Paul,
Refinable parameters are the model parameters that are varied to
optimize a goal function.
The "number of refinement parameters" is the number of refinable
parameters.
If you refine:
- individual coordinates only, then the number of refinable parameters
= number_of_atoms * 3;
- isotropic ADP only, then the number of refinable parameters =
number_of_atoms * 1;
- individual anisotropic ADP only, then the number of refinable
parameters = number_of_atoms * 6;
- isotropic ADP plus individual coordinates, then the number of
refinable parameters = number_of_atoms * (3 + 1);
- rigid bodies: number_of_rigid_groups * 6;
... etc, TLS, occupancies, hydrogens and other mixes can rise different
number of refinement parameters.
....
Add to this scale factor, 2 bulk solvent parameters and <= 6 anisotropic
scale parameters.
Pavel.
Paul Paukstelis wrote:
> A couple of basic questions concerning the number of refinement
> parameters:
>
> How do you come up with the number of parameters based on the number
> of atoms? I was under the impression it was simply the number of atoms
> times 4 (x,y,z,B), but I've seen some reported numbers that have left
> me confused.
>
> Second, am I correct in thinking that for situations with strict NCS
> the number of total parameters is decreased by the number of
> parameters for one protomer multiplied by N-1, with N being N-fold NCS?
>
> --paul
|