Data-driven machine learning models are increasingly being used to generate stable materials which form the basis for rational design of novel catalysts and battery materials. These models are trained on a database of stable materials computed using first principles methods. Based on this database, architectures such as the variational auto-encoder, generative adversarial networks and (more recently) the transformer learn an implicit probability distribution. This distribution is used to decides if a given material is stable. 

It is currently unclear if these data-driven machine learning methods outperform conventional global minimisation algorithms for the task of generating new stable structures for a fixed stoichiometry. In this work, we will apply our recently developed and implemented machine learning models and global optimization techniques to a few prototypical materials. We will benchmark stable structures generated by these methods and compare their time-to-solution.

This project is well suited for students interested in developing and applying machine learning models. If you are interested in this project, please follow the details here to find out how to meet to discuss the project further: https://colab.research.google.com/drive/1KrG4oz23Lr4tKocyfqtswtkUu15ogChM?usp=sharing

(Computational)