Abstract
For several decades, chemical modelling methods have underpinned the rapidly expanding and important field of computational chemistry. These methods have provided invaluable contributions to the understanding of asymmetric and catalytic reactions, in turn allowing the rational design of improved catalysts and reactions. However, recent developments in machine learning (ML) applied to reaction modelling are changing the shape of computational chemistry. ML models, once trained, could allow for much more rapid screening of chemical reactions. In this thesis, research into two distinct approaches to understandingorganic reactions, modelling and ML, are presented. Several examples of the successful application of conventional density functional theory modelling are provided, followed by details of a new ML methodology which improves on current standards for the prediction of reaction barriers, whilst also providing mechanistic insights. A particular emphasis is placed on the advantages of each approach with respect to modelling asymmetric and catalytic reactions, and their subsequent application to drug discovery and natural product synthesis.
Date of Award | 14 Sept 2022 |
---|---|
Original language | English |
Awarding Institution |
|
Supervisor | Matt Grayson (Supervisor) & Simon Lewis (Supervisor) |
Keywords
- Machine learning
- Reaction modelling
- Computational chemistry
- Density functional theory
- Semi empirical quantum chemistry
- Asymmetric catalysis
- Organocatalysis