Efficient Calculations of Dispersion Energies for Nanoscale Systems from Coupled Density Response Functions

Dispersion energies computed from coupled Kohn-Sham (CKS) dynamic density-density response functions are known to be highly accurate. At the same time, the computational algorithm is of only modest complexity compared to other accurate methods of dispersion energy calculation. We present a new implementation of this algorithm that removes several computational barriers present in current implementations and enables calculations of dispersion energies for systems with more than 200 atoms using more than 5000 basis functions. The improvements were mainly achieved by reorganizing the algorithm to minimize memory and disk usage. We present applications to two systems: the buckycatcher complex with fullerene and the vancomycin complex with a diacetyl-Lys-d-Ala-d-Ala bacterial wall precursor, both calculations performed with triple-ζ-quality basis sets. Our implementation makes it possible to use ab initio computed dispersion energies in popular "density functional theory plus dispersion" approaches.