+Q corrections for KRCI

[Peterson, Kirk]

Hi all,

 

I tried calculating +Q corrections in a couple of KRCI calculations this week following the procedure outlined in the tutorial. I thought I had done things correctly, but my resulting output was:

 

Coefficients entering the +Q equations

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

  c0^2 according to Eq. (43)   0.0000000000000000

  c0^2 according to Eq. (44)   0.0000000000000000

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

 

   MC/HF reference energy:       -30490.18301502

 + extra CI correlation energy:      -0.42593483

 = total CI energy:              -30490.60894985

 

And I ended up with “-Infinity” values for the various +Q contributions.  In the input given in the manual, the only one I didn’t understand and couldn’t seem to find an explanation for was for .CVORB.  In the water example it was set as:

 

.CVORB

3 0

 

Probably that is what I did wrong (I just left this option out).  Can someone define this for me?

 

Thanks in advance and best regards,

 

-Kirk

 

My input files were (first a MCSCF and then a KRCI. I’ve left out the extraneous parts):

 

*KRMCSCF

.CI PROGRAM

LUCIAREL

.INACTIVE

22 22

.GAS SHELLS

1

8 8 / 0 8

.SYMMETRY

8g

 

 

And then:

 

.CI PROGRAM

LUCIAREL

.INACTIVE

20 19

.GAS SHELLS

4

2  2 / 1 0

6  8 / 0 3

16 18 / 1 8

18 18 / 52 49

.CIROOTS

8g 1

.MAX CI

200

.MXCIVE

36

.NOOCCN

*QCORR

.REFFILE

 

[stefan]

Hi Kirk,

my apologies for the late reply and if my suggestions may not be helpful after all. I haven't touched the code since >10 years but if I recall correctly

.CVORB

x y

specifies for the KRCI run the number of core orbitals/Kramers pairs in the gerade/ungerade irreps that had been specified in the reference HF/CASSCF run. 

That is, in your case you might have to specifiy

.CVORB

22 22

The whole +Q code is in retrospective super clumsy and not very elegant but, as it often happens, there is no time to streamline it...

Hope, this helps. 

with my best regards 

Stefan

[peterso...@gmail.com]

Hi Stefan,

 

No worries and thank you for the reply. This is very helpful and certainly explains my result below when I omit the CVORB directive.

 

Best regards,

 

-Kirk

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/e7b6b46b-feb9-4d56-bc56-d801785bda17n%40googlegroups.com.

[Peterson, Kirk A]

Hi Stefan,

 

Sorry to bother you with this again, but I wonder if you  (or anyone) have gotten this to work for an atom?  The example job for H2O in the test directory works fine for me, but when I try it for any atom the values of c0^2 are essentially zero and the +q corrections of course are not correct. I’ve tried in both the default linear symmetry as well as forcing D2h, but with the same result.

 

Best regards,

 

-Kirk

 

From: dirac...@googlegroups.com <dirac...@googlegroups.com> on behalf of stefan <stefan...@phys.chem.ethz.ch>
Date: Tuesday, October 8, 2024 at 7:58 AM
To: dirac-users <dirac...@googlegroups.com>
Subject: [dirac-users] Re: +Q corrections for KRCI

--

[stefan]

Dear Kirk,

many thanks for your patience. I haven’t used the code in years and I am afraid that b/c tests might not have captured all possibilities, more recent restructuring of the code 

might have led to bugs in the +Q code. 

For example, I don’t know whether the c_0 coefficient (from SCF) are properly stored in a file at the moment when one does not run an MCSCF first, could be that this did not make it to the HDF5 file. 

Are you running an MCSCF first?

I know that atoms have worked in the past and there is nothing special about them compared to molecular systems.

As I am no longer able to maintain code and/or debug due to lack of time (in my out-of-academia job), I am hoping that someone else, 

most likely Hans Jørgen in his spare time perhaps, could be of help here. 

I know this isn’t possibly the answer you were hoping for but I am optimistic that the Dirac community might still be able to help. :) 

with my best regards 

Stefan

[peterso...@gmail.com]

Dear Stefan,

 

Thanks for taking the time to reply, I know that you are rather disconnected from the code these days. I hope that perhaps Hans Jørgen can help at some point. I’ve followed what I think is the same procedure as the example in the test folder (for water, which works correctly) where the MCSCF is run first and then the KRCI/qcorr. Even a simple test for Ne atom results in essentially zero c_0 coefficients, but presumably I’m doing something stupid. I’ve attached my input files in case someone can jump in on this.

 

Regards,

To view this discussion visit https://groups.google.com/d/msgid/dirac-users/57ae8735-d54b-4386-9521-060f59380f3an%40googlegroups.com.