Open shell SCF calculation for NH2

[Lorenzo Lodi]

Hello, I am trying to run some DIRAC calculations for the amino radical NH2. The ground electronic state of this molecule is a doublet B_2 (doublet a" in Cs symmetry), while the first excited state is doublet A_1.
As a preliminary test I would like to reproduce (if possible) the non-relativistic Restricted Open-shell Hartree-Fock energy I get from molpro.
For tests I used the C2V geometry r1=r2=  1.9464 a0, theta = 103.00, which is close to the ground-state equilibrium.
with molpro I get the following ROHF energies (cc-pVDZ basis set): ground=-55.56267191 ; excited=-55.47936319.
For DIRAC I tried the following input (see attachment for output):

**MOLECULE
*BASIS
.DEFAULT
cc-pVDZ
**INTEGRALS
.NUCMOD
 1
**DIRAC
.WAVE FUNCTION
**HAMILTONIAN
.LEVY-LEBLOND
**WAVE FUNCTION
.SCF
*SCF
.MAXITR
50
.CLOSED SHELL
8
.OPEN SHELL
1
1/2
*END OF INPUT

I am very unsure as to what I should put in the .OPEN SHELL section. The input above does not converge and after the first ~3 iterations the energies oscillate between ~-55.555 and -55.470, i.e. it seems it is oscillating between the ground and the excited state. Given that they have different C2V symmetries, I am a little suprised this can happen...
I would be very grateful if anyone had any suggestion.

Kind regards,
Lorenzo

[Hans Jørgen Aagaard Jensen]

Hi, it seems like the iterations are oscillating between two different sets of occupied orbitals. My next attempt would be to first converge the closed-shell cation, and then start from these orbitals for the open-shell calculation.

Hope it works!  -- Hans Jørgen.

Sendt fra min iPad

--
You received this message because you are subscribed to the Google Groups "dirac-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dirac-users...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

<NH2.out>

[Lorenzo Lodi]

Hello, I tried what you suggested (doing a calculation for NH2+, which converges without problems, and then using DFCOEF as infput) but I got the same behaviour (oscillation between two states). However, what did work was using a level shift:
.OLEVEL
0.2

Values for the shift in the range 0.10--0.30 seem to work best (convergence in 11-12 iterations).

Kind regards,
Lorenzo

[Lorenzo Lodi]

While trying to run NH2, I encounter a very strange situation. The attached output is a non-relativistic .LEVY-LEBLOND run.
In  the Hartree-Fock section of the output I find an energy -55.570679497973664, which is the same value I get with molpro.
However, in the following CCSD and CCSD(T) section of the output another value is displayed for Hartree-Fock!
 SCF energy :                                -55.413784778113950
 Overview of calculated energies
@ SCF energy :                               -55.413784778113950
@ Total CCSD energy :                        -55.770383699838774
@ Total CCSD(T) energy :                     -55.769566415734467

Why a different (wrong) value for Hartree-Fock in displayed in the final summary??
Do you think the CCSD and CCSD(T) are correct? They look more reasonable than the `second' Hartree-Fock energy but are still off by some 0.02 Eh with respect to CCSD or CCSD(T) energies obtained with other programs and with various references:

ROHF/RCCSD (molpro)                    -55.749852997355
ROHF/UCCSD (molpro)                    -55.749996722249
ROHF/CCSD (cfour, VCC)                -55.750002301405
UHF/ CCSD (mrcc)                            -55.750076535862

ROHF/RCCSD(T) (molpro)                -55.753971587135
ROHF/UCCSD(T)  (molpro)               -55.754092026432
ROHF/CCSD(T) (cfour, VCC)            -55.754097955796
UHF/ CCSD(T) (mrcc)                        -55.754113065912

Kind regards, Lorenzo

[Jeff Hammond]

Open shell CC has different flavors. You appear to be comparing apples and oranges here. The difference between ROHF-CCSD(T) in Molpro and CFOUR is discussed on their respective lists from time to time as new users learn about it. 

Jeff

Sent from my iPhone

--
You received this message because you are subscribed to the Google Groups "dirac-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dirac-users...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

<001_001.out>