diff --git a/src/ao_two_e_ints/cholesky.irp.f b/src/ao_two_e_ints/cholesky.irp.f
index f4746144..4702c850 100644
--- a/src/ao_two_e_ints/cholesky.irp.f
+++ b/src/ao_two_e_ints/cholesky.irp.f
@@ -29,7 +29,7 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
   double precision, pointer        :: L(:,:), L_old(:,:)
 
 
-  double precision, parameter :: s = 1.d-1
+  double precision :: s
   double precision, parameter :: dscale = 1.d0
 
   double precision, allocatable :: D(:), Delta(:,:), Ltmp_p(:,:), Ltmp_q(:,:)
@@ -43,16 +43,28 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
   double precision, external :: get_ao_two_e_integral
   logical, external :: ao_two_e_integral_zero
 
+  double precision, external      :: ao_two_e_integral
   integer :: block_size, iblock, ierr
 
   integer(omp_lock_kind), allocatable :: lock(:)
 
-  PROVIDE ao_two_e_integrals_in_map
+  double precision :: rss
+  double precision, external :: memory_of_double, memory_of_int
+
+
+  PROVIDE nucl_coord
+
+  if (.not.do_direct_integrals) then
+    PROVIDE ao_two_e_integrals_in_map
+  endif
   deallocate(cholesky_ao)
 
   ndim = ao_num*ao_num
   tau = ao_cholesky_threshold
 
+  rss = 6.d0 * memory_of_double(ndim) + &
+        6.d0 * memory_of_int(ndim)
+  call check_mem(rss, irp_here)
 
   allocate(L(ndim,1))
 
@@ -85,13 +97,22 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
     enddo
   enddo
 
-  !$OMP PARALLEL DO DEFAULT(SHARED) PRIVATE(i)
-  do i=1,ndim
-     D(i) = get_ao_two_e_integral(addr(1,i), addr(1,i), &
-                                  addr(2,i), addr(2,i), &
-                                  ao_integrals_map)
-  enddo
-  !$OMP END PARALLEL DO
+  if (do_direct_integrals) then
+    !$OMP PARALLEL DO DEFAULT(SHARED) PRIVATE(i)
+    do i=1,ndim
+       D(i) = ao_two_e_integral(addr(1,i), addr(2,i), &
+                                addr(1,i), addr(2,i))
+    enddo
+    !$OMP END PARALLEL DO
+  else
+    !$OMP PARALLEL DO DEFAULT(SHARED) PRIVATE(i) SCHEDULE(guided)
+    do i=1,ndim
+       D(i) = get_ao_two_e_integral(addr(1,i), addr(1,i), &
+                                    addr(2,i), addr(2,i), &
+                                    ao_integrals_map)
+    enddo
+    !$OMP END PARALLEL DO
+  endif
 
   Dmax = maxval(D)
 
@@ -117,21 +138,49 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
     ! a.
     i = i+1
 
-    ! b.
-     Dmin = max(s*Dmax,tau)
+    logical :: memory_ok
+    memory_ok = .False.
 
-    ! c.
-    nq=0
-    LDmap = 0
-    DLmap = 0
-    do p=1,np
-      if ( D(Lset(p)) > Dmin ) then
-        nq = nq+1
-        Dset(nq) = Lset(p)
-        Dset_rev(Dset(nq)) = nq
-        LDmap(p) = nq
-        DLmap(nq) = p
+    s = 0.1d0
+
+    ! Inrease s until the arrays fit in memory
+    do 
+
+      ! b.
+      Dmin = max(s*Dmax,tau)
+
+      ! c.
+      nq=0
+      LDmap = 0
+      DLmap = 0
+      do p=1,np
+        if ( D(Lset(p)) > Dmin ) then
+          nq = nq+1
+          Dset(nq) = Lset(p)
+          Dset_rev(Dset(nq)) = nq
+          LDmap(p) = nq
+          DLmap(nq) = p
+        endif
+      enddo
+
+      call resident_memory(rss)
+      rss = rss &
+            + np*memory_of_double(nq)               & ! Delta(np,nq)
+            + (rank+nq)* memory_of_double(ndim)     & ! L(ndim,rank+nq)
+            + (np+nq)*memory_of_double(block_size)    ! Ltmp_p(np,block_size)
+                                                      ! Ltmp_q(nq,block_size)
+
+      if (rss > qp_max_mem) then
+        s = s*2.d0
+      else
+        exit
       endif
+
+      if ((s > 1.d0).or.(nq == 0)) then
+        print *,  'Not enough memory. Reduce cholesky threshold'
+        stop -1
+      endif
+     
     enddo
 
     ! d., e.
@@ -170,10 +219,15 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
       stop -1
     endif
 
-    Delta(:,:) = 0.d0
 
     !$OMP PARALLEL DEFAULT(SHARED) PRIVATE(m,k,p,q,j)
 
+    !$OMP DO
+    do q=1,nq
+      Delta(:,q) = 0.d0
+    enddo
+    !$OMP ENDDO NOWAIT
+
     !$OMP DO 
     do k=1,N
       do p=1,np
@@ -183,9 +237,11 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
         Ltmp_q(q,k) = L(Dset(q),k)
       enddo
     enddo
-    !$OMP END DO
+    !$OMP END DO NOWAIT
 
-    !$OMP DO  SCHEDULE(dynamic,8)
+    !$OMP BARRIER
+
+    !$OMP DO  SCHEDULE(guided)
     do m=1,nq
 
       call omp_set_lock(lock(m))
@@ -196,8 +252,13 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
         if ((0 < q).and.(q < k)) cycle
         if (.not.ao_two_e_integral_zero( addr(1,Lset(k)), addr(1,Dset(m)), &
                                     addr(2,Lset(k)), addr(2,Dset(m)) ) ) then
-           Delta(k,m) = get_ao_two_e_integral( addr(1,Lset(k)), addr(1,Dset(m)), &
+           if (do_direct_integrals) then
+             Delta(k,m) = ao_two_e_integral(addr(1,Lset(k)), addr(2,Lset(k)), &
+                                            addr(1,Dset(m)), addr(2,Dset(m)))
+           else
+             Delta(k,m) = get_ao_two_e_integral( addr(1,Lset(k)), addr(1,Dset(m)), &
                            addr(2,Lset(k)), addr(2,Dset(m)), ao_integrals_map)
+           endif
            if (q /= 0) Delta(q,m) = Delta(k,m)
         endif
       enddo
@@ -218,8 +279,13 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
         if ((0 < q).and.(q < p)) cycle
         if (.not.ao_two_e_integral_zero( addr(1,Dset(k)), addr(1,Dset(m)), &
                                     addr(2,Dset(k)), addr(2,Dset(m)) ) ) then
-          Delta(p,m) = get_ao_two_e_integral( addr(1,Dset(k)), addr(1,Dset(m)), &
+          if (do_direct_integrals) then
+            Delta(p,m) = ao_two_e_integral(addr(1,Dset(k)), addr(2,Dset(k)), &
+                                           addr(1,Dset(m)), addr(2,Dset(m)))
+          else
+            Delta(p,m) = get_ao_two_e_integral( addr(1,Dset(k)), addr(1,Dset(m)), &
                              addr(2,Dset(k)), addr(2,Dset(m)), ao_integrals_map)
+          endif
           if (q   /= 0) Delta(q,m) = Delta(p,m)
           if (j   /= 0) Delta(p,j) = Delta(p,m)
           if (q*j /= 0) Delta(q,j) = Delta(p,m)
@@ -339,8 +405,16 @@ BEGIN_PROVIDER [ integer, cholesky_ao_num ]
     call omp_destroy_lock(lock(k))
   enddo
 
-  allocate(cholesky_ao(ao_num,ao_num,rank))
-  call dcopy(ndim*rank, L, 1, cholesky_ao, 1)
+  allocate(cholesky_ao(ao_num,ao_num,rank), stat=ierr)
+  if (ierr /= 0) then
+    print *,  irp_here, ': Allocation failed'
+    stop -1
+  endif
+  !$OMP PARALLEL DO PRIVATE(k)
+  do k=1,rank
+    call dcopy(ndim, L(1,k), 1, cholesky_ao(1,1,k), 1)
+  enddo
+  !$OMP END PARALLEL DO
   deallocate(L)
   cholesky_ao_num = rank
 
diff --git a/src/ao_two_e_ints/two_e_integrals.irp.f b/src/ao_two_e_ints/two_e_integrals.irp.f
index 85ff5bcf..f86fb269 100644
--- a/src/ao_two_e_ints/two_e_integrals.irp.f
+++ b/src/ao_two_e_ints/two_e_integrals.irp.f
@@ -460,7 +460,7 @@ BEGIN_PROVIDER [ double precision, ao_two_e_integral_schwartz, (ao_num, ao_num)
   !$OMP PARALLEL DO PRIVATE(i,k)                                     &
       !$OMP DEFAULT(NONE)                                            &
       !$OMP SHARED (ao_num,ao_two_e_integral_schwartz)              &
-      !$OMP SCHEDULE(dynamic)
+      !$OMP SCHEDULE(guided)
   do i=1,ao_num
     do k=1,i
       ao_two_e_integral_schwartz(i,k) = dsqrt(ao_two_e_integral(i,i,k,k))
@@ -975,7 +975,8 @@ recursive subroutine I_x1_pol_mult_recurs(a,c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt
 
 !  !DIR$ FORCEINLINE
 !  call multiply_poly(X,nx,B_10,2,d,nd)
-  call multiply_poly_c2(X,nx,B_10,d,nd)
+      !DIR$ FORCEINLINE
+  call multiply_poly_c2_inline_2e(X,nx,B_10,d,nd)
 
   nx = nd
   !DIR$ LOOP COUNT(8)
@@ -998,7 +999,8 @@ recursive subroutine I_x1_pol_mult_recurs(a,c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt
     endif
 !    !DIR$ FORCEINLINE
 !    call multiply_poly(X,nx,B_00,2,d,nd)
-    call multiply_poly_c2(X,nx,B_00,d,nd)
+      !DIR$ FORCEINLINE
+    call multiply_poly_c2_inline_2e(X,nx,B_00,d,nd)
   endif
 
   ny=0
@@ -1017,7 +1019,8 @@ recursive subroutine I_x1_pol_mult_recurs(a,c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt
 
 !  !DIR$ FORCEINLINE
 !  call multiply_poly(Y,ny,C_00,2,d,nd)
-  call multiply_poly_c2(Y,ny,C_00,d,nd)
+      !DIR$ FORCEINLINE
+  call multiply_poly_c2_inline_2e(Y,ny,C_00,d,nd)
 end
 
 recursive subroutine I_x1_pol_mult_a1(c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt_in)
@@ -1057,7 +1060,8 @@ recursive subroutine I_x1_pol_mult_a1(c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt_in)
 
 !  !DIR$ FORCEINLINE
 !  call multiply_poly(X,nx,B_00,2,d,nd)
-  call multiply_poly_c2(X,nx,B_00,d,nd)
+      !DIR$ FORCEINLINE
+  call multiply_poly_c2_inline_2e(X,nx,B_00,d,nd)
 
   ny=0
 
@@ -1069,7 +1073,8 @@ recursive subroutine I_x1_pol_mult_a1(c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt_in)
 
 !  !DIR$ FORCEINLINE
 !  call multiply_poly(Y,ny,C_00,2,d,nd)
-  call multiply_poly_c2(Y,ny,C_00,d,nd)
+      !DIR$ FORCEINLINE
+  call multiply_poly_c2_inline_2e(Y,ny,C_00,d,nd)
 
 end
 
@@ -1098,7 +1103,8 @@ recursive subroutine I_x1_pol_mult_a2(c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt_in)
 
