A master thesis from Aalborg University
Geometric Message Passing to utilize local and non-local information in 3D Graph Networks
[Geometric Message Passing to utilize local and non-local information in3D Graph Networks]
Author(s)
Term
4. Term
Publication year
2021
Submitted on
2021-09-14
Pages
11 pages
Abstract
To revolutionize computational chemistry within simulating and analysing molecular systems. A principled framework that captures the relative 3D information and long-range interactions is needed. In this work, we propose a generic framework to capture these interactions, known as the deterministic point graph network (DPGN). It provides a unified interface to interact with 3D graphs on different levels of bonds, angles, torsional and long-range effects. We then leverage this framework to add multiple structural improvements to propose the geometric message passing scheme (GMP) to realize DPGN. We demonstrate the benefit of the proposed changes in ablation studies. Finally, we validate this model by presenting results beating the predecessor MPNet on six properties on the QM9 data set, whereas one of them is state-of-art. Additionally, we demonstrate the models' ability within the field of molecular dynamics, where it beats state of the art on \(50\%\) of the targets on in MD17 data set.
Documents
Colophon: This page is part of the AAU Student Projects portal, which is run by Aalborg University. Here, you can find and download publicly available bachelor's theses and master's projects from across the university dating from 2008 onwards. Student projects from before 2008 are available in printed form at Aalborg University Library.
If you have any questions about AAU Student Projects or the research registration, dissemination and analysis at Aalborg University, please feel free to contact the VBN team. You can also find more information in the AAU Student Projects FAQs.