Crystal Structure Prediction using Graph Neural Combinatorial Optimization

Stavros Gerolymatos, J. Kyle Brubaker, Martin J. A. Schuetz, Vladimir V. Gusev · 2026

Read Crystal Structure Prediction using Graph Neural Combinatorial Optimization by Stavros Gerolymatos, J. Kyle Brubaker, Martin J. A. Schuetz, Vladimir V. Gusev (2026) 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-2604-23921.txt.

Description

Gerolymatos and colleagues tackle crystal structure search as a combinatorial optimisation problem, placing atoms on grid positions in a unit cell to cut interaction energy. Exact solvers grow costly at large scale, especially without symmetry constraints. Their neural approach uses graph neural networks to sample workable structures without supervision, with expander graphs capturing short- and long-range links and a Gumbel-Sinkhorn step fixing stoichiometry. The method beats older heuristics and rivals a commercial solver across chemical mixes, opening the way to GPU-scale crystal search.

Catalog record

CREATORStavros Gerolymatos, J. Kyle Brubaker, Martin J. A. Schuetz, Vladimir V. Gusev
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE78487
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.23921

Browse the full library · About the Tower of Babel