Installation¶
You can install geomeTRIC with conda
, with pip
, or by installing from source.
Conda¶
You can install or update geomeTRIC using the conda
package manager.
conda install geometric -c conda-forge
This installs geomeTRIC and its dependencies.
The geomeTRIC package is maintained on the conda-forge channel.
Pip¶
To install geomeTRIC with pip
pip install geometric
Install from Source¶
To install geomeTRIC from source, clone the repository from github:
git clone https://github.com/lpwgroup/geomeTRIC.git
cd geomeTRIC
python setup.py install
Dependencies¶
The required packages for geomeTRIC are as follows. Older versions of packages may work, no guarantees.
Python : Versions 2.7, 3.5, 3.6, 3.7 are supported. Basic functionalities tested on 3.6 to 3.10.
NumPy: Version 1.15 or above
NetworkX : Version 2.2 or above
You will also need at least one quantum chemistry (QM) or molecular mechanics (MM) package to evaluate energies and gradients. This requirement is already satisfied if you are using software like QCFractal or PySCF which uses geomeTRIC as a library.
Supported QM packages are as follows. Older versions of packages may work, no guarantees.
Q-Chem : Version 4.2 or above
TeraChem : Version 1.5 or above
Psi4 : Version 1.2 or above
Molpro : Version 2015 or 2019
Gaussian : Version 09 or 16
Supported MM packages are:
OpenMM : Version 6.3 or above
Gromacs : Version 4.6.7 or 5.1.4. Newer versions are not tested.
Additionally, a number of QM, MM, semiempirical and machine learning potentials are supported through the QCEngine package. Version 0.15.0 or above required.
A testing environment can be set up using conda
which includes OpenMM, Psi4, QCEngine, and RDKit. This can be accomplished by:
cd geomeTRIC
python devtools/scripts/create_conda_env.py -n=geo_test -p=3.7 devtools/conda-envs/omm_psi4_rdkit.yaml
Testing¶
Test geomeTRIC with pytest
:
cd geomeTRIC
pytest
More information is available in the developer documentation.
Installation of cctools¶
The Work Queue library in the CCTools package is utilized to provide distributed computing features in geomeTRIC, primarily the computation of numerical Hessian matrices.
Installation of cctools
can be done easily via conda as follows
conda install -c conda-forge ndcctools