openMP implementation of the VV block for ppBSE@GF2

mol/GW20/Ar.xyz

@@ -1,3 +0,0 @@
-1
-Argon; atom; s
-Ar 0.0 0.0 0.0

mol/GW20/BF.xyz

@@ -1,4 +0,0 @@
-2
-Fluoroborane; experimental structure from HCP92; s
-B 0.0000 0.0000 0.0000
-F 0.0000 0.0000 1.2626

mol/GW20/BH3.xyz

@@ -1,6 +0,0 @@
-4
-Borane; experimental structure from HCP92; s
-B  0.0000  0.0000  0.0000
-H   0.0000  0.0000  1.19
-H   0.0000  1.0306 -0.595
-H   0.0000 -1.0306 -0.595

mol/GW20/BN.xyz

@@ -1,4 +0,0 @@
-2
-Boron nitride; experimental structure from HCP92; s
-B 0.0000 0.0000 0.0000
-N 0.0000 0.0000 1.281

mol/GW20/BeO.xyz

@@ -1,4 +0,0 @@
-2
-Beryllium monoxide; experimental structure from HCP92; s
-Be 0.0000 0.0000 0.0000
-O  0.0000 0.0000 1.3308

mol/GW20/CH4.xyz

@@ -1,7 +0,0 @@
-5
-Methane; experimental structure from HCP92; m
-C  0.0000  0.0000  0.0000
-H  0.6276 -0.6275  0.6276
-H -0.6276  0.6276  0.6276
-H -0.6276 -0.6276 -0.6276
-H  0.6276  0.6276 -0.6276

mol/GW20/CO.xyz

@@ -1,4 +0,0 @@
-2
-Carbon monoxide; experimental structure from HCP92; s
-C 0.0000 0.0000 0.0000
-O 0.0000 0.0000 1.283

mol/GW20/Cu2.xyz

@@ -1,4 +0,0 @@
-2
-Copper dimer; experimental structure form HCP92; s
-Cu 0.0 0.0 0.0
-Cu 0.0 0.0 2.2197

mol/GW20/F2.xyz

@@ -1,4 +0,0 @@
-2
-Fluorine; experimental structure from HCP92; s
-F 0.0000 0.0000 0.0000
-F 0.0000 0.0000 1.4119

mol/GW20/H2.xyz

@@ -1,4 +0,0 @@
-2
-Hydrogen; experimental structure from HCP92; s
-H 0.0000 0.0000 0.0000
-H 0.0000 0.0000 0.74144

mol/GW20/H2O.xyz

@@ -1,5 +0,0 @@
-3
-Water; experimental structure from HCP92; s
-O  0.0000 0.0000 0.0000
-H  0.7571 0.0000 0.5861
-H -0.7571 0.0000 0.5861

mol/GW20/HCl.xyz

@@ -1,4 +0,0 @@
-2
-Hydrogen chloride; experimental structure from HCP92; s
-H  0.0000 0.0000 0.0000
-Cl 0.0000 0.0000 1.2746

mol/GW20/HF.xyz

@@ -1,4 +0,0 @@
-2
-Hydrogen fluoride; experimental structure from HCP92; s
-H 0.0000 0.0000 0.0000
-F 0.0000 0.0000 0.9169

mol/GW20/He.xyz

@@ -1,3 +0,0 @@
-1
-Helium; atom; s
-He 0.0 0.0 0.0

mol/GW20/Li2.xyz

@@ -1,4 +0,0 @@
-2
-Lithium dimer; experimental structure from HCP92; s
-Li 0.0000 0.0000 0.0000
-Li 0.0000 0.0000 2.6729

mol/GW20/LiF.xyz

@@ -1,4 +0,0 @@
-2
-Lithium fluoride; experimental structure from HCP92; s
-Li 0.0000 0.0000 0.0000
-F  0.0000 0.0000 1.5639

mol/GW20/LiH.xyz

@@ -1,4 +0,0 @@
-2
-Lithium hydride; experimental structure from HCP92; s
-Li 0.0000 0.0000 0.0000
-H  0.0000 0.0000 1.5949

