LCPQ/quantum_package
10
0
Fork 0
mirror of https://github.com/LCPQ/quantum_package synced 2024-06-02 11:25:26 +02:00
Code Issues Releases Wiki Activity

Merge pull request #51 from scemama/master

Browse Source 
merge
This commit is contained in:
Emmanuel Giner 2019-01-09 12:41:17 +01:00 committed by GitHub
commit 2efdecc167
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
31 changed files with 682 additions and 113 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
.gitignore vendored
View File

@ -1,4 +1,3 @@
quantum_package.rc
config/ifort.cfg
quantum_package_static.tar.gz
build.ninja

11
TODO
View File

@ -43,13 +43,6 @@
Refaire les benchmarks
# Commande qp avec completion dans le shell
* qp set => ezfio set
* qp get => ezfio get
* qp run => qp_run
* qp srun => qp_run
* qp mpirun => qp_run
* qp edit => qp_edit
* qp set_mo_class => qp_set_mo_class
* qp set_frozen_core => qp_set_frozen_core
# Documentation de qpsh
# Documentation de /etc

2
config/gfortran.cfg
View File

@ -35,7 +35,7 @@ OPENMP : 1 ; Append OpenMP flags
# -ffast-math and the Fortran-specific
# -fno-protect-parens and -fstack-arrays.
[OPT]
FCFLAGS : -Ofast
FCFLAGS : -Ofast -msse4.2
# Profiling flags
#################

2
config/gfortran_avx.cfg
View File

@ -35,7 +35,7 @@ OPENMP : 1 ; Append OpenMP flags
# -ffast-math and the Fortran-specific
# -fno-protect-parens and -fstack-arrays.
[OPT]
FCFLAGS : -Ofast
FCFLAGS : -Ofast -mavx
# Profiling flags
#################

2
config/ifort.cfg
View File

@ -32,7 +32,7 @@ OPENMP : 1 ; Append OpenMP flags
#
[OPT]
FC : -traceback
FCFLAGS : -xAVX -O2 -ip -ftz -g
FCFLAGS : -xSSE4.2 -O2 -ip -ftz -g
# Profiling flags
#################

63
config/ifort_avx.cfg Normal file
View File

@ -0,0 +1,63 @@
# Common flags
##############
#
# -mkl=[parallel|sequential] : Use the MKL library
# --ninja : Allow the utilisation of ninja. It is mandatory !
# --align=32 : Align all provided arrays on a 32-byte boundary
#
[COMMON]
FC : ifort
LAPACK_LIB : -mkl=parallel
IRPF90 : irpf90
IRPF90_FLAGS : --ninja --align=32
# Global options
################
#
# 1 : Activate
# 0 : Deactivate
#
[OPTION]
MODE : OPT ; [ OPT | PROFILE | DEBUG ] : Chooses the section below
CACHE : 0 ; Enable cache_compile.py
OPENMP : 1 ; Append OpenMP flags
# Optimization flags
####################
#
# -xHost : Compile a binary optimized for the current architecture
# -O2 : O3 not better than O2.
# -ip : Inter-procedural optimizations
# -ftz : Flushes denormal results to zero
#
[OPT]
FC : -traceback
FCFLAGS : -xAVX -O2 -ip -ftz -g
# Profiling flags
#################
#
[PROFILE]
FC : -p -g
FCFLAGS : -xSSE4.2 -O2 -ip -ftz
# Debugging flags
#################
#
# -traceback : Activate backtrace on runtime
# -fpe0 : All floating point exaceptions
# -C : Checks uninitialized variables, array subscripts, etc...
# -g : Extra debugging information
# -xSSE2 : Valgrind needs a very simple x86 executable
#
[DEBUG]
FC : -g -traceback
FCFLAGS : -xSSE2 -C -fpe0 -implicitnone
# OpenMP flags
#################
#
[OPENMP]
FC : -qopenmp
IRPF90_FLAGS : --openmp

