DIEinPEACE · Double Incremental Expansion in Potential Energies from Automized Computational Exploration

Modern computational methods of quantum chemistry are valuable and well-established tools for interpretations, refinements, and even predictions of experimental results. Recent advances within linear-scaling (with the system size) approaches allowed routine and efficient treatments of electronic structures of much larger molecular systems than those accessible in previous decades. This has the potential to extend the applicability of quantum chemistry to very large biomolecules. However, reaching a close to linear-scaling behavior for a single point calculation is by no means near to providing an efficient description of the total potential energy surface. Because potential energy surfaces are cornerstones for obtaining a detailed knowledge of reactivity, photochemical properties, vibrational motion, etc., development of a computationally inexpensive but accurate quantum chemical methodology for potential energy surface calculations of large biomolecules (such as proteins) is of extreme importance for chemical science. The proposed project aims at filling this gap by developing an ab initio, linear-scaling, and ""black-box"" machinery for protein potential energy surfaces calculations, where the linear-scaling refers to the total computational cost. This will be achieved by combining ideas of partitioning the total system into subsystems and incremental expansions of potential energy surfaces with efficient and accurate computational algorithms and modern concepts of machine leaning. The proposed strategy will enable theoretical spectra simulations for much larger biomolecules. This will significantly advance the current stage of the field and help to reveal many new and intricate details about structures and dynamics of proteins.""