run geo-opt by gpaw

[shixun sun]

Dear cp2k community,

  I was trying to run a geo-opt for a slab model of graphene(with a single water molecule on it) using GPAW method, but It seems something wrong with the output file. there isn't any force in the output file, and the SCF seems always not converge. I notice I set the "numerous outer SCF loop", 100, but it seems it's not the domain reason. I use 16 cores to run this, But after 1 hour later, there isn't any XYZ file be outputted. 

Can anyone please help me. Any help will be highly appreciated. 

best

sun shixun                                                       

[shixun sun]

Dear cp2k community,

  I thought maybe I‘ve added a 15Å vacuum of the slab to absorb the h2o molecule. and that makes it hard to converge. BUT I am not sure.  Do I need to fix my vacuum layer or something else?  Sincerely looking for your help

best wishes

sun shixun

shixun sun <suns...@gmail.com> 于2020年5月11日周一 下午3:06写道：

--
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/a419f7b4-b8d4-4a64-814d-452ceab29891%40googlegroups.com.

[Fabian Ducry]

Dear Sunshixun,

I am not very familiar with GAPW calculations but here are a couple of hints:

SCF/OUTER_SCF: EPS_SCF should always be the same in the inner and outer loop:

MAX_SCF 30

EPS_SCF 1e-7

&OUTER_SCF
         EPS_SCF 1e-7

         MAX_SCF 30

&END OUTER_SCF

With GAPW the recommended PRECONDITIONER is FULL_ALL

Increase EPS_SCF for the first couple of CELL_OPT iterations. The SCF does not converge to 1e-7 for the starting geometry.

You might also want to decrease EPS_DEFAULT to 1e-12

Best,

Fabian