accuracy of sTDA method

67 views
Skip to first unread message

Wei Li

unread,
Nov 15, 2023, 2:43:13 AM11/15/23
to cp2k
Hi CP2K forum,

I am trying to calculate the absorption spectrum using LR TDDFT plus sTDA by focusing on one water molecule in a big box.

The sTDA calculation was invoked by the following settings:

       KERNEL sTDA

The use of PBE functional gives negative excited state energy. Then I also included the following in the input:

       &STDA
          FRACTION X
       &END

Here I changed X to several values less than 1. Now the calculation can output positive excited state energy, but still, these tests gave much different results compared to PBE0 result.

Is there a way I can make the PBE/sTDA result better compared to the PBE0 result?

Wei
 

Jürg Hutter

unread,
Nov 15, 2023, 11:06:48 AM11/15/23
to cp...@googlegroups.com
Hi

you can shift the virtual states with
EV_SHIFT xxx
in
FORCE_EVAL / PROPERTIES / TDDFPT.

Have a look in the original sTDA papers for some more information.

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Wei Li <liwe...@gmail.com>
Sent: Wednesday, November 15, 2023 8:43 AM
To: cp2k
Subject: [CP2K:19504] accuracy of sTDA method
--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com<mailto:cp2k+uns...@googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/4d1fe1ff-1a44-4f90-afef-7472f608c1ban%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/4d1fe1ff-1a44-4f90-afef-7472f608c1ban%40googlegroups.com?utm_medium=email&utm_source=footer>.

Wei Li

unread,
Nov 15, 2023, 10:37:44 PM11/15/23
to cp...@googlegroups.com
Dear Prof. Hutter,

Thanks for your answer. One more question:

       &STDA
          FRACTION X
       &END

"FRACTION X", X is the Fock-exchange mixing parameter. 

If I am using sTDA/hybrid functional, X should be set according to the XC fraction of hybrid functional, or not?

But even if I am using PBE functional, I can still set X to different values to get different excited state energies.   

So, it is unclear to me whether the X can be chosen freely when combined with PBE or hybrid functional? Or there are some criteria to set the FRACTION X?

Wei

You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/IxXNAs-mV_o/unsubscribe.
To unsubscribe from this group and all its topics, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/ZR0P278MB07599906A9D158F2931AD9059FB1A%40ZR0P278MB0759.CHEP278.PROD.OUTLOOK.COM.
Reply all
Reply to author
Forward
0 new messages