 !  !DIR$ FORCEINLINE
 !  call multiply_poly(X,nx,B_10,2,d,nd)
-  call multiply_poly_c2(X,nx,B_10,d,nd)
+      !DIR$ FORCEINLINE
+  call multiply_poly_c2_inline_2e(X,nx,B_10,d,nd)
 
   nx = nd
   !DIR$ LOOP COUNT(8)
@@ -1118,7 +1124,8 @@ recursive subroutine I_x1_pol_mult_a2(c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt_in)
 
 !  !DIR$ FORCEINLINE
 !  call multiply_poly(X,nx,B_00,2,d,nd)
-  call multiply_poly_c2(X,nx,B_00,d,nd)
+      !DIR$ FORCEINLINE
+  call multiply_poly_c2_inline_2e(X,nx,B_00,d,nd)
 
   ny=0
   !DIR$ LOOP COUNT(8)
@@ -1130,7 +1137,8 @@ recursive subroutine I_x1_pol_mult_a2(c,B_10,B_01,B_00,C_00,D_00,d,nd,n_pt_in)
 
 !  !DIR$ FORCEINLINE
 !  call multiply_poly(Y,ny,C_00,2,d,nd)
-  call multiply_poly_c2(Y,ny,C_00,d,nd)
+      !DIR$ FORCEINLINE
+  call multiply_poly_c2_inline_2e(Y,ny,C_00,d,nd)
 end
 
 recursive subroutine I_x2_pol_mult(c,B_10,B_01,B_00,C_00,D_00,d,nd,dim)
@@ -1177,9 +1185,9 @@ recursive subroutine I_x2_pol_mult(c,B_10,B_01,B_00,C_00,D_00,d,nd,dim)
       Y(1) = D_00(1)
       Y(2) = D_00(2)
 
-!      !DIR$ FORCEINLINE
 !      call multiply_poly(Y,ny,D_00,2,d,nd)
-      call multiply_poly_c2(Y,ny,D_00,d,nd)
+      !DIR$ FORCEINLINE
+      call multiply_poly_c2_inline_2e(Y,ny,D_00,d,nd)
 
       return
 
@@ -1199,7 +1207,8 @@ recursive subroutine I_x2_pol_mult(c,B_10,B_01,B_00,C_00,D_00,d,nd,dim)
 
 !      !DIR$ FORCEINLINE
 !      call multiply_poly(X,nx,B_01,2,d,nd)
-      call multiply_poly_c2(X,nx,B_01,d,nd)
+      !DIR$ FORCEINLINE
+      call multiply_poly_c2_inline_2e(X,nx,B_01,d,nd)
 
       ny = 0
       !DIR$ LOOP COUNT(6)
@@ -1208,9 +1217,9 @@ recursive subroutine I_x2_pol_mult(c,B_10,B_01,B_00,C_00,D_00,d,nd,dim)
       enddo
       call I_x2_pol_mult(c-1,B_10,B_01,B_00,C_00,D_00,Y,ny,dim)
 
-!      !DIR$ FORCEINLINE
 !      call multiply_poly(Y,ny,D_00,2,d,nd)
-      call multiply_poly_c2(Y,ny,D_00,d,nd)
+      !DIR$ FORCEINLINE
+      call multiply_poly_c2_inline_2e(Y,ny,D_00,d,nd)
 
   end select
 end
@@ -1232,7 +1241,8 @@ subroutine compute_ao_integrals_jl(j,l,n_integrals,buffer_i,buffer_value)
   logical, external               :: ao_two_e_integral_zero
 
   integer                         :: i,k
-  double precision                :: ao_two_e_integral,cpu_1,cpu_2, wall_1, wall_2
+  double precision, external      :: ao_two_e_integral
+  double precision                :: cpu_1,cpu_2, wall_1, wall_2
   double precision                :: integral, wall_0
   double precision                :: thr
   integer                         :: kk, m, j1, i1
@@ -1299,3 +1309,56 @@ subroutine multiply_poly_local(b,nb,c,nc,d,nd)
 end
 
 
+!DIR$ FORCEINLINE
+subroutine multiply_poly_c2_inline_2e(b,nb,c,d,nd)
+  implicit none
+  BEGIN_DOC
+  ! Multiply two polynomials
+  ! D(t) += B(t)*C(t)
+  END_DOC
+
+  integer, intent(in)            :: nb
+  integer, intent(out)           :: nd
+  double precision, intent(in)   :: b(0:nb), c(0:2)
+  double precision, intent(inout) :: d(0:nb+2)
+
+  integer                        :: ndtmp
+  integer                        :: ib, ic, id, k
+  if(nb < 0) return !False if nb>=0
+
+  select case (nb)
+    case (0)
+      d(0) = d(0) + c(0) * b(0)
+      d(1) = d(1) + c(1) * b(0)
+      d(2) = d(2) + c(2) * b(0)
+
+    case (1)
+      d(0) = d(0) + c(0) * b(0)
+      d(1) = d(1) + c(0) * b(1) + c(1) * b(0)
+      d(2) = d(2) + c(1) * b(1) + c(2) * b(0)
+      d(3) = d(3) + c(2) * b(1)
+
+    case (2)
+      d(0) = d(0) + c(0) * b(0)
+      d(1) = d(1) + c(0) * b(1) + c(1) * b(0)
+      d(2) = d(2) + c(0) * b(2) + c(1) * b(1) + c(2) * b(0)
+      d(3) = d(3) + c(1) * b(2) + c(2) * b(1)
+      d(4) = d(4) + c(2) * b(2)
+
+    case default
+      d(0) = d(0) + c(0) * b(0)
+      d(1) = d(1) + c(0) * b(1) + c(1) * b(0)
+      do ib=2,nb
+        d(ib) = d(ib) + c(0) * b(ib) + c(1) * b(ib-1) + c(2) * b(ib-2)
+      enddo
+      d(nb+1) = d(nb+1) + c(1) * b(nb) + c(2) * b(nb-1)
+      d(nb+2) = d(nb+2) + c(2) * b(nb)
+
+  end select
+
+  do nd = nb+2,0,-1
+    if (d(nd) /= 0.d0) exit
+  enddo
+
+end
+
diff --git a/src/ccsd/ccsd_space_orb_sub.irp.f b/src/ccsd/ccsd_space_orb_sub.irp.f
index 1d77180e..76c9351e 100644
--- a/src/ccsd/ccsd_space_orb_sub.irp.f
+++ b/src/ccsd/ccsd_space_orb_sub.irp.f
@@ -85,13 +85,23 @@ subroutine run_ccsd_space_orb
 
   do while (not_converged)
 
-    call compute_H_oo(nO,nV,t1,t2,tau,H_oo)
-    call compute_H_vv(nO,nV,t1,t2,tau,H_vv)
-    call compute_H_vo(nO,nV,t1,t2,H_vo)
-
     ! Residue
-    call compute_r1_space(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r1,max_r1)
-    call compute_r2_space(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r2,max_r2)
+!    if (do_ao_cholesky) then
+    if (.False.) then
+      call compute_H_oo_chol(nO,nV,t1,t2,tau,H_oo)
+      call compute_H_vv_chol(nO,nV,t1,t2,tau,H_vv)
+      call compute_H_vo_chol(nO,nV,t1,t2,H_vo)
+
+      call compute_r1_space_chol(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r1,max_r1)
+      call compute_r2_space_chol(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r2,max_r2)
+    else
+      call compute_H_oo(nO,nV,t1,t2,tau,H_oo)
+      call compute_H_vv(nO,nV,t1,t2,tau,H_vv)
+      call compute_H_vo(nO,nV,t1,t2,H_vo)
+
+      call compute_r1_space(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r1,max_r1)
+      call compute_r2_space(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r2,max_r2)
+    endif
     max_r = max(max_r1,max_r2)
 
     ! Update
@@ -839,6 +849,10 @@ subroutine compute_r2_space(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r2,max_r2)
 
