Atomic

GNN-based approach to predict atomic properties (e.g. atomic stress/energy) from structural defects distribution for crystalline solids.

GPT-based and diffusion-based model for conditional generation of polymer electrolyte with high conductivity, validated with MD simulations.

Generative networks used to design diverse materials from architected materials to biological materials.

High-throughput (HT) generation of graphene foams and property quantification using both ML and DL.

Combining coarse-grained molecular dynamics simulation, machine learning, literature mining and atomic force microscopy to learn the self-assembly rule and mechanical properties of peptides.