[Wien] k points scaling

[Stefaan Cottenier]

>1. For the case with higher number of k points after the calculation for
>smaller number of k points converges, should I just increase the k points
>and run x dstart from the same directory? There must be something that I
>am missing here.
>  
>
Only 'x kgen' is needed. Do NOT run dstart again, as then you overwrite 
your (perhaps still rough) converged density by the original 
superposition of atomic densities. It will save you a lot of time if you 
start from the converged density of the coarser k-mesh.

>2. Regarding -in1new switch, how about it for higher number of k points?
>Should it be used for each calculation with different number of k points?
>  
>
Yes, once you have used it, keep using it. You might experience troubles 
with in1new if you use it right from the start, which is what you do if 
you ran dstart again. But if it behaved well for a coarser k-mesh, it 
for sure will behave well for a denser one (if dstart is not used).

>3. .machines files are not always created by PBS script in my setups. What
>could be the reason for this unpredictable thing?
>  
>
Probably something that depends on your particular computer system? I 
would insert some simple commands like 'echo "line 1" > dummyfile', 
'echo "line 2" >> dummyfile' after every line in the PBS script, such 
that you can follow afterwards which lines were really executed or not.

Stefaan


Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm