GSAS: spiraling in on the minimum and final structure
May have to refine in quite a spiral manner
Initially - have found circular refinement to be more controllable than combined refinement to be more controllable
- heavy atoms using X-ray data
- Oxygens using neutron data
Restraints - problem with cemented structure vs having the structure fall apart
- If restraints cement the structure, turning off restraints; doing small numbers of cycles and manually resetting errant atoms can get things rolling quite nicely.
- Relaxing restraints sometimes can improve the regular co-ordination. !?
If structure goes haywire, it can be easier to start again from the initial model
- Wall paper bubble effect - tighten one errant atom causes 1 to 4 more to go out