In the Hammer group at Aarhus University, we are currently developing a number of techniques for global optimization of atomistic structure. One method, GOFEE, employs Bayesian statistics to identify the most probable candidate among a set of candidates relaxed in a surrogate energy landscape. Another method, ASLA, constructs the candidates based on image recognition and reinforcement learning. Your role, if you were to come to Aarhus, will be to adopt these methods in search for challenging nano-scale structure in materials and surface science projects and to extend the methods with new intriguing techniques. The latter could be generative adversarial networks (GAN) or other established schemes from data science, or they could be your own new ideas that are perhaps more domain specific. See this web site for more about previous projects performed in the group.

Apply here