Interatomic potentials are derived from experimental data or quantum mechanical calculations. They are often parameterized to fit data from spectroscopy, diffraction experiments, or ab initio calculations. Machine learning techniques are increasingly being used to develop more accurate and complex potentials.
