Levy-Leblond initial guess

[Peterson, Kirk]

Hi,

 

is it possible to use a converged 4-component DHF calculation as an initial guess to a Levy-Leblond non-relativistic calculation?  I had hopes that the use of ACMOUT and ACMOIN would do the trick, and while the positronic solutions do get deleted, it immediately fails with :

 

*** ERROR (QSYMANA): MO-coefficient no.     1

   Large component basis functions of symmetry Ag  has   2 non-zero components

 

I'm trying to exploit linear symmetry for an atom (the .KPSELEC keyword is fantastic for forcing a particular occupation) by using a 4-c calculation with a very large value for the speed of light and then use this as an initial guess for an actual non-relativistic calculation with the LL Hamiltonian.

 

If this is not possible, are there any practical considerations for using a huge value for the speed of light to replicate the energy for a LL Hamiltonian result

 

best regards,

 

-Kirk

[Kenneth Dyall]

Hi Kirk,

One problem you might run into is numerical noise. 

But a question - what is your goal in trying to do a LL Leblond calculation with a DHF input? Is there another way to achieve that goal?

Ken.

--
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/24EED270-D30D-4028-A2AC-B0977218AE40%40wsu.edu.

[Peterson, Kirk]

Hi Ken,

 

I'm having a terrible time getting a normal Levy-Leblond calculation to converge to the correct, i.e., "clean", electronic configuration with all the correct orbitals occupied and resulting degeneracies.  This is for the superheavy element 126 where most of the usual initial guess routines are not yet available (I am seemingly restricted to .BNCORR).  If I do a DHF calculation, I can exploit linear symmetry and the .KPSELEC keyword. This makes all the difference in the world. At this point, I'm only interested in seeing how the lowest lying configurations change between DHF and LL calculations, which as you might expect is substantial.

 

best,  -Kirk

 

From: <dirac...@googlegroups.com> on behalf of Kenneth Dyall <diracso...@gmail.com>
Reply-To: "dirac...@googlegroups.com" <dirac...@googlegroups.com>
Date: Monday, February 20, 2023 at 10:00 PM
To: "dirac...@googlegroups.com" <dirac...@googlegroups.com>
Subject: Re: [dirac-users] Levy-Leblond initial guess

 

[EXTERNAL EMAIL]

To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/CAKEb7iXNgkT2t%2BaQk84HjPYnyrfGcnUxx%3D%3DEvoKZL0CoaNpFrw%40mail.gmail.com.

[Trond SAUE]

Hi Kirk,

send me a test case and I shall have a look.

All the best,

   Trond

--

Trond Saue
Laboratoire de Chimie et Physique Quantiques
UMR 5626 CNRS --- Université Toulouse III-Paul Sabatier
118 route de Narbonne, F-31062 Toulouse, France

Phone : +33/561556361 Fax: +33/561556065
Mail : trond...@irsamc.ups-tlse.fr
Web : https://dirac.ups-tlse.fr/saue
DIRAC : http://www.diracprogram.org/
ESQC : http://www.esqc.org/
ERC Advanced Grant: HAMP-vQED

Book: Principles and Practices of Molecular Properties: Theory, Modeling, and Simulations

[Knecht Stefan]

Hi,

do you think it would help to start from an X2C reference? On the public master (and the upcoming release) 

you can do X2C calculations with our new amf two-electron picture change corrections (and KPSELEC) [1] which reproduces exactly the 4c one. 

In this case there would be no small component so that maybe reading from these coefficients for LL might help. 

with my best regards

Stefan  

[1] https://doi.org/10.1063/5.0095112

--

-----------------------------------------------------
PD Dr. Stefan Knecht (he/him)
Lead QC Researcher, Algorithmiq Ltd
Kanavakatu 3C
00160 Helsinki
ste...@algorithmiq.fi
 
www.algorithmiq.fi
-----------------------------------------------------
Laboratory of Physical Chemistry
ETH Zuerich
HCIG 230
Vladimir-Prelog-Weg 2
CH-8093 Zuerich
Switzerland
-----------------------------------------------------

To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/3F2B0781-D142-446F-A0C0-A2633D9E300E%40wsu.edu.

[Peterson, Kirk]

Hi Stefan,

 

yes, that most certainly might do the trick. I had to do a few mods to get Dirac to work for Z>118, but perhaps now is the time to do those on the current version and push my private changes anyway.

 

best regards,

 

-Kirk

To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/B81A9C01-E96A-497D-BC1B-8E364975FC4F%40ethz.ch.

[Kenneth Dyall]

Hi Kirk,

I did DHF+CI calculations on a large number of configurations and J values for states of element 126 from +8 to neutral several years ago. These are unpublished, because I was intending to do the whole g block but could not complete the calculations for the lower ionization states around the half-filled g shell. The density of states is large, as you might expect for a system with open s, p, d, f, and g shells. The two configurations with the lowest states were 5g2 6f3 8s2 8p1 and 5g2 6f2 8s2 8p2, which are related by a 6f - 8p excitation. These will mix because they are of the same parity, and if you only have omega to go on, it's very likely that the 6f and 8p orbitals will mix. So you might even run into some problems in the DHF calculations. The 6f3 8p1 configuration has two half-filled shells, 6f5/2 and 8p1/2; the 6f2 8p2 configuration has a closed 8p1/2 shell. 

As you mention, relativistic effects are likely to be large. I never did nonrelativistic calculations (but could fairly easily do so) because I was interested in predicting the ground configurations.  In NR calculations, I would guess that the 8p is not occupied at all; the 8s would likely be the HOMO. As there is less screening by the core, I would guess that the 5g is even more compact in the NR calculations. Now you've got me interested! 

What basis set are you using? I used an even-tempered set which is roughly qz in quality - a family set, where the exponents are common across all l values.

 All the best,

Ken.

To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/3F2B0781-D142-446F-A0C0-A2633D9E300E%40wsu.edu.

[Trond Saue]

Note also this: https://arxiv.org/abs/2301.01740

Sent from my iPad

On 21 Feb 2023, at 10:58, Kenneth Dyall <diracso...@gmail.com> wrote:



To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/CAKEb7iWU%3DH%3D-S9VoKajT2tOPVU6DRO8ynjw5DOBz8A%3DwuWkoig%40mail.gmail.com.