tuthmose/IM_EA

Conformer search with different metaheuristics and wrappers to Electronic structure codes.

Both Simulated annealing and Evolutionary algorithm codes can be used to optimize arbitrary cost functions (which must be provided) and can operate on numpy arrays or arbitrary objects called specimens. In the latter case, genetic operations are still done on chromosomes; modifications to chromosomes are transmitted back and forth to specimens via the custom fitness function.

For the esploration of conformer space, the search can be made on rotatable bonds (with or without local convex optimization in G16) (zmat_specimen) or directly on cartesian coordinates (cluster_specimen; experimental).

• molecule_utils include a parser for Gaussian16 and a zmatrix module for internal coordinate manipulation and z-matrix/cartesian conversion.
• EvolutionaryAlgorithms contains ga_population and ga_evolution which implements an island $\lambda,\mu$ model.
• SimulatedAnnealing implements a simple SA algorithm; custom functions for the cooling scheme and generation of neighbours can be provided.
• test containts a few examples jupyter notebook and shell scripts with submission and post processing pipelines

please cite one or more of the following studies:

1. Mancini, G.; Fusè, M.; Lazzari, F.; Chandramouli, B.; Barone, V., Fast exploration of potential energy surfaces with a joint venture of quantum chemistry, evolutionary algorithms and unsupervised learning, Digital Discovery, 2022,1, 790-805, https://doi.org/10.1039/D2DD00070A
2. Mancini, G.; Fusè, M.; Lipparini, F.; Nottoli, M.; Scalmani, G.; Barone, V. Molecular Dynamics Simulations Enforcing Nonperiodic Boundary Conditions: New Developments and Application to the Solvent Shifts of Nitroxide Magnetic Parameters. J. Chem. Theory Comput. 2022, acs.jctc.2c00046. https://doi.org/10.1021/acs.jctc.2c00046.
3. Mancini, G.; Fusè, M.; Lazzari, F.; Chandramouli, B.; Barone, V. Unsupervised Search of Low-Lying Conformers with Spectroscopic Accuracy: A Two-Step Algorithm Rooted into the Island Model Evolutionary Algorithm. J. Chem. Phys. 2020, 153 (12), 124110. https://doi.org/10.1063/5.0018314.
4. Mancini, G.; Del Galdo, S.; Chandramouli, B.; Pagliai, M.; Barone, V. Computational Spectroscopy in Solution by Integration of Variational and Perturbative Approaches on Top of Clusterized Molecular Dynamics. J. Chem. Theory Comput. 2020, 16 (9), 5747–5761. https://doi.org/10.1021/acs.jctc.0c00454.