Harnessing Machine Learning Potentials to Understand the Functional Properties of Phase-Change Materials

Taking stock and looking ahead

At the very beginning of my PhD (2009, I believe?!) I developed a machine learning potential for the phase change material GeTe. Phase change materials for data storage are quite an essential ingredient of future memory devices – and little did I know at the time that such a potential would be at the heart of so many computational works (more than 15 as we speak) from then on. This paper has been conceived and written by myself and my PhD supervisor – a pivotal figure/influence in my academic career. It is thus with some satisfaction that I would encourage you to have a look at the paper online, or grab a .pdf. You will find a hopefully balanced discussion of what machine learning can do for us in the field of phase change materials, with quite a few references to recent advances and a honest take on the limitations we will have to face in the near future.