mol/GW20/N2.xyz

@@ -1,4 +0,0 @@
-2
-Nitrogen; experimental structure from HCP92; s
-N 0.0000 0.0000 0.0000
-N 0.0000 0.0000 1.0977

mol/GW20/NH3.xyz

@@ -1,6 +0,0 @@
-4
-Amonia; experimental structure from HCP92; s
-N  0.0000  0.0000  0.0000
-H  0.0000 -0.9377 -0.3816
-H  0.8121  0.4689 -0.3816
-H -0.8121  0.4689 -0.3816

mol/GW20/Ne.xyz

@@ -1,3 +0,0 @@
-1
-Neon; atom; s
-Ne 0.0 0.0 0.0

mol/GW20/O3.xyz

@@ -1,5 +0,0 @@
-3
-Ozon; experimental structure from HCP92; s
-O  0.0000 0.0000 0.0000
-O  1.0869 0.0000 0.6600
-O -1.0869 0.0000 0.6600

mol/GW20/SH2.xyz

@@ -1,5 +0,0 @@
-3
-Hydrogen sulfide; experimental structure from HCP92; s
-S  0.0000 0.0000 0.0000
-H  0.9617 0.0000 0.9268
-H -0.9617 0.0000 0.9268

src/GF/RGF2_ppBSE2_static_kernel_C.f90

@@ -25,8 +25,11 @@ subroutine RGF2_ppBSE2_static_kernel_C(ispin,eta,nBas,nC,nO,nV,nR,nVV,lambda,ERI
   double precision              :: dem,num
   integer                       :: m
   integer                       :: a,b,c,d,e
-  integer                       :: ab,cd
+  integer                       :: a0,aa,ab,cd
 
+  double precision              :: eta2
+  double precision, allocatable :: Om_tmp(:,:)
+  
 ! Output variables
 
   double precision,intent(out)  :: KC_sta(nVV,nVV)
@@ -34,15 +37,39 @@ subroutine RGF2_ppBSE2_static_kernel_C(ispin,eta,nBas,nC,nO,nV,nR,nVV,lambda,ERI
 ! Initialization
 
   KC_sta(:,:) = 0d0
+  eta2 = eta * eta
 
+  allocate(Om_tmp(nO,nV))
+
+  ! Compute the energy differences and denominator once and store them in a temporary array
+  !$OMP PARALLEL DEFAULT(NONE) PRIVATE(m,e,dem) SHARED(nC,nO,nBas,nR, eta2, eGF, Om_tmp)
+  !$OMP DO
+  do m=nC+1,nO
+     do e=nO+1,nBas-nR
+        dem = eGF(m) - eGF(e)
+        Om_tmp(m,e) = dem / (dem*dem + eta2)
+     enddo
+  enddo
+  !$OMP END DO
+  !$OMP END PARALLEL
+  
 ! Second-order correlation kernel for the block C of the singlet manifold
 
+!  --- --- ---
+!    OpenMP implementation
+!  --- --- ---
+
   if(ispin == 1) then
 
-    ab = 0
+    a0 = nBas - nR - nO
+    !$OMP PARALLEL DEFAULT(NONE)                              &
+    !$OMP          PRIVATE(a, b, aa, ab, c, d, cd, m, e, num) &
+    !$OMP          SHARED(nO, nBas, nR, nC, a0, ERI, Om_tmp, KC_sta, lambda)
+    !$OMP DO
     do a=nO+1,nBas-nR
+      aa = a0 * (a - nO - 1) - (a - nO - 1) * (a - nO) / 2 - nO
       do b=a,nBas-nR
-        ab = ab + 1
+        ab = aa + b
 
         cd = 0
         do c=nO+1,nBas-nR
@@ -52,39 +79,91 @@ subroutine RGF2_ppBSE2_static_kernel_C(ispin,eta,nBas,nC,nO,nV,nR,nVV,lambda,ERI
             do m=nC+1,nO
               do e=nO+1,nBas-nR
      
