1
0
mirror of https://github.com/TREX-CoE/qmckl.git synced 2024-11-19 12:32:40 +01:00
Go to file
2022-10-13 11:53:49 +02:00
.github/workflows Fix Verificarlo CI by setting the safe.directory git variable 2022-09-23 12:02:14 +02:00
cmake Create FindQMCKL module 2022-03-04 11:27:59 +01:00
include Added new test 2021-10-13 10:10:09 +02:00
m4 Add SWIG and pyqmckl ingredients for Autotools 2022-05-03 16:48:27 +02:00
man Add doc in dist, and text doc 2021-04-21 12:44:03 +02:00
org Openmp in AOs if more than 16 points to compute 2022-10-13 11:53:49 +02:00
pkgconfig Clean configure 2021-06-03 01:32:50 +02:00
python Add qmckl_adjugate_safe proxy function 2022-08-22 17:31:15 +02:00
share Add emacs-htmlize as git submodule 2022-07-28 11:11:14 +02:00
src Cleaning 2021-05-12 02:20:39 +02:00
to_be_processed Added a to_be_processes directory 2020-10-31 19:01:26 +01:00
tools Added dependency to config.h in makefiles 2022-10-11 17:56:28 +02:00
.gitignore Add the qmckl_probes interface (#2) 2021-07-23 12:01:14 +02:00
.gitmodules Add emacs-htmlize as git submodule 2022-07-28 11:11:14 +02:00
AUTHORS Automake working 2021-05-10 23:56:26 +02:00
autogen.sh python -> python3 2022-06-07 15:55:48 +02:00
ChangeLog Automake working 2021-05-10 23:56:26 +02:00
configure.ac Add options for PGCC and PGF90 compilers 2022-10-13 10:30:56 +02:00
COPYING Working on automake 2021-05-11 11:45:49 +02:00
INSTALL Automake working 2021-05-10 23:56:26 +02:00
LICENSE Initial commit 2020-10-13 15:08:11 +02:00
Makefile.am Moved fortran source in include 2022-10-07 14:33:27 +02:00
NEWS Automake working 2021-05-10 23:56:26 +02:00
README Working on automake 2021-05-11 11:45:49 +02:00
README.md Add the python API section 2022-08-31 16:45:11 +02:00
vfc_tests_config.json Modify the vfc_ci wrapper and its config 2022-09-27 13:37:06 +00:00

QMCkl: Quantum Monte Carlo Kernel Library

Build Status

The domain of quantum chemistry needs a library in which the main kernels of Quantum Monte Carlo (QMC) methods are implemented. In the library proposed in this project, we expose the main algorithms in a simple language and provide a standard API and tests to enable the development of high-performance QMCkl implementations taking advantage of modern hardware.

See the source code to read the documentation.

To clone the repository, use:

git clone https://github.com/TREX-CoE/qmckl.git

Installation

The simplest way to obtain the source files of QMCkl is to download a source distribution. This particular repository is for maintainers, who write the kernels in org-mode files and produce the source code and the documentation from these files.

For maintainers

./autogen.sh
./configure --prefix=$PWD/_install

make
make check

For users

Obtain a source distribution.

To build the documentation version:

./configure

To build an optimized version with Intel compilers:

./configure \
   --with-icc \
   --with-ifort \
   --enable-hpc \
   --with-openmp

To build an optimized version with GCC:

./configure \
  CC=gcc \
  CFLAGS="-g -O2 -march=native  -flto -fno-trapping-math -fno-math-errno -ftree-vectorize" \
  FC=gfortran \
  FCFLAGS="-g -O2 -march=native  -flto -ftree-vectorize" \
  --enable-hpc \
  --with-openmp

Then, compile with:

make -j
make -j check
sudo make install
sudo make installcheck

Python API

  • SWIG (>= 4.0) is required to build the Python API for maintainers

In order to install the qmckl Python package, first install the shared C library libqmckl following the installation guide above and then run the following command:

make python-install

To test the installation, run

make python-test

Minimal example demonstrating the use of the qmckl Python API can be found in the test_api.py file.

We highly recommend to use virtual environments to avoid compatibility issues and to improve reproducibility.

Installation procedure for Guix users

QMCkl can be installed with the GNU Guix functional package manager. The qmckl.scm Schema file contains the manifest specification for the qmckl installations. It can be installed within the selected $GUIX_PROFILE as follows:

guix package \
    --profile=$GUIX_PROFILE         \
    --load-path=<path_to_trexio_scm>    \
    --cores=<n_cores>           \
    --install-from-file=qmckl.scm

where <path_to_trexio_scm> should point to a folder, which contains the TREXIO manifest file trexio.scm (e.g. ~/trexio/tools/ if TREXIO repository was cloned under $HOME).

Installation procedures for both development version (qmckl-dev) and stable releases (qmckl-hpc) are provided. One can switch between them using the return value (last line) in the qmckl.scm file.

Linking to your program

The make install command takes care of installing the QMCkl shared library on the user machine. Once installed, add -lqmckl to the list of compiler options.

In some cases (e.g. when using custom prefix during configuration), the QMCkl library might end up installed in a directory, which is absent in the default $LIBRARY_PATH. In order to link the program against QMCkl, the search paths can be modified as follows:

export LIBRARY_PATH=$LIBRARY_PATH:<path_to_qmckl>/lib

(same holds for $LD_LIBRARY_PATH). The <path_to_qmckl> has to be replaced with the prefix used during the installation.

If your project relies on the CMake build system, feel free to use the FindQMCKL.cmake module to find and link the QMCkl library automatically.

Verificarlo CI

Since Verificarlo should not be a dependency of QMCkl, all Verificarlo functions are called only when the support is explicitely enabled (and ignored by the preprocessor otherwise). To enable vfc_ci support, the library should be configured with the following command :

./configure \
  CC="verificarlo-f" \
  FC="verificarlo-f" \
  --prefix=$PWD/_install \
  --enable-vfc_ci \
  --host=x86_64 \

where CC and FC are set to verificarlo-f, and support is explicitely enabled with the enable-vfc_ci flag. Configuring the library with the “standard” command will cause all calls to Verificarlo related functions to be ignored, and the library will be built as usual.


European flag TREX: Targeting Real Chemical Accuracy at the Exascale project has received funding from the European Unions Horizon 2020 - Research and Innovation program - under grant agreement no. 952165. The content of this document does not represent the opinion of the European Union, and the European Union is not responsible for any use that might be made of such content.