 !  allocate(B1(nV,nV,nV,nV))
 !  call compute_B1(nO,nV,t1,t2,B1)
+!  call dgemm('N','N',nO*nO,nV*nV,nV*nV, &
+!              1d0, tau, size(tau,1) * size(tau,2), &
+!                   B1 , size(B1_gam,1) * size(B1_gam,2), &
+!              1d0, r2, size(r2,1) * size(r2,2))
   allocate(B1_gam(nV,nV,nV))
   do gam=1,nV
     call compute_B1_gam(nO,nV,t1,t2,B1_gam,gam)
diff --git a/src/ccsd/ccsd_space_orb_sub_chol.irp.f b/src/ccsd/ccsd_space_orb_sub_chol.irp.f
new file mode 100644
index 00000000..190c163b
--- /dev/null
+++ b/src/ccsd/ccsd_space_orb_sub_chol.irp.f
@@ -0,0 +1,1395 @@
+subroutine ccsd_energy_space_chol(nO,nV,tau,t1,energy)
+
+  implicit none
+
+  integer, intent(in)           :: nO, nV
+  double precision, intent(in)  :: tau(nO,nO,nV,nV)
+  double precision, intent(in)  :: t1(nO,nV)
+  double precision, intent(out) :: energy
+
+  ! internal
+  integer :: i,j,a,b
+  double precision :: e
+
+  energy = 0d0
+  !$omp parallel &
+  !$omp shared(nO,nV,energy,tau,t1,&
+  !$omp cc_space_f_vo,cc_space_w_oovv) &
+  !$omp private(i,j,a,b,e) &
+  !$omp default(none)
+  e = 0d0
+  !$omp do
+  do a = 1, nV
+    do i = 1, nO
+      e = e + 2d0 * cc_space_f_vo(a,i) * t1(i,a)
+    enddo
+  enddo
+  !$omp end do nowait
+  !$omp do
+  do b = 1, nV
+    do a = 1, nV
+      do j = 1, nO
+        do i = 1, nO
+          e = e + tau(i,j,a,b) * cc_space_w_oovv(i,j,a,b)
+       enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do nowait
+  !$omp critical
+  energy = energy + e
+  !$omp end critical
+  !$omp end parallel
+
+end
+
+! Tau
+
+subroutine update_tau_space_chol(nO,nV,t1,t2,tau)
+
+  implicit none
+
+  ! in
+  integer, intent(in)           :: nO, nV
+  double precision, intent(in)  :: t1(nO,nV), t2(nO,nO,nV,nV)
+
+  ! out
+  double precision, intent(out) :: tau(nO,nO,nV,nV)
+
+  ! internal
+  integer                       :: i,j,a,b
+
+  !$OMP PARALLEL &
+  !$OMP SHARED(nO,nV,tau,t2,t1) &
+  !$OMP PRIVATE(i,j,a,b) &
+  !$OMP DEFAULT(NONE)
+  !$OMP DO
+  do b = 1, nV
+    do a = 1, nV
+      do j = 1, nO
+        do i = 1, nO
+          tau(i,j,a,b) = t2(i,j,a,b) + t1(i,a) * t1(j,b)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$OMP END DO
+  !$OMP END PARALLEL
+
+end
+
+! R1
+
+subroutine compute_r1_space_chol(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r1,max_r1)
+
+  implicit none
+
+  ! in
+  integer, intent(in)           :: nO, nV
+  double precision, intent(in)  :: t1(nO,nV), t2(nO,nO,nV,nV), tau(nO,nO,nV,nV)
+  double precision, intent(in)  :: H_oo(nO,nO), H_vv(nV,nV), H_vo(nV,nO)
+
+  ! out
+  double precision, intent(out) :: r1(nO,nV), max_r1
+
+  ! internal
+  integer                       :: u,i,j,beta,a,b
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r1,cc_space_f_ov) &
+  !$omp private(u,beta) &
+  !$omp default(none)
+  !$omp do
+  do beta = 1, nV
+    do u = 1, nO
+      r1(u,beta) = cc_space_f_ov(u,beta)
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  double precision, allocatable :: X_oo(:,:)
+  allocate(X_oo(nO,nO))
+  call dgemm('N','N', nO, nO, nV, &
+             -2d0, t1    , size(t1,1), &
+                   cc_space_f_vo, size(cc_space_f_vo,1), &
+              0d0, X_oo  , size(X_oo,1))
+
+  call dgemm('T','N', nO, nV, nO, &
+             1d0, X_oo, size(X_oo,2), &
+                  t1  , size(t1,1), &
+             1d0, r1  , size(r1,1))
+  deallocate(X_oo)
+
+  call dgemm('N','N', nO, nV, nV, &
+             1d0, t1  , size(t1,1), &
+                  H_vv, size(H_vv,1), &
+             1d0, r1  , size(r1,1))
+
+  call dgemm('N','N', nO, nV, nO, &
+             -1d0, H_oo, size(H_oo,1), &
+                   t1  , size(t1,1), &
+              1d0, r1, size(r1,1))
+
+  double precision, allocatable :: X_voov(:,:,:,:)
+  allocate(X_voov(nV, nO, nO, nV))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,X_voov,t2,t1) &
+  !$omp private(u,beta,i,a) &
+  !$omp default(none)
+  !$omp do
+  do beta = 1, nV
+    do u = 1, nO
+      do i = 1, nO
+        do a = 1, nV
+          X_voov(a,i,u,beta) = 2d0 * t2(i,u,a,beta) - t2(u,i,a,beta) + t1(u,a) * t1(i,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemv('T', nV*nO, nO*nV, &
+             1d0, X_voov, size(X_voov,1) * size(X_voov,2), &
+                  H_vo  , 1, &
+             1d0, r1    , 1)
+
+  deallocate(X_voov)
+
+  double precision, allocatable :: X_ovov(:,:,:,:)
+  allocate(X_ovov(nO, nV, nO, nV))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,cc_space_v_ovov,cc_space_v_voov,X_ovov) &
+  !$omp private(u,beta,i,a) &
+  !$omp default(none)
+  !$omp do
+  do beta = 1, nV
+    do u = 1, nO
+      do a = 1, nv
+        do i = 1, nO
+          X_ovov(i,a,u,beta) = 2d0 * cc_space_v_voov(a,u,i,beta) - cc_space_v_ovov(u,a,i,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemv('T', nO*nV, nO*nV, &
+             1d0, X_ovov, size(X_ovov,1) * size(X_ovov,2), &
+                  t1     , 1, &
+             1d0, r1     , 1)
+
+  deallocate(X_ovov)
+
+  double precision, allocatable :: W_vvov(:,:,:,:), T_vvoo(:,:,:,:)
+  allocate(W_vvov(nV,nV,nO,nV), T_vvoo(nV,nV,nO,nO))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,cc_space_v_vvov,W_vvov,T_vvoo,tau) &
+  !$omp private(b,beta,i,a) &
+  !$omp default(none)
+  !$omp do
+  do beta = 1, nV
+    do i = 1, nO
+      do b = 1, nV
+        do a = 1, nV
+          W_vvov(a,b,i,beta) = 2d0 * cc_space_v_vvov(a,b,i,beta) - cc_space_v_vvov(b,a,i,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do nowait
+
+  !$omp do
+  do u = 1, nO
+    do i = 1, nO
+      do b = 1, nV
+        do a = 1, nV
+          T_vvoo(a,b,i,u) = tau(i,u,a,b)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do nowait
+  !$omp end parallel
+
+  call dgemm('T','N',nO,nV,nO*nV*nV, &
+             1d0, T_vvoo, size(T_vvoo,1) * size(T_vvoo,2) * size(T_vvoo,3), &
+                  W_vvov, size(W_vvov,1) * size(W_vvov,2) * size(W_vvov,3), &
+             1d0, r1    , size(r1,1))
+
+  deallocate(W_vvov,T_vvoo)
+
+  double precision, allocatable :: W_oovo(:,:,:,:)
+  allocate(W_oovo(nO,nO,nV,nO))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,cc_space_v_vooo,W_oovo) &
+  !$omp private(u,a,i,j) &
+  !$omp default(none)
+  do u = 1, nO
+    !$omp do
+    do a = 1, nV
+      do j = 1, nO
+        do i = 1, nO
+          W_oovo(i,j,a,u) = 2d0 * cc_space_v_vooo(a,u,i,j) - cc_space_v_vooo(a,u,j,i)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+  !$omp end parallel
+
+  call dgemm('T','N', nO, nV, nO*nO*nV, &
+             -1d0, W_oovo, size(W_oovo,1) * size(W_oovo,2) * size(W_oovo,3), &
+                   tau   , size(tau,1) * size(tau,2) * size(tau,3), &
+              1d0, r1    , size(r1,1))
+
+  deallocate(W_oovo)
+
+  max_r1 = 0d0
+  do a = 1, nV
+    do i = 1, nO
+      max_r1 = max(dabs(r1(i,a)), max_r1)
+    enddo
+  enddo
+
+  ! Change the sign for consistency with the code in spin orbitals
+  !$omp parallel &
+  !$omp shared(nO,nV,r1) &
+  !$omp private(a,i) &
+  !$omp default(none)
+  !$omp do
+  do a = 1, nV
+    do i = 1, nO
+      r1(i,a) = -r1(i,a)
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+end
+
+! H_oo
+
+subroutine compute_H_oo_chol(nO,nV,t1,t2,tau,H_oo)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV
+  double precision, intent(in)  :: t1(nO, nV)
+  double precision, intent(in)  :: t2(nO, nO, nV, nV)
+  double precision, intent(in)  :: tau(nO, nO, nV, nV)
+  double precision, intent(out) :: H_oo(nO, nO)
+
+  integer :: a,tmp_a,k,b,l,c,d,tmp_c,tmp_d,i,j,u
+
+  ! H_oo(u,i) = cc_space_f_oo(u,i)
+  !$omp parallel &
+  !$omp shared(nO,H_oo,cc_space_f_oo) &
+  !$omp private(i,u) &
+  !$omp default(none)
+  !$omp do
+  do i = 1, nO
+    do u = 1, nO
+      H_oo(u,i) = cc_space_f_oo(u,i)
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  ! H_oo(u,i) += cc_space_w_oovv(i,j,a,b) * tau(u,j,a,b)
+  ! H_oo(u,i) += tau(u,j,a,b) * cc_space_w_oovv(i,j,a,b)
+  call dgemm('N','T', nO, nO, nO*nV*nV,       &
+             1d0, tau     , size(tau,1),      &
+                  cc_space_w_oovv, size(cc_space_w_oovv,1), &
+             1d0, H_oo    , size(H_oo,1))
+
+end
+
+! H_vv
+
+subroutine compute_H_vv_chol(nO,nV,t1,t2,tau,H_vv)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV
+  double precision, intent(in)  :: t1(nO, nV)
+  double precision, intent(in)  :: t2(nO, nO, nV, nV)
+  double precision, intent(in)  :: tau(nO, nO, nV, nV)
+  double precision, intent(out) :: H_vv(nV, nV)
+
+  integer :: a,tmp_a,b,k,l,c,d,tmp_c,tmp_d,i,j,u, beta
+
+  double precision, allocatable :: tmp_tau(:,:,:,:)
+
+  allocate(tmp_tau(nV,nO,nO,nV))
+
+  ! H_vv(a,beta) = cc_space_f_vv(a,beta)
+  !$omp parallel &
+  !$omp shared(nV,nO,H_vv,cc_space_f_vv,tmp_tau,tau) &
+  !$omp private(a,beta,i,j,b) &
+  !$omp default(none)
+  !$omp do
+  do beta = 1, nV
+    do a = 1, nV
+      H_vv(a,beta) = cc_space_f_vv(a,beta)
+    enddo
+  enddo
+  !$omp end do nowait
+
+  !$omp do
+  do beta = 1, nV
+    do j = 1, nO
+      do i = 1, nO
+        do b = 1, nV
+          tmp_tau(b,i,j,beta) = tau(i,j,beta,b)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N',nV,nV,nO*nO*nV,           &
+             -1d0, cc_space_w_vvoo, size(cc_space_w_vvoo,1), &
+                   tmp_tau , size(tmp_tau,1) * size(tmp_tau,2) * size(tmp_tau,3), &
+              1d0, H_vv    , size(H_vv,1))
+
+  deallocate(tmp_tau)
+
+end
+
+! H_vo
+
+subroutine compute_H_vo_chol(nO,nV,t1,t2,H_vo)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV
+  double precision, intent(in)  :: t1(nO, nV)
+  double precision, intent(in)  :: t2(nO, nO, nV, nV)
+  double precision, intent(out) :: H_vo(nV, nO)
+
+  integer :: a,tmp_a,b,k,l,c,d,tmp_c,tmp_d,i,j,u, beta
+
+  double precision, allocatable :: w(:,:,:,:)
+
+  allocate(w(nV,nO,nO,nV))
+
+  !$omp parallel &
+  !$omp shared(nV,nO,H_vo,cc_space_f_vo,w,cc_space_w_vvoo,t1) &
+  !$omp private(a,beta,i,j,b) &
+  !$omp default(none)
+  !$omp do
+  do i = 1, nO
+    do a = 1, nV
+      H_vo(a,i) = cc_space_f_vo(a,i)
+    enddo
+  enddo
+  !$omp end do nowait
+
+  ! H_vo(a,i) = H_vo(a,i) + cc_space_w_vvoo(a,b,i,j) * t1(j,b)
+  ! H_vo(a,i) = H_vo(a,i) + w(a,i,j,b) * t1(j,b)
+
+  !$omp do
+  do b = 1, nV
+    do j = 1, nO
+      do i = 1, nO
+        do a = 1, nV
+          w(a,i,j,b) = cc_space_w_vvoo(a,b,i,j)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemv('N',nV*nO, nO*nV, &
+             1d0, w   , size(w,1) * size(w,2), &
+                  t1  , 1, &
+             1d0, H_vo, 1)
+
+  deallocate(w)
+
+end
+
+! R2
+
+subroutine compute_r2_space_chol(nO,nV,t1,t2,tau,H_oo,H_vv,H_vo,r2,max_r2)
+
+  implicit none
+
+  ! in
+  integer, intent(in)           :: nO, nV
+  double precision, intent(in)  :: t1(nO,nV), t2(nO,nO,nV,nV), tau(nO,nO,nV,nV)
+  double precision, intent(in)  :: H_oo(nO,nO), H_vv(nV,nV), H_vo(nV,nO)
+
+  ! out
+  double precision, intent(out) :: r2(nO,nO,nV,nV), max_r2
+
+  ! internal
+  double precision, allocatable :: g_occ(:,:), g_vir(:,:), J1(:,:,:,:), K1(:,:,:,:)
+  double precision, allocatable :: A1(:,:,:,:)
+  integer                       :: u,v,i,j,beta,gam,a,b
+
+  allocate(g_occ(nO,nO), g_vir(nV,nV))
+  allocate(J1(nO,nV,nV,nO), K1(nO,nV,nO,nV))
+  allocate(A1(nO,nO,nO,nO))
+
+  call compute_g_occ_chol(nO,nV,t1,t2,H_oo,g_occ)
+  call compute_g_vir_chol(nO,nV,t1,t2,H_vv,g_vir)
+  call compute_A1_chol(nO,nV,t1,t2,tau,A1)
+  call compute_J1_chol(nO,nV,t1,t2,cc_space_v_ovvo,cc_space_v_ovoo, &
+       cc_space_v_vvvo,cc_space_v_vvoo,J1)
+  call compute_K1_chol(nO,nV,t1,t2,cc_space_v_ovoo,cc_space_v_vvoo, &
+       cc_space_v_ovov,cc_space_v_vvov,K1)
+
+  ! Residual
+  !r2 = 0d0
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,cc_space_v_oovv) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+         r2(u,v,beta,gam) = cc_space_v_oovv(u,v,beta,gam)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N',nO*nO,nV*nV,nO*nO, &
+             1d0, A1, size(A1,1) * size(A1,2), &
+                  tau, size(tau,1) * size(tau,2), &
+             1d0, r2, size(r2,1) * size(r2,2))
+
+  double precision, dimension(:,:,:,:), allocatable :: r2_chem, tmp, tau_chem
+  double precision, dimension(:,:,:,:), allocatable :: B1
+
+  allocate(B1(nV,nV,nV,nV))
+  call compute_B1_chol(nO,nV,t1,B1,cholesky_ao_num)
+  call dgemm('N','N',nO*nO,nV*nV,nV*nV, &
+              1d0, tau, size(tau,1) * size(tau,2), &
+                   B1 , size(B1 ,1) * size(B1 ,2), &
+              1d0, r2,  size(r2 ,1) * size(r2 ,2))
+
+  double precision, allocatable :: X_oovv(:,:,:,:),Y_oovv(:,:,:,:)
+  allocate(X_oovv(nO,nO,nV,nV),Y_oovv(nO,nO,nV,nV))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,t2,X_oovv) &
+  !$omp private(u,v,gam,a) &
+  !$omp default(none)
+  !$omp do
+  do a = 1, nV
+    do gam = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+          X_oovv(u,v,gam,a) = t2(u,v,gam,a)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N',nO*nO*nV,nV,nV, &
+             1d0, X_oovv, size(X_oovv,1) * size(X_oovv,2) * size(X_oovv,3), &
+                  g_vir, size(g_vir,1), &
+             0d0, Y_oovv, size(Y_oovv,1) * size(Y_oovv,2) * size(Y_oovv,3))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,Y_oovv) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+          r2(u,v,beta,gam) = r2(u,v,beta,gam) + Y_oovv(u,v,beta,gam) + Y_oovv(v,u,gam,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N',nO,nO*nV*nV,nO, &
+             1d0, g_occ , size(g_occ,1), &
+                  t2    , size(t2,1), &
+             0d0, X_oovv, size(X_oovv,1))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,X_oovv) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+          r2(u,v,beta,gam) = r2(u,v,beta,gam) - X_oovv(u,v,beta,gam) - X_oovv(v,u,gam,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  deallocate(X_oovv)
+
+  double precision, allocatable :: X_vovv(:,:,:,:)
+  allocate(X_vovv(nV,nO,nV,nV))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,X_vovv,cc_space_v_ovvv) &
+  !$omp private(u,a,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do u = 1, nO
+        do a = 1, nV
+          X_vovv(a,u,beta,gam) = cc_space_v_ovvv(u,a,beta,gam)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N',nO,nO*nV*nV,nV, &
+             1d0, t1    , size(t1,1), &
+                  X_vovv, size(X_vovv,1), &
+             0d0, Y_oovv, size(Y_oovv,1))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,Y_oovv) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+          r2(u,v,beta,gam) = r2(u,v,beta,gam) + Y_oovv(v,u,beta,gam) + Y_oovv(u,v,gam,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  double precision, allocatable :: X_vovo(:,:,:,:), Y_vovv(:,:,:,:)
+  allocate(X_vovo(nV,nO,nV,nO), Y_vovv(nV,nO,nV,nV),X_oovv(nO,nO,nV,nV))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,X_vovo,cc_space_v_ovov) &
+  !$omp private(u,v,gam,i) &
+  !$omp default(none)
+  do i = 1, nO
+    !$omp do
+    do gam = 1, nV
+      do u = 1, nO
+        do a = 1, nV
+          X_vovo(a,u,gam,i) = cc_space_v_ovov(u,a,i,gam)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+  !$omp end parallel
+
+  call dgemm('N','N',nV*nO*nV,nV,nO, &
+              1d0, X_vovo, size(X_vovo,1) * size(X_vovo,2) * size(X_vovo,3), &
+                   t1    , size(t1,1), &
+              0d0, Y_vovv, size(Y_vovv,1) * size(Y_vovv,2) * size(Y_vovv,3))
+
+  call dgemm('N','N',nO,nO*nV*nV,nV, &
+             1d0, t1, size(t1,1), &
+                  Y_vovv, size(Y_vovv,1), &
+             0d0, X_oovv, size(X_oovv,1))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,X_oovv) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+          r2(u,v,beta,gam) = r2(u,v,beta,gam) - X_oovv(v,u,gam,beta) - X_oovv(u,v,beta,gam)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  deallocate(X_vovo,Y_vovv)
+
+  call dgemm('N','N',nO*nO*nV,nV,nO, &
+             1d0, cc_space_v_oovo, size(cc_space_v_oovo,1) * size(cc_space_v_oovo,2) * size(cc_space_v_oovo,3), &
+                  t1 , size(t1,1), &
+             0d0, X_oovv, size(X_oovv,1) * size(X_oovv,2) * size(X_oovv,3))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,X_oovv) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+           r2(u,v,beta,gam) = r2(u,v,beta,gam) - X_oovv(u,v,beta,gam) - X_oovv(v,u,gam,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  double precision, allocatable :: Y_oovo(:,:,:,:)
+  allocate(X_vovo(nV,nO,nV,nO), Y_oovo(nO,nO,nV,nO))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,X_vovo,cc_space_v_ovvo) &
+  !$omp private(a,v,gam,i) &
+  !$omp default(none)
+  do i = 1, nO
+    !$omp do
+    do gam = 1, nV
+      do v = 1, nO
+        do a = 1, nV
+          X_vovo(a,v,gam,i) = cc_space_v_ovvo(v,a,gam,i)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+  !$omp end parallel
+
+  call dgemm('N','N',nO,nO*nV*nO,nV, &
+             1d0, t1, size(t1,1), &
+                  X_vovo, size(X_vovo,1), &
+             0d0, Y_oovo, size(Y_oovo,1))
+
+  call dgemm('N','N',nO*nO*nV, nV, nO, &
