| 5:50:43 PM PDT - Mon, Jul 28th 2014 |
|
How badly do they diverge? Can you provide input and output files for a pair of divergent calculations? I do not have access to Molpro, but I have been in the unpleasant position of trying to harmonize an underspecified method between two different quantum chemistry packages.
Some things to consider:
-Are both packages using the exact same basis set definitions? Specify basis set definitions inline with your input file or check your basis set libraries carefully to be sure that e.g. "6-311+G*" really means what you expect in both calculations.
-Are any orbitals frozen in the calcuations? If so, are they the same orbitals? Gaussian applies orbital freezing schemes by default for post-HF calculations while NWChem does not; I don't know what Molpro does.
-Are convergence criteria sufficiently similar for both calculations? This can be a very hard question to answer a priori, as different programs specify convergence criteria in different ways and not just with different numbers. It's a question best investigated after you have eliminated more obvious culprits.
|