Two general questions about cell optimization in cp2k

[Torstein Fjermestad]

Dear all,

I have two general questions about optimization of cell parameters in cp2k.

1) How can the Pulay stress be reduced in an optimization of the cell parameters?  In VASP, the recommendation  is to increase the kinetic cut-off energy by approximately 1.5 times the default kinetic cut-off energy. Are there any similar guidelines for cp2k?

2) I see that the default value for the external pressure is set to 100 bar. Why is this a reasonable default value? If we are comparing the computationally optimized cell parameters with the experimentally determined cell parameters, should we then change the value of the external pressure to the pressure for which the experiment was carried out?

Thanks for your help.

Regards,

Torstein

 

[Matthias Krack]

Dear Torstein

1) CP2K provides the &CELL_REF section in the &CELL section which allows for the definition of a larger reference cell. The reference cell can be 5-10% larger. This results basically in a larger cutoff value employed by CP2K. So you can also request just a larger cutoff (and no reference cell) like with VASP.

2) I select usually the pressure from experiment. There is nothing special about the default value of 100 bar. The value for the pressure_tolerance is more interesting for cell optimizations. I employ a pressure tolerance of 10 bar with DFT usually.

HTH

Matthias

[Torstein Fjermestad]

Dear Matthias,

Thanks for your help. 

Concerning 1), is there any criterion for choosing the size of the reference cell?

Wouldn't it be necessary to do some kind of convergence test similarly to what we normally do for the energy cutoff?

Thanks and regards,

Torstein

[Krack Matthias (PSI)]

Dear Torstein

 

Your reference cell should cover in size at least the largest fluctuation of your cell, for instance during an npT run, if you want to keep the accuracy for the selected cell and cutoff. If you use a conservative cutoff anyway, the choice of the reference cell will become less important. Though it is always good to perform convergence tests, the reference cell approach allows you to stay on the safe side, if you apply a conservative estimate for the maximum expected cell size during a CELL_OPT or npT run.

 

Best regards

 

Matthias

--
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/74fad6e5-a673-4bb7-a69c-5ff5f9a92f04%40googlegroups.com.