+             1d0, Y_oovo, size(Y_oovo,1) * size(Y_oovo,2) * size(Y_oovo,3), &
+                  t1    , size(t1,1), &
+             0d0, X_oovv, size(X_oovv,1) * size(X_oovv,2) * size(X_oovv,3))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,X_oovv) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+           r2(u,v,beta,gam) = r2(u,v,beta,gam) - X_oovv(u,v,gam,beta) - X_oovv(v,u,beta,gam)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  deallocate(X_vovo,Y_oovo)
+
+  double precision, allocatable :: X_ovvo(:,:,:,:), Y_voov(:,:,:,:), Z_ovov(:,:,:,:)
+  allocate(X_ovvo(nO,nV,nV,nO), Y_voov(nV,nO,nO,nV),Z_ovov(nO,nV,nO,nV))
+  !$omp parallel &
+  !$omp shared(nO,nV,X_ovvo,Y_voov,K1,J1,t2) &
+  !$omp private(u,v,gam,beta,i,a) &
+  !$omp default(none)
+  do i = 1, nO
+    !$omp do
+    do a = 1, nV
+      do beta = 1, nV
+        do u = 1, nO
+          X_ovvo(u,beta,a,i) = (J1(u,a,beta,i) - 0.5d0 * K1(u,a,i,beta))
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+
+  !$omp do
+  do gam = 1, nV
+    do v = 1, nO
+      do i = 1, nO
+        do a = 1, nV
+          Y_voov(a,i,v,gam) = 2d0 * t2(i,v,a,gam) - t2(i,v,gam,a)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N', nO*nV,nO*nV,nV*nO, &
+             1d0, X_ovvo, size(X_ovvo,1) * size(X_ovvo,2), &
+                  Y_voov, size(Y_voov,1) * size(Y_voov,2), &
+             0d0, Z_ovov, size(Z_ovov,1) * size(Z_ovov,2))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,Z_ovov) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+           r2(u,v,beta,gam) = r2(u,v,beta,gam) + Z_ovov(u,beta,v,gam) + Z_ovov(v,gam,u,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  deallocate(X_ovvo,Y_voov)
+
+  double precision, allocatable :: X_ovov(:,:,:,:),Y_ovov(:,:,:,:)
+  allocate(X_ovov(nO,nV,nO,nV),Y_ovov(nO,nV,nO,nV))
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,K1,X_ovov,Y_ovov,t2) &
+  !$omp private(u,a,i,beta,gam) &
+  !$omp default(none)
+  !$omp do
+  do beta = 1, nV
+    do u = 1, nO
+      do a = 1, nV
+        do i = 1, nO
+          X_ovov(i,a,u,beta) = 0.5d0 * K1(u,a,i,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do nowait
+
+  !$omp do
+  do gam = 1, nV
+    do v = 1, nO
+      do a = 1, nV
+        do i = 1, nO
+          Y_ovov(i,a,v,gam) = t2(i,v,gam,a)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('T','N',nO*nV,nO*nV,nO*nV, &
+             1d0, X_ovov, size(X_ovov,1) * size(X_ovov,2), &
+                  Y_ovov, size(Y_ovov,1) * size(Y_ovov,2), &
+             0d0, Z_ovov, size(Y_ovov,1) * size(Y_ovov,2))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,Z_ovov) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+           r2(u,v,beta,gam) = r2(u,v,beta,gam) - Z_ovov(u,beta,v,gam) - Z_ovov(v,gam,u,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  !$omp parallel &
+  !$omp shared(nO,nV,K1,X_ovov,Z_ovov,t2) &
+  !$omp private(u,v,gam,beta,i,a) &
+  !$omp default(none)
+  !$omp do
+  do a = 1, nV
+    do i = 1, nO
+      do gam = 1, nV
+        do u = 1, nO
+          X_ovov(u,gam,i,a) = K1(u,a,i,gam)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do nowait
+
+  !$omp do
+  do beta = 1, nV
+    do v = 1, nO
+      do a = 1, nV
+        do i = 1, nO
+          Z_ovov(i,a,v,beta) = t2(i,v,beta,a)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N',nO*nV,nO*nV,nO*nV, &
+             1d0, X_ovov, size(X_ovov,1) * size(X_ovov,2), &
+                  Y_ovov, size(Y_ovov,1) * size(Y_ovov,2), &
+             0d0, Z_ovov, size(Y_ovov,1) * size(Y_ovov,2))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,r2,Z_ovov) &
+  !$omp private(u,v,gam,beta) &
+  !$omp default(none)
+  !$omp do
+  do gam = 1, nV
+    do beta = 1, nV
+      do v = 1, nO
+        do u = 1, nO
+           r2(u,v,beta,gam) = r2(u,v,beta,gam) - Z_ovov(u,gam,v,beta) - Z_ovov(v,beta,u,gam)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  deallocate(X_ovov,Y_ovov,Z_ovov)
+
+  ! Change the sign for consistency with the code in spin orbitals
+  !$omp parallel &
+  !$omp shared(nO,nV,r2) &
+  !$omp private(i,j,a,b) &
+  !$omp default(none)
+  !$omp do
+  do b = 1, nV
+    do a = 1, nV
+      do j = 1, nO
+        do i = 1, nO
+          r2(i,j,a,b) = -r2(i,j,a,b)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  max_r2 = 0d0
+  do b = 1, nV
+    do a = 1, nV
+      do j = 1, nO
+        do i = 1, nO
+          max_r2 = max(r2(i,j,a,b), max_r2)
+        enddo
+      enddo
+    enddo
+  enddo
+
+  deallocate(g_occ,g_vir,J1,K1,A1)
+
+end
+
+! A1
+
+subroutine compute_A1_chol(nO,nV,t1,t2,tau,A1)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV
+  double precision, intent(in)  :: t1(nO, nV)
+  double precision, intent(in)  :: t2(nO, nO, nV, nV)
+  double precision, intent(in)  :: tau(nO, nO, nV, nV)
+  double precision, intent(out) :: A1(nO, nO, nO, nO)
+
+  integer :: a,tmp_a,b,k,l,c,d,tmp_c,tmp_d,i,j,u,v, beta
+
+  double precision, allocatable :: X_vooo(:,:,:,:), Y_oooo(:,:,:,:)
+  allocate(X_vooo(nV,nO,nO,nO), Y_oooo(nO,nO,nO,nO))
+
+  ! A1(u,v,i,j) = cc_space_v_oooo(u,v,i,j)
+  !$omp parallel &
+  !$omp shared(nO,nV,A1,cc_space_v_oooo,cc_space_v_ovoo,X_vooo) &
+  !$omp private(u,v,i,j) &
+  !$omp default(none)
+  !$omp do collapse(2)
+  do j = 1, nO
+    do i = 1, nO
+      do v = 1, nO
+        do u = 1, nO
+          A1(u,v,i,j) = cc_space_v_oooo(u,v,i,j)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do nowait
+
+  ! A1(u,v,i,j) += cc_space_v_ovoo(u,a,i,j) * t1(v,a) &
+
+  !$omp do collapse(2)
+  do j = 1, nO
+    do i = 1, nO
+      do u = 1, nO
+        do a = 1, nV
+          X_vooo(a,u,i,j) = cc_space_v_ovoo(u,a,i,j)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N', nO, nO*nO*nO, nV, &
+             1d0, t1    , size(t1,1), &
+                  X_vooo, size(X_vooo,1), &
+             0d0, Y_oooo, size(Y_oooo,1))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,A1,Y_oooo) &
+  !$omp private(u,v,i,j) &
+  !$omp default(none)
+  !$omp do collapse(2)
+  do j = 1, nO
+    do i = 1, nO
+      do v = 1, nO
+        do u = 1, nO
+          A1(u,v,i,j) = A1(u,v,i,j) + Y_oooo(v,u,i,j)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  deallocate(X_vooo,Y_oooo)
+
+  ! A1(u,v,i,j) += cc_space_v_vooo(a,v,i,j) * t1(u,a)
+  call dgemm('N','N', nO, nO*nO*nO, nV, &
+             1d0, t1      , size(t1,1), &
+                  cc_space_v_vooo, size(cc_space_v_vooo,1), &
+             1d0, A1      , size(A1,1))
+
+  ! A1(u,v,i,j) += cc_space_v_vvoo(a,b,i,j) * tau(u,v,a,b)
+  call dgemm('N','N', nO*nO, nO*nO, nV*nV, &
+             1d0, tau     , size(tau,1) * size(tau,2), &
+                  cc_space_v_vvoo, size(cc_space_v_vvoo,1) * size(cc_space_v_vvoo,2), &
+             1d0, A1      , size(A1,1) * size(A1,2))
+
+end
+
+! B1
+subroutine compute_B1_chol(nO,nV,t1,B1,ldb)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV,ldb
+  double precision, intent(in)  :: t1(nO, nV)
+  double precision, intent(out) :: B1(nV, nV, nV, nV)
+
+  integer :: a,tmp_a,b,k,l,c,d,tmp_c,tmp_d,i,j,u,v, beta, gam
+
+  do gam = 1, nV
+    do beta = 1, nV
+      do b = 1, nV
+        do a = 1, nV
+          B1(a,b,beta,gam) = cc_space_v_vvvv(a,b,beta,gam)
+
+          do i = 1, nO
+            B1(a,b,beta,gam) = B1(a,b,beta,gam) &
+            - cc_space_v_vvvo(a,b,beta,i) * t1(i,gam) &
+            - cc_space_v_vvov(a,b,i,gam) * t1(i,beta)
+          enddo
+  
+        enddo
+      enddo
+    enddo
+  enddo
+
+end
+
+! g_occ
+
+subroutine compute_g_occ_chol(nO,nV,t1,t2,H_oo,g_occ)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV
+  double precision, intent(in)  :: t1(nO, nV), H_oo(nO, nO)
+  double precision, intent(in)  :: t2(nO, nO, nV, nV)
+  double precision, intent(out) :: g_occ(nO, nO)
+
+  integer :: a,tmp_a,b,k,l,c,d,tmp_c,tmp_d,i,j,u,v, beta, gam
+
+  call dgemm('N','N',nO,nO,nV, &
+             1d0, t1, size(t1,1), &
+                  cc_space_f_vo, size(cc_space_f_vo,1), &
+             0d0, g_occ, size(g_occ,1))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,g_occ,H_oo, cc_space_v_ovoo,t1) &
+  !$omp private(i,j,a,u) &
+  !$omp default(none)
+  !$omp do
+  do i = 1, nO
+    do u = 1, nO
+      g_occ(u,i) = g_occ(u,i) + H_oo(u,i)
+    enddo
+  enddo
+  !$omp end do
+
+  !$omp do
+  do i = 1, nO
+    do j = 1, nO
+      do a = 1, nV
+        do u = 1, nO
+          g_occ(u,i) = g_occ(u,i) + (2d0 * cc_space_v_ovoo(u,a,i,j) - cc_space_v_ovoo(u,a,j,i)) * t1(j,a)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+end
+
+! g_vir
+
+subroutine compute_g_vir_chol(nO,nV,t1,t2,H_vv,g_vir)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV
+  double precision, intent(in)  :: t1(nO, nV), H_vv(nV, nV)
+  double precision, intent(in)  :: t2(nO, nO, nV, nV)
+  double precision, intent(out) :: g_vir(nV, nV)
+
+  integer :: a,tmp_a,b,k,l,c,d,tmp_c,tmp_d,i,j,u,v, beta, gam
+
+  call dgemm('N','N',nV,nV,nO, &
+             -1d0, cc_space_f_vo , size(cc_space_f_vo,1), &
+                   t1   , size(t1,1), &
+              0d0, g_vir, size(g_vir,1))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,g_vir,H_vv, cc_space_v_vvvo,t1) &
+  !$omp private(i,b,a,beta) &
+  !$omp default(none)
+  !$omp do
+  do beta = 1, nV
+    do a = 1, nV
+      g_vir(a,beta) = g_vir(a,beta) + H_vv(a,beta)
+    enddo
+  enddo
+  !$omp end do
+
+  !$omp do
+  do beta = 1, nV
+    do i = 1, nO
+      do b = 1, nV
+        do a = 1, nV
+          g_vir(a,beta) = g_vir(a,beta) + (2d0 * cc_space_v_vvvo(a,b,beta,i) - cc_space_v_vvvo(b,a,beta,i)) * t1(i,b)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+end
+
+! J1
+
+subroutine compute_J1_chol(nO,nV,t1,t2,v_ovvo,v_ovoo,v_vvvo,v_vvoo,J1)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV
+  double precision, intent(in)  :: t1(nO, nV)
+  double precision, intent(in)  :: t2(nO, nO, nV, nV)
+  double precision, intent(in)  :: v_ovvo(nO,nV,nV,nO), v_ovoo(nO,nV,nO,nO)
+  double precision, intent(in)  :: v_vvvo(nV,nV,nV,nO), v_vvoo(nV,nV,nO,nO)
+  double precision, intent(out) :: J1(nO, nV, nV, nO)
+
+  integer :: a,tmp_a,b,k,l,c,d,tmp_c,tmp_d,i,j,u,v, beta, gam
+
+  double precision, allocatable :: X_ovoo(:,:,:,:), Y_ovov(:,:,:,:)
+  allocate(X_ovoo(nO,nV,nO,nO),Y_ovov(nO,nV,nO,nV))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,J1,v_ovvo,v_ovoo,X_ovoo) &
+  !$omp private(i,j,a,u,beta) &
+  !$omp default(none)
+  do i = 1, nO
+    !$omp do
+    do beta = 1, nV
+      do a = 1, nV
+        do u = 1, nO
+          J1(u,a,beta,i) = v_ovvo(u,a,beta,i)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+
+  !$omp do collapse(2)
+  do j = 1, nO
+    do i = 1, nO
+      do a = 1, nV
+        do u = 1, nO
+          X_ovoo(u,a,i,j) = v_ovoo(u,a,j,i)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','N',nO*nV*nO,nV,nO, &
+            -1d0, X_ovoo, size(X_ovoo,1) * size(X_ovoo,2) * size(X_ovoo,3), &
+                  t1    , size(t1,1), &
+             0d0, Y_ovov, size(Y_ovov,1) * size(Y_ovov,2) * size(Y_ovov,3))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,J1,Y_ovov) &
+  !$omp private(i,beta,a,u) &
+  !$omp default(none)
+  do i = 1, nO
+    !$omp do
+    do beta = 1, nV
+      do a = 1, nV
+        do u = 1, nO
+          J1(u,a,beta,i) = J1(u,a,beta,i) + Y_ovov(u,a,i,beta)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+  !$omp end parallel
+  deallocate(X_ovoo)
+
+  ! v_vvvo(b,a,beta,i) * t1(u,b)
+  call dgemm('N','N',nO,nV*nV*nO,nV, &
+             1d0, t1    , size(t1,1), &
+                  v_vvvo, size(v_vvvo,1), &
+             1d0, J1    , size(J1,1))
+
+  !- cc_space_v_vvoo(a,b,i,j) * (0.5d0 * t2(u,j,b,beta) + t1(u,b) * t1(j,beta)) &
+  double precision, allocatable :: X_voov(:,:,:,:), Z_ovvo(:,:,:,:)
+  allocate(X_voov(nV,nO,nO,nV), Z_ovvo(nO,nV,nV,nO))
+  !$omp parallel &
+  !$omp shared(nO,nV,t2,t1,Y_ovov,X_voov,v_vvoo) &
+  !$omp private(i,beta,a,u,b,j) &
+  !$omp default(none)
+  !$omp do
+  do b = 1, nV
+    do j = 1, nO
+      do beta = 1, nV
+        do u = 1, nO
+          Y_ovov(u,beta,j,b) = 0.5d0 * t2(u,j,b,beta) + t1(u,b) * t1(j,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do nowait
+
+  !$omp do
+  do b = 1, nV
+    do j = 1, nO
+      do i = 1, nO
+        do a = 1, nV
+          X_voov(a,i,j,b) = v_vvoo(a,b,i,j)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  call dgemm('N','T',nO*nV,nV*nO,nO*nV, &
+             -1d0, Y_ovov, size(Y_ovov,1) * size(Y_ovov,2), &
+                   X_voov, size(X_voov,1) * size(X_voov,2), &
+              0d0, Z_ovvo, size(Z_ovvo,1) * size(Z_ovvo,2))
+  deallocate(X_voov)
+
+  double precision, allocatable :: X_ovvo(:,:,:,:), Y_vovo(:,:,:,:)
+  allocate(X_ovvo(nO,nV,nV,nO),Y_vovo(nV,nO,nV,nO))
+  !$omp parallel &
+  !$omp shared(nO,nV,J1,Z_ovvo,t2,Y_vovo,v_vvoo,X_ovvo) &
+  !$omp private(i,beta,a,u,j,b) &
+  !$omp default(none)
+  do i = 1, nO
+    !$omp do
+    do beta = 1, nV
+      do a = 1, nV
+        do u = 1, nO
+          J1(u,a,beta,i) = J1(u,a,beta,i) + Z_ovvo(u,beta,a,i)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+
+  !+ 0.5d0 * (2d0 * cc_space_v_vvoo(a,b,i,j) - cc_space_v_vvoo(b,a,i,j)) * t2(u,j,beta,b)
+  do j = 1, nO
+    !$omp do
+    do b = 1, nV
+      do i = 1, nO
+        do a = 1, nV
+          Y_vovo(a,i,b,j) = 0.5d0 * (2d0 * v_vvoo(a,b,i,j) - v_vvoo(b,a,i,j))
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+
+  do j = 1, nO
+    !$omp do
+    do b = 1, nV
+      do beta = 1, nV
+        do u = 1, nO
+          X_ovvo(u,beta,b,j) = t2(u,j,beta,b)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+  !$omp end parallel
+
+  call dgemm('N','T',nO*nV,nV*nO,nV*nO, &
+             1d0, X_ovvo, size(X_ovvo,1) * size(X_ovvo,2), &
+                  Y_vovo, size(Y_vovo,1) * size(Y_vovo,2), &
+             0d0, Z_ovvo, size(Z_ovvo,1) * size(Z_ovvo,2))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,J1,Z_ovvo) &
+  !$omp private(i,beta,a,u) &
+  !$omp default(none)
+  do i = 1, nO
+    !$omp do
+    do beta = 1, nV
+      do a = 1, nV
+        do u = 1, nO
+          J1(u,a,beta,i) = J1(u,a,beta,i) + Z_ovvo(u,beta,a,i)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+  !$omp end parallel
+
+  deallocate(X_ovvo,Z_ovvo,Y_ovov)
+
+end
+
+! K1
+
+subroutine compute_K1_chol(nO,nV,t1,t2,v_ovoo,v_vvoo,v_ovov,v_vvov,K1)
+
+  implicit none
+
+  integer, intent(in)           :: nO,nV
+  double precision, intent(in)  :: t1(nO, nV)
+  double precision, intent(in)  :: t2(nO, nO, nV, nV)
+  double precision, intent(in)  :: v_vvoo(nV,nV,nO,nO), v_ovov(nO,nV,nO,nV)
+  double precision, intent(in)  :: v_vvov(nV,nV,nO,nV), v_ovoo(nO,nV,nO,nO)
+  double precision, intent(out) :: K1(nO, nV, nO, nV)
+
+  double precision, allocatable :: X(:,:,:,:), Y(:,:,:,:), Z(:,:,:,:)
+
+  integer :: a,tmp_a,b,k,l,c,d,tmp_c,tmp_d,i,j,u,v, beta, gam
+
+  allocate(X(nV,nO,nV,nO),Y(nO,nV,nV,nO),Z(nO,nV,nV,nO))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,K1,X,Y,v_vvoo,v_ovov,t1,t2) &
+  !$omp private(i,beta,a,u,j,b) &
+  !$omp default(none)
+  !$omp do
+  do beta = 1, nV
+    do i = 1, nO
+      do a = 1, nV
+        do u = 1, nO
+          K1(u,a,i,beta) = v_ovov(u,a,i,beta)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do nowait
+
+  do i = 1, nO
+    !$omp do
+    do a = 1, nV
+      do j = 1, nO
+        do b = 1, nV
+          X(b,j,a,i) = - v_vvoo(b,a,i,j)
+        enddo
+      enddo
+    enddo
+    !$omp end do nowait
+  enddo
+
+  do j = 1, nO
+    !$omp do
+    do b = 1, nV
+      do beta = 1, nV
+        do u = 1, nO
+          Y(u,beta,b,j) = 0.5d0 * t2(u,j,b,beta) + t1(u,b) * t1(j,beta)
+        enddo
+      enddo
+    enddo
+    !$omp end do
+  enddo
+  !$omp end parallel
+
+  call dgemm('N','N',nO*nV*nO,nV,nO, &
+            -1d0, v_ovoo, size(v_ovoo,1) * size(v_ovoo,2) * size(v_ovoo,3), &
+                  t1    , size(t1,1), &
+            1d0, K1    , size(K1,1) * size(K1,2) * size(K1,3))
+
+  call dgemm('N','N',nO,nV*nO*nV,nV, &
+             1d0, t1    , size(t1,1), &
+                  v_vvov, size(v_vvov,1), &
+             1d0, K1    , size(K1,1))
+
+  ! Y(u,beta,b,j) * X(b,j,a,i) = Z(u,beta,a,i)
+  call dgemm('N','N',nV*nO,nO*nV,nV*nO, &
+             1d0, Y, size(Y,1) * size(Y,2), &
+                  X, size(X,1) * size(X,2), &
+             0d0, Z, size(Z,1) * size(Z,2))
+
+  !$omp parallel &
+  !$omp shared(nO,nV,K1,Z) &
+  !$omp private(i,beta,a,u) &
+  !$omp default(none)
+  !$omp do
+   do beta = 1, nV
+    do i = 1, nO
+      do a = 1, nV
+        do u = 1, nO
+          K1(u,a,i,beta) = K1(u,a,i,beta) + Z(u,beta,a,i)
+        enddo
+      enddo
+    enddo
+  enddo
+  !$omp end do
+  !$omp end parallel
+
+  deallocate(X,Y,Z)
+
+end
diff --git a/src/ccsd/ccsd_t_space_orb_abc.irp.f b/src/ccsd/ccsd_t_space_orb_abc.irp.f
index 1aab6bd7..12a71045 100644
--- a/src/ccsd/ccsd_t_space_orb_abc.irp.f
+++ b/src/ccsd/ccsd_t_space_orb_abc.irp.f
@@ -101,7 +101,7 @@ subroutine ccsd_par_t_space_v3(nO,nV,t1,t2,f_o,f_v,v_vvvo,v_vvoo,v_vooo,energy)
 
