Skip to main content

How HPC and Machine Learning are improving the analysis of renewable materials

Dan won the machine learning prize at the 2018 HPC symposium. He continues to use HPC to reduce the amount of time it takes to analyse new sustainable materials

Machine Learning

Dan Davies won the mechanical learning prize at the High Performance Computing (HPC) symposium in June 2018 for his talk “Supercharging the materials discovery process with machine learning.” His work used machine learning techniques to improve how renewable energy applications, such as solar cell devices, are understood. Dan used the Balena HPC system to create intuitive models built from materials data, this sidesteps previous lengthy approaches for one which takes a fraction of the time. The model has been successful in large projects, such as analysing the properties of 400,000 crystal structures in just two days. Since then, Dan has completed his PhD in the Centre Sustainable Chemical Technologies and will be joining Imperial College as a post-doc in their materials department in collaboration with the Faraday Institution, where he will continue his research on sustainable technologies.

Background into Dan’s work

Dan’s work has developed as part of the Centre of Sustainable Chemical Technologies with his supervisors Dr Ben Morgan and Prof Aron Walsh. A vast majority of the materials that we use in batteries are not renewable and heavily rely on fossil fuels. Dan’s research develops and improves new renewable materials that could be less toxic to the environment. His research focuses on analysing the properties of new materials to determine whether they are suitable to be used in batteries.

Role of HPC

The Balena HPC system allowed Dan to run thousands of quantum mechanical calculations to explore the properties of these crystal structures. With this data, Dan was able to determine whether the materials have the correct properties to conduct light and electricity. Balena is an essential productivity tool for a project with complex data requirements which are too big for a desktop. Dan has been able to move his analysis from his workstation to running in parallel on the Balena HPC system. HPC has allowed Dan to not only reduce his lab time but also to decrease the number of materials he has to use. In addition, using Balena has aided him to learn new programmes and has given him the experience of using National HPC systems (Tier1) such as Archer.

Final Thoughts

Dan recommends getting involved with HPC to students of all levels, especially those who look at sustainable technologies in science and engineering. In addition to this, he encourages students to become involved with the HPC and the HPC Symposium as it will help improve your future employment opportunities.

Get involved in this year's Symposium

Find out more about the Symposium Like Dan, you can showcase your computational research in this year's HPC Symposium