64
config/ifort_avx_mpi.cfg Normal file
View File

@ -0,0 +1,64 @@
# Common flags
##############
#
# -mkl=[parallel|sequential] : Use the MKL library
# --ninja : Allow the utilisation of ninja. It is mandatory !
# --align=32 : Align all provided arrays on a 32-byte boundary
#
[COMMON]
FC : mpiifort
LAPACK_LIB : -mkl=parallel
IRPF90 : irpf90
IRPF90_FLAGS : --ninja --align=32 -DMPI
# Global options
################
#
# 1 : Activate
# 0 : Deactivate
#
[OPTION]
MODE : OPT ; [ OPT | PROFILE | DEBUG ] : Chooses the section below
CACHE : 0 ; Enable cache_compile.py
OPENMP : 1 ; Append OpenMP flags
# Optimization flags
####################
#
# -xHost : Compile a binary optimized for the current architecture
# -O2 : O3 not better than O2.
# -ip : Inter-procedural optimizations
# -ftz : Flushes denormal results to zero
#
[OPT]
FCFLAGS : -xAVX -O2 -ip -ftz -g -traceback
# Profiling flags
#################
#
[PROFILE]
FC : -p -g
FCFLAGS : -xSSE4.2 -O2 -ip -ftz
# Debugging flags
#################
#
# -traceback : Activate backtrace on runtime
# -fpe0 : All floating point exaceptions
# -C : Checks uninitialized variables, array subscripts, etc...
# -g : Extra debugging information
# -xSSE2 : Valgrind needs a very simple x86 executable
#
[DEBUG]
FC : -g -traceback
FCFLAGS : -xSSE2 -C -fpe0 -implicitnone
# OpenMP flags
#################
#
[OPENMP]
FC : -qopenmp
IRPF90_FLAGS : --openmp

2
config/ifort_debug.cfg
View File

@ -32,7 +32,7 @@ OPENMP : 1 ; Append OpenMP flags
#
[OPT]
FC : -traceback
FCFLAGS : -xAVX -O2 -ip -ftz -g
FCFLAGS : -xSSE4.2 -O2 -ip -ftz -g
# Profiling flags

1
config/ifort_mpi.cfg
View File

@ -33,7 +33,6 @@ OPENMP : 1 ; Append OpenMP flags
[OPT]
FCFLAGS : -xSSE4.2 -O2 -ip -ftz -g -traceback
# !xAVX
# Profiling flags
#################
#

32
configure vendored
View File

@ -3,6 +3,10 @@
# Quantum Package configuration script
#
export QP_ROOT="$( cd "$(dirname "$0")" ; pwd -P )"
echo "QP_ROOT="$QP_ROOT
help()
{
cat <<EOF
@ -17,14 +21,16 @@ Options:
-c <file>, --config <file> define a configuration file, in
"${QP_ROOT}/config/".
-h, --help print the help message
-i <package>, --install <package> install <package>. Use at your own risk.
-i <package>, --install <package> install <package>. Use at your own
risk: no support will be provided
for the installation of dependencies.
Example:
./configure -c config/gfortran.cfg
Note:
- Using different configuration files for installing dependencies and
compiling QP can improve the user experience.
- Users are encouraged to create their own configuration files instead of
modifying the existing ones.
EOF
exit
@ -42,7 +48,7 @@ while : ; do
-i|--install)
PACKAGES="${PACKAGE} $2"
shift;;
-h|--help)
-h|-help|--help)
help;;
*)
echo "unknown option $1, try --help"
@ -54,20 +60,12 @@ done
# Trim leading and trailing spaces
PACKAGES=$(echo $PACKAGES | xargs)
sed "s!^export QP_ROOT=.*\$!export QP_ROOT=\"$PWD\"!" \
quantum_package.rc.default > quantum_package.rc
echo "quantum_package.rc created."
echo "export QP_ROOT=\"$QP_ROOT\"" > ${QP_ROOT}/etc/00.qp_root
source quantum_package.rc
function fail() {
echo "Please refer to INSTALL.rst to install the missing dependencies."
exit 1
@ -151,7 +149,7 @@ for PACKAGE in ${PACKAGES} ; do
elif [[ ${PACKAGE} = irpf90 ]] ; then
download \
"https://gitlab.com/scemama/irpf90/-/archive/v1.7.3/irpf90-v1.7.3.tar.gz" \
"https://gitlab.com/scemama/irpf90/-/archive/v1.7.4/irpf90-v1.7.4.tar.gz" \
"${QP_ROOT}"/external/irpf90.tar.gz
(
cd "${QP_ROOT}"/external
@ -380,10 +378,14 @@ echo " ||----w | "
echo " || || "
echo "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"
echo ""
echo "Now:"
echo "Now, if you want to use the current shell,"
echo ""
echo " source $QP_ROOT/quantum_package.rc"
echo ""
echo "Or if you want to use a Bash shell adapted for QP:"
echo ""
echo " $QP_ROOT/qpsh"
echo ""

1
etc/.gitignore vendored Normal file
View File

@ -0,0 +1 @@
00.qp_root

1
etc/README.rst Normal file
View File

@ -0,0 +1 @@
This directory contains all the files to be sourced when the |QP| environment is loaded.

