CASPT2 Calculations on HF Dimer

[thdun...@mac.com]

I recently optimized the geometry of the HF dimer using the valence CASPT2 method with an aug-cc-pVDZ basis set.  The optimization gave an energy of -176.52034681 hartrees.  When I used the optimized z-matrix coordinates, to all of the significant figures given in the optg output, in a subsequent valence CASPT2 calculation, I obtained an energy of -176.52026788 hartrees.  The difference between the two energies is much, much larger than I expected.  Does anyone know why?