Merge pull request #346 from QuantumPackage/dev-stable

Dev stable

config/gfortran_debug_mkl.cfg

@@ -0,0 +1,63 @@
+# Common flags
+##############
+#
+# -ffree-line-length-none : Needed for IRPF90 which produces long lines
+# -lblas -llapack         : Link with libblas and liblapack libraries provided by the system
+# -I .                    : Include the curent directory (Mandatory)
+#
+# --ninja                 : Allow the utilisation of ninja. (Mandatory)
+# --align=32              : Align all provided arrays on a 32-byte boundary
+#
+#
+[COMMON]
+FC           : gfortran -g -ffree-line-length-none -I . -fPIC -std=legacy
+LAPACK_LIB   : -I${MKLROOT}/include -L${MKLROOT}/lib/intel64 -Wl,--no-as-needed -lmkl_gf_lp64 -lmkl_core -lpthread -lm -ldl -lmkl_gnu_thread -lgomp -fopenmp
+IRPF90       : irpf90
+IRPF90_FLAGS : --ninja --align=32 --assert -DSET_NESTED
+
+# Global options
+################
+#
+# 1 : Activate
+# 0 : Deactivate
+#
+[OPTION]
+MODE    : DEBUG      ; [ OPT | PROFILE | DEBUG ] : Chooses the section below
+CACHE   : 0          ; Enable cache_compile.py
+OPENMP  : 1          ; Append OpenMP flags
+
+# Optimization flags
+####################
+#
+# -Ofast                  : Disregard strict standards compliance. Enables all -O3 optimizations.
+#                           It also enables optimizations that are not valid
+#                           for all standard-compliant programs.  It turns on
+#                           -ffast-math and the Fortran-specific
+#                           -fno-protect-parens and -fstack-arrays.
+[OPT]
+FCFLAGS : -Ofast
+
+# Profiling flags
+#################
+#
+[PROFILE]
+FC      : -p -g
+FCFLAGS : -Ofast
+
+# Debugging flags
+#################
+#
+# -fcheck=all  : Checks uninitialized variables,  array subscripts, etc...
+# -g           : Extra debugging information
+#
+[DEBUG]
+#FCFLAGS : -g -msse4.2  -fcheck=all -Waliasing -Wampersand -Wconversion -Wsurprising -Wintrinsics-std -Wno-tabs -Wintrinsic-shadow -Wline-truncation -Wreal-q-constant -Wuninitialized  -fbacktrace -ffpe-trap=zero,overflow,underflow -finit-real=nan
+FCFLAGS : -g -mavx  -fcheck=all -Waliasing -Wampersand -Wconversion -Wsurprising -Wintrinsics-std -Wno-tabs -Wintrinsic-shadow -Wline-truncation -Wreal-q-constant -Wuninitialized  -fbacktrace -ffpe-trap=zero,overflow -finit-real=nan
+
+# OpenMP flags
+#################
+#
+[OPENMP]
+FC           : -fopenmp
+IRPF90_FLAGS : --openmp
+

configure

@@ -40,14 +40,16 @@ Usage:
   $(basename $0) -c <file>    
   $(basename $0) -h           
   $(basename $0) -i <package> 
+  $(basename $0) -g [nvidia|intel|none]
 
 Options:
-  -c  <file>    Define a COMPILATION configuration file,
-                in "${QP_ROOT}/config/".
-  -h            Print the HELP message
-  -i <package>  INSTALL <package>. Use at your OWN RISK:
-                no support will be provided for the installation of
-                dependencies.
+  -c  <file>        Define a COMPILATION configuration file,
+                    in "${QP_ROOT}/config/".
+  -h                Print the HELP message
+  -i <package>      INSTALL <package>. Use at your OWN RISK:
+                    no support will be provided for the installation of
+                    dependencies.
+  -g [nvidia|intel|none]  Choose GPU acceleration
 
 Example:
   ./$(basename $0) -c config/gfortran.cfg
@@ -83,7 +85,7 @@ function execute () {
 PACKAGES=""
 
 
-while getopts "d:c:i:h" c ; do
+while getopts "d:c:i:g:h" c ; do
     case "$c" in
         c)
             case "$OPTARG" in
@@ -100,6 +102,9 @@ while getopts "d:c:i:h" c ; do
                 "") help ; break;;
                 *) PACKAGES="${PACKAGE} $OPTARG"
             esac;;
+        g)
+            GPU=$OPTARG;
+            break;;
         h)
             help
             exit 0;;
@@ -109,6 +114,27 @@ while getopts "d:c:i:h" c ; do
     esac
 done
 
