Suggestions on how to develop restraints on torsion angle.

[MatthewKAIST]

I think one of the most important parts of SDPD for organic molecules is to set up restraints appropriate to the shape of the molecule.
In particular, by setting information about a well-known torsion angle in the form of a restraint, the system can be prevented from staying in the local minima.
However, there may be many torsional angles preferred by some structural motifs.
This information can also be easily obtained based on the CCDC database.

However, the restaint for the dihedral angle currently used in F.O.X. can be set only for a single range.
If the restraint range can be set to more than one, and the dichotomy can be made to effectively determine the angle of the two ranges in the global optimization process, the FOX program is likely to get better.

[vincefn]

This would be quite difficult to add to objcryst/Fox. The dihedral restraints are not just used as a penalty affecting the molecule's likelihood (and therefore the likelihood of accepting a given model).
In addition to this penalty, the part of the algorithm which generates random conformation changes relies explicitly on the quadratic form of the restraints. Changing this would be quite complicated.
Another important point is that when searching 'ab initio' for a solution, often too many restraints can be detrimental to finding the global minimum. This, because going from one incorrect conformation to the correct one may require breaking some restraints..