ReGNet: Reciprocal Space-Aware Long-Range Modeling and Multi-Property Prediction for Crystals

2025-02-04

AI Summary

Accurately modeling long-range interactions remains challenging in multi-property prediction for crystalline materials. Method: We propose ReGNet-MT, a novel architecture integrating real-space geometric graph neural networks with a learnable Fourier filtering module operating in reciprocal space. To our knowledge, this is the first work to introduce learnable Fourier transforms for reciprocal-space modeling, enabling joint capture of short-range bonding and long-range electronic interactions. A dedicated Reciprocal Block facilitates reciprocal-space-aware long-range modeling, while a Mixture-of-Experts (MoE)-driven multi-task framework encourages positive cross-property transfer. Results: On the JARVIS and Materials Project benchmarks, ReGNet-MT achieves state-of-the-art performance—particularly on critical tasks such as bandgap prediction—with both high accuracy and computational efficiency, establishing a new paradigm for crystal property prediction.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Predicting properties of crystals from their structures is a fundamental yet challenging task in materials science. Unlike molecules, crystal structures exhibit infinite periodic arrangements of atoms, requiring methods capable of capturing both local and global information effectively. However, most current works fall short of capturing long-range interactions within periodic structures. To address this limitation, we leverage reciprocal space to efficiently encode long-range interactions with learnable filters within Fourier transforms. We introduce Reciprocal Geometry Network (ReGNet), a novel architecture that integrates geometric GNNs and reciprocal blocks to model short-range and long-range interactions, respectively. Additionally, we introduce ReGNet-MT, a multi-task extension that employs mixture of experts (MoE) for multi-property prediction. Experimental results on the JARVIS and Materials Project benchmarks demonstrate that ReGNet achieves significant performance improvements. Moreover, ReGNet-MT attains state-of-the-art results on two bandgap properties due to positive transfer, while maintaining high computational efficiency. These findings highlight the potential of our model as a scalable and accurate solution for crystal property prediction. The code will be released upon paper acceptance.

Problem

• Predict crystal properties effectively
• Model long-range crystal interactions
• Enable multi-property prediction efficiently

Innovation

• Reciprocal space encoding
• Geometric GNNs integration
• Multi-task mixture experts