atomic supersymmetry in Dirac21
Peterson, Kirk
Hi all,
with Dirac21, while now having the ability to exploit atomic supersymmetry for atoms is great, it seems this now causes problems if I try to take an initial guess from a SF calculation. Is there a way to turn this off?
regards,
-Kirk
* REACMO: Coefficients read from file DFCOEF - Total energy: -10009.9272264661140
* Heading :Pm neutral 1s2 2s2 2p6 3s2 3p6 3d10 4s2 4p6 4d10 5Tue Sep 7 12:03:28 2021
Master node : --- SEVERE ERROR, PROGRAM WILL BE ABORTED ---
Date and time (Linux) : Wed Sep 8 17:34:02 2021
REACMO: ERROR reading irrep supersymmetry KAPPA
Traceback from this point (reacmo_supersymmetry_+0x220): 12 frames.
11 function name: reacmo_supersymmetry_+0x220
10 function name: reacmo_+0x329
9 function name: atomic_reacmo_+0xd9
Ayaki Sunaga
A simple way to turn off the atomic supersymmetry is specifying the symmetry by using *.mol format.
http://www.diracprogram.org/doc/master/molecule_and_basis/molecule_using_mol.html
The atomic supersymmetry code does not work with an initial guess from SF (Spin-free) calculation, because the symmetry is different.
Are there any reasons for taking the initial guess from SF calculation?
According to my experience, the default initial guess (LDA potential) is sufficient, and a special initial guess (in this case, SF calculation) is not needed to reach convergence.
Best regards,
Ayaki
--
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/9C25DF55-C523-4524-93CE-B9AB2C62D029%40wsu.edu.
Peterson, Kirk
Dear Ayaki,
thanks, I should have thought of that. With complicated atoms (lanthanides/actinides), we've found starting from a converged SF solution has some advantages. In particular the SF calculation makes it trivial to see if you have all the correct orbital degeneracies as well as if one is getting the correct scalar relativistic ground state for a particular choice of open-shell definition. In addition, our goal is generally to use the difference between SO and SF calculations (either DHF or correlated) to obtain the effects of SO, so we tend to always do the SF calculation first anyway.
regards,
-Kirk
To view this discussion on the web visit
https://groups.google.com/d/msgid/dirac-users/CAHv9aY6NPpsafrR1-QZZBQ%3DQTCE%3DxaYcAdW1CeJdjQ9BVGn%3D_Q%40mail.gmail.com.
Ayaki Sunaga
Thank you for your explanation.
For the particular choice of the open-shell occupation of an atom, .KPSELE option should be useful.
http://diracprogram.org/doc/master/manual/wave_function/scf.html#kpsele
.KPSELE improves the convergence of complicated atoms as SF's initial guess or more, and can specify the target occupation.
Ayaki
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/E4210AF5-9443-434E-8726-57A5FDA34CEC%40wsu.edu.
Peterson, Kirk
Dear Ayaki,
yes, I had been looking at that yesterday and could see how it could be really useful.
best regards,
To view this discussion on the web visit
https://groups.google.com/d/msgid/dirac-users/CAHv9aY6X_YAaEZgwO1M6EEE3pYYwQt2fgZpStpRm_J3ntr684Q%40mail.gmail.com.
Peterson, Kirk
Hi,
revisiting this thread today. I'm attempting to calculate an atom with KRCI but when I explicitly specify D2h symmetry to get around the Kappa issue outlined below, linear symmetry is not detected and my specification of 2xOmega for the state requested in Luciarel is not accepted and the program crashes. Is there a way around this?
best regards,
-Kirk
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/6397C289-EF1E-46DD-A237-D52B8903F92A%40wsu.edu.
Ayaki Sunaga
I hope you have a wonderful year.
>when I explicitly specify D2h symmetry to get around the Kappa issue outlined below, linear symmetry is not detected and my specification of 2xOmega for the state requested in Luciarel is not accepted and the program crashes.
You should specify the "irrep" of your target instead of 2xOmega in the KRCI input.
The example for calculating 5 roots in irrep X = 1 is shown in the following link.
http://www.diracprogram.org/doc/master/manual/wave_function/krci.html#ciroots
Best regards
Ayaki
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/MWHPR01MB323161A6203355F9F69B716DD6469%40MWHPR01MB3231.prod.exchangelabs.com.
