[CP2K-user] [CP2K:12146] Re: Problems in geometry optimization

[Patrick Gono]

Dear Chen,

A CELL_OPT calculation automatically finds the optimal atomic configuration
as well. In general, the DFT lattice parameter will not be identical to the
experimental one. Therefore, it makes sense to run a CELL_OPT calculation
to determine the correct unit cell parameters and the atomic structure. In
practice, such calculations do not take much more time than a GEO_OPT
simulation. Note that in the case of bulk Pt, you know that the cell is
cubic, and thus can constrain the angles during CELL_OPT by switching the
option KEEP_ANGLES in the &CELL_OPT subsection of the &MOTION section to
true.

But regarding your original issue: CP2K, by default, only uses a single
k-point, the Gamma point. Whereas in a plane-wave code you would converge
any results with respect to the k-point mesh density, in CP2K you must
converge the results with respect to the supercell size. I would suggest
running a CELL_OPT calculation with at least a 2x2x2, preferably a 3x3x3
supercell, and compare the lattice parameters with literature (still, the
number will be off due to the functional used, so compare with PBE results).

Note that you must specify the MULTIPLE_UNIT_CELL keyword in both the
&TOPOLOGY and the &CELL subsections!

Of course, get rid of that extra charge, you most probably want to study a
neutral system. And if you are dealing with bulk systems, keeping fixed the
position of an atom is not necessary, as the system will not float around.
This might only be relevant in the case of e.g. molecules or 2D sheets in
vacuum.

Yours sincerely,
Patrick Gono

On Wed, 28 Aug 2019 at 12:54, Chn <chen... at gmail.com> wrote:

> Dear Daniele,
>
> Thanks for your answer! I annotated the keyword "CHARGE" and the
> constraint part as you mentioned, but it still doesn't work. Could you
> please tell me some reasons that why does this situation happen in general?
> I am wandering that whether it is caused by the input file.
> Does the variation of cell volume you mentioned above mean that I need to
> do the CELL_OPT before GEO_OPT?
>
> thanks again,
> chn
>
> 在 2019年8月28日星期三 UTC+8下午5:26:58，Daniele Passerone写道：
>>
>> Dear Chn
>>
>> Apart from the size of the cell that could give you some trouble, the
>> problem is the keyword
>>
>> CHARGE 1
>>
>> that you added in the &DFT section, meaning that you are simulating a
>> charged system. Removing that would work.
>>
>> Indeed in the output we see:
>>
>>                           Hirshfeld Charges
>>
>>
>>   #Atom  Element  Kind  Ref Charge     Population                    Net
>> charge
>>
>>       1       Pt     1      18.000         17.750
>>   0.250
>>
>>       2       Pt     1      18.000         17.750
>>   0.250
>>
>>       3       Pt     1      18.000         17.750
>>   0.250
>>
>>       4       Pt     1      18.000         17.750
>>   0.250
>>
>>
>>   Total Charge
>>   1.000
>>
>>
>>
>>
>>
>>
>> Also not necessary is the constraint on the position of one atom that you
>> have in the optimization section.
>> In general you should also vary the cell volume and find what is the
>> equilibrium volume for the PBE functional.
>> Hope this helps
>>
>> Daniele
>>
>>
>> On Wednesday, August 28, 2019 at 10:43:05 AM UTC+2, Chn wrote:
>>>
>>> Dear experts in CP2K,
>>>
>>> I am a learner of CP2K package and I am really thankful if my naive
>>> question can be explained.
>>> I tried to do geometry optimization toward Pt bulk or Pt(111) by set the
>>> keyword "RUN_TYPE" as GEO_OPT and the *.inp file was referred to the
>>> exercises "Adsorption of acetylene on an intermetallic surface" about PdGa
>>> at CP2K website. But when the calculation done after several hours, I found
>>> that the atoms in cell distorted seriously. I think it should be obvious
>>> incorrect but I have no idea about how to fix it. Many thanks in advance.
>>> The imported .xyz was prepared by using materials studio and VESTA. I
>>> have tried to use larger supercell (3*3*3) or to set the keyword
>>> "MULTIPLE_UNIT_CELL" as 3 3 3 and tried to optimize other structures like
>>> TiO2 but no use..is it that the cell I used still too small?
>>> The attachments were the input files I used.
>>>
>>> Regards,
>>> chn
>>>
>>> [image: o.png][image: d.png]
>>>
>>> --
> 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 cp... at googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/cp2k/7092cb9a-be30-4f63-8246-b988dea87c67%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/7092cb9a-be30-4f63-8246-b988dea87c67%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190828/b76f2bc1/attachment.htm>