SCF run not converged

[Puneet Chandel]

Hello Everyone

 

I have done geometry optimization using the crystal structure of the electrolyte, but it does not get converged in any step and shows  "Leaving inner SCF loop after reaching  1500 steps". My system contains a total of 100 atoms. During the GO some bonds between the molecules also break, can someone help me with this?

Thanks a lot

[Krack Matthias]

Hi

 

A smaller ALPHA value, e.g. 0.1 instead of the default  value 0.4, might help.

 

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/13734f6c-54a4-4b2e-b83c-bb225809ce99n%40googlegroups.com.

[Puneet Chandel]

Hi

Thanku for ypur respons sir, I"ve reduced the ALPHA value to 0.1 but the results are still same. Even after 1500 SCF loops it does not get converged. I"ve attached the input and output files here. someone help me with this.

Thanks a lot

[Lobna Saeed]

you added LDA and PBE in the kind block and in the xc_functional block you wrote PBE !! I am not so sure if this is correct   

try to use either LDA or PBE as in the xc_functional block. 

In addition, LDA never converges with me as well. 

Also 1500 as max iteration is just too much. why wait for 1500 iterations??,  is this even possible, that a system converges after 1500 iterations ?? and if it does I don't think It will give you precise numbers; I only use 300. 

Also I think the CG minimizer is not always good to use because the line search is not always stable. I prefer BFGS. 

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/81bf50b8-266a-47f6-88a7-75fce40a70fcn%40googlegroups.com.

[Krack Matthias]

There are more issues in the input:

• The Na-q9 GTH-PP requires a CUTOFF of at least 800 Ry
• Use only MOLOPT-SR basis sets. I suggest to start with DZVP-MOLOPT-SR_GTH.
• Try OT CG with PRECONDITIONER FULL_ALL instead of diagonalization
• Check again the atomic coordinates and the cell parameters. It seems the cell does not match.

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CAFNXbSyTGp0hhFuP83g4WrrYf%2B0hnJA94Ba6VdqzDgmRbgVJBw%40mail.gmail.com.