[CP2K-user] [CP2K:21445] Hotelling Inversion Did Not Converge — Metal/Water Interface

[Jürg Hutter]

Hi

you can reduce EPS_DEFAULT to 10^-14 (if not yet done).
If that doesn't help, your basis set might not be adequate for your problem (linear dependencies).

regards
JH

________________________________________
From: cp2k at googlegroups.com <cp2k at googlegroups.com> on behalf of Lang Li <lilang8497 at gmail.com>
Sent: Tuesday, May 6, 2025 4:10 PM
To: cp2k
Subject: Re: [CP2K:21442] Hotelling Inversion Did Not Converge — Metal/Water Interface

Dear Jürg,

Thank you for your suggestion.

I followed your advice and adjusted my input as follows:

&SCF
  SCF_GUESS RESTART
  EPS_SCF 3.0E-7
  MAX_SCF 50
  MAX_SCF_HISTORY 5
  &OUTER_SCF
    EPS_SCF 1.0E-6
    MAX_SCF 20
  &END

  &OT
    MINIMIZER DIIS
    PRECONDITIONER FULL_ALL
    STEPSIZE 0.01
  &END
&END SCF

However, I now encounter the following error:

Cholesky decompose failed: the matrix is not positive definite or ill-conditioned.


Do you have any further suggestions on how to address this issue?

Best regards,
Lang

On Tuesday, 22 April 2025 at 10:13:05 UTC+2 Jürg Hutter wrote:
Hi

try direct methods instead of

CHOLESKY INVERSE_DBCSR

and

PRECOND_SOLVER INVERSE_UPDATE

maybe that helps (at least it should avoid Hotelling's method)

regards
JH

________________________________________
From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Lang Li <lilan... at gmail.com>
Sent: Monday, April 21, 2025 11:59 PM
To: cp2k
Subject: [CP2K:21398] Hotelling Inversion Did Not Converge — Metal/Water Interface

Dear CP2K Developers and Users,

I’m encountering an issue while running a simulation involving a metal/water interface. Specifically, I fixed all metal atoms (while keeping the water molecules free) and received the following error during the SCF procedure:

Hotelling inversion did not converge

I'm using the 2nd generation CPMD method. Below is the relevant part of my input file:

&SCF
SCF_GUESS RESTART
EPS_SCF 3.0E-7
MAX_SCF 50
MAX_SCF_HISTORY 5
CHOLESKY INVERSE_DBCSR
&OUTER_SCF
EPS_SCF 1.0E-6
MAX_SCF 20
&END
&END SCF

&OT
MINIMIZER DIIS
PRECOND_SOLVER INVERSE_UPDATE
PRECONDITIONER FULL_SINGLE_INVERSE
STEPSIZE 0.01
&END OT

This issue seems to be related to the SCF convergence when all metal atoms are constrained. I’d appreciate any suggestions or insights from others who may have encountered something similar.

Thank you in advance for your help!

Best regards,
Lang

--
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... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/5d7b8639-475c-4b6c-b109-e1b6dd7b1293n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/5d7b8639-475c-4b6c-b109-e1b6dd7b1293n%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<mailto:cp2k+unsubscribe at googlegroups.com>.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/91f7110c-11b5-44c8-9301-5e591e801f2dn%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/91f7110c-11b5-44c8-9301-5e591e801f2dn%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/ZR0P278MB075956505D02F19CC6F435549F89A%40ZR0P278MB0759.CHEP278.PROD.OUTLOOK.COM.