Physics Practical Magic

Scientists created a 'crystal ball' for chemistry that can predict rare, one-in-a-billion molecular events without any data.

April 15, 2026

Original Paper

Differentiable free energy surface: a variational approach to directly observing rare events using generative deep-learning models

arXiv · 2604.09769

The Takeaway

Some chemical reactions are so rare that you could run a computer simulation for a year and never see them happen once. This paper introduces a generative deep-learning model (VaFES) that can 'dream up' these rare transitions—like a protein folding or a battery degrading—without needing any previous data to learn from. It uses the laws of physics to find the 'rare paths' that atoms take during these critical moments. For regular people, this means we can finally see exactly how drugs interact with cells or why a specific battery fails, all without needing millions of hours of expensive supercomputer time. It’s a shortcut to seeing the 'unseeable' in chemistry.

From the abstract

Rare events are central to the evolution of complex many-body systems, characterized as key transitional configurations on the free energy surface (FES). Conventional methods require adequate sampling of rare event transitions to obtain the FES, which is computationally very demanding. Here we introduce the variational free energy surface (VaFES), a dataset-free framework that directly models FESs using tractable-density generative models. Rare events can then be immediately identified from the