We describe recent efforts to improve the computational tractability and chemical utility of symmetry adapted perturbation theory (SAPT). SAPT produces an ab initio decomposition of the interaction energy between non-bonded molecules into chemically-intuitive electrostatic, steric repulsion, polarization, and London dispersion components. On one front, we have developed the world's fastest code for the semi-quantitative SAPT0 decomposition, using a combination of density fitting and Laplace transformation tensor compression techniques. This SAPT0 code can be applied to systems as large as a few hundred atoms, and has revealed a number of extremely interesting chemical findings, such as the critical influence of the backbone on the geometry of DNA-proflavine intercalation, and the transient nature of isotopic effects in CH-πinteractions. On another front, we have developed the world's fastest code for the highly accurate SAPT2+3(CCD) decomposition, using natural orbital compression techniques. This SAPT2+3(CCD) code can be applied to systems as large as a few dozen atoms, and reveals a number of important cautionary notes regarding the accuracy of lower-order SAPT methods. Finally, we have very recently developed a rigorous spatial partition of the SAPT0 decomposition, yielding a method known as atomic SAPT (A-SAPT). A-SAPT goes beyond the overall decomposition of standard SAPT methods by allowing one to see which atoms or functional groups are favorable or unfavorable in each interaction energy component. The result is a visualization system that largely matches chemical intuition, but is derived wholly from first principles. This method may have significant application in drug design, crystal packing, molecular recognition, and chemical education.
Prof. David Sherrill (404-894-4037)