Machine Learning of Dynamic Electron Correlation Energies from Topological Atoms

Author(s)

McDonagh, James L. ^[1] ; Silva, Arnaldo F. ^[1] ; Vincent, Mark A. ^[2] ; Popelier, Paul L. A. ^{[1, 2]}

Affiliation

^[1] Univ Manchester, Manchester Inst Biotechnol, 131 Princess St, Manchester M1 7DN, Lancs - England

^[2] Univ Manchester, Sch Chem, Oxford Rd, Manchester M13 9PL, Lancs - England

• Document type: Journal article
• Source: JOURNAL OF CHEMICAL THEORY AND COMPUTATION; v. 14, n. 1, p. 216-224, JAN 2018.
• Web of Science Citations: 7

Abstract

We present an innovative method for predicting the dynamic electron correlation energy of an atom or a bond in a molecule utilizing topological atoms. Our approach uses the machine learning method Kriging (Gaussian Process Regression with a non-zero mean function) to predict these dynamic electron correlation energy contributions. The true energy values are calculated by partitioning the MP2 two-particle density-matrix via the Interacting Quantum Atoms (IQA) procedure. To our knowledge, this is the first time such energies have been predicted by a machine learning technique. We present here three important proof-of-concept cases: the water monomer, the water dimer, and the van der Waals complex H-2 center dot center dot center dot He. These cases represent the final step toward the design of a full IQA potential for molecular simulation. This final piece will enable us to consider situations in which dispersion is the dominant intermolecular interaction. The results from these examples suggest a new method by which dispersion potentials for molecular simulation can be generated. (AU)

• FAPESP's process: 14/21241-9 - The inclusion of polarization effects in the description of amino acids and peptides through the use of atomic multipoles obtained from electron densities
• Grantee: Arnaldo Fernandes da Silva Filho
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• FAPESP's process: 15/22247-3 - Using the Quantum Chemical Topology theory for modeling force fields for peptides using electron densities
• Grantee: Arnaldo Fernandes da Silva Filho
• Support Opportunities: Scholarships abroad - Research Internship - Post-doctor