12
etc/ezfio.rc Normal file
View File

@ -0,0 +1,12 @@
# Configuration of EZFIO package
export QP_EZFIO=${QP_ROOT}/external/ezfio
if [[ -f ${QP_EZFIO}/Bash/ezfio.sh ]]; then
if [[ "$(ps -p $$ -ocomm=)" == "zsh" ]] ; then
autoload bashcompinit
bashcompinit
fi
source ${QP_EZFIO}/Bash/ezfio.sh
fi

10
etc/irpf90.rc Normal file
View File

@ -0,0 +1,10 @@
# Configuration of IRPF90 package
export IRPF90=${QP_ROOT}/bin/irpf90
# Load irpman shell completion
(
irpman=$(tail -1 "${QP_ROOT}/bin/irpman" | cut --delimiter " " --field=2)
source $(dirname ${irpman})/../irpman-completions.bash
)

8
etc/libraries.rc Normal file
View File

@ -0,0 +1,8 @@
# Configuration of additional libraries required for modules
QP_LIB=""
# Include here the optional external libraries to link with your binaries,
# for example :
# export QP_LIB="$QP_LIB -lint"

15
etc/local.rc Normal file
View File

@ -0,0 +1,15 @@
# Configuration specific to the local machine
# Maximum allowed memory per node
# export QP_MAXMEM=64
# Target number of threads for Davidson's algorithm
# export QP_NTHREADS_DAVIDSON=32
# Target number of threads for the computation of the PT2
# export QP_NTHREADS_PT2=32
# Name of the network interface to be chosen
# export QP_NIC=ib0

3
etc/ninja.rc Normal file
View File

@ -0,0 +1,3 @@
# Configuration for the Ninja package
export NINJA=${QP_ROOT}/bin/ninja

12
etc/ocaml.rc Normal file
View File

@ -0,0 +1,12 @@
# Configuration for the OCaml compiler
if [[ -z $OPAMROOT ]]
then
# Comment these lines if you have a system-wide OCaml installation
export OPAMROOT=${QP_ROOT}/external/opam
fi
source ${OPAMROOT}/opam-init/init.sh > /dev/null 2> /dev/null || true

43
etc/paths.rc Normal file
View File

@ -0,0 +1,43 @@
# Configuration of all the paths to executables and libraries
if [[ -z $QP_PYTHON ]]
then
# Load dependencies
for i in ezfio.rc irpf90.rc network.rc ninja.rc ocaml.rc
do
source $i
done
export QP_PYTHON=${QP_PYTHON}:${QP_EZFIO}/Python
export QP_PYTHON=${QP_PYTHON}:${QP_ROOT}/scripts
export QP_PYTHON=${QP_PYTHON}:${QP_ROOT}/scripts/ezfio_interface
export QP_PYTHON=${QP_PYTHON}:${QP_ROOT}/scripts/utility
export QP_PYTHON=${QP_PYTHON}:${QP_ROOT}/scripts/module
export QP_PYTHON=${QP_PYTHON}:${QP_ROOT}/scripts/compilation
export QP_PYTHON=${QP_PYTHON}:${QP_ROOT}/external/Python
export QP_PYTHON=${QP_PYTHON}:${QP_ROOT}/external/Python/resultsFile
function qp_prepend_export () {
eval "value_1="\${$1}""
if [[ -z $value_1 ]] ; then
echo "${2}:"
else
echo "${2}:${value_1}"
fi
}
export PYTHONPATH=$(qp_prepend_export "PYTHONPATH" "${QP_EZFIO}/Python":"${QP_PYTHON}")
export PATH=$(qp_prepend_export "PATH" "${QP_PYTHON}":"${QP_ROOT}"/bin:"${QP_ROOT}"/ocaml)
export LD_LIBRARY_PATH=$(qp_prepend_export "LD_LIBRARY_PATH" "${QP_ROOT}"/lib:"${QP_ROOT}"/lib64)
export LIBRARY_PATH=$(qp_prepend_export "LIBRARY_PATH" "${QP_ROOT}"/lib:"${QP_ROOT}"/lib64)
export C_INCLUDE_PATH=$(qp_prepend_export "C_INCLUDE_PATH" "${QP_ROOT}"/include)
fi