-                dem = eGF(m) - eGF(e)
                 num = 2d0*ERI(a,m,c,e)*ERI(b,e,d,m) -     ERI(a,m,c,e)*ERI(b,e,m,d) & 
                     -     ERI(a,m,e,c)*ERI(b,e,d,m) -     ERI(a,m,e,c)*ERI(b,e,m,d)
 
-                KC_sta(ab,cd) = KC_sta(ab,cd) + num*dem/(dem**2 + eta**2)
+                KC_sta(ab,cd) = KC_sta(ab,cd) + num * Om_tmp(m,e)
             
-                dem = eGF(m) - eGF(e)
                 num = 2d0*ERI(b,m,c,e)*ERI(a,e,d,m) -     ERI(b,m,c,e)*ERI(a,e,m,d) & 
                     -     ERI(b,m,e,c)*ERI(a,e,d,m) -     ERI(b,m,e,c)*ERI(a,e,m,d)
 
-                KC_sta(ab,cd) = KC_sta(ab,cd) + num*dem/(dem**2 + eta**2)
+                KC_sta(ab,cd) = KC_sta(ab,cd) + num * Om_tmp(m,e)
             
               end do
-           end do
+            end do
 
-           KC_sta(ab,cd) = 2d0*lambda*KC_sta(ab,cd)/sqrt((1d0 + Kronecker_delta(a,b))*(1d0 + Kronecker_delta(c,d)))
+            KC_sta(ab,cd) = 2d0*lambda*KC_sta(ab,cd)/sqrt((1d0 + Kronecker_delta(a,b))*(1d0 + Kronecker_delta(c,d)))
             
-          end do
+           end do
-        end do
+         end do
 
       end do
     end do
+    !$OMP END DO
+    !$OMP END PARALLEL
 
   end if
 
+!  --- --- ---
+!  Naive implementation
+!  --- --- ---
+!   if(ispin == 1) then
+   
+!     ab = 0
+!     do a=nO+1,nBas-nR
+!       do b=a,nBas-nR
+!         ab = ab + 1
+   
+!         cd = 0
+!         do c=nO+1,nBas-nR
+!           do d=c,nBas-nR
+!             cd = cd + 1
+   
+!             do m=nC+1,nO
+!               do e=nO+1,nBas-nR
+     
+!                 dem = eGF(m) - eGF(e)
+!                 num = 2d0*ERI(a,m,c,e)*ERI(b,e,d,m) -     ERI(a,m,c,e)*ERI(b,e,m,d) & 
+!                     -     ERI(a,m,e,c)*ERI(b,e,d,m) -     ERI(a,m,e,c)*ERI(b,e,m,d)
+   
+!                 KC_sta(ab,cd) = KC_sta(ab,cd) + num*dem/(dem**2 + eta**2)
+            
+!                 dem = eGF(m) - eGF(e)
+!                 num = 2d0*ERI(b,m,c,e)*ERI(a,e,d,m) -     ERI(b,m,c,e)*ERI(a,e,m,d) & 
+!                     -     ERI(b,m,e,c)*ERI(a,e,d,m) -     ERI(b,m,e,c)*ERI(a,e,m,d)
+   
+!                 KC_sta(ab,cd) = KC_sta(ab,cd) + num*dem/(dem**2 + eta**2)
+            
+!               end do
+!            end do
+   
+!            KC_sta(ab,cd) = 2d0*lambda*KC_sta(ab,cd)/sqrt((1d0 + Kronecker_delta(a,b))*(1d0 + Kronecker_delta(c,d)))
+            
+!           end do
+!         end do
+   
+!       end do
+!     end do
+   
+!   end if
+
 ! Second-order correlation kernel for the block C of the triplet manifold
 
+!  --- --- ---
+!  OpenMP implementation
+!  --- --- ---
+  
   if(ispin == 2) then
 
