# CrystalDiT: A Diffusion Transformer for Crystal Generation

arXiv: https://arxiv.org/abs/2508.16614 · HTML: https://mlsanity.com/paper/2508.16614/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Flow matching, time series and science, one of 14 directions in Diffusion models.
- Builds on 13 of its 41 references that are in the map.
- First posted on arXiv 13 Aug 2025.

## Placed in

- Diffusion models / [Flow matching, time series and science](https://mlsanity.com/diffusion/flow-matching-time-series-and-science/)

## Builds on (most cited first, 13 of 13)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Generative Modeling by Estimating Gradients of the Data Distribution](https://mlsanity.com/paper/1907.05600/) (2019)
- [Structured Denoising Diffusion Models in Discrete State-Spaces](https://mlsanity.com/paper/2107.03006/) (2021)
- [Scaling Diffusion Language Models via Adaptation from Autoregressive Models](https://mlsanity.com/paper/2410.17891/) (2024)
- [Crystal Diffusion Variational Autoencoder for Periodic Material Generation](https://mlsanity.com/paper/2110.06197/) (2021)
- [FlowMM: Generating Materials with Riemannian Flow Matching](https://mlsanity.com/paper/2406.04713/) (2024)
- [All-atom Diffusion Transformers: Unified generative modelling of molecules and materials](https://mlsanity.com/paper/2503.03965/) (2025)
- [FlowLLM: Flow Matching for Material Generation with Large Language Models as Base Distributions](https://mlsanity.com/paper/2410.23405/) (2024)
- [SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Models](https://mlsanity.com/paper/2502.03638/) (2025)
- [Wyckoff Transformer: Generation of Symmetric Crystals](https://mlsanity.com/paper/2503.02407/) (2025)
- [NTPP: Generative Speech Language Modeling for Dual-Channel Spoken Dialogue via Next-Token-Pair Prediction](https://mlsanity.com/paper/2506.00975/) (2025)
- [MSDformer: Multi-scale Discrete Transformer For Time Series Generation](https://mlsanity.com/paper/2505.14202/) (2025)
- [ETDock: A Novel Equivariant Transformer for Protein-Ligand Docking](https://mlsanity.com/paper/2310.08061/) (2023)

## Built on by (most cited first, 6 of 6)

- [Crystalite: A Lightweight Transformer for Efficient Crystal Modeling](https://mlsanity.com/paper/2604.02270/) (2026)
- [Multimodal Crystal Flow: Any-to-Any Modality Generation for Unified Crystal Modeling](https://mlsanity.com/paper/2602.20210/) (2026)
- [Crys-JEPA: Accelerating Crystal Discovery via Embedding Screening and Generative Refinement](https://mlsanity.com/paper/2605.14759/) (2026)
- [Fourier Transformers for Latent Crystallographic Diffusion and Generative Modeling](https://mlsanity.com/paper/2602.12045/) (2026)
- [Generating Symmetric Materials using Latent Flow Matching](https://mlsanity.com/paper/2605.10115/) (2026)
- [GLASS: Global Latent Aggregation with Slot-based Set Decoding for Scalable All-Atom Crystal Generation](https://mlsanity.com/paper/2609.37158/) (2026)

## Abstract (arXiv)

We present CrystalDiT, a diffusion transformer for crystal structure generation that achieves state-of-the-art performance by challenging the trend of architectural complexity. Instead of intricate, multi-stream designs, CrystalDiT employs a unified transformer that imposes a powerful inductive bias: treating lattice and atomic properties as a single, interdependent system. Combined with a periodic table-based atomic representation and a balanced training strategy, our approach achieves 8.78% SUN (Stable, Unique, Novel) rate on MP-20, substantially outperforming recent methods including FlowMM (4.21%) and MatterGen (3.66%). Notably, CrystalDiT generates 63.28% unique and novel structures while maintaining comparable stability rates, demonstrating that architectural simplicity can be more effective than complexity for materials discovery. Our results suggest that in data-limited scientific domains, carefully designed simple architectures outperform sophisticated alternatives that are prone to overfitting.