[molpro-user] MCSCF: random orbital re-ordering

[Benjamin LASORNE]

Dear Molpro users and developers,

I am experiencing troubles with the MCSCF program regarding the orbital
order.

Here is a summary of the calculation:
1) I run a selected-active-space MCSCF calculation using the 6-31+G* basis
set using explicit D2h symmetry;
2) I redo a sort of CI calculation at the same geometry, using the same
configuration space (MCSCF with frozen orbitals) with no symmetry whereby
the basis set is projected to a d-aug-cc-pVDZ basis set (this, for a
further MRCI calculation);
3) I make a displacement that lowers the symmetry to C2v(x,z);
4) same as step 1;
5) same as step 2.

In step 5, the orbital order changes after the MCSCF calculation is done
and before the natural orbitals and CI coefficients are dumped, which
means
that the recalculated eigenenergies are now wrong (see attached files
containing "bu" in the name):
 !MCSCF STATE 1.1 Energy              -78.034988934370
 !MCSCF STATE 2.1 Energy              -77.790576418421
 !MCSCF STATE 3.1 Energy              -77.776569867301
 !MCSCF STATE 4.1 Energy              -77.770175274891
 !MCSCF STATE 5.1 Energy              -77.746019952410
 !MCSCF STATE 6.1 Energy              -77.736348142733
 !MCSCF STATE 7.1 Energy              -77.728770331613
 !MCSCF STATE 8.1 Energy              -77.720523555372
 !MCSCF STATE 9.1 Energy              -77.611544652211
 !MCSCF STATE10.1 Energy              -77.523124911923
 !MCSCF STATE11.1 Energy              -77.513605654585
 !MCSCF STATE12.1 Energy              -77.498030488204
 !MCSCF STATE13.1 Energy              -77.489481474961
 !MCSCF STATE14.1 Energy              -77.456571842883
 !MCSCF STATE15.1 Energy              -77.452342849268
 !MCSCF STATE16.1 Energy              -77.442297990325
 !MCSCF STATE17.1 Energy              -77.357517519296
(...)
 TOTAL ENERGIES                       -77.73604715   -77.66139422  
-77.53906468
                                      -77.51659277   -77.50032678  
-77.49348794
                                      -77.46393142   -77.45788528  
-77.44374369
                                      -77.37335967   -77.33353957  
-77.27269877
                                      -77.27185508   -77.25463919  
-77.22869609
                                      -77.22798260   -77.21017155

This is because active orbitals previously numbered 11.1 and 15.1 with
respect to the "select con" procedure have become 10.1 and 11.1
(occupancy:
0.58824).

However, this does not occur for other kinds of deformation (see attached
files containing "ba" in the name, where the symmetry is lowered from D2h
to C2v(x,y), i.e., another principal axis than the previous case).
Also, this does not occur either when symmetry is not lowered (i.e., when
explicit C2v symmetry is used).

Any suggestion?

With many thanks.

Best regards,

Benjamin Lasorne
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Error_Molpro.tar.gz
Type: application/x-gzip
Size: 92660 bytes
Desc: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20111017/adad319b/attachment.gz>