Back when I was first learning ab initio methods in Cliff Dykstra’s lab, I played a bit with the post-HF method CEPA (couple electron pair approximation). This method fell out of favor over the years with the rise of MP theory and then with DFT. Now, Neese and Grimme and co-workers are resurrecting it.1 Their Accounts article provides a series of tests of CEPA/1 against benchmark computations (typically CCSD(T)) and lo and behold, CEPA performs remarkably well! It bests B3LYP (no surprise there), B2LYP and MP2 in virtually every category, ranging from reaction energies, hydrogen bond energies, van der Waals interaction energies, and activation barrier heights. As an example, for the isomerization energy of toluene to norbornadiene, CCSD(T) estimates the energy is 42.79 kcal mol-1. B3LYP does miserably, with an error of nearly 14 kcal mol-1, but the CEPA/1 estimate is off by only 0.04 kcal mol-1. Since the computational time of CEPA/1 is competitive with MP2, the authors conclude that CEPA/1 is well-worth reinvestigating as an alternative post-HF methodology.


(1) Neese, F.; Hansen, A.; Wennmohs, F.; Grimme, S., "Accurate Theoretical Chemistry with Coupled Pair Models," Acc. Chem. Res. 2009, 42, 641-648 DOI: 10.1021/ar800241t.