[CP2K-user] How to get a equilibrium structure quickly with a random configuration of different atoms

[Yiwei Sun]

Hello everyone, 

AIMD for the silicate melts based on cp2k was performed.  The simulation 
box of melt has about 300 atoms, and the simulation temperature is 1800K.  

Many examples in the literatures show that the initial structure for AIMD 
is from the equilibrium structure of classical MD, which can greatly reduce 
the  AIMD time for the system to reach equilibrium.  Due to lack of 
classical force field in the present studied system, I can only start AIMD 
with a random configuration with 300 atoms, and it would spend a long time 
to reach equilibrium.   

In order to get a equilibrium structure quickly, can I run AIMD with some 
paremeters (like EPS_SCF ) with a low precision at first, when it is nearly 
equilibrium, then I change them to high precision. Does this procedure can 
get a reasonable outcome?

 Another idea on reducing the AIMD time is to using a initial structure 
optimized by classical forcefield, but no specified forcefield for present 
systems is developed, so is there another method to optimized the random 
configuration? If not, how to get a equilibrium structure quickly with a 
random configuration quickly?

Best wishes
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20210707/4dd3037f/attachment.htm>