A low resources space time approach to the GW approximation

Abstract The GW approximation for electronic excitations is considered computationally hard because its CPU time scales as O ( N 4 ) with the number N of atoms in the unit cell. The space time approach to this approximation scales, in principle, as O ( N 3 ) , but for crystals it is nearly impossible to exploit this feature because it requires too much computer memory. Here we remove this memory bottleneck by restricting the screened Coulomb interaction in the GW self-energy to its Nyquist content. This reduces both the needed memory and the CPU time by several orders of magnitude, while keeping satisfactory agreement with experimentally measured gaps. Our method allows exploiting the O ( N 3 ) scaling feature of the GW space time approach with modest computational resources and it will facilitate the computation of the fundamental gaps of organic semiconductors.