   !$OMP PARALLEL PRIVATE(a,b,c,e) DEFAULT(SHARED)
   e = 0d0
-  !$OMP DO SCHEDULE(dynamic)
+  !$OMP DO SCHEDULE(guided)
   do a = 1, nV
     do b = a+1, nV
       do c = b+1, nV
diff --git a/src/determinants/density_matrix.irp.f b/src/determinants/density_matrix.irp.f
index 1a1d92b5..ce4d96c2 100644
--- a/src/determinants/density_matrix.irp.f
+++ b/src/determinants/density_matrix.irp.f
@@ -117,7 +117,7 @@ END_PROVIDER
       !$OMP  N_det_alpha_unique,N_det_beta_unique,irp_here)
   allocate(tmp_a(mo_num,mo_num,N_states), tmp_b(mo_num,mo_num,N_states) )
   tmp_a = 0.d0
-  !$OMP DO SCHEDULE(dynamic,64)
+  !$OMP DO SCHEDULE(guided)
   do k_a=1,N_det
     krow = psi_bilinear_matrix_rows(k_a)
     ASSERT (krow <= N_det_alpha_unique)
@@ -173,7 +173,7 @@ END_PROVIDER
   deallocate(tmp_a)
 
   tmp_b = 0.d0
-  !$OMP DO SCHEDULE(dynamic,64)
+  !$OMP DO SCHEDULE(guided)
   do k_b=1,N_det
     krow = psi_bilinear_matrix_transp_rows(k_b)
     ASSERT (krow <= N_det_alpha_unique)
diff --git a/src/determinants/h_apply.irp.f b/src/determinants/h_apply.irp.f
index 078c2104..65f1a832 100644
--- a/src/determinants/h_apply.irp.f
+++ b/src/determinants/h_apply.irp.f
@@ -250,7 +250,7 @@ subroutine remove_duplicates_in_psi_det(found_duplicates)
   enddo
   !$OMP END DO
 
