[CP2K-user] [CP2K:12592] Choosing cutoff for AIMD
Ramanish Singh
sing... at umn.edu
Wed Dec 4 16:56:41 UTC 2019
Thank you so much Patrick and Travis!
On Wednesday, December 4, 2019 at 10:19:46 AM UTC-6, Travis wrote:
>
> Hi,
>
> Just to add, the 40 referenced in "Now assuming this is the highest among
> all elements present, take this number and multiply it by 40" should be set
> to whatever REL_CUTOFF is. I typically use a value of 50 for static cell
> calculations and 80 for variable cell ones. These figures are based on my
> experience and from literature on condensed phase simulations. Using the
> basis set listed in Patrick's post, this would give cutoffs of 520 Ry and
> 850 Ry with some rounding up to 'nicer' numbers. The point of this is to
> ensure that the tightest orbitals make it to the finest grid. Importantly,
> this assumes you don't use the smoothing function in
> https://manual.cp2k.org/trunk/CP2K_INPUT/ATOM/METHOD/XC/XC_GRID.html.
>
> -T
>
>
>
> On Wednesday, December 4, 2019 at 6:27:18 AM UTC-4, Patrick Gono wrote:
>>
>> Dear Ramanasish,
>>
>> You are attempting to do ab-initio molecular dynamics. At each step, the
>> forces are calculated by performing a single point energy calculation of
>> your system. If your CUTOFF is too small, the single point calculation
>> leads to the wrong electron density, and hence your forces will also be
>> wrong.
>>
>> Choosing CUTOFF, you only need to make sure that your energy in a single
>> point calculation is converged. To be completely clear: the latter of the
>> two options you propose.
>>
>> In general, look at the highest exponent of the basis you use. For
>> example, if your system contains oxygen, and you use the double-zeta MOLOPT
>> basis set:
>> O DZVP-MOLOPT-SR-GTH DZVP-MOLOPT-SR-GTH-q6
>> 1
>> 2 0 2 5 2 2 1
>> 10.389228018317 0.126240722900 0.069215797900 -0.061302037200
>> -0.026862701100 0.029845227500
>> 3.849621072005 0.139933704300 0.115634538900 -0.190087511700
>> -0.006283021000 0.060939733900
>> 1.388401188741 -0.434348231700 -0.322839719400 -0.377726982800
>> -0.224839187800 0.732321580100
>> 0.496955043655 -0.852791790900 -0.095944016600 -0.454266086000
>> 0.380324658600 0.893564918400
>> 0.162491615040 -0.242351537800 1.102830348700 -0.257388983000
>> 1.054102919900 0.152954188700
>>
>> The highest exponent is 10.389... in the first row of values, on the very
>> left.
>>
>> Now assuming this is the highest among all elements present, take this
>> number and multiply it by 40. That should give you some idea of the
>> appropriate value of CUTOFF, in this case CUTOFF = 400. Run a few ENERGY
>> calculations with values, say, between 350 and 600, and check that the
>> energy is, indeed, converged.
>>
>> Yours sincerely,
>> Patrick Gono
>>
>> On Tue, Dec 3, 2019 at 10:42 PM Ramanish Singh <si... at umn.edu> wrote:
>>
>>> Hi,
>>>
>>> I have a question regarding the plane wave CUTOFF in the MGRID section.
>>> I am doing NVT MD simulations with a liquid box.
>>>
>>> For deciding the CUTOFF value do I need to run the MD simulations at
>>> different values of CUTOFF and then decide the appropriate CUTOFF based on
>>> evolution of energy with time for different CUTOFF values
>>>
>>> OR
>>>
>>> a single point energy of the system (RUN_TYPE ENERGY) is fine? (as
>>> given here: https://www.cp2k.org/howto:converging_cutoff )
>>>
>>> Please also give any relevant references for the same if available.
>>>
>>>
>>> Thanks,
>>> Ramanish Singh
>>>
>>> --
>>> 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 c... at googlegroups.com.
>>> To view this discussion on the web visit
>>> https://groups.google.com/d/msgid/cp2k/722d97d6-2566-4562-a571-c38dec41ed81%40googlegroups.com
>>> <https://groups.google.com/d/msgid/cp2k/722d97d6-2566-4562-a571-c38dec41ed81%40googlegroups.com?utm_medium=email&utm_source=footer>
>>> .
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191204/314e3e21/attachment.htm>
More information about the CP2K-user
mailing list