[CP2K-user] [CP2K:21218] Parallel efficiency and post-processing of localization

[Jürg Hutter]

Hi

1) Yes, parallel efficiency of the localization algorithms is rather low. Different
     algorithms available show also rather different efficiencies. However, I can't
     remember which one is best.
2) Unfortunately, this is not possible.

regards
JH

________________________________________
From: cp2k at googlegroups.com <cp2k at googlegroups.com> on behalf of Chiahsin Chu <xmuchiahsin at gmail.com>
Sent: Friday, February 21, 2025 10:06 AM
To: cp2k
Subject: [CP2K:21185] Parallel efficiency and post-processing of localization

Dear cp2k developers and users,

I have two questions about the Wannier calculation in cp2k.
1. Parallel efficiency of the algorithm.
I found that the efficiency of Wannier calculation is almost independent of the core number used for calculation. I wonder if this calculation is implemented in parallel and what is the expected parallel efficiency. The relevant section (using Jacobi algorithm for robustness):
&LOCALIZE
METHOD JACOBI
&PRINT
&WANNIER_CENTERS
FILENAME =wannier.xyz
&END WANNIER_CENTERS
&END PRINT
&END LOCALIZE
2. Localization based on wfn or other output of cp2k SCF calculation.
If the parallel efficiency of the localisation algorithm is expected to be low, I prefer to use fewer cores to perform the localisation after SCF calculation. In other words, I would like to perform localisation only for a SCF-converged calculation. Is there any way to achieve this?

Thanks in advance,
Jiaxin

--
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+unsubscribe at googlegroups.com<mailto:cp2k+unsubscribe at googlegroups.com>.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/73a21bac-536f-4ce4-9c1e-122617b239ffn%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/73a21bac-536f-4ce4-9c1e-122617b239ffn%40googlegroups.com?utm_medium=email&utm_source=footer>.

-- 
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+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/ZR0P278MB0759A2DFD7F6A6047B7F98C39FC22%40ZR0P278MB0759.CHEP278.PROD.OUTLOOK.COM.