-  !$OMP DO schedule(dynamic,1024)
+  !$OMP DO schedule(guided,64)
   do i=1,N_det-1
     if (duplicate(i)) then
       cycle
diff --git a/src/determinants/s2.irp.f b/src/determinants/s2.irp.f
index 2c1a8757..6dc49526 100644
--- a/src/determinants/s2.irp.f
+++ b/src/determinants/s2.irp.f
@@ -317,7 +317,7 @@ subroutine get_uJ_s2_uI(psi_keys_tmp,psi_coefs_tmp,n,nmax_coefs,nmax_keys,s2,nst
           !$OMP SHARED (ll,jj,psi_keys_tmp,psi_coefs_tmp,N_int,n,nstates)&
           !$OMP REDUCTION(+:accu)
       allocate(idx(0:n))
-      !$OMP DO SCHEDULE(dynamic)
+      !$OMP DO SCHEDULE(guided)
       do i = n,1,-1   ! Better OMP scheduling
         call get_s2(psi_keys_tmp(1,1,i),psi_keys_tmp(1,1,i),N_int,s2_tmp)
         accu += psi_coefs_tmp(i,ll) * s2_tmp * psi_coefs_tmp(i,jj)
diff --git a/src/hartree_fock/fock_matrix_hf.irp.f b/src/hartree_fock/fock_matrix_hf.irp.f
index 8c6658c5..a5ab6a60 100644
--- a/src/hartree_fock/fock_matrix_hf.irp.f
+++ b/src/hartree_fock/fock_matrix_hf.irp.f
@@ -190,47 +190,75 @@ END_PROVIDER
 
  deallocate(X)
 
- ao_two_e_integral_beta_chol = ao_two_e_integral_alpha_chol
+ if (elec_alpha_num > elec_beta_num) then
+   ao_two_e_integral_beta_chol = ao_two_e_integral_alpha_chol
+ endif
 
 
- allocate(X2(ao_num,ao_num,cholesky_ao_num,2))
+ double precision :: rss
+ double precision :: memory_of_double
 
+ integer :: iblock
+ integer, parameter :: block_size = 32
+
+ rss = memory_of_double(ao_num*ao_num)
+ call check_mem(2.d0*block_size*rss, irp_here)
+ allocate(X2(ao_num,ao_num,block_size,2))
+ allocate(X3(ao_num,block_size,ao_num,2))
+    
 ! ao_two_e_integral_alpha_chol (l,s) -= cholesky_ao(l,m,j) * SCF_density_matrix_ao_beta (m,n) * cholesky_ao(n,s,j)
 
- call dgemm('N','N',ao_num,ao_num*cholesky_ao_num,ao_num, 1.d0,      &
-     SCF_density_matrix_ao_alpha, ao_num,                            &
-     cholesky_ao, ao_num, 0.d0,                                      &
-     X2(1,1,1,1), ao_num)
+ do iblock=1,cholesky_ao_num,block_size
 
- call dgemm('N','N',ao_num,ao_num*cholesky_ao_num,ao_num, 1.d0,      &
-     SCF_density_matrix_ao_beta, ao_num,                             &
-     cholesky_ao, ao_num, 0.d0,                                      &
-     X2(1,1,1,2), ao_num)
+   call dgemm('N','N',ao_num,ao_num*min(cholesky_ao_num-iblock+1,block_size),ao_num, 1.d0,      &
+       SCF_density_matrix_ao_alpha, ao_num,    &
+       cholesky_ao(1,1,iblock), ao_num, 0.d0,  &
+       X2(1,1,1,1), ao_num)
 
- allocate(X3(ao_num,cholesky_ao_num,ao_num,2))
+   if (elec_alpha_num > elec_beta_num) then
+     call dgemm('N','N',ao_num,ao_num*min(cholesky_ao_num-iblock+1,block_size),ao_num, 1.d0,      &
+       SCF_density_matrix_ao_beta, ao_num,     &
+       cholesky_ao(1,1,iblock), ao_num, 0.d0,  &
+       X2(1,1,1,2), ao_num)
+
+     do s=1,ao_num
+      do j=1,min(cholesky_ao_num-iblock+1,block_size)
+       do m=1,ao_num
+        X3(m,j,s,1) = X2(m,s,j,1)
+        X3(m,j,s,2) = X2(m,s,j,2)
+       enddo
+      enddo
+     enddo
+
+   else
+
+     do s=1,ao_num
+      do j=1,min(cholesky_ao_num-iblock+1,block_size)
+       do m=1,ao_num
+        X3(m,j,s,1) = X2(m,s,j,1)
+       enddo
+      enddo
+     enddo
+   endif
+
+   call dgemm('N','N',ao_num,ao_num,ao_num*min(cholesky_ao_num-iblock+1,block_size), -1.d0,     &
+       cholesky_ao(1,1,iblock), ao_num,       &
+       X3(1,1,1,1), ao_num*block_size, 1.d0,  &
+       ao_two_e_integral_alpha_chol, ao_num)
+
+   if (elec_alpha_num > elec_beta_num) then
+     call dgemm('N','N',ao_num,ao_num,ao_num*min(cholesky_ao_num-iblock+1,block_size), -1.d0,     &
+       cholesky_ao(1,1,iblock), ao_num,        &
+       X3(1,1,1,2), ao_num*block_size, 1.d0,   &
+       ao_two_e_integral_beta_chol, ao_num)
+   endif
 
