Proceedings of the National Academy of Sciences · 24 July 2026 · From issue 11
A unified machine-learning framework for ab initio multiscale modeling of liquids
A computer model built from quantum physics predicted how liquids act when squeezed into tiny spaces.
Anna T. Bui, Stephen J. Cox.
A new computer method predicts how liquids like water and carbon dioxide behave. It starts only from the basic laws of quantum physics, with no extra shortcuts. The method links two types of artificial intelligence. One learns how atoms push and pull on each other. The other learns how liquid thickness changes near walls or in tight spaces. This combined tool matched real equations for water and carbon dioxide, and even predicted new effects, like how squeezing water into tiny spaces changes when it freezes or boils.
Integrity screen: passed (3 checks) Checked 26 July 2026. Read the source
The plain-language reading of this paper is in the full issue, for subscribers.
Read the full issue