159
etc/qp.rc Normal file
View File

@ -0,0 +1,159 @@
# Configuration of the qp shell command
if [[ "$(ps -p $$ -ocomm=)" == "zsh" ]] ; then
autoload bashcompinit
bashcompinit
fi
source ${QP_ROOT}/etc/ezfio.rc
function _qp_usage()
{
echo "
Usage:
qp set_file EZFIO_DIRECTORY
qp unset_file
qp has DIRECTORY ITEM
qp get DIRECTORY ITEM
qp set DIRECTORY ITEM VALUE : Scalar values
qp set DIRECTORY ITEM : Array values read from stdin
qp run PROGRAM
qp srun PROGRAM
qp mpirun PROGRAM
qp set_frozen_core
qp create_ezfio_from_xyz -help
qp set_mo_class -help
"
}
function qp()
{
case $1 in
"has"|"set"|"get"|"set_file"|"unset_file")
ezfio $@
;;
"set_frozen_core")
shift
qp_set_frozen_core ${EZFIO_FILE}
;;
"create_ezfio_from_xyz")
shift
[[ -n $EZFIO_FILE ]] && ezfio unset_file
NAME=$(qp_create_ezfio_from_xyz $@)
ezfio set_file $NAME
;;
"set_mo_class")
shift
qp_set_mo_class ${EZFIO_FILE} $@
;;
"edit")
shift
qp_edit ${EZFIO_FILE}
;;
"run")
shift
qp_run $@ ${EZFIO_FILE}
;;
"srun")
shift
qp_srun $@ ${EZFIO_FILE}
;;
"mpirun")
shift
qp_mpirun $@ ${EZFIO_FILE}
;;
*)
_qp_usage
;;
esac
}
_Complete()
{
local cur
COMPREPLY=()
cur="${COMP_WORDS[COMP_CWORD]}"
prev="${COMP_WORDS[COMP_CWORD-1]}"
prev2="${COMP_WORDS[COMP_CWORD-2]}"
if [[ -n ${EZFIO_FILE} && -d ${EZFIO_FILE} ]]
then
case "${prev2}" in
"set"|has|get)
if [[ ${prev} == "qp" ]] ; then
COMPREPLY=( $(compgen -W "set set_frozen_core set_mo_class" -- $cur ) )
elif [[ ! -d ${EZFIO_FILE}/${prev} ]] ; then
COMPREPLY=( $(compgen -W "" -- $cur ) )
else
COMPREPLY=( $(compgen -W "$(cd ${EZFIO_FILE}/${prev} ; ls | sed 's/\.gz//' )" -- $cur ) )
fi
return 0
;;
*)
COMPREPLY=( $(compgen -W "$(\ls)" -- $cur ) )
esac
case "${prev}" in
run|srun|mpirun)
COMPREPLY=( $(compgen -W "$(cat ${QP_ROOT}/data/executables | cut -d ' ' -f 1)" -- $cur ) )
return 0
;;
unset_file|edit|set_frozen_core)
COMPREPLY=()
return 0
;;
set_mo_class)
COMPREPLY=( $(compgen -W "-h -core -inact -act -virt -del" -- $cur ) )
return 0
;;
set|has|get)
COMPREPLY=( $(compgen -W "$(cd ${EZFIO_FILE} ; \ls -d */ | sed 's|/||g')" -- $cur ) )
return 0
;;
*)
COMPREPLY=( $(compgen -W 'has get set unset_file edit \
run srun mpirun set_frozen_core \
set_mo_class create_ezfio_from_xyz \
-h' -- $cur ) )
return 0
;;
esac
else
case "${prev}" in
set_file)
COMPREPLY=( $(compgen -W "$(\ls -d */ | sed 's|/||g')" -- ${cur} ) )
return 0
;;
create_ezfio_from_xyz)
COMPREPLY=( $(compgen -W "$(\ls)" -- ${cur} ) )
return 0
;;
*)
COMPREPLY=( $(compgen -W 'set_file \
create_ezfio_from_xyz \
-h' -- $cur ) )
return 0
;;
esac
fi
}
complete -F _Complete qp

2
ocaml/qp_create_ezfio_from_xyz.ml
View File

@ -660,7 +660,7 @@ let run ?o b au c d m p cart xyz_file =
end;
raise ex;
end
in ()
in print_endline ezfio_file

11
qpsh Executable file
View File

@ -0,0 +1,11 @@
#!/bin/bash
export QP_ROOT=$(dirname $0)
exec bash --init-file <(cat << EOF
PS1="[\u@\h \W] |\\\${EZFIO_FILE}> "
source $QP_ROOT/quantum_package.rc
EOF
)

