spinors not in canonical order
Peterson, Kirk
Dear Dirac experts,
I recently ran into a weird instance where the resulting spinors from a DHF calculation are not in canonical order. This seems to completely spoil a subsequent CCSD calculation. The case is Th atom using the .KPSELEC keyword with a 2/10,0 open shell definition. Convergence is just fine but in the final spinors the degenerate J=3/2 components of the valence 6d orbital do not occur one after the other. The Mj=1/2 is first, followed by all the J=5/2 spinors and then finally the Mj=-3/2 component of the J=3/2 level. If I calculate Th-, which has 3 electrons in the 6d, the resulting spinors are in canonical order. I could do a reordering of the spinors post-dhf, but it's my understanding that this will not help as subsequent CCSD calculation. Suggestions? I've attached the output.
best regards,
-Kirk
Ayaki Sunaga
I reproduce your problem when I use your input (lshift_1_Th.out), but I can get the correct order (j=3/2 --> j=5/2) by turning off .LSHIFT (non_level_Th.out).
According to my experience, if one works with .KPSELE, the default convergence scheme (DIIS) is enough, and one does not have to add any additional options such as .LSHIFT.
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/C9A69A9D-6D43-4C4B-A5A7-685FF04BB995%40wsu.edu.
Peterson, Kirk
Dear Ayaki,
thank you! Admittedly I had just left the LSHIFT option in as a pseudo-default option. I had no idea it could cause something like this.
best regards,
-Kirk
From: <dirac...@googlegroups.com> on behalf of Ayaki Sunaga <sunag...@gmail.com>
Reply-To: "dirac...@googlegroups.com" <dirac...@googlegroups.com>
Date: Monday, April 17, 2023 at 5:11 PM
To: "dirac...@googlegroups.com" <dirac...@googlegroups.com>
Subject: Re: [dirac-users] spinors not in canonical order
[EXTERNAL EMAIL]
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/CAHv9aY5Rv%2B%2Bk-eHEC8uc3L%3De-h3OXnZTWQGj7x2MTvL3yZrRSw%40mail.gmail.com.
Peterson, Kirk
Dear Ayaki,
do you understand why it does this? As it turns out, I have another case where this same issue actually allowed me to converge my CCSD to state that I shouldn't have been able to converge on. It would be nice to be able to exploit this when I want to.
best regards,
-Kirk
From: <dirac...@googlegroups.com> on behalf of Ayaki Sunaga <sunag...@gmail.com>
Reply-To: "dirac...@googlegroups.com" <dirac...@googlegroups.com>
Date: Monday, April 17, 2023 at 5:11 PM
To: "dirac...@googlegroups.com" <dirac...@googlegroups.com>
Subject: Re: [dirac-users] spinors not in canonical order
[EXTERNAL EMAIL]
Dear Kirk,
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/CAHv9aY5Rv%2B%2Bk-eHEC8uc3L%3De-h3OXnZTWQGj7x2MTvL3yZrRSw%40mail.gmail.com.
Ayaki Sunaga
I do not know the mechanism. When I implemented .KPSELE, I did not test the combination between .KPSELE and .LSHIFT.
A quick practical tip is that .LSHIFT is the level shift for the closed shell system. In the case of the open shell, .OLEVEL should be used.
https://www.diracprogram.org/doc/master/manual/wave_function/scf.html#olevel
.OLEVEL
0.2
I reached the convergence with the correct order (attachment).
Best regards,
Ayaki
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/571D8C3A-F6BA-4BF0-9784-F2B4FEE1BB22%40wsu.edu.