> A reasonable approximation for a crystalline structure is balls – point masses – connected by springs, where the springs are covalent bonds, plus electrostatic effects between point charges.
This is the model used by most chemical simulation codes (that don't account for quantum chemistry). See for example the LAMMPS simulation software (https://lammps.sandia.gov/)
