The influence of electronic temperature on interatomic forces in warm dense matter
File(s)
Author(s)
Yates, Thomas
Type
Thesis
Abstract
In this thesis, commonly used methods of calculating free energies and forces in warm
dense matter are discussed. Mermin-Hohenberg-Kohn-Sham DFT (MHKS-DFT) is expected to be the most accurate, but is computationally prohibitive for large system sizes and high temperatures. Orbital-free DFT can partially overcome these computational limitations; however, the accuracy of the method is severely restricted by the requirement of a local pseudopotential and the choice of the approximation for the non-interacting free energy. Interatomic potentials are the most computationally efficient method, but require fitting to accurate ab initio data and are often restricted to a single electronic temperature.
It is then proposed that the Atomic Cluster Expansion can be fit to MHKS-DFT data at finite electronic temperature and this is shown to predict free energies and forces accurate to a few percent of ab initio results for a fraction of the cost. This accuracy is maintained for all densities bounded by the training dataset and, by use of a simple linear
interpolation scheme, intermediate electronic temperatures can also be probed without signifiant loss of accuracy. The many-body expansion, the form of which inspired the Atomic Cluster Expansion, is then used to investigate the importance of many-body interactions in warm dense aluminium. Finally, an experimentally realisable nonequilibrium system in which electronic-temperature dependence has appreciable effects is presented.
dense matter are discussed. Mermin-Hohenberg-Kohn-Sham DFT (MHKS-DFT) is expected to be the most accurate, but is computationally prohibitive for large system sizes and high temperatures. Orbital-free DFT can partially overcome these computational limitations; however, the accuracy of the method is severely restricted by the requirement of a local pseudopotential and the choice of the approximation for the non-interacting free energy. Interatomic potentials are the most computationally efficient method, but require fitting to accurate ab initio data and are often restricted to a single electronic temperature.
It is then proposed that the Atomic Cluster Expansion can be fit to MHKS-DFT data at finite electronic temperature and this is shown to predict free energies and forces accurate to a few percent of ab initio results for a fraction of the cost. This accuracy is maintained for all densities bounded by the training dataset and, by use of a simple linear
interpolation scheme, intermediate electronic temperatures can also be probed without signifiant loss of accuracy. The many-body expansion, the form of which inspired the Atomic Cluster Expansion, is then used to investigate the importance of many-body interactions in warm dense aluminium. Finally, an experimentally realisable nonequilibrium system in which electronic-temperature dependence has appreciable effects is presented.
Version
Open Access
Date Issued
2023-12
Date Awarded
2024-05
Copyright Statement
Creative Commons Attribution NonCommercial Licence
License URL
Advisor
Foulkes, William
Lee, Derek
Publisher Department
Physics
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)