Graph-theoretic active learning for the closed-loop discovery of stochastic heterogeneous composites

Lei Qiu, Shuxin Zhang, Yongbin Yang, Mengdie Wang · 2026

Read Graph-theoretic active learning for the closed-loop discovery of stochastic heterogeneous composites by Lei Qiu, Shuxin Zhang, Yongbin Yang, Mengdie Wang (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/plos-pone-0353692.txt.

Description

Lei Qiu and three colleagues speed up the search for self-healing construction materials in this 2026 paper. Their target is a cement composite laced with tiny capsules that release healing agents when cracks reach them. Simulating how fluids move and cracks spread through such disordered material is accurate but painfully slow. The team instead represents scans of the material as graphs. Grains, pores, and capsules become nodes, their physical connections edges. Graph neural networks then train as fast stand-ins for the heavy simulations.

The graph models predicted permeability more than a hundred thousand times faster than the standard simulation while staying highly accurate. Guided by this active learning loop, the system found a surprising best design: elongated capsules tilted at an angle to the likely crack direction, within a few dozen expensive checks. The authors present the framework as a general shortcut for designing complex materials.

Catalog record

CREATORLei Qiu, Shuxin Zhang, Yongbin Yang, Mengdie Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE70828
LICENSECC_BY
RIGHTSCC BY
SOURCEPLOS

Source: https://doi.org/10.1371/journal.pone.0353692

Browse the full library · About the Tower of Babel