[molpro-user] CCSDT(Q) energy calculation using MRCC Kallay on Molpro

[Shiblee Ratan]

Dear Kirk,

That was a great hack! I followed your suggestion and it seems like UHF
maxit=1 retains the preceding ROHF orbital properties; i.e. the orbital
occupation and orbital energies for ROHF and UHF look identical in the
output file. I think this is a good guess reference for the mrcc CCSDT(Q)
job that comes later.

Also, since I will eventually be calculating relative energies, any error
arising from the hack is expected to cancel out.

I have included the input here for people who might be interested:

------------------------------------------------------------------------------------------------------------------------------------------------------------
memory,768,m
gthresh,energy=1.0d-12,orbital=1.0d-12,oneint=1.0d-16,twoint=1.0d-16
Geometry = {
N
}

basis=cc-pVDZ

{rhf;maxit,100
wf,nelec=7,sym=8,spin=3
occ,2,1,1,0,1;closed,2}

{uhf;maxit,1
wf,nelec=7,sym=8,spin=3
occ,2,1,1,0,1;closed,2}

!NOTE: MRCC works only with UHF reference. So, here we will trick MRCC by
doing only one UHF iteration right after ROHF.

{mrcc,method=ccsdt(q)
orbital,ignore_error=1}

-----------------------------------------------------------------------------------------------------------------------------------------------------------------


Thanks,
Shiblee


-----------------------------------------------------------------------------------------

*Dr. Shiblee Ratan Barua*
*Postdoctoral Research Fellow*Rm S162, Bld 34
NASA Goddard Space Flight Center
8800 Greenbelt Rd, Greenbelt, MD USA 20771
*Email*: shibleeratan at gmail.com, shiblee.barua at nasa.gov

On Mon, Dec 11, 2017 at 11:27 AM, Peterson, Kirk <kipeters at wsu.edu> wrote:

> Hi,
>
> one hack you can use is to run your ROHF orbitals through 1 iteration of
> the UHF program (maxit,1) and then use these in mrcc. You’ll need to add
> the ignore_error option to the orbital directive in mrcc. This is not
> entirely rigorous since some terms are then missing in the (Q), but it does
> generally give very reasonable results.
>
> regards,
>
> -Kirk
>
> On Dec 8, 2017, at 4:54 PM, Shiblee Ratan <shibleeratan at gmail.com> wrote:
>
> Hello Molpro Experts,
>
> Is it possible to do an *MRCC Kallay CCSDT(Q) energy* calculation for an *open-shell
> ROHF reference*? The error message below keeps giving me hope:
>
>
>
>
>
>
>
>
>
> *Executing mrcc...  Approximate CC methods are not implemented for
> standard ROHF orbitals!  Use semicanonical orbitals!  Fatal error in mrcc.
>  ? Error  ?  Fatal error in mrcc.  ? The problem occurs in MRCC*
>
>
> I am trying to calculate the CCSDT(Q) energy for a (doublet-B1) NH2
> molecule. Tried a lot of things, but it seems like there is no option to
> force Molpro to generate *semicanonical ROHF orbitals* as the error
> message above wants. ROHF keeps generating canonical/standard orbitals in
> record file 2100.2.
>
> CCSDT(Q) energy works fine with *UHF reference* though.
>
> Any help or suggestion to make MRCC Kallay's *CCSDT(Q)* work with the *ROHF
> reference* will be greatly appreciated.
>
> Input file I used is:
>
> ------------------------------------------------------------
> -----------------------------------------------
> memory,768,m
> gthresh,energy=1.0d-12,orbital=1.0d-12,oneint=1.0d-16,twoint=1.0d-16
> Geometry = {
> H1
> N  ,  H1  , rNH
> H2 ,  N   , rNH  ,  H1  ,  aHNH
> }
>
> rNH   =    1.04100000 Ang
> aHNH  =  100.72710000 Degree
>
> basis=cc-pvdz
>
> {rhf;maxit,100
> wf,nelec=9,sym=2,spin=1
> *orbital,type=semicanonical or pseudo-canonical*              <--- Tried
> these and many other combinations but did not make any difference
> occ,3,1,1,0;closed,3,0,1,0}
>
> mrcc,method=ccsdt(q)
>
>
>
> ------------------------------------------------------------
> -----------------------------
>
> *Dr. Shiblee Ratan Barua *
> *Postdoctoral Research Fellow *Rm S162, Bld 34
> NASA Goddard Space Flight Center
> 8800 Greenbelt Rd, Greenbelt, MD USA 20771
> <https://maps.google.com/?q=8800+Greenbelt+Rd,+Greenbelt,+MD+USA+20771&entry=gmail&source=g>
> * Email*: shibleeratan at gmail.com, shiblee.barua at nasa.gov
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> https://urldefense.proofpoint.com/v2/url?u=http-3A__www.
> molpro.net_mailman_listinfo_molpro-2Duser&d=DwICAg&c=C3yme8gMkxg_
> ihJNXS06ZyWk4EJm8LdrrvxQb-Je7sw&r=mqaXWsZ0ghQR-pXvPLjY_g&m=
> 0PlxeCyyU5FG9MdyeiJ6ccPXXIcRFB0sYsuIef6mpi4&s=
> XG89y43F9VuDOaIy_vg5Rf6NpSGr-JbVcLt56_gobkg&e=
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20171215/a0a4f2dc/attachment.html>