The cutoff unit in CP2K

[mshakiba.k...@gmail.com]

Hello everyone,

I wanted to ask a question about the cutoff value unit. The unit in the manual is in Ry and the default value is 280 Ry and for some systems, we may need to increase this value. I wanted to know why would we need such huge values? In Quantum ESPRESSO 60-80 Ry is fine for some calculations and adopting higher values is computationally expensive. Can you please explain a bit more about the unit of the cutoff in CP2K? and why in CP2K a larger value is required than QE?

Thank you very much in advance.

[Marcella Iannuzzi]

In GPW the electron density, and not the wave functions, is expanded in PW.

Therefore the PW cutoff is four time as large as a the one you would use to expand the wave functions, as it is done in Quantum ESPRESSO.

Whether or not the grids and PW parts of one GPW calculation become expensive and dominant depends on the system and on the method you are using. Hence increasing the PW cutoff, within a reasonable range, is often not too problematic.

Regards

Marcella

[mshakiba.k...@gmail.com]

Dear Marcella,

Thank you very much for your response. I have another question about this, what is the effect of NGRIDS keyword on the calculations. I see increasing the NGRIDS increase the speed of calculations but there are small differences in the total energy calculations. How does NGRIDS affect the calculations?

Thank you so much again.

[Marcella Iannuzzi]

Coarser grids are used to collocate contributions to the density from smoother functions. 

The total density is then obtained on the finest grids using interpolation operators.

This operation can lead to differences depending on the choice of the grids. 

Please find more details in 

VandeVondele, J.; Krack, M.; Mohamed, F.; Parrinello, M.; Chassaing, T.; Hutter, J. Quickstep: Fast and Accurate Density Functional Calculations Using a Mixed Gaussian and Plane Waves Approach. Comput Phys Commun 2005, 167 (2), 103–128. https://doi.org/10.1016/j.cpc.2004.12.014.

Regards

Marcella

[mshakiba.k...@gmail.com]

Dear Marcella, Thank you very much again for your response.