MRCI & natural orbitals

[Jussi Eloranta]

Dear netters,


I have the following input for molpro2000.1:

<MCSCF part etc. removed>

mrci
states,3;
wf,17,1,1;
save,5000.2;
natorb,5400.2;

Then I later try to access the natural orbitals for these three states:

put,molden,tr-plot-1.1;orbital,5400.2,state=1.1;
put,molden,tr-plot-2.1;orbital,5400.2,state=2.1;
put,molden,tr-plot-3.1;orbital,5400.2,state=3.1;

In the first put statement I get error message:

?REQUESTED ORBITALS NOT FOUND IN RECORD    5400.2 FOR STATE=  1.1

 THE RECORD CONTAINS THE FOLLOWING MRCI DATA:
 DENSITY/CHARGE      SET= 1  STATE=  1.1  MS2=1  NELEC= 17
 ORBITALS/NATURAL    SET= 1  STATE=  3.1  MS2=1  NELEC= 17
 OCC/NATURAL         SET= 1  STATE=  3.1  MS2=1  NELEC= 17
 DENSITY/CHARGE      SET= 2  STATE=  2.1  MS2=1  NELEC= 17
 ORBITALS/NATURAL    SET= 2  STATE=  3.1  MS2=1  NELEC= 17
 OCC/NATURAL         SET= 2  STATE=  3.1  MS2=1  NELEC= 17
 DENSITY/CHARGE      SET= 3  STATE=  3.1  MS2=1  NELEC= 17
 ORBITALS/NATURAL    SET= 3  STATE=  3.1  MS2=1  NELEC= 17
 OCC/NATURAL         SET= 3  STATE=  3.1  MS2=1  NELEC= 17
 DENSITY/TRANSITION  SET= 4  STATE=201.1  MS2=1  NELEC= 17
 DENSITY/TRANSITION  SET= 5  STATE=301.1  MS2=1  NELEC= 17
 DENSITY/TRANSITION  SET= 6  STATE=302.1  MS2=1  NELEC= 17

 ERROR EXIT

The state labels look strange. They should have been 1.1, 2.1, 3.1 ?

Also when molpro is dumping the natural orbitals it says:

 Natural orbitals saved on    5400.2 (orbital set 3)
 Natural orbitals saved on    5400.2 (orbital set 6)
 Natural orbitals saved on    5400.2 (orbital set 9)

The set numbers look quite different from the file summary above.
Or maybe these are just some other indices?

Any ideas what is going on?

---

Another problem. Is there any way to perform geometry optimization
using spin-orbit corrected energies? I tried but the optimizer seems to
get confused about which parts of the input should be used for 
evaluating the energy. I am obviously using finite differences for 
forces. BTW related to the optimization with finite differences it would
be nice to be able to optimize parameters that are given in basis set
input (ie. simple exponent / contraction coeff. optimization). Variable
statement does not accept such variables. The finite differences should
not care about the variable being in the basis set.

---

Best regards,

Jussi Eloranta
Department of Chemistry
University of Jyvaskyla
Finland