37
quantum_package.rc Normal file
View File

@ -0,0 +1,37 @@
#!/bin/bash
# This script loads the Quantum Package environment. It should be invoked as
# source quantum_package.rc
QP_ROOT="/NOT/FOUND"
case "$(ps -p $$ -ocomm=)" in
"zsh")
QP_ROOT=$(dirname $0)
;;
*)
QP_ROOT=$(dirname ${BASH_SOURCE}[0])
;;
esac
export QP_ROOT="$( cd $QP_ROOT ; pwd -P )"
WD=$PWD
cd $QP_ROOT/etc
for SCRIPT in *.rc
do
source $SCRIPT
done
cd $WD
if [[ -z ${QP_ROOT} ]]
then
echo The QP_ROOT environment variable is not set. Run
./configure --help
fi

57
quantum_package.rc.default
View File

@ -1,57 +0,0 @@
# This line should contain the location of your quantum_package directory
export QP_ROOT=
# Include here the optional external libraries to link with your binaries
# export QP_LIB=" -lint"
#### BEGIN Don't modify
export IRPF90=${QP_ROOT}/bin/irpf90
export NINJA=${QP_ROOT}/bin/ninja
export QP_EZFIO=${QP_ROOT}/external/ezfio
export QP_PYTHON=${QP_ROOT}/scripts:${QP_ROOT}/scripts/ezfio_interface:${QP_ROOT}/scripts/utility:${QP_ROOT}/scripts/module:${QP_ROOT}/scripts/pseudo:${QP_ROOT}/scripts/compilation:${QP_EZFIO}/Python:${QP_ROOT}/external/Python:${QP_ROOT}/external/Python/resultsFile
function qp_prepend_export () {
eval "value_1="\${$1}""
if [[ -z $value_1 ]] ; then
echo "${2}:"
else
echo "${2}:${value_1}"
fi
}
export PYTHONPATH=$(qp_prepend_export "PYTHONPATH" "${QP_EZFIO}/Python":"${QP_PYTHON}")
export PATH=$(qp_prepend_export "PATH" "${QP_PYTHON}":"${QP_ROOT}"/bin:"${QP_ROOT}"/ocaml)
export LD_LIBRARY_PATH=$(qp_prepend_export "LD_LIBRARY_PATH" "${QP_ROOT}"/lib:"${QP_ROOT}"/lib64)
export LIBRARY_PATH=$(qp_prepend_export "LIBRARY_PATH" "${QP_ROOT}"/lib:"${QP_ROOT}"/lib64)
export C_INCLUDE_PATH=$(qp_prepend_export "C_INCLUDE_PATH" "${QP_ROOT}"/include)
function qp() {
ezfio $@
}
# EZFIO completion
if [[ $SHELL == "/bin/bash" ]] ; then
if [[ -f ${QP_EZFIO}/Bash/ezfio.sh ]]; then
source ${QP_EZFIO}/Bash/ezfio.sh
else
echo "EZFIO is not installed."
fi
fi
#### END Don't modify
# Choose the correct network interface if the default one is incorrect
# export QP_NIC=ib0
# export QP_NIC=eth0
# Comment these lines if you have a system-wide OCaml installation
export OPAMROOT=${OPAMROOT:-${QP_ROOT}/external/opam}
source ${OPAMROOT}/opam-init/init.sh > /dev/null 2> /dev/null || true

3
src/cipsi/cipsi.irp.f
View File

@ -46,6 +46,9 @@ subroutine run_cipsi
psi_coef = psi_coef_sorted
N_det = N_det_max
soft_touch N_det psi_det psi_coef
if (s2_eig) then
call make_s2_eigenfunction
endif
call diagonalize_CI
call save_wavefunction
endif

