Reply to the comments on 'Efficient calculation of canonical MP2 energies' by A. Kohn and C. Hattig
by Pulay, P.; Baker, J.; Wolinski, K.
It is argued that appropriate methods for timing comparisons with the resolution of identity second-order Moller-Plesset (RI-MP2) technique are local MP2 and dual-basis MP2. not full-accuracy MP2. Like RI-MP2, these methods accelerate MP2 energy evaluation for large molecules and large basis sets, respectively, and generate smooth energy surfaces. at the cost of a small deviation from the canonical results. RI-MP2 is an excellent technique for moderately large molecules and large basis sets. However, its asymptotic scaling with molecular size is much steeper than local MP2, and scaling with basis set size is also worse than dual-basis MP2. We take issue with the claim of Kohn and Hattig that the precision of the programs should reflect the accuracy provided by the one-electron basis set. Three incorrect MP2 correlation energy values in our previous paper are corrected.