AI3SD Project: Interpretable crystal descriptions across length scales for materials discovery

Cumby, James , Kanza, Samantha and Frey, Jeremy G. (eds.) (2021) AI3SD Project: Interpretable crystal descriptions across length scales for materials discovery (AI3SD-Project-Series, 6) Southampton. University of Southampton (doi:10.5258/SOTON/P0038).

Record type: Monograph (Project Report)

Abstract

Most technological devices depend in some way on crystalline inorganic materials, from the perovskite oxides found in the capacitors underpinning phones and computers through to the ceramic materials used to insulate ovens and hobs. Future technologies will require new materials with different properties, but discovering these is a significant challenge; trial and error is simply too complex and time-consuming. An alternative approach is to harness our knowledge of the crystalline structure of existing materials in order to predict the properties of new ones, using machine learning (ML). Unfortunately, the conventional way in which we represent crystalstructures is unsuitable for current ML methods. This project aims to develop new ways to represent structures as an input for ML, and ultimately to predict physical properties (such as
how hard a material is) based on its atomic structure.

Text

AI3SD-Project-Series_Report_7_Cumby_FinalReport - Version of Record

Available under License Creative Commons Attribution.

Download (3MB)

Text

AI3SD_006_interim_report

Available under License Creative Commons Attribution.

Download (690kB)

More information

Published date: 16 July 2021

Additional Information: Correction: James Cumby was originally listed as an editor instead of the author, and Victoria Hooper was mistakenly listed as the author. We apologise for this error.

Keywords: AI3SD, Funded Project