New AI Method Enables Reinforcement Learning for Crystal Structure Prediction
Researchers have introduced OMatG-IRL, a policy-gradient reinforcement learning framework that operates on learned velocity fields in flow-based generative models to guide crystal structure prediction toward target material properties. Prior RL approaches for generative models required access to a mathematical quantity called the score, which flow-based models do not compute, making this the first application of RL to crystal structure prediction. The method achieves order-of-magnitude improvements in sampling efficiency while maintaining structural diversity, potentially accelerating the discovery of new materials with desired properties.
A team of researchers has developed Open Materials Generation with Inference-time Reinforcement Learning (OMatG-IRL), a framework that applies policy-gradient reinforcement learning directly to flow-based generative models used for predicting stable crystal structures. The core technical challenge addressed is that standard RL alignment methods require access to a score function, which flow-based models—learning only velocity fields—do not provide; OMatG-IRL circumvents this by using stochastic perturbations of the generation dynamics to enable exploration and gradient estimation without explicit score computation. The framework reinforces an energy-based objective to steer generation toward stable, low-energy crystal configurations while preserving compositional diversity through conditioning. Benchmarks show performance competitive with existing score-based RL approaches, validating the method's effectiveness despite its different mathematical foundation. Notably, OMatG-IRL learns time-dependent velocity-annealing schedules that reduce the number of sampling steps required by roughly an order of magnitude, substantially cutting generation time. The code has been released as part of the Open Materials Generation (OMatG) framework, making the approach accessible to the broader materials science and machine learning communities. The work spans 25 pages with 12 figures and 6 tables and is available as a preprint on arXiv.
What's missing
The paper does not report experimental synthesis or laboratory validation of the crystal structures generated, leaving open the question of whether computationally predicted structures translate to physically realizable materials. Generalization beyond the specific energy-based objective to other material property targets (e.g., bandgap, conductivity) is not fully demonstrated.
What different sources said
- arXiv cs.LGCenter
Open Materials Generation with Inference-Time Reinforcement Learning
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.