Abstract

Today many quantum chemists are inspired by tasks such as overcoming the complexity of the electronic structure problem or bypassing DFT /wavefunction-based computations by leveraging machine learning (ML) techniques. Reported applications span from simple atomization energies to complex mathematical objects, such as the many body wavefunction. This talk discusses our strategies exploiting the interplay between quantum chemistry and statistical learning techniques.[1-4] We will discuss a ML workflow specifically adapted to predict the primary output of quantum chemical computations (e.g., ρ®, particle- hole densities, on-top pair density, etc.) from which many properties could be derived. These properties range from the electrostatic potential of a protein to a real-space indicator of electron correlation.[2-4] Emphasis will also be placed on the importance of the quantum chemical metrics chosen for the regression.[1] Finally and if time remains, we will discussed atomistic ML models aiming at more efficiently learning challenging energetic properties (e.g., the enantiomeric excess).

[1] Briling, K. R.; Fabrizio, A.; Corminboeuf, C. J. Chem. Phys. 2021, 155, 024107. [2] Vela, S.; Fabrizio, A.; Briling, K. R.; Corminboeuf, C. J. Phys. Chem. Lett. 2021, 12, 5957. [3] Fabrizio, A.; Briling K. R.; Girardier, D. D.; Corminboeuf, C. J. Chem. Phys. 2020, 153, 204111. [4] Fabrizio, A.; Grisafi, A.; Meyer, B.; Ceriotti, M.; Corminboeuf, C. Chem. Sci. 2019, 10, 9424.