SCF cycle energy going crazy in QM/MM cal with hybrids

[Kushagra Agrawal]

Hi

I am trying to run a DFT calculation for the QM part of a QM/MM region with hybrid functional PBE0.

When I run it, the energy of SCF gives random value for each iteration instead of systematically going down as expected. Naturally, the system also does not converge after long cycles. can someone point to what could be wrong in my file. I am attaching my input file here for reference.

Regards

Kush

[Fabian Ducry]

Hi Kush,

Your eps values are probably not tight enough, try lowering them:

     EPS_DEFAULT 1.0E-10 ->     EPS_DEFAULT 1.0E-12
     EPS_PGF_ORB 1.0E-03 ->      EPS_PGF_ORB 1.0E-06 or lower (this probably even triggers a warning in your output!)

     EPS_SCHWARZ 1.0E-6 -> EPS_SCHWARZ 1.0E-7 or lower

Also MAX_SCF for the inner scf loop should not be larger than 30:

MAX_SCF 300 -> MAX_SCF 30

Cheers,

Fabian

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/771f2659-5377-4cfd-b3b4-34c0c51d504an%40googlegroups.com.

[Kushagra Agrawal]

Hi Fabian

Thanks for your response. I tried your suggestions, but that did not help me. My SCF cycle energies are still going haywire. Any other thoughts?

Regards

Kush

[Mrinmoy Mandal]

Hi Kush

I am also facing similar problem.  How did you solve your problem?

Best,

mrinmoy