Segmentation fault: Linear scaling SCF (LS_SCF) and linear response properties

[Johannes Stöckelmaier]

Dear cp2k developers!

Since a few weeks I try to learn how to apply cp2k to investigate biomolecules.

One feature that I'm very interested in is linear scaling SCF (LS_SCF).

I try to calculate properties of my system using the RUN_TYPE LINEAR_RESPONSE.method. In my minimum (not-)working example attached, cp2k crashes with segmentation fault once it attempts to work with the &LINRES block at the very end of the *.inp file. Deleting the &LINRES block allows the calculation to pass (but then I can't access f.ex. NMR properties). Moreover, normal energy and force calculations seem to work.

Should it be possible to use RUN_TYPE LINEAR_RESPONSE, to calculate NMR properties, together with linear scaling SCF or is it expected to crash?

I tested the behavior with version 9.1 and 2023.2

I wish you a happy new year!

Thank you for providing the software to the community!

Johannes

[Jürg Hutter]

Hi

Unfortunately, many options are not compatible with LS_SCF (calculates
the density matrix only, no orbitals).
The linear response options are not possible and also GAPW is not
available. GAPW is needed in your case if you want to use an all
electron setup (POTENTIAL ALL).

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Johannes Stöckelmaier <j.stoec...@gmail.com>
Sent: Tuesday, January 2, 2024 1:32 AM
To: cp2k
Subject: [CP2K:19688] Segmentation fault: Linear scaling SCF (LS_SCF) and linear response properties

--
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/da12f008-7c68-48ba-b573-7be8298d4275n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/da12f008-7c68-48ba-b573-7be8298d4275n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Johannes Stöckelmaier]

Dear Professor Hutter!

Thank you for answering my question! This was very helpful for me to safe time.

You write that the combination of GAPW and LS_SCF methods is not valid but it seems that the cp2k software produces some results with these settings. 

There is no segmentation fault or abort when for example calculating energies (RUN_TYPE ENERGY).

Thank you again and I would like to pass on best greetings in the name of the Oostenbrink group from Vienna to you.

Best regards,

Johannes

[Jürg Hutter]

Hi

probably my mistake. Good to know that GAPW with LS_SCF also works.

Thank you

JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Johannes Stöckelmaier <j.stoec...@gmail.com>

Sent: Friday, January 12, 2024 1:47 PM
To: cp2k
Subject: Re: [CP2K:19736] Segmentation fault: Linear scaling SCF (LS_SCF) and linear response properties

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/c46e3a12-7bea-4ab5-91e7-9a57338535bfn%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/c46e3a12-7bea-4ab5-91e7-9a57338535bfn%40googlegroups.com?utm_medium=email&utm_source=footer>.