Published in JACS on October 8, the study combines a shear-bond descriptor with universal machine-learning potentials to predict 3,665 negative-thermal-expansion crystals, with experimental synthesis and characterization of 11 previously unreported compounds.
A study in ACS Nano connects potential-based molecular dynamics with flexibility in solid–liquid nanofluidic friction.
A global machine-learning potential and multiobjective optimization workflow explores new superhard compounds.
Machine learning broadens the search for two-dimensional materials with unusual thermal expansion behaviour.
A combined strategy brings learned structural representations and global optimization into one prediction workflow.
First-principles analysis follows a structural pathway relevant to the stability of perovskite materials.
A computational study links lattice softness and carrier dynamics in hybrid perovskites.
Data-driven symbolic regression is used to search for compact, interpretable physical descriptors.
The work examines how alloying changes the atomistic pathways that shape carrier behaviour.
