Study Compares Reliability of Two Approaches for Probabilistic Forecasting of Physical Systems
Researchers from the Alan Turing Institute and collaborating institutions have published a systematic comparison of two leading approaches for probabilistic forecasting of physical systems, finding that CRPS-trained ensembles generally produce more reliable uncertainty estimates than latent-space generative models. The study evaluated generative models (such as diffusion and flow matching) against ensembles trained with the continuous ranked probability score loss across multiple 2D spatiotemporal systems under matched computational budgets. The findings matter because reliable uncertainty quantification is critical for deploying machine learning emulators in high-stakes scientific and engineering applications.
A team from the Alan Turing Institute, Autodesk Research, PhysicsX, and academic partners has released a preprint systematically benchmarking two dominant paradigms for probabilistic emulation of physical systems: generative models and CRPS-trained deterministic ensembles. The study finds that CRPS-trained ensembles achieve better empirical coverage of predictive intervals on both single-step predictions and autoregressive rollouts compared to generative models trained in a compressed latent space, which is the most common practical setting for high-dimensional problems. Generative models trained in ambient (uncompressed) space show comparable coverage to CRPS ensembles, but this approach is often computationally infeasible at scale. Notably, CRPS-trained ensembles do not suffer significant coverage degradation when moved to latent space, giving them a practical advantage. Beyond reliability, CRPS ensembles also offer substantially faster inference times. Both approaches demonstrate strong predictive accuracy overall. To support reproducibility and future research, the authors release two open tools: AutoCast, a modular benchmarking framework, and AutoSim, a dataset generation package for rapid prototyping.
What's missing
The study is a preprint and has not yet undergone peer review. The evaluation is limited to 2D spatiotemporal systems, and it remains an open question whether the findings generalize to higher-dimensional or real-world physical systems such as full 3D climate or fluid dynamics models. The paper does not address how results may vary across different generative model architectures beyond diffusion and flow matching, nor does it examine sensitivity to hyperparameter choices or training data size.
What different sources said
- arXiv stat.MLCenter
Reliability of Probabilistic Emulation of Physical Systems
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.