Symposium MT03: Synthesis of 2D Materials—Theory and Simulation
December 05, 2024
Daniel Palmer, University of Illinois at Urbana-Champaign
Uncertainty Quantification of Total Energy Tight Binding Models for van der Waals Materials
Written by Andrew M. Fitzgerald
Daniel Palmer and colleagues from the University of Illinois at Urbana-Champaign have developed methods for uncertainty quantification (UQ) in Total Energy Tight Binding (TETB) models, addressing a gap in understanding the accuracy of semi-empirical interatomic potentials. TETB models, used to describe potential energy in materials with electronic degrees of freedom, offer a computationally scalable alternative to fully ab initio methods; however, the uncertainty associated with these models had remained largely unexplored. Palmer and colleagues used Markov-chain Monte Carlo (MCMC) methods to quantify uncertainties in TETB models, focusing on twisted bilayer graphene, a system known for its complex lattice reconstruction and multiscale energy properties. Across various models, their MCMC approach to UQ outperformed more traditional methods of UQ, namely cross validation. This work provides an excellent new approach to evaluating uncertainties in TETB models, making their use in analyzing van der Waals materials more effective.
Comments