[CP2K-user] Constrained DFT (CDFT) Errors

[Brian Day]

Hi Nico,

I have some follow up questions about CDFT if you don't mind. I figured I'd 
post them here so that many CDFT questions were addressed in the same 
thread. 

1. In a few mixed CDFT calculations, none of the following files were 
printed: *.inverseJacobian, *-JacobianInfo.out, and -LineSearch.out. 
Fortunately, there were also no warnings in the *mixed-run.out, but I was 
curious as to why these files never got printed. In other runs, where there 
are errors such as LS not converging, these files all get printed. Do these 
files only get printed when there are errors?

2. With regards to the coupling reliability metric, it seems that the paper 
implies that roughly anything less than 0.10 can be considered reliable - 
Would you agree that this in an acceptable cutoff? Additionally, how 
serious are more moderate coupling metrics, such as 0.15-0.25? Lastly, is 
there anything to your knowledge that can be changed in the code to reduce 
this value towards 0, or it it just system dependent? (I realize there is 
probably not a clean-cut answer to any of these questions, but any extra 
insight would be a huge help.)

Best,
     Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20181119/215b14ec/attachment.htm>