Topology-driven exploration of high-dimensional potential energy hypersurfaces for targeted materials engineering

Research questions

Recent advancements in machine learning and GPU acceleration have revolutionized materials science, enabling the atomic-scale exploration of the enormous space of possible materials. This pilot project aims to integrate topological data analysis with machine learning accelerated structure prediction to develop computational tools for exploring potential energy hypersurfaces of complex materials, understanding their stability under varying conditions, and creating human-interpretable visualizations of high-dimensional data. By leveraging these innovative techniques, we hope to overcome current limitations and provide deeper insights into the design and synthesis of new materials.

Sustainability aspects

This project seeks to develop foundational analytical methods and tools for investigating and understanding high-dimensional potential energy hypersurfaces in complex materials. These approaches will have an immediate impact on the search and discovery of novel sustainable materials. Our ultimate goal is to provide high-level predictions that facilitate experimental access to advanced sustainable materials by addressing key challenges in high-dimensional data analysis, storage, and computational efficiency related to computational materials exploration.