Quotient Complex Transformer (QCformer) for Perovskite Data Analysis

2025-05-14

AI Summary

Traditional graph neural networks (GNNs) struggle to model the periodic crystal structures and higher-order many-body interactions inherent in hybrid organic–inorganic perovskites (HOIPs), limiting predictive accuracy for key material properties. To address this, we propose a higher-order geometric representation framework based on quotient complexes (QCs), which uniformly encodes HOIPs as simplicial complexes capturing pairwise and multi-body atomic interactions. We further introduce the first Simplex Transformer tailored for material simplicial complexes, integrating simplex embeddings, geometrically invariant attention mechanisms, and a pretraining–fine-tuning paradigm (Materials Project/JARVIS → HOIP). Our approach overcomes fundamental GNN limitations in modeling periodicity and high-dimensional correlations. On the HOIP benchmark, it achieves state-of-the-art performance, reducing prediction errors for bandgap and formation energy by 23.6% and 18.4%, respectively. The method delivers both high accuracy and intrinsic interpretability, establishing a new paradigm for rational perovskite design.

Technology Category

• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Machine Learning: Mixture of Experts (MoE)
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

The discovery of novel functional materials is crucial in addressing the challenges of sustainable energy generation and climate change. Hybrid organic-inorganic perovskites (HOIPs) have gained attention for their exceptional optoelectronic properties in photovoltaics. Recently, geometric deep learning, particularly graph neural networks (GNNs), has shown strong potential in predicting material properties and guiding material design. However, traditional GNNs often struggle to capture the periodic structures and higher-order interactions prevalent in such systems. To address these limitations, we propose a novel representation based on quotient complexes (QCs) and introduce the Quotient Complex Transformer (QCformer) for material property prediction. A material structure is modeled as a quotient complex, which encodes both pairwise and many-body interactions via simplices of varying dimensions and captures material periodicity through a quotient operation. Our model leverages higher-order features defined on simplices and processes them using a simplex-based Transformer module. We pretrain QCformer on benchmark datasets such as the Materials Project and JARVIS, and fine-tune it on HOIP datasets. The results show that QCformer outperforms state-of-the-art models in HOIP property prediction, demonstrating its effectiveness. The quotient complex representation and QCformer model together contribute a powerful new tool for predictive modeling of perovskite materials.

Problem

• Predicting material properties of perovskites using deep learning
• Capturing periodic structures and higher-order interactions in materials
• Improving accuracy in hybrid organic-inorganic perovskite property prediction

Innovation

• Quotient Complex Transformer for material analysis
• Encodes many-body interactions via simplices
• Pretrains on benchmarks, fine-tunes on HOIPs