+# Handle GPU acceleration
+rm -f ${QP_ROOT}/src/gpu_arch
+case "$GPU" in
+  amd) # AMD
+     echo "Activating AMD GPU acceleration"
+     ln -s ${QP_ROOT}/plugins/local/gpu_amd ${QP_ROOT}/src/gpu_arch
+     ;;
+  intel) # Intel
+     echo "Activating Intel GPU acceleration (EXPERIMENTAL)"
+     ln -s ${QP_ROOT}/plugins/local/gpu_intel ${QP_ROOT}/src/gpu_arch
+     ;;
+  nvidia) # Nvidia
+     echo "Activating Nvidia GPU acceleration"
+     ln -s ${QP_ROOT}/plugins/local/gpu_nvidia ${QP_ROOT}/src/gpu_arch
+     ;;
+  *) # No Acceleration
+     echo "Disabling GPU acceleration"
+     ln -s ${QP_ROOT}/plugins/local/gpu_x86 ${QP_ROOT}/src/gpu_arch
+     ;;
+esac
+
 # Trim leading and trailing spaces
 PACKAGES=$(echo $PACKAGES | xargs)
 

etc/paths.rc

@@ -28,6 +28,15 @@ function qp_prepend_export () {
   fi
 }
 
+function qp_append_export () {
+  eval "value_1="\${$1}""
+  if [[ -z $value_1 ]] ; then
+      echo "${2}:"
+  else
+      echo "${value_1}:${2}"
+  fi
+}
+
 export PYTHONPATH=$(qp_prepend_export "PYTHONPATH" "${QP_EZFIO}/Python":"${QP_PYTHON}")
 
 export PATH=$(qp_prepend_export "PATH" "${QP_PYTHON}":"${QP_ROOT}"/bin:"${QP_ROOT}"/ocaml)

plugins/local/basis_correction/51.basis_c.bats

@@ -37,14 +37,6 @@ function run_sd() {
   eq $energy1 $1 $thresh
 }
 
-@test "O2 CAS" {  
-  qp set_file o2_cas.gms.ezfio
-  qp set_mo_class -c "[1-2]" -a "[3-10]" -d "[11-46]"
-  run -149.72435425 3.e-4 10000
-  qp set_mo_class -c "[1-2]" -a "[3-10]" -v "[11-46]"
-  run_md -0.1160222327 1.e-6 
-}
-
 
 @test "LiF RHF" {  
   qp set_file lif.ezfio 

plugins/local/basis_correction/basis_correction.irp.f

@@ -7,10 +7,6 @@ program basis_correction
   touch read_wf
   no_core_density = .True.
   touch no_core_density
-  if(io_mo_two_e_integrals .ne. "Read")then
-   provide ao_two_e_integrals_in_map
-  endif
-  provide mo_two_e_integrals_in_map
   call print_basis_correction
 end
 

plugins/local/basis_correction/print_routine.irp.f

@@ -22,7 +22,7 @@ subroutine print_basis_correction
  print*, '****************************************'
  print*, '****************************************'
  print*, 'mu_of_r_potential = ',mu_of_r_potential
- if(mu_of_r_potential.EQ."hf")then
+ if(mu_of_r_potential.EQ."hf".or.mu_of_r_potential.EQ."hf_old".or.mu_of_r_potential.EQ."hf_sparse")then
    print*, ''
    print*,'Using a HF-like two-body density to define mu(r)'
    print*,'This assumes that HF is a qualitative representation of the wave function '

plugins/local/basis_correction/test_chol_bas.irp.f

@@ -0,0 +1,18 @@
+program pouet
+ implicit none
+ call test
+end
+subroutine test
+ implicit none
+! provide mos_times_cholesky_r1
+! provide mos_times_cholesky_r2
+ integer :: ipoint
+ double precision :: accu,weight
+ accu = 0.d0
+ do ipoint = 1, n_points_final_grid
+  weight = final_weight_at_r_vector(ipoint)
+!  accu += dabs(mu_of_r_hf(ipoint) - mu_of_r_hf_old(ipoint)) * weight
+  accu += dabs(f_hf_cholesky_sparse(ipoint) - f_hf_cholesky(ipoint)) * weight
+ enddo
+ print*,'accu = ',accu
+end

plugins/local/bi_ort_ints/bi_ort_ints.irp.f

@@ -17,12 +17,15 @@ program bi_ort_ints
 ! call test_3e
 ! call test_5idx
 ! call test_5idx2
-  call test_4idx()
+!  call test_4idx()
   !call test_4idx_n4()
   !call test_4idx2()
   !call test_5idx2
   !call test_5idx
 
