Closed ExpHP closed 5 years ago
In the time since this issue was made, automatic drift cancellation was added and then removed. The idea interacts awkwardly with normalization of the direction vector during CG. If support for fixing one atom is desired in the future, it can be added, simple as that.
It occurred to me that overall drift of the structure could be occurring during minimization, making the direction in state-space (and thus the definition of alpha and beta) less meaningful.
The easy and obvious fix, if I ever want to bother with it, is to fix a single atom.