Hi All,

We are supporting a SIESTA user at our computer centre where a single CG step 
can take up to or more than 12 hours complete.  Restarting these jobs incurs a 
very high startup cost as the restart is not seamless (they are using 
UseSaveData).  Restarting using this method also introduces convergence issues. 
 The 3.2 manual I think summarizes this well on page 90.

Is there are roadmap to support seamless restart in the CG case, as is done 
with the other MD methods?  Or perhaps there are other ways to work around this 
issues.

Thanks,

Paul


Responder a