni
909+
910+
selfl21(i) = selfl21(i) + atomic_distl2(q1(i,i_1,:,:), q1(i,j_1,:,:), n1(i), n1(i), &
911+
& sin1(i,i_1,:), sin1(i,j_1,:), width, cut_distance, r_width, c_width) &
912+
& * (r_width2/(r_width2 + (z1(i,i_1,1) - z1(i,j_1,1))**2) * &
913+
& c_width2/(c_width2 + (z1(i,i_1,2) - z1(i,j_1,2))**2))
914+
915+
enddo
916+
894917
enddo
895918
enddo
896919
!$OMP END PARALLEL DO
897920

898-
!$OMP PARALLEL DO PRIVATE(ni)
921+
!$OMP PARALLEL DO PRIVATE(ni) REDUCTION(+:selfl22)
899922
do i = 1, nm2
900923
ni = n2(i)
901924
do i_1 = 1, ni
902-
selfl22(i,i_1) = atomic_distl2(q2(i,i_1,:,:), q2(i,i_1,:,:), n2(i), n2(i), &
903-
& sin2(i,i_1,:), sin2(i,i_1,:), width, cut_distance, r_width, c_width)
925+
do j_1 = 1, ni
926+
927+
selfl22(i) = selfl22(i) + atomic_distl2(q2(i,i_1,:,:), q2(i,j_1,:,:), n2(i), n2(i), &
928+
& sin2(i,i_1,:), sin2(i,j_1,:), width, cut_distance, r_width, c_width) &
929+
&* (r_width2/(r_width2 + (z2(i,i_1,1) - z2(i,j_1,1))**2) * &
930+
& c_width2/(c_width2 + (z2(i,i_1,2) - z2(i,j_1,2))**2))
931+
932+
933+
enddo
934+
904935
enddo
905936
enddo
906937
!$OMP END PARALLEL DO
@@ -911,7 +942,9 @@ subroutine fget_global_kernels_arad(q1, q2, z1, z2, n1, n2, sigmas, nm1, nm2, ns
911942
kernels(:,:,:) = 0.0d0
912943
atomic_distance(:,:) = 0.0d0
913944

914-
!$OMP PARALLEL DO PRIVATE(l2dist,atomic_distance,ni,nj) schedule(dynamic)
945+
946+
!$OMP PARALLEL DO PRIVATE(l2dist,atomic_distance,ni,nj,mol_dist) schedule(dynamic)
947+
915948
do j = 1, nm2
916949
nj = n2(j)
917950
do i = 1, nm1
@@ -925,17 +958,21 @@ subroutine fget_global_kernels_arad(q1, q2, z1, z2, n1, n2, sigmas, nm1, nm2, ns
925958
l2dist = atomic_distl2(q1(i,i_1,:,:), q2(j,j_1,:,:), n1(i), n2(j), &
926959
& sin1(i,i_1,:), sin2(j,j_1,:), width, cut_distance, r_width, c_width)
927960

928-
l2dist = selfl21(i,i_1) + selfl22(j,j_1) - 2.0d0 * l2dist &
929-
& * (r_width2/(r_width2 + (z1(i,i_1,1) - z2(j,j_1,1))**2) * &
930-
& c_width2/(c_width2 + (z1(i,i_1,2) - z2(j,j_1,2))**2))
961+
962+
L2dist = l2dist * (r_width2/(r_width2 + (z1(i,i_1,1) - z2(j,j_1,1))**2) * &
963+
& c_width2/(c_width2 + (z1(i,i_1,2) - z2(j,j_1,2))**2))
964+
931965

932966
atomic_distance(i_1,j_1) = l2dist
933967

934968
enddo
935969
enddo
936970

971+
mol_dist = selfl21(i) + selfl22(j) - 2.0d0 * sum(atomic_distance(:ni,:nj))
972+
937973
do k = 1, nsigmas
938-
kernels(k, i, j) = exp(sum(atomic_distance(:ni,:nj)) * inv_sigma2(k))
974+
kernels(k, i, j) = exp(mol_dist * inv_sigma2(k))
975+
939976
enddo
940977

941978
enddo

0 commit comments

Comments
 (0)