Researchers from Skoltech developed a machine learning method to calculate the mechanical properties of solids like polycrystals and composites with near-quantum mechanical accuracy. This method, based on Moment Tensor Potentials (MTP), actively learns from local atomic environments to improve calculations for materials with large numbers of atoms. It’s shown to be especially effective for studying materials that are not perfectly crystalline, such as diamond polycrystals and tungsten carbide with cobalt, which are commonly used in industrial applications. For more details, you can read the full article on Phys.org: