Space Group Conditional Flow Matching

2025-09-28

AI Summary

Existing inorganic crystal generation models often neglect symmetry constraints—such as space groups and Wyckoff positions—leading to low-symmetry or physically unrealistic structures. To address this, we propose the first generative framework integrating space-group conditioning with flow matching. Our method introduces a conditional symmetric noise distribution and a group-equivariant vector field, explicitly models Wyckoff positions, and employs an efficient group-averaging mechanism to enforce strict crystallographic symmetry throughout atomic coordinate generation. This ensures full adherence to crystallographic symmetry at every step, significantly improving accuracy in generating high-symmetry structures. Evaluated on crystal structure prediction and *de novo* generation tasks, our approach achieves state-of-the-art performance: generated distributions closely match those of real, stable high-symmetry crystals, while incurring minimal computational overhead.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation
• Computer Vision: Generative Adversarial Networks (GANs) for Vision

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Web Mining and Content Analysis: Web data generation and simulation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Inorganic crystals are periodic, highly-symmetric arrangements of atoms in three-dimensional space. Their structures are constrained by the symmetry operations of a crystallographic emph{space group} and restricted to lie in specific affine subspaces known as emph{Wyckoff positions}. The frequency an atom appears in the crystal and its rough positioning are determined by its Wyckoff position. Most generative models that predict atomic coordinates overlook these symmetry constraints, leading to unrealistically high populations of proposed crystals exhibiting limited symmetry. We introduce Space Group Conditional Flow Matching, a novel generative framework that samples significantly closer to the target population of highly-symmetric, stable crystals. We achieve this by conditioning the entire generation process on a given space group and set of Wyckoff positions; specifically, we define a conditionally symmetric noise base distribution and a group-conditioned, equivariant, parametric vector field that restricts the motion of atoms to their initial Wyckoff position. Our form of group-conditioned equivariance is achieved using an efficient reformulation of emph{group averaging} tailored for symmetric crystals. Importantly, it reduces the computational overhead of symmetrization to a negligible level. We achieve state of the art results on crystal structure prediction and de novo generation benchmarks. We also perform relevant ablations.

Problem

• Generating inorganic crystals with proper symmetry constraints
• Addressing unrealistic symmetry in existing crystal structure models
• Sampling highly-symmetric stable crystals using space group conditioning

Innovation

• Conditions generation on space group symmetry
• Uses group-conditioned equivariant vector field
• Reduces symmetrization computational overhead significantly