Guiding generative models to uncover diverse and novel crystals via reinforcement learning
- 1Department of Materials, Imperial College London, London, UK
Abstract
Discovering functional crystalline materials entails navigating an immense combinatorial design space. Although recent advances in generative artificial intelligence have enabled the sampling of chemically plausible compositions and structures, a fundamental challenge remains: the objective misalignment between the likelihood-based sampling in generative modelling and the targeted focus on underexplored regions where novel compounds reside. Here we introduce a reinforcement learning framework that guides latent denoising diffusion models in finding diverse and novel, yet thermodynamically viable, crystalline compounds. Our approach integrates group-relative policy optimization with verifiable, multi-objective rewards that jointly balance creativity, stability and diversity. Beyond de novo generation, we demonstrate enhanced property-guided design that preserves chemical validity while targeting desired functional properties. This approach establishes a modular foundation for controllable AI-driven inverse design that addresses the novelty–validity trade-off across the scientific discovery applications of generative models.
Related
- Continuous SUN (stable, unique, and novel) metric for generative modeling of inorganic crystals
- Stacking ensemble learning and SHAP analysis for predicting CO2 solubility and guiding ionic liquid design
- Designing organic photoredox catalysts as a fragment-based puzzle solved by reinforcement learning
- Mapping Hierarchical Motifs in disorder-order transitions: learning structural motifs in supercooled liquids and crystals
- Reinforcement learning guided robotic platform that investigates and discovers asymmetric reactions
BibTeX
@article{park2026guiding,
title = {{Guiding generative models to uncover diverse and novel crystals via reinforcement learning}},
author = {Hyunsoo Park and Aron Walsh},
journal = {Nature Machine Intelligence},
year = {2026},
month = jul,
doi = {10.1038/s42256-026-01262-4},
url = {https://doi.org/10.1038/s42256-026-01262-4},
note = {CC BY},
}
RIS
TY - JOUR
TI - Guiding generative models to uncover diverse and novel crystals via reinforcement learning
AU - Hyunsoo Park
AU - Aron Walsh
PY - 2026
DA - 2026/07/06
JO - Nature Machine Intelligence
DO - 10.1038/s42256-026-01262-4
UR - https://doi.org/10.1038/s42256-026-01262-4
AB - Discovering functional crystalline materials entails navigating an immense combinatorial design space. Although recent advances in generative artificial intelligence have enabled the sampling of chemically plausible compositions and structures, a fundamental challenge remains: the objective misalignment between the likelihood-based sampling in generative modelling and the targeted focus on underexplored regions where novel compounds reside. Here we introduce a reinforcement learning framework that guides latent denoising diffusion models in finding diverse and novel, yet thermodynamically viable, crystalline compounds. Our approach integrates group-relative policy optimization with verifiable, multi-objective rewards that jointly balance creativity, stability and diversity. Beyond de novo generation, we demonstrate enhanced property-guided design that preserves chemical validity while targeting desired functional properties. This approach establishes a modular foundation for controllable AI-driven inverse design that addresses the novelty–validity trade-off across the scientific discovery applications of generative models.
SN - 2522-5839
ER -