convergence issue with diagonalisation

[Hepsibahpriyadarshini C]

Hi,

I tried out optimising the geometry of Ag2O using OT I obtained convergence since I require the usage of K-points for my study I had to switch over to diagonalisation. When I started with 6 atoms I was able to obtain convergence with 4X4X4 K-points. When I started to increase the no.of.atoms that is 12 I am not able to obtain convergence. Inspite of using BFGS as the optimizer, changing EPS_SCF values and playing with mixing parameters I am not able to obtain convergence. Should ELPA be compiled for this. Can band structure,PDOS calculations and phonon calculations be performed with CP2K-5.1 version

I have attached my input file for your kind reference. 

Regards

Hepsibah.C

[Fabian Ducry]

Hi

You have to remove WAVEFUNCTIONS REAL from the DFT%KPOINTS section. You can also delete SYMMETRY TRUE because this is disabled by FULL_GRID .TRUE.

I would also recommend you to reduce DFT%SCF%MIXING%ALPHA to 0.15

I doubt that you can perform these calculations with cp2k 5.1, you should download a newer version (https://www.cp2k.org/download).

Fabian

[Hepsibahpriyadarshini C]

Thank you so much for your valuable feedback. I have tried out by increasing the cell along 3 directions and Ive removed wavefunctions, symmetry under K-points section and changed the alpha value. I am able to obtain convergence. Can you please suggest me the CP2K version to be used for carrying out PDOS, Band structure and phonon calculations.

Regards

Hepsibah.C

On Wed, May 6, 2020 at 6:50 PM Travis <polla...@gmail.com> wrote:

Hi,

Your cell has 12 atoms, the CIF should only define 6 for the unit cell. So you doubled the number of atoms in only one direction but you gave a cell size for a 2x2x2 supercell. Your calculation is slow to converge because ~3/4 of the volume of your cell is vacuum and this has to collapse before you can get a meaningful structure. 

-T

--
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/2e60da34-c63c-458d-9120-99baa01b7c94%40googlegroups.com.

[Fabian Ducry]

I would go with the newest version you can find for your system. cp2k 7.1 can perform these calculations, 6.1 probably as well.

Fabian

[Hepsibahpriyadarshini C]

Thank you 

Regards

Hepsibah

On Thu, May 7, 2020 at 1:00 PM Fabian Ducry <fabia...@gmail.com> wrote:

I would go with the newest version you can find for your system. cp2k 7.1 can perform these calculations, 6.1 probably as well.

Fabian

--
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/e2ef3c8f-605a-4930-a6d6-c8da9bbcf402%40googlegroups.com.