Abstract
Are the currently used reference methods to approximately solve the many-electron Schrödinger equation accurate enough? Here, we investigate recently reported discrepancies of noncovalent interaction energies for large molecules predicted by two of the most widely-trusted many-electron theories: diffusion quantum Monte Carlo and coupled-cluster theory. We are able to unequivocally pin down the main source of the puzzling discrepancies and present modifications to widely-used coupled-cluster methods needed for more accurate noncovalent interaction energies of large molecules on the hundred-atom scale. This is of critical impact for a wide range of applications which rely on highly-accurate interaction energies between large and polarizable molecules.