The fruitful interplay of high-resolution spectroscopy and quantum chemistry has a long history, especially in the field of small, semi-rigid molecules. However, in recent years, the targets of spectroscopic studies are shifting toward flexible molecules, characterized by a large number of closely spaced energy minima, all contributing to the overall spectrum. Here, artificial intelligence comes into play since it is at the basis of powerful unsupervised techniques for the exploration of soft degrees of freedom. Integration of such algorithms with a two-stage QM/QM′ (Quantum Mechanical) exploration/refinement strategy driven by a user-friendly graphical interface is the topic of the present paper. We will address in particular: (i) the performances of different semi-empirical methods for the exploration step and (ii) the comparison between stochastic and meta-heuristic algorithms in achieving a cheap yet complete exploration of the conformational space for medium sized chromophores. As test cases, we choose three amino acids of increasing complexity, whose full conformer enumeration has been reached only very recently. Next, we show that systems in condensed phases can be treated at the same level and with the same efficiency when employing a polarizable continuum description of the solvent. Finally, the challenging issue represented by the vibrational circular dichroism spectra of some rhodium complexes with flexible ligands has been addressed, showing that our fully unsupervised approach leads to remarkable agreement with the experiment.
Unsupervised search of low-lying conformers with spectroscopic accuracy: A two-step algorithm rooted into the island model evolutionary algorithm
Note: This paper is part of the JCP Special Topic on Machine Learning Meets Chemical Physics.
Giordano Mancini, Marco Fusè, Federico Lazzari, Balasubramanian Chandramouli, Vincenzo Barone; Unsupervised search of low-lying conformers with spectroscopic accuracy: A two-step algorithm rooted into the island model evolutionary algorithm. J. Chem. Phys. 28 September 2020; 153 (12): 124110. https://doi.org/10.1063/5.0018314
Download citation file: