Delves into predicting non-scalar properties beyond energies in scientific machine learning, focusing on dipole moments, polarizability, and dielectric response.
Explores equivariant structural representations in atomistic machine learning, emphasizing the importance of representing target properties in the spherical basis.