59
src/cipsi/pt2_stoch_routines.irp.f
View File

@ -125,7 +125,7 @@ subroutine ZMQ_pt2(E, pt2,relative_error, error, variance, norm, N_in)
error(:) = 0.d0
else
N = max(N_in,1)
N = max(N_in,1) * N_states
state_average_weight_save(:) = state_average_weight(:)
call create_selection_buffer(N, N*2, b)
ASSERT (associated(b%det))
@ -185,7 +185,7 @@ subroutine ZMQ_pt2(E, pt2,relative_error, error, variance, norm, N_in)
ipos=1
do i= 1, N_det_generators
do j=1,pt2_F(pt2_J(i))
write(task(ipos:ipos+30),'(I9,1X,I9,1X,I9,''|'')') j, pt2_J(i), N
write(task(ipos:ipos+30),'(I9,1X,I9,1X,I9,''|'')') j, pt2_J(i), N_in
ipos += 30
if (ipos > 300000-30) then
if (add_task_to_taskserver(zmq_to_qp_run_socket,trim(task(1:ipos))) == -1) then
@ -213,15 +213,49 @@ subroutine ZMQ_pt2(E, pt2,relative_error, error, variance, norm, N_in)
endif
double precision :: mem_collector, mem, rss
call resident_memory(rss)
mem_collector = 8.d0 * & ! bytes
( 1.d0*pt2_n_tasks_max & ! task_id, index
+ 0.635d0*N_det_generators & ! f,d
+ 3.d0*N_det_generators*N_states & ! eI, vI, nI
+ 3.d0*pt2_n_tasks_max*N_states & ! eI_task, vI_task, nI_task
+ 4.d0*(pt2_N_teeth+1) & ! S, S2, T2, T3
+ 1.d0*(N_int*2.d0*N + N) & ! selection buffer
+ 1.d0*(N_int*2.d0*N + N) & ! sort selection buffer
) / 1024.d0**3
integer :: nproc_target
nproc_target = nproc
double precision :: mem
mem = 8.d0 * N_det * (N_int * 2.d0 * 3.d0 + 3.d0 + 5.d0) / (1024.d0**3)
call write_double(6,mem,'Estimated memory/thread (Gb)')
if (qp_max_mem > 0) then
nproc_target = max(1,int(dble(qp_max_mem)/mem))
nproc_target = min(nproc_target,nproc)
endif
nproc_target = nthreads_pt2
do
mem = mem_collector + & !
nproc_target * 8.d0 * & ! bytes
( 0.5d0*pt2_n_tasks_max & ! task_id
+ 64.d0*pt2_n_tasks_max & ! task
+ 3.d0*pt2_n_tasks_max*N_states & ! pt2, variance, norm
+ 1.d0*pt2_n_tasks_max & ! i_generator, subset
+ 2.d0*(N_int*2.d0*N_in + N_in) & ! selection buffers
+ 1.d0*(N_int*2.d0*N_in + N_in) & ! sort/merge selection buffers
) / 1024.d0**3
if (nproc_target == 0) then
call check_mem(mem,irp_here)
nproc_target = 1
exit
endif
if (mem+rss < qp_max_mem) then
exit
endif
nproc_target = nproc_target - 1
enddo
call write_int(6,nproc_target,'Number of threads for PT2')
call write_double(6,mem,'Memory (Gb)')
call omp_set_nested(.false.)
@ -253,8 +287,11 @@ subroutine ZMQ_pt2(E, pt2,relative_error, error, variance, norm, N_in)
FREE pt2_stoch_istate
if (N_in > 0) then
b%cur = min(N_in,b%cur)
if (s2_eig) then
call make_selection_buffer_s2(b)
else
call remove_duplicates_in_selection_buffer(b)
endif
call fill_H_apply_buffer_no_selection(b%cur,b%det,N_int,0)
endif
@ -326,6 +363,8 @@ subroutine pt2_collector(zmq_socket_pull, E, relative_error, pt2, error, &
rss += memory_of_double(pt2_N_teeth+1)*4.d0
call check_mem(rss,irp_here)
! If an allocation is added here, the estimate of the memory should also be
! updated in ZMQ_pt2
allocate(task_id(pt2_n_tasks_max), index(pt2_n_tasks_max), f(N_det_generators))
allocate(d(N_det_generators+1))
allocate(eI(N_states, N_det_generators), eI_task(N_states, pt2_n_tasks_max))