+  call test_mos_in_r()
+  call test_int2_grad1_u12_bimo_t()
+
 end
 
 subroutine test_5idx2
@@ -472,4 +475,94 @@ subroutine test_4idx()
   return
 end
 
+! ---
+
+subroutine test_mos_in_r()
+
+  implicit none
+
+  integer          :: i, j
+  double precision :: err_tot, nrm_tot, err_loc, acc_thr
+
+  PROVIDE mos_l_in_r_array_transp_old mos_r_in_r_array_transp_old
+  PROVIDE mos_l_in_r_array_transp mos_r_in_r_array_transp
+
+  acc_thr = 1d-13
+
+  err_tot = 0.d0
+  nrm_tot = 0.d0
+  do i = 1, mo_num
+    do j = 1, n_points_final_grid
+      err_loc = dabs(mos_l_in_r_array_transp_old(j,i) - mos_l_in_r_array_transp(j,i))
+      if(err_loc > acc_thr) then
+        print*, " error on", j, i
+        print*, " old res", mos_l_in_r_array_transp_old(j,i)
+        print*, " new res", mos_l_in_r_array_transp    (j,i)
+        stop
+      endif
+      err_tot = err_tot + err_loc
+      nrm_tot = nrm_tot + dabs(mos_l_in_r_array_transp_old(j,i))
+    enddo
+  enddo
+  print *, ' absolute accuracy on mos_l_in_r_array_transp (%) =', 100.d0 * err_tot / nrm_tot
+
+  err_tot = 0.d0
+  nrm_tot = 0.d0
+  do i = 1, mo_num
+    do j = 1, n_points_final_grid
+      err_loc = dabs(mos_r_in_r_array_transp_old(j,i) - mos_r_in_r_array_transp(j,i))
+      if(err_loc > acc_thr) then
+        print*, " error on", j, i
+        print*, " old res", mos_r_in_r_array_transp_old(j,i)
+        print*, " new res", mos_r_in_r_array_transp    (j,i)
+        stop
+      endif
+      err_tot = err_tot + err_loc
+      nrm_tot = nrm_tot + dabs(mos_r_in_r_array_transp_old(j,i))
+    enddo
+  enddo
+  print *, ' absolute accuracy on mos_r_in_r_array_transp (%) =', 100.d0 * err_tot / nrm_tot
+
+  return
+end
+
+! ---
+
+subroutine test_int2_grad1_u12_bimo_t()
+
+  implicit none
+  integer          :: i, j, ipoint, m
+  double precision :: err_tot, nrm_tot, err_loc, acc_thr
+
+  PROVIDE int2_grad1_u12_bimo_t_old
+  PROVIDE int2_grad1_u12_bimo_t
+
+  acc_thr = 1d-13
+
+  err_tot = 0.d0
+  nrm_tot = 0.d0
+  do i = 1, mo_num
+    do j = 1, mo_num
+      do m = 1, 3
+        do ipoint = 1, n_points_final_grid
+          err_loc = dabs(int2_grad1_u12_bimo_t_old(ipoint,m,j,i) - int2_grad1_u12_bimo_t(ipoint,m,j,i))
+          if(err_loc > acc_thr) then
+            print*, " error on", ipoint, m, j, i
+            print*, " old res", int2_grad1_u12_bimo_t_old(ipoint,m,j,i)
+            print*, " new res", int2_grad1_u12_bimo_t    (ipoint,m,j,i)
+            stop
+          endif
+          err_tot = err_tot + err_loc
+          nrm_tot = nrm_tot + dabs(int2_grad1_u12_bimo_t_old(ipoint,m,j,i))
+        enddo
+      enddo
+    enddo
+  enddo
+  print *, ' absolute accuracy on int2_grad1_u12_bimo_t (%) =', 100.d0 * err_tot / nrm_tot
+
+  return
+end
+
+! ---
+
 

plugins/local/bi_ort_ints/no_dressing.irp.f

@@ -322,6 +322,12 @@
 
 BEGIN_PROVIDER [double precision, noL_0e]
 
+  BEGIN_DOC
+  ! 
+  ! < Phi_left | L | Phi_right >
+  !
+  END_DOC 
+
   implicit none
   integer                       :: i, j, k, ipoint
   double precision              :: t0, t1
@@ -330,10 +336,6 @@
   double precision, allocatable :: tmp_M(:,:), tmp_S(:), tmp_O(:), tmp_J(:,:)
   double precision, allocatable :: tmp_M_priv(:,:), tmp_S_priv(:), tmp_O_priv(:), tmp_J_priv(:,:)
 
-
-  call wall_time(t0)
-  print*, " Providing noL_0e ..."
-
   if(elec_alpha_num .eq. elec_beta_num) then
 
     allocate(tmp(elec_beta_num))
@@ -708,10 +710,7 @@
 
   endif
 
-  call wall_time(t1)
-  print*, " Wall time for noL_0e (min) = ", (t1 - t0)/60.d0
-
-  print*, " noL_0e = ", noL_0e
+  print*, " noL_0e =", noL_0e
 
 END_PROVIDER