Doctor of Philosophy (PhD)
Physics and Astronomy
First Committee Member
Qiang Q. Zhu
Second Committee Member
Third Committee Member
Fourth Committee Member
Fifth Committee Member
Sixth Committee Member
Number of Pages
Atomistic modeling methods such as molecular dynamics play important roles in investigating time-dependent physical and chemical processes at the microscopic level. In the simulations, energy and forces, sometimes including stress tensor, need to be recalculated iteratively as the atomic configuration evolves. Consequently, atomistic simulations crucially depend on the accuracy of the underlying potential energy surface. Modern quantum mechanical modeling based on density functional theory can consistently generate an accurate description of the potential energy surface. In most cases, molecular dynamics simulations based on density functional theory suffer from highly demanding computational costs. On the other hand, atomistic simulations based on classical force fields have proven to be essential in the computational modeling community due to their unrivaled computational efficiency. However, classical force fields are only useful for inspecting the qualitative insights because they fail to provide confidence in the quantitative results for a lot of cases. In this thesis, I will show the power of machine learning potentials, resolving the predicaments described above. First, the machine learning potential methods will be applied to SiO2 for investigating the implications of different machine learning potentials. Second, the machine learning potentials will be extended to predict physical properties of crystalline silicon, Ni-Mo system, high entropy alloy of NbMoTaW, and Pt for nanoparticle systems. Finally, the diffusion barriers of Pt adsorptions on Pt(111) and Pt(100) surfaces will be examined in detail.
Atomistic simulations; Density functional theory; Gaussian process regression; Machine learning interatomic potentials; Neural networks; Nudged elastic band
Engineering Science and Materials | Materials Science and Engineering | Physical Chemistry
University of Nevada, Las Vegas
Yanxon, Howard, "Developments of Machine Learning Potentials for Atomistic Simulations" (2020). UNLV Theses, Dissertations, Professional Papers, and Capstones. 4093.
IN COPYRIGHT. For more information about this rights statement, please visit http://rightsstatements.org/vocab/InC/1.0/