Poster
in
Workshop: AI4Mat-2024: NeurIPS 2024 Workshop on AI for Accelerated Materials Design
A Geometric Foundation Model for Crystalline Material Discovery
Shengchao Liu · Divin Yan · weitao Du · Zhuoxinran Li · Zhiling Zheng · Omar Yaghi · Christian Borgs · Hongyu Guo · Animashree Anandkumar · Jennifer Chayes
Keywords: [ flow matching ] [ foundation model ] [ geometric modeling ] [ geometric pretraining ] [ periodic invariance ] [ property prediction ] [ SE(3)-equivariance ]
The use of artificial intelligence in crystalline material discovery is gaining significant attention from both the machine learning and chemistry communities. In this work, we present NeuralCrystal, a foundation model specifically designed to push the boundaries of material discovery by combining cutting-edge geometric modeling and large-scale pretraining techniques. The model ensures rotational and translational equivariance by using a vector frame basis, while projecting the coordinate system into the Fourier domain to capture the periodic symmetries and long-range interactions characteristic of crystalline materials. For geometric pretraining, we adopt an equivariant denoising approach by constructing dual views of crystalline structures from the Cambridge Structural Database. NeuralCrystal was rigorously tested on eight MatBench property prediction tasks, outperforming six, and demonstrating its strong potential to significantly accelerate the discovery of new materials.