Comment on `Efficient calculation of canonical MP2 energiesʹ [P. Pulay, S. Saebø, K. Wolinski, Chem. Phys. Lett. 344 (2001) 543–552]

Accuracy and efficiency of the resolution of the identity approximation in second-order Møller–Plesset perturbation theory (RI-MP2) are reassessed in the light of the new MP2 code by Pulay et al. It is demonstrated that with the RI approximation considerable computational savings can be achieved, especially for larger basis sets of triple-zeta and higher quality. The error introduced by the RI approximation has to be viewed in comparison with the one-electron basis set error. It is shown that this error is insignificant if optimized auxiliary basis sets are used.