Ar atom and CI - surprising result


[ Molcas user's WWWBoard ]

Posted by Pawel Wielgus on February 06, 2007 at 17:53:13:

Dear Molcasers!

To make the story as short as possible:

I take Ar atom, ANO-L basis set, with the default contraction pattern, and then 3 times perform single reference CI using different orbitals. The results are quite astonishing for me...

1) HF >> SRCI(8),
that is single reference CI on top of the HF. In CI I excite only 8 electrons, 1s2s2p are frozen in Motra. The energy is -257.0435... and is quite similar to that reported by prof. Roos in 1991 ...

2) HF >> CAS(8,8) with 1s2s2p FROZEN >> SRCI(8),
that is the same CI, but after HF I make CAS, where 1s2s2p are treated as totally frozen orbitals... The energy is -257.0435... Looks fine.

3) HF >> CAS(8,8) with 1s2s2p INACTIVE >> SRCI(8),
which basically is exactly the same as 2), but I wanted to have even better orbitals, so I didn't freeze 1s2s2p but rather made them inactive. The energy is -257.017....


What's wrong with 3) ? Why I get so artificial energy?
For 2) and 3) I use exactly the same input files which have only one difference: for 2) I have 5 Frozen and 0 Inactive, and for 3) I have 0 Frozen and 5 Inactive in CAS...In Motra there are always 5 Frozen.

Shouldn't I get the best orbitals for CI from number 3?

Regards,

Pawel Wielgus




Follow Ups:



Post a Followup

Name:
E-Mail:

Subject:

if B is 1s22s22p1, what is Li?

Passfield:

Comments:


[ Follow Ups ] [ Post Followup ] [ Molcas user's WWWBoard ]