-    ab = 0
+    a0 = nBas - nR - nO - 1
-    do a=nO+1,nBas-nR
+    !$OMP PARALLEL DEFAULT(NONE)                              &
-      do b=a+1,nBas-nR
+    !$OMP          PRIVATE(a, b, aa, ab, c, d, cd, m, e, num) &
-        ab = ab + 1
+    !$OMP          SHARED(nO, nBas, nR, nC, a0, ERI, Om_tmp, KC_sta)
+    !$OMP DO
+    do a = nO+1, nBas-nR
+      aa = a0 * (a - nO - 1) - (a - nO - 1) * (a - nO) / 2 - nO - 1
+      do b = a+1, nBas-nR
+        ab = aa + b
 
         cd = 0
         do c=nO+1,nBas-nR
@@ -94,17 +173,15 @@ subroutine RGF2_ppBSE2_static_kernel_C(ispin,eta,nBas,nC,nO,nV,nR,nVV,lambda,ERI
             do m=nC+1,nO
               do e=nO+1,nBas-nR
      
-                dem = eGF(m) - eGF(e)
                 num = 2d0*ERI(a,m,c,e)*ERI(b,e,d,m) -     ERI(a,m,c,e)*ERI(b,e,m,d) &
                     -     ERI(a,m,e,c)*ERI(b,e,d,m) +     ERI(a,m,e,c)*ERI(b,e,m,d)
 
-                KC_sta(ab,cd) = KC_sta(ab,cd) + 2d0*num*dem/(dem**2 + eta**2)
+                KC_sta(ab,cd) = KC_sta(ab,cd) + 2d0 * num * Om_tmp(m,e)
             
-                dem = eGF(m) - eGF(e)
                 num = 2d0*ERI(b,m,c,e)*ERI(a,e,d,m) -     ERI(b,m,c,e)*ERI(a,e,m,d) &
                     -     ERI(b,m,e,c)*ERI(a,e,d,m) +     ERI(b,m,e,c)*ERI(a,e,m,d)
 
-                KC_sta(ab,cd) = KC_sta(ab,cd) - 2d0*num*dem/(dem**2 + eta**2)
+                KC_sta(ab,cd) = KC_sta(ab,cd) - 2d0 * num * Om_tmp(m,e)
             
               end do
             end do
@@ -114,7 +191,54 @@ subroutine RGF2_ppBSE2_static_kernel_C(ispin,eta,nBas,nC,nO,nV,nR,nVV,lambda,ERI
 
       end do
     end do
+    !$OMP END DO
+    !$OMP END PARALLEL
 
   end if
 
+  
+!  --- --- ---
+!  Naive implementation
+!  --- --- ---
+!   if(ispin == 2) then
+   
+!     ab = 0
+!     do a=nO+1,nBas-nR
+!       do b=a+1,nBas-nR
+!         ab = ab + 1
+   
+!         cd = 0
+!         do c=nO+1,nBas-nR
+!           do d=c+1,nBas-nR
+!             cd = cd + 1
+   
+!             do m=nC+1,nO
+!               do e=nO+1,nBas-nR
+     
+!                 dem = eGF(m) - eGF(e)
+!                 num = 2d0*ERI(a,m,c,e)*ERI(b,e,d,m) -     ERI(a,m,c,e)*ERI(b,e,m,d) &
+!                     -     ERI(a,m,e,c)*ERI(b,e,d,m) +     ERI(a,m,e,c)*ERI(b,e,m,d)
+   
+!                 KC_sta(ab,cd) = KC_sta(ab,cd) + 2d0*num*dem/(dem**2 + eta**2)
+            
+!                 dem = eGF(m) - eGF(e)
+!                 num = 2d0*ERI(b,m,c,e)*ERI(a,e,d,m) -     ERI(b,m,c,e)*ERI(a,e,m,d) &
+!                     -     ERI(b,m,e,c)*ERI(a,e,d,m) +     ERI(b,m,e,c)*ERI(a,e,m,d)
+   
+!                 KC_sta(ab,cd) = KC_sta(ab,cd) - 2d0*num*dem/(dem**2 + eta**2)
+            
+!               end do
+!             end do
+   
+!           end do
+!         end do
+   
+!       end do
+!     end do
+   
+  !   end if
+  
+
+  deallocate(Om_tmp)
+
 end subroutine 

src/GF/ufRG0F02.f90

@@ -71,7 +71,7 @@ subroutine ufRG0F02(dotest,nBas,nC,nO,nV,nR,nS,ENuc,ERHF,ERI,eHF)
 ! Main loop over orbitals !
 !-------------------------!
 
