[phenixbb] Refinement cycles

Pavel Afonine pafonine at lbl.gov
Thu Jul 17 08:28:12 PDT 2014

Hi George,

3 is a compromise between runtime and chances to reach convergence. If 
initial model is good this may be enough, otherwise 5-10 is better. Look 
how R-factors change between macro-cycles and this is the best guide: 
once they reach a plateau that's the point to stop.


On 7/17/14, 7:49 AM, George Devaniranjan wrote:
> Hi,
> I was wondering if there is a reason for setting the default number of 
> refinement cycles to 3 in Phenix refine?
> is there a number of cycles at which point it becomes unacceptable to 
> run so many ?
> Thank you,
> George

More information about the phenixbb mailing list