Disputation
Due to the ongoing Pandemic, the Disputation will be held digital using Zoom. The Chair of Defence will lead the Disputation and the Defence technician will solve technical issues.
Ex auditorio questions: The Chair of Defence will invite the audience to ex auditorio questions. These can be asked orally, by clicking "Participants - Raise hand" in the Zoom menu. The Zoom-host will grant you to speak in the meeting.
Order the Dissertation as PDF from this email address. Please add name of Candidate: a.c.gartner@kjemi.uio.no
Trial lecture
8th. of April 2021 at 10.30 AM, Zoom
Title:"Machine learning in electronic-structure theory"
Conferral summary
I sin avhandling presenterer kandidaten et rammeverk for å tilpasse kvantekjemiske metoder til periodiske systemer som krystaller, og demonstrerer hvordan dette muliggjør nøyaktige simuleringer av den elektroniske bølgefunksjonen til disse systemene.
Main research findings
Quantum Chemistry constitutes an extensive framework for accurately simulating the electronic wavefunction of atoms and molecules. The same framework may in principle be applied to the domain of periodic structures such as crystals, but is in practice severely limited by the infinite nature of these structures in conjunction with the computational complexity of quantum chemical methods. In his thesis, the candidate utilizes a mathematical structure known as bi-infinite block-Toeplitz matrices in order to smoothly transition between the molecular and periodic realm. Furthermore, he extends the divide-expand-consolidate methods originally devised for molecules to the periodic case, and demonstrates that this procedure can reduce the computational scaling of the simulation while retaining systematic control over the error.