- do s=1,ao_num
-  do j=1,cholesky_ao_num
-   do m=1,ao_num
-    X3(m,j,s,1) = X2(m,s,j,1)
-    X3(m,j,s,2) = X2(m,s,j,2)
-   enddo
-  enddo
  enddo
 
- deallocate(X2)
-
- call dgemm('N','N',ao_num,ao_num,ao_num*cholesky_ao_num, -1.d0,     &
-     cholesky_ao, ao_num,                                            &
-     X3(1,1,1,1), ao_num*cholesky_ao_num, 1.d0,                      &
-     ao_two_e_integral_alpha_chol, ao_num)
-
- call dgemm('N','N',ao_num,ao_num,ao_num*cholesky_ao_num, -1.d0,     &
-     cholesky_ao, ao_num,                                            &
-     X3(1,1,1,2), ao_num*cholesky_ao_num, 1.d0,                      &
-     ao_two_e_integral_beta_chol, ao_num)
-
- deallocate(X3)
+ if (elec_alpha_num == elec_beta_num) then
+   ao_two_e_integral_beta_chol = ao_two_e_integral_alpha_chol
+ endif
+ deallocate(X2,X3)
 
 END_PROVIDER
 
diff --git a/src/mo_two_e_ints/cholesky.irp.f b/src/mo_two_e_ints/cholesky.irp.f
index 32c0dccd..3a868cbe 100644
--- a/src/mo_two_e_ints/cholesky.irp.f
+++ b/src/mo_two_e_ints/cholesky.irp.f
@@ -4,16 +4,18 @@ BEGIN_PROVIDER [ double precision, cholesky_mo, (mo_num, mo_num, cholesky_ao_num
  ! Cholesky vectors in MO basis
  END_DOC
 
- integer :: k
+ integer :: k, i, j
 
  call set_multiple_levels_omp(.False.)
- print *, 'AO->MO Transformation of Cholesky vectors'
  !$OMP PARALLEL DO PRIVATE(k)
  do k=1,cholesky_ao_num
-  call ao_to_mo(cholesky_ao(1,1,k),ao_num,cholesky_mo(1,1,k),mo_num)
+  do j=1,mo_num
+    do i=1,mo_num
+      cholesky_mo(i,j,k) = cholesky_mo_transp(k,i,j)
+    enddo
+  enddo
  enddo
  !$OMP END PARALLEL DO
- print *, ''
 
 END_PROVIDER
 
@@ -23,27 +25,19 @@ BEGIN_PROVIDER [ double precision, cholesky_mo_transp, (cholesky_ao_num, mo_num,
  ! Cholesky vectors in MO basis
  END_DOC
 
- integer :: i,j,k
- double precision, allocatable :: buffer(:,:)
+ double precision, allocatable :: X(:,:,:)
+ integer :: ierr
+ print *, 'AO->MO Transformation of Cholesky vectors'
 
- print *, 'AO->MO Transformation of Cholesky vectors  .'
-
- call set_multiple_levels_omp(.False.)
- !$OMP PARALLEL PRIVATE(i,j,k,buffer)
- allocate(buffer(mo_num,mo_num))
- !$OMP DO SCHEDULE(static)
- do k=1,cholesky_ao_num
-  call ao_to_mo(cholesky_ao(1,1,k),ao_num,buffer,mo_num)
-  do j=1,mo_num
-    do i=1,mo_num
-      cholesky_mo_transp(k,i,j) = buffer(i,j)
-    enddo
-  enddo
- enddo
- !$OMP END DO
- deallocate(buffer)
- !$OMP END PARALLEL
- print *, ''
+ allocate(X(mo_num,cholesky_ao_num,ao_num), stat=ierr)
+ if (ierr /= 0) then
+   print *, irp_here, ': Allocation failed'
+ endif
+ call dgemm('T','N', ao_num*cholesky_ao_num, mo_num, ao_num, 1.d0, &
+     cholesky_ao, ao_num, mo_coef, ao_num, 0.d0, X, ao_num*cholesky_ao_num)
+ call dgemm('T','N', cholesky_ao_num*mo_num, mo_num, ao_num, 1.d0, &
+     X, ao_num, mo_coef, ao_num, 0.d0, cholesky_mo_transp, cholesky_ao_num*mo_num)
+ deallocate(X)
 
 END_PROVIDER
 
diff --git a/src/mo_two_e_ints/mo_bi_integrals.irp.f b/src/mo_two_e_ints/mo_bi_integrals.irp.f
index a461504e..0d3fe176 100644
--- a/src/mo_two_e_ints/mo_bi_integrals.irp.f
+++ b/src/mo_two_e_ints/mo_bi_integrals.irp.f
@@ -37,7 +37,9 @@ BEGIN_PROVIDER [ logical, mo_two_e_integrals_in_map ]
     call map_load_from_disk(trim(ezfio_filename)//'/work/mo_ints',mo_integrals_map)
     print*, 'MO integrals provided'
     return
-  else
+  endif
+
+  if (.not. do_direct_integrals) then
     PROVIDE ao_two_e_integrals_in_map
   endif
 
@@ -90,6 +92,10 @@ subroutine four_idx_dgemm
   double precision, allocatable :: a1(:,:,:,:)
   double precision, allocatable :: a2(:,:,:,:)
 
+  if (ao_num > 1289) then
+    print *,  irp_here, ': Integer overflow in ao_num**3'
+  endif
+
   allocate (a1(ao_num,ao_num,ao_num,ao_num))
 
   print *, 'Getting AOs'
@@ -103,6 +109,7 @@ subroutine four_idx_dgemm
   enddo
   !$OMP END PARALLEL DO
 
+
   print *, '1st transformation'
   ! 1st transformation
   allocate (a2(ao_num,ao_num,ao_num,mo_num))
@@ -166,11 +173,9 @@ subroutine four_idx_dgemm
 
   deallocate (a1)
 
+  call map_sort(mo_integrals_map)
   call map_unique(mo_integrals_map)
 
-  integer*8                      :: get_mo_map_size, mo_map_size
-  mo_map_size = get_mo_map_size()
-
 end subroutine
 
 subroutine add_integrals_to_map(mask_ijkl)
@@ -250,7 +255,7 @@ subroutine add_integrals_to_map(mask_ijkl)
 
   call wall_time(wall_1)
 
-  size_buffer = min(ao_num*ao_num*ao_num,8000000)
+  size_buffer = min(ao_num*ao_num,8000000)
   print*, 'Buffers : ', 8.*(mo_num*(n_j)*(n_k+1) + mo_num+&
       ao_num+ao_num*ao_num+ size_buffer*3)/(1024*1024), 'MB / core'
 
@@ -443,11 +448,6 @@ subroutine add_integrals_to_map(mask_ijkl)
   !$OMP END PARALLEL
   call map_merge(mo_integrals_map)
 
-  call wall_time(wall_2)
-  call cpu_time(cpu_2)
-  integer*8                      :: get_mo_map_size, mo_map_size
-  mo_map_size = get_mo_map_size()
-
   deallocate(list_ijkl)
 
 
@@ -465,51 +465,53 @@ subroutine add_integrals_to_map_cholesky
   integer :: size_buffer, n_integrals
   size_buffer = min(mo_num*mo_num*mo_num,16000000)
 
-  double precision, allocatable :: Vtmp(:,:,:,:)
+  double precision, allocatable :: Vtmp(:,:,:)
   integer(key_kind)  , allocatable :: buffer_i(:)
   real(integral_kind), allocatable :: buffer_value(:)
 
-  if (.True.) then
-    ! In-memory transformation
+  call set_multiple_levels_omp(.False.)
 
-    allocate (Vtmp(mo_num,mo_num,mo_num,mo_num))
+  !$OMP PARALLEL DEFAULT(SHARED) &
+  !$OMP PRIVATE(i,j,k,l,n_integrals,buffer_value, buffer_i, Vtmp)
+  allocate (buffer_i(size_buffer), buffer_value(size_buffer))
+  allocate (Vtmp(mo_num,mo_num,mo_num))
+  n_integrals = 0
 
-    call dgemm('N','T',mo_num*mo_num,mo_num*mo_num,cholesky_ao_num,1.d0, &
-       cholesky_mo, mo_num*mo_num, &
-       cholesky_mo, mo_num*mo_num, 0.d0, &
+  !$OMP DO SCHEDULE(dynamic)
+  do l=1,mo_num
+    call dgemm('T','N',mo_num*mo_num,mo_num,cholesky_ao_num,1.d0, &
+       cholesky_mo_transp, cholesky_ao_num, &
+       cholesky_mo_transp(1,1,l), cholesky_ao_num, 0.d0, &
        Vtmp, mo_num*mo_num)
 
-    !$OMP PARALLEL PRIVATE(i,j,k,l,n_integrals,buffer_value, buffer_i)
-    allocate (buffer_i(size_buffer), buffer_value(size_buffer))
-    n_integrals = 0
-    !$OMP DO
-    do l=1,mo_num
-      do k=1,l
-        do j=1,mo_num
-          do i=1,j
-            if (abs(Vtmp(i,j,k,l)) > mo_integrals_threshold) then
-              n_integrals += 1
-              buffer_value(n_integrals) = Vtmp(i,j,k,l)
-              !DIR$ FORCEINLINE
-              call mo_two_e_integrals_index(i,k,j,l,buffer_i(n_integrals))
-              if (n_integrals == size_buffer) then
-                call map_append(mo_integrals_map, buffer_i, buffer_value, n_integrals)
-                n_integrals = 0
-              endif
+    do k=1,l
+      do j=1,mo_num
+        do i=1,j
+          if (dabs(Vtmp(i,j,k)) > mo_integrals_threshold) then
+            n_integrals = n_integrals + 1
+            buffer_value(n_integrals) = Vtmp(i,j,k)
+            !DIR$ FORCEINLINE
+            call mo_two_e_integrals_index(i,k,j,l,buffer_i(n_integrals))
+            if (n_integrals == size_buffer) then
+              call map_append(mo_integrals_map, buffer_i, buffer_value, n_integrals)
+              n_integrals = 0
             endif
-          enddo
+          endif
         enddo
       enddo
     enddo
-    !$OMP END DO
+  enddo
+  !$OMP END DO NOWAIT
+
+  if (n_integrals > 0) then
     call map_append(mo_integrals_map, buffer_i, buffer_value, n_integrals)
-    deallocate(buffer_i, buffer_value)
-    !$OMP END PARALLEL
-
-    deallocate(Vtmp)
-    call map_unique(mo_integrals_map)
-
   endif
+  deallocate(buffer_i, buffer_value, Vtmp)
+  !$OMP BARRIER
+  !$OMP END PARALLEL
+
+  call map_sort(mo_integrals_map)
+  call map_unique(mo_integrals_map)
 
 end
 
@@ -580,6 +582,9 @@ subroutine add_integrals_to_map_three_indices(mask_ijk)
     return
   endif
 
+  if (ao_num > 1289) then
+    print *,  irp_here, ': Integer overflow in ao_num**3'
+  endif
   size_buffer = min(ao_num*ao_num*ao_num,16000000)
   print*, 'Providing the molecular integrals '
   print*, 'Buffers : ', 8.*(mo_num*(n_j)*(n_k+1) + mo_num+&
@@ -855,6 +860,9 @@ subroutine add_integrals_to_map_no_exit_34(mask_ijkl)
   call bitstring_to_list( mask_ijkl(1,3), list_ijkl(1,3), n_k, N_int )
   call bitstring_to_list( mask_ijkl(1,4), list_ijkl(1,4), n_l, N_int )
 
+  if (ao_num > 1289) then
+    print *,  irp_here, ': Integer overflow in ao_num**3'
+  endif
   size_buffer = min(ao_num*ao_num*ao_num,16000000)
   print*, 'Providing the molecular integrals '
   print*, 'Buffers : ', 8.*(mo_num*(n_j)*(n_k+1) + mo_num+&
@@ -1350,16 +1358,29 @@ END_PROVIDER
   ! mo_two_e_integrals_jj_anti(i,j) = J_ij - K_ij
   END_DOC
 
