## Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

Re^8: MRCI

Maksim Shundalau
m.shundalau@gmail.com

Dear Alex,
I thank you for your help. All these days I’m trying to solving the problem with XMCQDPT2. I had some problems with CASSCF for ground and several excited states with 10 active orbitals. And now with 14 active orbitals I have smooth curves and negligible (about 0.2 cm-1) splittings for doubly degenerated Pi-terms.
But with XMCQDPT2 I have the next problem: At some distance from the minimum (~ 4.09 A) curves lose smoothness and doubly degenerated Pi-terms are split into dozens cm-1. The distance at which this happens depends on the number of states averaged (in XMCQDPT2). If this number (N) is equal to 7 (number averaged singlets or triplets in CASSCF), then the breaks of the curve start near 4.3 A, with N=21 – between 5.5 and 6 A, with N=28 – between 6.5 and 7 A, with N=35 – near 7 A, with N=90 – between 7.5 and 8 A.
Could you explain to me where I'm wrong?

Best regards,
Maksim.

On Tue Nov 11 '14 2:09pm, Alex Granovsky wrote
----------------------------------------------
>Dear Maksim,

>>Then, in Molpro for MRCI level of theory I have the energies:
>>-52.338309 (4.00 A), -52.338409 (4.10 A) and -52.338304 (4.20 A) and
>>energy differences:
>>-0.000100 [E(4.10)-E(4.00)] and +0.000105 [E(4.20)-E(4.10)].

>>Thus, on passing from CASSCF to MRCI the energy decreases approximately 0.5 a.u. at each point, and minimum shifts from approximately 4.30 A to 4.00 A.

>This is only possible if Molpro includes 8 lowest occupied orbitals
>as double occupied inactive orbitals in the MRCI procedure.
>This means that these orbitals are *not* frozen and that the single
>and double excitations from these orbitals are allowed. This is