University of Twente Student Theses
Superionic Conduction simulated by Neural Network Potentials trained on On-The-Fly Force Fields
Kuipers, T.P.W. (2022) Superionic Conduction simulated by Neural Network Potentials trained on On-The-Fly Force Fields.
|
PDF
1MB |
| Abstract: | We present a method to train High-Dimensional Neural Network Potentials on small (~10^2 structures) training sets with significantly reduced computation time. This is achieved using On-The-Fly Machine Learning Force Fields for data set generation. This method is analysed on diamond, a simple solid well-described by harmonic lattice dynamics, and lithium nitride, both below and above its superionic phase transition, exhibiting difficult-to-capture anharmonic lattice dynamics. The High-Dimensional Neural Network Potential is shown to work well for diamond, but fails to capture lithium diffusion in lithium nitride well enough to perform molecular dynamics above the phase transition. We conclude with some promising improvements that might yet lead to a correct description of superionic lithium nitride with a significantly reduced training set. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 31 mathematics, 33 physics, 51 materials science |
| Programme: | Applied Mathematics BSc (56965) |
| Link to this item: | https://purl.utwente.nl/essays/92384 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page