-  do p=nO-1,nO
+  do p=nO-3,nO
 
      H(:,:) = 0d0
      Reigv(:,:) = 0d0

src/GW/RGW_ppBSE.f90

@@ -61,7 +61,7 @@ subroutine RGW_ppBSE(TDA_W,TDA,dBSE,dTDA,singlet,triplet,eta,nBas,nC,nO,nV,nR,nS
   double precision,allocatable  :: KB_sta(:,:)
   double precision,allocatable  :: KC_sta(:,:)
   double precision,allocatable  :: KD_sta(:,:)
-
+  
 ! Output variables
 
   double precision,intent(out)  :: EcBSE(nspin)
@@ -114,7 +114,7 @@ subroutine RGW_ppBSE(TDA_W,TDA,dBSE,dTDA,singlet,triplet,eta,nBas,nC,nO,nV,nR,nS
     call RGW_ppBSE_static_kernel_C(ispin,eta,nBas,nC,nO,nV,nR,nS,nVV,1d0,ERI,OmRPA,rho_RPA,KC_sta)
     call RGW_ppBSE_static_kernel_D(ispin,eta,nBas,nC,nO,nV,nR,nS,nOO,1d0,ERI,OmRPA,rho_RPA,KD_sta)
     if(.not.TDA) call RGW_ppBSE_static_kernel_B(ispin,eta,nBas,nC,nO,nV,nR,nS,nOO,nVV,1d0,ERI,OmRPA,rho_RPA,KB_sta)
-
+    
     call ppLR_C(ispin,nBas,nC,nO,nV,nR,nVV,1d0,eGW,ERI,Cpp)
     call ppLR_D(ispin,nBas,nC,nO,nV,nR,nOO,1d0,eGW,ERI,Dpp)
     if(.not.TDA) call ppLR_B(ispin,nBas,nC,nO,nV,nR,nOO,nVV,1d0,ERI,Bpp)

src/LR/ppLR_transition_vectors.f90

@@ -117,8 +117,8 @@ subroutine ppLR_transition_vectors(spin_allowed,nBas,nC,nO,nV,nR,nOO,nVV,dipole_
 
 ! Thomas-Reiche-Kuhn sum rule
 
-  if(nVV > 0) write(*,'(A50,F10.6)') 'Thomas-Reiche-Kuhn sum rule for p-p sector = ',sum(os1(:))
+!  if(nVV > 0) write(*,'(A50,F10.6)') 'Thomas-Reiche-Kuhn sum rule for p-p sector = ',sum(os1(:))
-  write(*,*)
+!  write(*,*)
 
 !-----------------------------------------------!
 ! Print details about excitations for hh sector !
@@ -188,7 +188,7 @@ subroutine ppLR_transition_vectors(spin_allowed,nBas,nC,nO,nV,nR,nOO,nVV,dipole_
 
 ! Thomas-Reiche-Kuhn sum rule
 
-  if(nOO > 0) write(*,'(A50,F10.6)') 'Thomas-Reiche-Kuhn sum rule for h-h sector = ',sum(os2(:))
+!  if(nOO > 0) write(*,'(A50,F10.6)') 'Thomas-Reiche-Kuhn sum rule for h-h sector = ',sum(os2(:))
-  write(*,*)
+!  write(*,*)
 
 end subroutine 

src/make_ninja.py

@@ -88,9 +88,9 @@ FIX_ORDER_OF_LIBS=-Wl,--start-group
 """
 
 if sys.platform in ["linux", "linux2"]:
-#    compiler = compile_gfortran_linux
+#   compiler = compile_gfortran_linux
-   compiler = compile_ifort_linux
+#   compiler = compile_ifort_linux 
-#   compiler = compile_olympe
+    compiler = compile_olympe
 elif sys.platform == "darwin":
   compiler = compile_gfortran_mac
 else: