[phenixbb] Cross Validation Reflection Set question

[Pavel Afonine]

>> Here is the situation, I have data to 2 A.
>> I refined the model to 3 A first with S.A.

Is this really necessary in this case (refine against lower res data, 
then continue with the whole data set)?
I can see why it may be good to do if you say have 0.7A resolution data 
set and you just solved a MR problem and have Rs~40%, then yes, getting 
some more or less good model using ~2A data is a good first step before 
jumping into modeling finer details using all data.

> (use_lattice_symmetry=True, which
> is the default - I'm not sure why this is even optional).

To be able to prove that it is a good idea.

Pavel