Dispersion-corrected Machine Learning Potentials for 2D van der Waals Materials

Mikkel Ohm Sauer 1, Peder Meisner Lyngby 1, Kristian Sommer Thygesen 1 · 2025

Read Dispersion-corrected Machine Learning Potentials for 2D van der Waals Materials by Mikkel Ohm Sauer 1, Peder Meisner Lyngby 1, Kristian Sommer Thygesen 1 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-05754.txt.

Description

Mikkel Ohm Sauer, Peder Meisner Lyngby and Kristian Sommer Thygesen benchmarked in 2025 six machine-learned atomistic potentials on stacked two-dimensional materials. Such learned force fields promise fast simulations with near first-principles accuracy, but weak van der Waals forces between layers are easy to miss. Each model was corrected for those dispersion forces and tested on bilayer pairs whose moiré cells hold from 4 to 300 atoms. Relaxed structures were compared against reference quantum calculations.

With a few exceptions, the learned models matched the reference structures about as well as the reference methods agree among themselves. The team also traced how small structure errors flow into electronic band energies, finding average errors as low as 35 milli-electron volts. The authors conclude that corrected learned potentials now rival direct quantum calculation for these stacks. That verdict licenses their use on complex, realistic layered materials.

Catalog record

CREATORMikkel Ohm Sauer 1, Peder Meisner Lyngby 1, Kristian Sommer Thygesen 1
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-05754

Browse the full library · About the Tower of Babel