110
src/cipsi/selection_buffer.irp.f
View File

@ -10,7 +10,7 @@ subroutine create_selection_buffer(N, siz_, res)
siz = max(siz_,1)
double precision :: rss
double precision, external :: memory_of_double, memory_of_int
double precision, external :: memory_of_double
rss = memory_of_double(siz)*(N_int*2+1)
call check_mem(rss,irp_here)
@ -171,7 +171,7 @@ subroutine make_selection_buffer_s2(b)
n_d = b%cur
double precision :: rss
double precision, external :: memory_of_double, memory_of_int
double precision, external :: memory_of_double
rss = (4*N_int+4)*memory_of_double(n_d)
call check_mem(rss,irp_here)
allocate(o(N_int,2,n_d), iorder(n_d), duplicate(n_d), bit_tmp(n_d), &
@ -294,8 +294,112 @@ subroutine make_selection_buffer_s2(b)
if (k > n_d) exit
enddo
deallocate(o)
b%N = n_d
b%cur = n_d
b%N = n_d
end
subroutine remove_duplicates_in_selection_buffer(b)
use selection_types
type(selection_buffer), intent(inout) :: b
integer(bit_kind), allocatable :: o(:,:,:)
double precision, allocatable :: val(:)
integer :: n_d
integer :: i,k,sze,n_alpha,j,n
logical :: dup
! Sort
integer, allocatable :: iorder(:)
integer*8, allocatable :: bit_tmp(:)
integer*8, external :: det_search_key
integer(bit_kind), allocatable :: tmp_array(:,:,:)
logical, allocatable :: duplicate(:)
n_d = b%cur
logical :: found_duplicates
double precision :: rss
double precision, external :: memory_of_double
rss = (4*N_int+4)*memory_of_double(n_d)
call check_mem(rss,irp_here)
found_duplicates = .False.
allocate(iorder(n_d), duplicate(n_d), bit_tmp(n_d), &
tmp_array(N_int,2,n_d), val(n_d) )
do i=1,n_d
iorder(i) = i
bit_tmp(i) = det_search_key(b%det(1,1,i),N_int)
enddo
call i8sort(bit_tmp,iorder,n_d)
do i=1,n_d
do k=1,N_int
tmp_array(k,1,i) = b%det(k,1,iorder(i))
tmp_array(k,2,i) = b%det(k,2,iorder(i))
enddo
val(i) = b%val(iorder(i))
duplicate(i) = .False.
enddo
! Find duplicates
do i=1,n_d-1
if (duplicate(i)) then
cycle
endif
j = i+1
do while (bit_tmp(j)==bit_tmp(i))
if (duplicate(j)) then
j+=1
if (j>n_d) then
exit
endif
cycle
endif
dup = .True.
do k=1,N_int
if ( (tmp_array(k,1,i) /= tmp_array(k,1,j)) &
.or. (tmp_array(k,2,i) /= tmp_array(k,2,j)) ) then
dup = .False.
exit
endif
enddo
if (dup) then
duplicate(j) = .True.
found_duplicates = .True.
endif
j+=1
if (j>n_d) then
exit
endif
enddo
enddo
if (found_duplicates) then
! Copy filtered result
integer :: n_p
n_p=0
do i=1,n_d
if (duplicate(i)) then
cycle
endif
n_p = n_p + 1
do k=1,N_int
b%det(k,1,n_p) = tmp_array(k,1,i)
b%det(k,2,n_p) = tmp_array(k,2,i)
enddo
val(n_p) = val(i)
enddo
b%cur=n_p
b%N=n_p
endif
end

3
src/cipsi/stochastic_cipsi.irp.f
View File

@ -48,6 +48,9 @@ subroutine run_stochastic_cipsi
psi_coef = psi_coef_sorted
N_det = N_det_max
soft_touch N_det psi_det psi_coef
if (s2_eig) then
call make_s2_eigenfunction
endif
call diagonalize_CI
call save_wavefunction
endif

4
src/davidson/davidson_parallel.irp.f
View File

@ -452,10 +452,10 @@ BEGIN_PROVIDER [ integer, nthreads_davidson ]
END_DOC
nthreads_davidson = nproc
character*(32) :: env
call getenv('NTHREADS_DAVIDSON',env)
call getenv('QP_NTHREADS_DAVIDSON',env)
if (trim(env) /= '') then
read(env,*) nthreads_davidson
call write_int(6,nthreads_davidson,'Number of threads for <Psi|H|Psi>')
call write_int(6,nthreads_davidson,'Target number of threads for <Psi|H|Psi>')
endif
END_PROVIDER

