SCF Convergence Issues in CP2K – Oscillating Energy Behavior

[hamid zabihi]

Hello CP2K Community,

I’m running a calculation in CP2K (input and coordinates attached) and encountering unusual SCF behavior. The SCF wavefunction optimization seems to oscillate rather than converge steadily

I’ve attached the full input file (cr202.inp), the log, and the coordinates (cr202.xyz) for reference. Any advice on improving SCF convergence would be greatly appreciated.

Thank you,
Hamid 

[Johann Pototschnig]

There are several methods for convergence, you could try one of the others. i.e. OT (https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCF/OT.html)

A lower ALPHA value of i.e. 0.1 slows down convergence, but it can help with oscillations. 

You could also try to decrease  EPS_DEFAULT 1.0E-16. 

I higher temperature for your FERMI smearing could also improve convergence. 

best,

Johann

[Frederick Stein]

Smearing is not compatible with OT. Considering your output, you do not need to turn off the CHOLESKY decomposition, just comment out the line. To check the effect of smearing, check the orbital energies (see https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/PRINT/MO.html).