Random phase approximation with exchange for an accurate description of crystalline polymorphism

We determine the correlation energy of BN, ${\mathrm{SiO}}_{2}$, and ice polymorphs employing a recently developed random phase approximation with exchange (RPAx) approach. The RPAx provides larger more accurate polarizabilities as compared to (RPA), captures effects anisotropy. In turn, energy, defined an integral over density-density response function, gives improved binding energies without need for error cancellation. Here, we demonstrate that these features are crucial predicting relative between low- high-pressure different coordination number as, e.g., $\ensuremath{\alpha}$-quartz stishovite in layered cubic BN. Furthermore, reliable ${({\mathrm{H}}_{2}\mathrm{O})}_{2}$ potential surface is obtained, necessary describing various phases ice. results comparable other high-level methods such coupled cluster quantum Monte Carlo, also cases where RPA breaks down. Although higher computational cost than RPA, observe faster convergence respect eigenvalues function.