-  integer                        :: i,j
+  integer                        :: i,j,k
   double precision               :: get_two_e_integral
 
 
   if (do_ao_cholesky) then
+    double precision, allocatable :: buffer(:,:)
+    allocate (buffer(cholesky_ao_num,mo_num))
+    do k=1,cholesky_ao_num
+      do i=1,mo_num
+        buffer(k,i) = cholesky_mo_transp(k,i,i)
+      enddo
+    enddo
+    call dgemm('T','N',mo_num,mo_num,cholesky_ao_num,1.d0, &
+      buffer, cholesky_ao_num, buffer, cholesky_ao_num, 0.d0, mo_two_e_integrals_jj, mo_num)
+    deallocate(buffer)
+
     do j=1,mo_num
       do i=1,mo_num
-        !TODO: use dgemm
-        mo_two_e_integrals_jj(i,j) = sum(cholesky_mo_transp(:,i,i)*cholesky_mo_transp(:,j,j))
-        mo_two_e_integrals_jj_exchange(i,j) = sum(cholesky_mo_transp(:,i,j)*cholesky_mo_transp(:,j,i))
+          mo_two_e_integrals_jj_exchange(i,j) = 0.d0
+          do k=1,cholesky_ao_num
+            mo_two_e_integrals_jj_exchange(i,j) = mo_two_e_integrals_jj_exchange(i,j) + &
+                          cholesky_mo_transp(k,i,j)*cholesky_mo_transp(k,j,i)
+          enddo
       enddo
     enddo
 
diff --git a/src/utils/fortran_mmap.c b/src/utils/fortran_mmap.c
index 52df2476..71426002 100644
--- a/src/utils/fortran_mmap.c
+++ b/src/utils/fortran_mmap.c
@@ -22,11 +22,7 @@ void* mmap_fortran(char* filename, size_t bytes, int* file_descr, int read_only)
             perror("Error opening mmap file for reading");
             exit(EXIT_FAILURE);
         }
-        map = mmap(NULL, bytes, PROT_READ, MAP_SHARED | MAP_HUGETLB, fd, 0);
-        if (map == MAP_FAILED) {
-          /* try again without huge pages */
-          map = mmap(NULL, bytes, PROT_READ, MAP_SHARED, fd, 0);
-        }
+        map = mmap(NULL, bytes, PROT_READ, MAP_SHARED, fd, 0);
     }
     else
     {
@@ -53,16 +49,12 @@ void* mmap_fortran(char* filename, size_t bytes, int* file_descr, int read_only)
             exit(EXIT_FAILURE);
         }
 
-        map = mmap(NULL, bytes, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_HUGETLB, fd, 0);
-        if (map == MAP_FAILED) {
-          /* try again without huge pages */
-          map = mmap(NULL, bytes, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
-        }
+        map = mmap(NULL, bytes, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
     }
 
     if (map == MAP_FAILED) {
         close(fd);
-        printf("%s:\n", filename);
+        printf("%s: %lu\n", filename, bytes);
         perror("Error mmapping the file");
         exit(EXIT_FAILURE);
     }
diff --git a/src/utils/map_functions.irp.f b/src/utils/map_functions.irp.f
index cd3b28a8..97d0e8bf 100644
--- a/src/utils/map_functions.irp.f
+++ b/src/utils/map_functions.irp.f
@@ -11,6 +11,10 @@ subroutine map_save_to_disk(filename,map)
 
   integer*8 :: n_elements
   n_elements = int(map % n_elements,8)
+  if (n_elements <= 0) then
+    print *,  'Unable to write map to disk: n_elements = ', n_elements
+    stop -1
+  endif
 
 
   if (map % consolidated) then
diff --git a/src/utils/memory.irp.f b/src/utils/memory.irp.f
index 115b2cbe..7da283ec 100644
--- a/src/utils/memory.irp.f
+++ b/src/utils/memory.irp.f
@@ -4,8 +4,10 @@ BEGIN_PROVIDER [ integer, qp_max_mem ]
  ! Maximum memory in Gb
  END_DOC
  character*(128) :: env
+ integer, external :: get_total_available_memory
 
- qp_max_mem = 2000
+ qp_max_mem = get_total_available_memory()
+ call write_int(6,qp_max_mem,'Total available memory (GB)')
  call getenv('QP_MAXMEM',env)
  if (trim(env) /= '') then
     call lock_io()
@@ -122,3 +124,35 @@ subroutine print_memory_usage()
     '.. >>>>> [ RES  MEM : ', rss , &
         ' GB ] [ VIRT MEM : ', mem, ' GB ] <<<<< ..'
 end
+
+integer function get_total_available_memory() result(res)
+  implicit none
+  BEGIN_DOC
+! Returns the total available memory on the current machine
+  END_DOC
+
+  character(len=128) :: line
+  integer :: status
+  integer :: iunit
+  integer*8, parameter :: KB = 1024
+  integer*8, parameter :: GiB = 1024**3
+  integer, external :: getUnitAndOpen
+
+  iunit = getUnitAndOpen('/proc/meminfo','r')
+
+  res = 512
+  do
+      read(iunit, '(A)', END=10) line
+      if (line(1:10) == "MemTotal: ") then
+        read(line(11:), *, ERR=20) res
+        res = int((res*KB) / GiB,4)
+        exit
+ 20     continue
+      end if
+  end do
+ 10 continue
+  close(iunit)
+
+end function get_total_available_memory
+
+
diff --git a/src/utils/mmap.f90 b/src/utils/mmap.f90
index 49147283..41e60224 100644
--- a/src/utils/mmap.f90
+++ b/src/utils/mmap.f90
@@ -46,7 +46,13 @@ module mmap_module
       integer(c_size_t)              :: length
       integer(c_int)                 :: fd_
 
-      length = PRODUCT( shape(:) ) * bytes
+      integer :: i
+
+      length = int(bytes,8)
+      do i=1,size(shape)
+        length = length * shape(i)
+      enddo
+
       if (read_only) then
           map = c_mmap_fortran( trim(filename)//char(0), length, fd_, 1)
       else
@@ -66,7 +72,12 @@ module mmap_module
       integer(c_size_t)              :: length
       integer(c_int)                 :: fd_
 
-      length = PRODUCT( shape(:) ) * bytes
+      integer :: i
+
+      length = int(bytes,8)
+      do i=1,size(shape)
+        length = length * shape(i)
+      enddo
       fd_ = fd
       call c_munmap_fortran( length, fd_, map)
   end subroutine
@@ -82,7 +93,12 @@ module mmap_module
       integer(c_size_t)              :: length
       integer(c_int)                 :: fd_
 
-      length = PRODUCT( shape(:) ) * bytes
+      integer :: i
+
+      length = int(bytes,8)
+      do i=1,size(shape)
+        length = length * shape(i)
+      enddo
       fd_ = fd
       call c_msync_fortran( length, fd_, map)
   end subroutine
diff --git a/src/utils_cc/mo_integrals_cc.irp.f b/src/utils_cc/mo_integrals_cc.irp.f
index 2db614b4..62237229 100644
--- a/src/utils_cc/mo_integrals_cc.irp.f
+++ b/src/utils_cc/mo_integrals_cc.irp.f
@@ -53,33 +53,8 @@ subroutine gen_v_space(n1,n2,n3,n4,list1,list2,list3,list4,v)
     allocate(v1(cholesky_ao_num,n1,n3), v2(cholesky_ao_num,n2,n4))
     allocate(buffer(n1,n3,n2,n4))
 
-    !$OMP PARALLEL PRIVATE(i1,i2,i3,i4,idx1,idx2,idx3,idx4,k)
-    !$OMP DO
-    do i3=1,n3
-      idx3 = list3(i3)
-      do i1=1,n1
-        idx1 = list1(i1)
-        do k=1,cholesky_ao_num
-          v1(k,i1,i3) = cholesky_mo_transp(k,idx1,idx3)
-        enddo
-      enddo
-    enddo
-    !$OMP END DO NOWAIT
-
-    !$OMP DO
-    do i4=1,n4
-      idx4 = list4(i4)
-      do i2=1,n2
-        idx2 = list2(i2)
-        do k=1,cholesky_ao_num
-          v2(k,i2,i4) = cholesky_mo_transp(k,idx2,idx4)
-        enddo
-      enddo
-    enddo
-    !$OMP END DO NOWAIT
-
-    !$OMP BARRIER
-    !$OMP END PARALLEL
+    call gen_v_space_chol(n1,n3,list1,list3,v1,cholesky_ao_num)
+    call gen_v_space_chol(n2,n4,list2,list4,v2,cholesky_ao_num)
 
     call dgemm('T','N', n1*n3, n2*n4, cholesky_ao_num, 1.d0, &
          v1, cholesky_ao_num, &
@@ -129,6 +104,30 @@ subroutine gen_v_space(n1,n2,n3,n4,list1,list2,list3,list4,v)
 
 end
 
+subroutine gen_v_space_chol(n1,n3,list1,list3,v,ldv)
+
+  implicit none
+
+  integer, intent(in)           :: n1,n3,ldv
+  integer, intent(in)           :: list1(n1),list3(n3)
+  double precision, intent(out) :: v(ldv,n1,n3)
+
+  integer                       :: i1,i3,idx1,idx3,k
+
+  !$OMP PARALLEL DO PRIVATE(i1,i3,idx1,idx3,k)
+  do i3=1,n3
+    idx3 = list3(i3)
+    do i1=1,n1
+      idx1 = list1(i1)
+      do k=1,cholesky_ao_num
+        v(k,i1,i3) = cholesky_mo_transp(k,idx1,idx3)
+      enddo
+    enddo
+  enddo
+  !$OMP END PARALLEL DO
+
+end
+
 ! full
 
 BEGIN_PROVIDER [double precision, cc_space_v, (mo_num,mo_num,mo_num,mo_num)]
@@ -345,6 +344,38 @@ BEGIN_PROVIDER [double precision, cc_space_v_vvvv, (cc_nVa, cc_nVa, cc_nVa, cc_n
 
 END_PROVIDER
 
+BEGIN_PROVIDER [double precision, cc_space_v_vv_chol, (cholesky_ao_num, cc_nVa, cc_nVa)]
+
+  implicit none
+
+  call gen_v_space_chol(cc_nVa, cc_nVa, cc_list_vir, cc_list_vir, cc_space_v_vv_chol, cholesky_ao_num)
+
+END_PROVIDER
+
+BEGIN_PROVIDER [double precision, cc_space_v_vo_chol, (cholesky_ao_num, cc_nVa, cc_nOa)]
+
+  implicit none
+
+  call gen_v_space_chol(cc_nVa, cc_nOa, cc_list_vir, cc_list_occ, cc_space_v_vo_chol, cholesky_ao_num)
+
+END_PROVIDER
+
+BEGIN_PROVIDER [double precision, cc_space_v_ov_chol, (cholesky_ao_num, cc_nOa, cc_nVa)]
+
+  implicit none
+
+  call gen_v_space_chol(cc_nOa, cc_nVa, cc_list_occ, cc_list_vir, cc_space_v_ov_chol, cholesky_ao_num)
+
+END_PROVIDER
+
+BEGIN_PROVIDER [double precision, cc_space_v_oo_chol, (cholesky_ao_num, cc_nOa, cc_nOa)]
+
+  implicit none
+
+  call gen_v_space_chol(cc_nOa, cc_nOa, cc_list_occ, cc_list_occ, cc_space_v_oo_chol, cholesky_ao_num)
+
+END_PROVIDER
+
 ! ppqq
 
 BEGIN_PROVIDER [double precision, cc_space_v_ppqq, (cc_n_mo, cc_n_mo)]