State-averaged CASSCF for OH

Posted by Lorenzo Lodi on March 23, 2012 at 16:33:52:

Hello, I'm new to MOLCAS. I'm using version 7.6. I'm trying to use RASSCF to get a few low-lying curves for the OH radical. I already did the same calculations with Molpro and I'm trying to reproduce the same results. Unfortunately, I've been unsuccessful so far. I have a couple of questions.

1) I tried to specify the use of spherical harmonic basis by writing
BASIS SET
O.cc-pVTZ.Dunning.10s5p2d1f.4s3p2d1f..
O 0.000000 0.000000 0.000000
spherical all
END OF BASIS

and similarly for the H atom.
However, the command seems to be ignored and I can see in the output that the s and p shells use Cartesian functions. On the other hand if I specify
cartesian all
all shells use Cartesian functions, as expected. Is it possible to use spherical harmonics for the s,p shells?
And by the way, is there any (speed?) advantage in using one form or the other?

2) To begin with I would like to do a state-averaged CASSCF calculation of the two doublet PI components (PI_x and PI_y) of the ground state.
If I specify C2v symmetry as far as I understand one cannot do a state average over the PI_x and PI_y states because they have different symmetry (b1 and b2 respectively) and Molcas cannot do state-averaged calculations for states with different symmetry... or can it? I found an old message (Posted by D. M. Hirst on July 27, 1999, "state averaging in RASSCF") with somewhat contradictory information. The "AVERAGE" keywork is mentioned but in the pdf manual such option it is not listed in the CASSCF section and I could not understand how to use it.
With Molpro such a state-averaged CASSCF calculation (valence CAS) in the cc-pVTZ basis set (and cartesian basis functions) gives an energy of -75.43793718

With Molcas, in c2v symmetry, the RASSCF part of my input is
&GUESSORB
&RASSCF
SYMMETRY = 3
SPIN = 2
NACTEL = 7 0 0
INACTIVE = 1 0 0 0
RAS1 = 0 0 0 0
RAS2 = 3 1 1 0
RAS3 = 0 0 0 0
TIGHT = 1.0e-7 1.0e-5
LINEAR
END OF INPUT

Which gives an energy of -75.43616902; specifying SYMMETRY=2 instead gives -75.43799041. I tried to do the same calculation with deduced symmetry (or no symmetry) so that I am allowed to state-average but still could not reproduce the Molpro result.
I would be very grateful for any suggestions!

Lorenzo

Follow Ups:

Re: State-averaged CASSCF for OH Ivan Cernusak 14:03:36 4/16/12 (0)

Post a Followup

Name:
E-Mail:

Subject:

if B is 1s22s22p1, what is Li?

Passfield:

Comments:
: Hello, I'm new to MOLCAS. I'm using version 7.6. I'm trying to use RASSCF to get a few low-lying curves for the OH radical. I already did the same calculations with Molpro and I'm trying to reproduce the same results. Unfortunately, I've been unsuccessful so far. I have a couple of questions. : 1) I tried to specify the use of spherical harmonic basis by writing : BASIS SET : O.cc-pVTZ.Dunning.10s5p2d1f.4s3p2d1f.. : O 0.000000 0.000000 0.000000 : spherical all : END OF BASIS : : and similarly for the H atom. : However, the command seems to be ignored and I can see in the output that the s and p shells use Cartesian functions. On the other hand if I specify : cartesian all : all shells use Cartesian functions, as expected. Is it possible to use spherical harmonics for the s,p shells? : And by the way, is there any (speed?) advantage in using one form or the other? : 2) To begin with I would like to do a state-averaged CASSCF calculation of the two doublet PI components (PI_x and PI_y) of the ground state. : If I specify C2v symmetry as far as I understand one cannot do a state average over the PI_x and PI_y states because they have different symmetry (b1 and b2 respectively) and Molcas cannot do state-averaged calculations for states with different symmetry... or can it? I found an old message (Posted by D. M. Hirst on July 27, 1999, "state averaging in RASSCF") with somewhat contradictory information. The "AVERAGE" keywork is mentioned but in the pdf manual such option it is not listed in the CASSCF section and I could not understand how to use it. : With Molpro such a state-averaged CASSCF calculation (valence CAS) in the cc-pVTZ basis set (and cartesian basis functions) gives an energy of -75.43793718 : With Molcas, in c2v symmetry, the RASSCF part of my input is : &GUESSORB : &RASSCF : SYMMETRY = 3 : SPIN = 2 : NACTEL = 7 0 0 : INACTIVE = 1 0 0 0 : RAS1 = 0 0 0 0 : RAS2 = 3 1 1 0 : RAS3 = 0 0 0 0 : TIGHT = 1.0e-7 1.0e-5 : LINEAR : END OF INPUT : : Which gives an energy of -75.43616902; specifying SYMMETRY=2 instead gives -75.43799041. I tried to do the same calculation with deduced symmetry (or no symmetry) so that I am allowed to state-average but still could not reproduce the Molpro result. : I would be very grateful for any suggestions! : Lorenzo

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