Soft and transferable pseudopotentials from multi-objective optimization

Ab initio pseudopotentials are a linchpin of modern molecular and condensed matter electronic structure calculations. In this work, we employ multi-objective optimization to maximize pseudopotential softness while maintaining high accuracy transferability. To accomplish this, develop formulation in which simultaneously maximized, with determined by the ability reproduce all-electron energy differences between Bravais lattice structures, whereupon resulting Pareto frontier is scanned for softest that provides desired established transferability tests. We an evolutionary algorithm solve problem apply it generate comprehensive table optimized norm-conserving Vanderbilt (ONCV) (https://github.com/SPARC-X/SPMS-psps). show softer than existing tables comparable accuracy, more accurate softness. The potentials thus afford possibility speed up calculations broad range applications areas accuracy.