I am running single point RI-MP2/aug-cc-pvdz calculations on a variety of dinucleotides. I've noticed that the correlation energy is vastly different when the calculation must do 2 integral passes rather than 1. Is this a bug? Any ideas? Thanks --Niel