[molpro-user] Why {MRCI; ... NOEXC} energies differ from input CASSCF ones?

[Lorenzo Lodi]

I've noticed that in state-averaged calculations MRCI/NOEXC energies
sometimes comes out exactly the same as the MCSCF given in input, but
sometimes they are substantially different.

For example, consider the attached inputs for two electronic
terms of the scandium atom (runs in ~5 sec).
>From state-averaged CASSCF I get from the (ground state) 2D term
E=-759.754308 and for the 2F term E=-759.678567.
I then have blocks of MRCI using the  NOEXC option. The energies are now
substantially higher, E(2D)=-759.725614, E(4F)=-759.660592.

Can anybody explain the reason for this behaviour?

Thanks.
Lorenzo



-------------- next part --------------
***,Calculation for Sc atom  === Input for MOLPRO 2010.1
MEMORY, 250.00, M
GEXPEC, REL, DARWIN, MASSV

basis={default, cc-pVDZ}
geometry={Sc}


! state averaged for ground 2D and 2F terms
{CASSCF; OCC, 6,3,3,1,3,1,1,0; CLOSED,3,2,2,0,2,0,0,0; FROZEN,0,0,0,0,0,0,0,0
WF,elec=21,sym=1,spin=1;state,3
WF,elec=21,sym=4,spin=1;state,3
WF,elec=21,sym=6,spin=1;state,3
WF,elec=21,sym=7,spin=1;state,3
 TRAN , DM, LX, LY,LZ;
 TRAN2, LXY, LXZ, LYZ, LXX,LYY,LZZ; ORBPRINT, 20
}

{MRCI; OCC, 6,2,2,1,2,1,1,0; CLOSED,3,2,2,0,2,0,0,0; CORE,3,2,2,0,2,0,0,0; WF,elec=21,sym=1,spin=1;state,3; SAVE, 3050.1; NOEXC}
{MRCI; OCC, 6,2,2,1,2,1,1,0; CLOSED,3,2,2,0,2,0,0,0; CORE,3,2,2,0,2,0,0,0; WF,elec=21,sym=4,spin=1;state,3; SAVE, 3060.1; NOEXC}
{MRCI; OCC, 6,2,2,1,2,1,1,0; CLOSED,3,2,2,0,2,0,0,0; CORE,3,2,2,0,2,0,0,0; WF,elec=21,sym=6,spin=1;state,3; SAVE, 3070.1; NOEXC}
{MRCI; OCC, 6,2,2,1,2,1,1,0; CLOSED,3,2,2,0,2,0,0,0; CORE,3,2,2,0,2,0,0,0; WF,elec=21,sym=7,spin=1;state,3; SAVE, 3080.1; NOEXC}