Stefan Knecht
---------------------------------------------------------------------------
PD Dr. Stefan Knecht
---------------------------------------------------------------------------
GSI Helmholtzzentrum für Schwerionenforschung GmbH
Abteilung SHE Chemie
Planckstr. 1
64291 Darmstadt
Germany
room: SB3 3.163
phone: +49 6159 71 2588
web: http://stefanknecht.xyz
email: s.kn...@gsi.de
---------------------------------------------------------------------------
GSI Helmholtzzentrum für Schwerionenforschung GmbH
Planckstraße 1, 64291 Darmstadt, Germany, www.gsi.de
Commercial Register / Handelsregister: Amtsgericht Darmstadt, HRB 1528
Managing Directors / Geschäftsführung:
Professor Dr. Paolo Giubellino, Dr. Ulrich Breuer, Jörg Blaurock
Chairman of the GSI Supervisory Board / Vorsitzender des GSI-Aufsichtsrats:
Ministerialdirigent Dr. Volkmar Dietz
---------------------------------------------------------------------------
Johannes-Gutenberg Universität Mainz
Department Chemie
Duesbergweg 10-14
55128 Mainz
Germany
---------------------------------------------------------------------------
On 1 Jan 2022, at 12:01, Ayaki Sunaga <sunag...@gmail.com> wrote:
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/CAHv9aY7vx%2BQXGCA7fE29%2BpGjPDnO%2BbXO%3Dxo0XBsg%3D1i40iindA%40mail.gmail.com.
Peterson, Kirk
Dear Ayaki and Stefan,
Happy New Year wishes to both of you and thanks for the answers on New Years Day! I should have thought to just use the irrep, but admittedly I was too lazy to figure out what that number would be and was hoping there was a simple flag I could turn on :) Thanks for the ghost atom workaround as well Stefan.
best regards,
-Kirk
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/00EA8FF6-EA9B-4D25-A26F-3E166FC1D5A0%40phys.chem.ethz.ch.
Ayaki Sunaga
>In any case, I propose that we add an option to turn off atomic symmetry and to use as fallback option linear symmetry. It should be easy to do, do you agree @Ayaki?
Best regards,
Ayaki
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/MWHPR01MB3231788FD6EED5A41C22C457D6479%40MWHPR01MB3231.prod.exchangelabs.com.
Trond SAUE
Hi Kirk,
I am sorry for answering this one late. I am a bit puzzled by your queries. At some stage linear supersymmetry was introduced in DIRAC. This implies that all orbitals have well-defined m_j values, and this is stored in an integer array in the coefficient file. With atomic supersymmetry orbitals are well defined in terms of both m_j and kappa values. During the SCF run this information is packed into a single integer array. However, two arrays are now stored in the coefficient file. Whereas the full atomic symmetry can be reasonably easy exploited at the SCF level, this is not the case at the correlated level. Therefore, in the code, atomic supersymmetry is downgraded to linear symmetry in the WFT-based correlation methods, which for instance means that only information from m_j integer array is used. It also means that one should specify symmetry with respect to linear and not atomic in such calculations. Therefore it is now clear to me what causes problems for you....
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/
Book: Principles and Practices of Molecular Properties: Theory, Modeling, and Simulations
Peterson, Kirk
Hi Trond,
if we were only doing calculations with a spin-orbit Hamiltonian, you're right this would not cause us any problems and introduction of symmetry is always a good thing. However we generally use Dirac to calculate the effects of SO, so our SO calculations are always accompanied by analogous calculations with a spin-free Hamiltonian. Before doing our Dirac calculations we generally have lots of experience (with Molpro) on the scalar-only side of things, so our first step is to do the spin-free calculations in Dirac. We then will generally use those DFCOEF files as initial guesses for the SO calculations. The latter step is where we run into problems since the atomic supersymmetry labels aren't contained in the spin-free DFCOEF file and everything crashes.
Does that help?
best wishes,
-Kirk
From:
dirac...@googlegroups.com <dirac...@googlegroups.com> on behalf of Trond SAUE <trond...@irsamc.ups-tlse.fr>
Date: Thursday, January 20, 2022 at 9:23 AM
To: dirac...@googlegroups.com <dirac...@googlegroups.com>
Subject: Re: [dirac-users] atomic supersymmetry in Dirac21
--
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/ec1623ae-678d-4564-6a08-e1d2acb008a0%40irsamc.ups-tlse.fr.