Skip to main content
Skip to main menu Skip to spotlight region Skip to secondary region Skip to UGA region Skip to Tertiary region Skip to Quaternary region Skip to unit footer


Machine Learning Methods for Studying Complex Potential Energy Surfaces

Anthony Schaefer, speaker
Anthony Schaefer
Graduate Student, Department of Chemistry
University of Georgia
Chemistry Building, Room 553
Physical Seminar

For complex molecular systems, computational methods can provide fine details that are not readily available from experimental methods. However, some insight can only come from thorough exploration of the potential energy surface (PES) for a molecular system. Exhaustive PES exploration using quantum mechanical methods is very time- and resource-consuming.

Empirical reactive force fields can provide a more cost-effective alternative to quantum mechanical methods. However, parameterizing more general force fields is typically a very involved process. Machine-learned force fields are another low-cost method for exploring PES’s, relative to quantum mechanical methods.1 I will discuss how neural network algorithms have been used to parameterize reactive force fields with minimal user input.2 I will also discuss a Gaussian Process Regression model of a PES for a system that is relevant to astrochemistry and combustion chemistry.3 This PES has been used to predict the rate constant of a reaction.

Additionally, I will discuss a Gaussian Process Regression method that is transferable to larger systems.4 This method has recently been extended to inter- and intramolecular potentials separately, adaptively learn a PES, and has been applied to study the dimerization of cyclopentadiene.5


  1. Deringer, V. L.; Bartók, A. P.; Bernstein, N.; Wilkins, D. M.; Ceriotti, M.; Csányi, G. Gaussian Process Regression for Materials and Molecules. Chem. Rev. 2021, 121, 10073-10141.

  2. Guo, F.; Wen, Y.; Feng, S.; Li, X.; Li, H.; Cui, S.; Zhang, Z.; Hu, H.; Zhang, G.; Cheng, X. Intelligent-ReaxFF: Evaluating the reactive force field parameters with machine learning. Computational Materials Science 2020, 172, 109393.

  3. Song, Q.; Zhang, Q.; Meng, Q. Revisiting the Gaussian process regression for fittinghigh-dimensional potential energy surface and its application to the OH + HO2 → O2 + H2O reaction. J. Chem. Phys. 2020, 152, 134309.

  4. Bartók, A. P.; Payne, M. C.; Kondor, R.; Csányi, G. Gaussian Approximation Potentials: The Accuracy of Quantum Mechanics, without the Electrons. Phys. Rev. Lett. 2010, 104, 136403.

  5. Young, T. A.; Johnston-Wood, T.; Deringer, V. L.; Duarte, F. A transferable active-learning strategy for reactive molecular force fields. Chem. Sci. 2021, 12, 10944-10955.

Support Us

We appreciate your financial support. Your gift is important to us and helps support critical opportunities for students and faculty alike, including lectures, travel support, and any number of educational events that augment the classroom experience. Click here to learn more about giving.

Every dollar given has a direct impact upon our students and faculty.

Got More Questions?

Undergraduate inquiries: 

Registration and credit

AP Credit, Section Changes, Overrides,

Graduate inquiries:

Contact Us!

Assistant to the Department Head: Donna Spotts, 706-542-1919 

Main office phone: 706-542-1919 

Main Email:

Head of Chemistry: Prof. Jason Locklin