Code repo for convenient usage of methods developed in the chemspacelab group. Successor to qmlcode/qml repo that also draws a lot of inspiration from qmllib and qml-lightning repos.
Install with:
pip install .or, if makefile is installed,
make installSome parts of the code depend on additional dependencies that can be installed with the corresponding optional dependecies flag. The three defined for this package are
-
orb_ml- for FJK (machine learning from orbital information) -
msorf- for MSORF (everything inqml2.multilevel_sorf) -
torch- Torch functionality (efficiency questionable right now TBH)
For example, to use the orb_ml and msorf optional dependecy flags in your installation use
pip install .[orb_ml,msorf]or, if makefile is installed,
make install OPT=[orb_ml,msorf]
To check that the installed repo works correctly run
make testNOTE: The command assumes that python environmental variable points towards a valid executable. If you use an environment alias change definition of the python variable in the beginning of the Makefile.
API documentation can be generated with Doxygen by running
make docsor, if Makefile is not installed, running
python doxygen/run_doxygen.pyThis will create manual.html file that can be opened with an Internet browser.
NUMBA_NUM_THREADS - since the code is written in Numba OpenMP parallelization is mostly controlled via this variable.
QML2_NUM_PROCS - number of processes spawned by parts of the code parallelized via python.multiprocessing (training set representations in model-related classes, pyscf calculations made by OML_Compound_list attribute calls). For limiting number of OpenMP threads spawned in turn by these processes use suppress options (such as KRRModel class's training_reps_suppress_openmp option). Also see parallelization.set_default_num_procs.
QML2_DEFAULT_JIT - setting to NUMBA (default) or TORCH (both are case insensitive) determines whether Numba or TorchScript JIT compilation is used. Also see jit_interfaces.set_default_jit.
QML2_DEBUG - if 1 add debug=True to all Numba @njit instances.
QML2_SKIP_JIT - if 1 do not use JIT.
QML2_SKIP_JIT_FAILURES - if 1 TorchScript does not terminate when encountering uncompilable part of the code.
We use several packages that maximize code readability, listed in requirements-dev.txt; hence should you decide to commit make sure you have a conda environment you are prepared to modify. Having, for example, created a fresh conda environment named qml2dev with
conda create --name qml2dev
conda activate qml2devone prepares the environment and the pre-commit scripts in the repository with
make dev-setupThis allows automatic formatting/readability checks for the committed code. It is also possible to enforce adherence to Conventional Commits format of commit messages inside your fork with
make conventional-commitsIf you plan to use qml2 regularly consider contacting the developers to join their Slack channel for quick means of communication.
qml2-dev is the private "developer" version of qml2 repository created to allow package developers creation and easy management of private forks, as well as sharing messier code drafts. If you interact with the code and one of the package authors on a regular basis consider asking them to be added; though in general we try to prevent qml2 from lagging behind qml2-dev for more than two months.