Keeping Squares Square*

First principles calculations for materials use ~25% of the world’s supercomputing resources. We propose a method for reducing the computational cost for these types of calculations without losing accuracy. Because crystalline structures and their properties are symmetric, it is only necessary to compute their properties at symmetrically unique points. Others have utilized these symmetric properties to create grids of symmetrically unique points. However these grids are restricted to integer divisions of the lattice vectors that define the crystal. Our method explores all possible grids that have the same symmetries of the crystal. By utilizing all the symmetries of a crystalline structure, a material’s properties can be computed more efficiently, reducing the amount of computation needed to predict new materials.