Diffusion probabilistic models enhance variational autoencoder for crystal structure generative modeling

Teerachote Pakornchote,Natthaphon Choomphon-anomakhun,Sorrjit Arrerut,Chayanon Atthapak,Sakarn Khamkaeo,Thiparat Chotibut,Thiti Bovornratanaraks

DOI: https://doi.org/10.1038/s41598-024-51400-4

IF: 4.6

2024-01-14

Scientific Reports

Abstract:The crystal diffusion variational autoencoder (CDVAE) is a machine learning model that leverages score matching to generate realistic crystal structures that preserve crystal symmetry. In this study, we leverage novel diffusion probabilistic (DP) models to denoise atomic coordinates rather than adopting the standard score matching approach in CDVAE. Our proposed DP-CDVAE model can reconstruct and generate crystal structures whose qualities are statistically comparable to those of the original CDVAE. Furthermore, notably, when comparing the carbon structures generated by the DP-CDVAE model with relaxed structures obtained from density functional theory calculations, we find that the DP-CDVAE generated structures are remarkably closer to their respective ground states. The energy differences between these structures and the true ground states are, on average, 68.1 meV/atom lower than those generated by the original CDVAE. This significant improvement in the energy accuracy highlights the effectiveness of the DP-CDVAE model in generating crystal structures that better represent their ground-state configurations.

multidisciplinary sciences

What problem does this paper attempt to address?

The paper aims to address the following issues:

1. **Improving Crystal Structure Generation Models**: By combining Diffusion Probability Models (DP) with Variational Autoencoders (VAE), a new crystal structure generation model (DP-CDVAE) is proposed. This model aims to generate structures that are closer to the symmetry of real crystals and shows a significant improvement in energy accuracy when generating carbon structures.

2. **Enhancing the Quality of Generated Structures**: Compared to traditional CDVAE, DP-CDVAE can better approximate the ground state configuration when generating structures. Specifically, the energy difference between the generated structures and those optimized by Density Functional Theory (DFT) is significantly reduced.

3. **Boosting Generation Performance**: The performance of the DP-CDVAE model has been validated on multiple datasets, particularly in terms of the physical properties of the generated structures, such as density and formation energy, showing significant improvements.

In summary, the goal of the paper is to improve existing crystal structure generation techniques by introducing a method based on Diffusion Probability Models, thereby generating higher quality and more ground-state-like crystal structures.