51
src/davidson/diagonalization_hs2_dressed.irp.f
View File

@ -115,7 +115,7 @@ subroutine davidson_diag_hjj_sjj(dets_in,u_in,H_jj,s2_out,energies,dim_in,sze,N_
integer :: iter2, itertot
double precision, allocatable :: W(:,:), U(:,:), S(:,:), overlap(:,:)
double precision, allocatable :: y(:,:), h(:,:), lambda(:), s2(:)
double precision, allocatable :: c(:), s_(:,:), s_tmp(:,:)
double precision, allocatable :: s_(:,:), s_tmp(:,:)
double precision :: diag_h_mat_elem
double precision, allocatable :: residual_norm(:)
character*(16384) :: write_buffer
@ -137,6 +137,13 @@ subroutine davidson_diag_hjj_sjj(dets_in,u_in,H_jj,s2_out,energies,dim_in,sze,N_
itermax = max(3,min(davidson_sze_max, sze/N_st_diag))
itertot = 0
if (state_following) then
allocate(overlap(N_st_diag*itermax, N_st_diag*itermax))
else
allocate(overlap(1,1)) ! avoid 'if' for deallocate
endif
overlap = 0.d0
PROVIDE nuclear_repulsion expected_s2 psi_bilinear_matrix_order psi_bilinear_matrix_order_reverse
call write_time(6)
@ -149,25 +156,51 @@ subroutine davidson_diag_hjj_sjj(dets_in,u_in,H_jj,s2_out,energies,dim_in,sze,N_
call write_int(6,N_st,'Number of states')
call write_int(6,N_st_diag,'Number of states in diagonalization')
call write_int(6,sze,'Number of determinants')
! Find max number of cores to fit in memory
! -----------------------------------------
nproc_target = nproc
double precision :: rss
integer :: maxab
maxab = max(N_det_alpha_unique, N_det_beta_unique)+1
call resident_memory(rss)
r1 = 8.d0*(3.d0*dble(sze*N_st_diag*itermax+5.d0*(N_st_diag*itermax)**2 &
+ 3.d0*(N_st_diag*itermax)+nproc*(4.d0*N_det_alpha_unique+2.d0*N_st_diag*sze)))/(1024.d0**3)
do while (r1+rss > qp_max_mem)
nproc_target = nproc_target - 1
r1 = 8.d0*(3.d0*dble(sze*N_st_diag*itermax+5.d0*(N_st_diag*itermax)**2 &
+ 3.d0*(N_st_diag*itermax)+nproc_target*(4.d0*N_det_alpha_unique+2.d0*N_st_diag*sze)))/(1024.d0**3)
do
r1 = 8.d0 * &! bytes
( 3.d0*(dble(sze)*(N_st_diag*itermax)) &! W,U,S
+ 4.d0*(N_st_diag*itermax)**2 &! h,y,s_,s_tmp
+ 2.d0*(N_st_diag*itermax) &! s2,lambda
+ 1.d0*(N_st_diag) &! residual_norm
! In H_S2_u_0_nstates_zmq
+ 3.d0*(N_st_diag*N_det) &! u_t, v_t, s_t on collector
+ 3.d0*(N_st_diag*N_det) &! u_t, v_t, s_t on slave
+ 0.5d0*maxab &! idx0 in H_S2_u_0_nstates_openmp_work_*
+ nproc_target * &! In OMP section
( 1.d0*(N_int*maxab) &! buffer
+ 3.5d0*(maxab) ) &! singles_a, singles_b, doubles, idx
) / 1024.d0**3
if (nproc_target == 0) then
call check_mem(r1,irp_here)
nproc_target = 1
exit
endif
if (r1+rss < qp_max_mem) then
exit
endif
nproc_target = nproc_target - 1
enddo
nthreads_davidson = nproc_target
TOUCH nthreads_davidson
call write_int(6,nproc_target,'Number of threads for diagonalization')
call write_double(6, r1, 'Memory(Gb)')
!---------------
write(6,'(A)') ''
write_buffer = '====='
do i=1,N_st
@ -198,9 +231,7 @@ subroutine davidson_diag_hjj_sjj(dets_in,u_in,H_jj,s2_out,energies,dim_in,sze,N_
s_(N_st_diag*itermax,N_st_diag*itermax), &
s_tmp(N_st_diag*itermax,N_st_diag*itermax), &
residual_norm(N_st_diag), &
c(N_st_diag*itermax), &
s2(N_st_diag*itermax), &
overlap(N_st_diag*itermax, N_st_diag*itermax), &
lambda(N_st_diag*itermax))
h = 0.d0
@ -503,7 +534,7 @@ subroutine davidson_diag_hjj_sjj(dets_in,u_in,H_jj,s2_out,energies,dim_in,sze,N_
deallocate ( &
W, residual_norm, &
U, overlap, &
c, S, &
S, &
h, &
y, s_, s_tmp, &
lambda &

14
src/fci/environment.irp.f Normal file
View File

@ -0,0 +1,14 @@
BEGIN_PROVIDER [ integer, nthreads_pt2 ]
implicit none
BEGIN_DOC
! Number of threads for Davidson
END_DOC
nthreads_pt2 = nproc
character*(32) :: env
call getenv('QP_NTHREADS_PT2',env)
if (trim(env) /= '') then
read(env,*) nthreads_pt2
call write_int(6,nthreads_pt2,'Target number of threads for PT2')
endif
END_PROVIDER