Overcoming the Combinatorial Bottleneck in Symmetry-Driven Crystal Structure Prediction

Researchers developed a new AI-powered framework for crystal structure prediction that uses large language models and symmetry-driven generation to overcome computational bottlenecks. The approach achieves state-of-the-art performance in discovering new materials without relying on existing databases, potentially accelerating materials science research.